Homework answers / question archive / In this article, two text-mining-related projects from different industries with different challenges are discussed to identify standard procedures and methodologies that can be used

In this article, two text-mining-related projects from different industries with different challenges are discussed to identify standard procedures and methodologies that can be used

Writing

Share With

In this article, two text-mining-related projects from different industries with different challenges are discussed to identify standard procedures and methodologies that can be used. In the gig work sector, one of the partners offers a platform solution for employee recruitment for temporary work. The work assessment is performed using short reviews for which a method for sentiment assessment based on machine learning has been developed. The other partner creates an information extraction service for business documents in the financial advice sector, including insurance policies.

Studies in Systems, Decision and Control 294 Rolf Dornberger Editor New Trends in Business Information Systems and Technology Digital Innovation and Digital Business Transformation Studies in Systems, Decision and Control Volume 294 Series Editor Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland The series “Studies in Systems, Decision and Control” (SSDC) covers both new developments and advances, as well as the state of the art, in the various areas of broadly perceived systems, decision making and control–quickly, up to date and with a high quality. The intent is to cover the theory, applications, and perspectives on the state of the art and future developments relevant to systems, decision making, control, complex processes and related areas, as embedded in the ?elds of engineering, computer science, physics, economics, social and life sciences, as well as the paradigms and methodologies behind them. The series contains monographs, textbooks, lecture notes and edited volumes in systems, decision making and control spanning the areas of Cyber-Physical Systems, Autonomous Systems, Sensor Networks, Control Systems, Energy Systems, Automotive Systems, Biological Systems, Vehicular Networking and Connected Vehicles, Aerospace Systems, Automation, Manufacturing, Smart Grids, Nonlinear Systems, Power Systems, Robotics, Social Systems, Economic Systems and other. Of particular value to both the contributors and the readership are the short publication timeframe and the world-wide distribution and exposure which enable both a wide and rapid dissemination of research output. ** Indexing: The books of this series are submitted to ISI, SCOPUS, DBLP, Ulrichs, MathSciNet, Current Mathematical Publications, Mathematical Reviews, Zentralblatt Math: MetaPress and Springerlink. More information about this series at http://www.springer.com/series/13304 Rolf Dornberger Editor New Trends in Business Information Systems and Technology Digital Innovation and Digital Business Transformation 123 Editor Rolf Dornberger School of Business Institute for Information Systems, FHNW University of Applied Sciences and Arts Northwestern Switzerland Basel, Switzerland ISSN 2198-4182 ISSN 2198-4190 (electronic) Studies in Systems, Decision and Control ISBN 978-3-030-48331-9 ISBN 978-3-030-48332-6 (eBook) https://doi.org/10.1007/978-3-030-48332-6 © Springer Nature Switzerland AG 2021 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, speci?cally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on micro?lms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a speci?c statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional af?liations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Preface What is digital transformation? This question is often asked because no one knows what digital transformation and its impact on our lives are and will be in the future. Therefore, it is a highly demanding challenge to describe in detail what constitutes digital transformation. The term digital transformation is often referred to as digitalization (not to be confused with digitization, the process of converting analog data into a digital counterpart). In this book, we use the word digitalization to describe the evolution process, in which digital technologies (meaning electronic, information, communication, etc. technologies) enable changes in business and society. These changes can range from continuous and smooth to erratic and disruptive. On the one hand, new digital technologies are the drivers for changes in classical processes (e.g., text-processing software has made the old mechanical typewriters obsolete). On the other hand, a need for business or organizational processes is looking for the best available digital technology solution to overcome old processes (e.g., a suitable voice-over-IP protocol is in the process of replacing the traditional telephone transmission). In this book, in the ?rst case, we call it digital innovation, in the second case, digital business transformation. In general, however, both tendencies are often interconnected and not so clearly separated. In order to provide a selection of the diversity of what digital transformation means for our daily lives, this book presents many different digitalization cases—some more focused on digital innovation and some on digital business transformation. This book was written by professors and researchers who work, research, and teach daily in the areas of (Business) Information Systems, (Business) Information Technology, Computer Science, Business Administration and Management at the School of Business of the University of Applied Sciences and Arts Northwestern Switzerland FHNW. This university of applied sciences focuses on applied research and development while retaining a high level of practical relevance. Thus, the writing style of the chapters of this book follows the philosophy of reconciling the degree of depth and rigor in research with a meaningful translation into relevance in practice. v vi Preface While the Institute for Information Systems of the School of Business originally focused on bridging the gap between business and IT, thus aligning business and IT in the context of organizations, its topics are now expanding extensively to cover a wider range of digitalization: The impact of new ICT-related technologies and IT-supported methods in business and society is investigated. In addition to the classic domains of business information technology, our areas of interest are, for example, agile process management in a digital economy, cloud computing and related digital transformations, cybersecurity and resilience, digital supply chain management, arti?cial intelligence including computational intelligence (i.e., nature-inspired AI) and symbolic/sub-symbolic AI, Internet-of-Things, human– machine interaction, (social) robotics, management of complex systems, humans in technical systems and organizations, etc. Our passion for the different facets of digitalization eventually led to the idea of writing this second book (after the Springer book “Business Information Systems and Technology 4.0” in 2018). As the editor of the book, I would like to thank our employer, the University of Applied Sciences and Arts Northwestern Switzerland FHNW and particularly the School of Business for supporting the writing of the book by granting additional hours to the authors. My immense and heartfelt thanks also go to all our authors, who have made excellent contributions to this book with their view of digitalization. My special thanks go to Prof. Dr. Thomas Hanne, Prof. Dr. Uwe Leimstoll, and Prof. Dr. Michael von Kutzschenbach, who have organized a thorough and independent scienti?c peer review of all the chapters according to internationally recognized high quality standards. Additionally, I would like to thank Vivienne Jia Zhong for coordinating the process of elaborating the contributions of 42 authors, compiled in 21 chapters, ef?ciently and competently. I would also like to thank Christine Lorgé and Natalie Jonkers for their diligence, competence, and commitment to language and readability. Finally, I would like to thank all our families for their continued patience and great support in allowing us to write this book on weekends and during the night. Basel, Switzerland January 2020 Prof. Dr. Rolf Dornberger Contents Digital Innovation and Digital Business Transformation in the Age of Digital Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rolf Dornberger and Dino Schwaferts 1 Digital Innovation A Survey of State of the Art Methods Employed in the Of?ine Signature Veri?cation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Michael Stauffer, Paul Maergner, Andreas Fischer, and Kaspar Riesen 17 Agile Visualization in Design Thinking . . . . . . . . . . . . . . . . . . . . . . . . . Emanuele Laurenzi, Knut Hinkelmann, Devid Montecchiari, and Mini Goel 31 Text Mining Innovation for Business . . . . . . . . . . . . . . . . . . . . . . . . . . . Ela Pustulka and Thomas Hanne 49 Using Mobile Sensing on Smartphones for the Management of Daily Life Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dilip Menon, Safak Korkut, Terry Inglese, and Rolf Dornberger 63 A Dialog-Based Tutoring System for Project-Based Learning in Information Systems Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hans Friedrich Witschel, Prajakta Diwanji, and Knut Hinkelmann 81 A Human Aptitude Test for Object-Oriented Programming in the Context of AI and Machine Learning . . . . . . . . . . . . . . . . . . . . . Rainer Telesko and Stephan Jüngling 97 Adapting the Teaching of Computational Intelligence Techniques to Improve Learning Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Thomas Hanne and Rolf Dornberger vii viii Contents Automatic Programming of Cellular Automata and Arti?cial Neural Networks Guided by Philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 Patrik Christen and Olivier Del Fabbro Facial Recognition and Path?nding on the Humanoid Robot Pepper as a Starting Point for Social Interaction . . . . . . . . . . . . . . . . . . . . . . . . 147 Achim Dannecker and Daniel Hertig Digital Business Transformation Social Robots in Organizational Contexts: The Role of Culture and Future Research Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 Theresa Schmiedel, Janine Jäger, and Vivienne Jia Zhong Digital Transformation for Sustainability: A Necessary Technical and Mental Revolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 Michael von Kutzschenbach and Claus-Heinrich Daub Visualization of Patterns for Hybrid Learning and Reasoning with Human Involvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 Hans Friedrich Witschel, Charuta Pande, Andreas Martin, Emanuele Laurenzi, and Knut Hinkelmann Modeling the Instructional Design of a Language Training for Professional Purposes, Using Augmented Reality . . . . . . . . . . . . . . . 205 Terry Inglese and Safak Korkut Frictionless Commerce and Seamless Payment . . . . . . . . . . . . . . . . . . . . 223 Michael H. Quade Direct to Consumer (D2C) E-Commerce: Goals and Strategies of Brand Manufacturers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 Uwe Leimstoll and Ralf Wölfle The Digital Marketing Toolkit: A Literature Review for the Identi?cation of Digital Marketing Channels and Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 Marc K. Peter and Martina Dalla Vecchia How the Internet of Things Drives Innovation for the Logistics of the Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 Herbert Ruile Recommendations for Conducting Service-Dominant Logic Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 Joachim C. F. Ehrenthal, Thomas W. Gruen, and Joerg S. Hofstetter Contents ix Blockchain Technologies Towards Data Privacy—Hyperledger Sawtooth as Unit of Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 Pascal Moriggl, Petra Maria Asprion, and Bettina Schneider Leadership in the Age of Arti?cial Intelligence—Exploring Links and Implications in Internationally Operating Insurance Companies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315 Sarah-Louise Richter and Dörte Resch Digital Innovation and Digital Business Transformation in the Age of Digital Change Rolf Dornberger and Dino Schwaferts Abstract This chapter introduces the central theme of the book and the following chapters. In our book, we present 20 selected examples of digitalization in the age of digital change. The chapters are divided into two parts, called “Digital Innovation” and “Digital Business Transformation”. On the one hand, digital innovation includes cases where a new technology stimulates or enables new business opportunities. On the other hand, digital business transformation collects examples of digitalization in which business or management concepts exploit certain technological solutions for their practical implementation. To share this mind-set with the reader, quantumcomputing technology will be taken as an example of digital innovation. This will be followed by reflections on digital business transformation, which seeks and uses digital technologies. Finally, the next chapters will be presented accordingly. Keywords Digital innovation · Digital business transformation · Digitalization · Quantum computing · Digital eco-system 1 Introduction It is widely predicted that digitalization will change the world. Already in 2018, the book “Business Information Systems and Technology 4.0: New Trends in the Age of Digital Change” [1] has presented selected cases which show how quickly the world changes and transforms. That book has provided explanations of key principles such as digitization, digitalization, and digital transformation. Therefore, the model in Fig. 1 “Digitalization: Yesterday, today and tomorrow” [2] was introduced. This R. Dornberger (B) School of Business, Institute for Information Systems, FHNW University of Applied Sciences and Arts Northwestern, Peter Merian-Strasse 86, 4002 Basel, Switzerland e-mail: rolf.dornberger@fhnw.ch D. Schwaferts School of Business, Institute for Information Systems, FHNW University of Applied Sciences and Arts Northwestern, Riggenbachstrasse 16, 4600 Olten, Switzerland e-mail: dino.schwaferts@fhnw.ch © Springer Nature Switzerland AG 2021 R. Dornberger (ed.), New Trends in Business Information Systems and Technology, Studies in Systems, Decision and Control 294, https://doi.org/10.1007/978-3-030-48332-6_1 1 2 R. Dornberger and D. Schwaferts Fig. 1 Digitalization: yesterday, today and tomorrow [2] model discusses the development of the information and communication technology over four stages: The sequence begins with (1) Information Systems in the 1980s and 1990s and (2) E-Business Applications in the late 1990s and early 2000s. In addition, the rise of web applications and smartphones from the mid to the end of the first decade of the 21st century, launched the (3) Web 2.0 Revolution. The sequence concludes with the renaissance of (4) Artificial Intelligence from the mid to the end of the second decade of the 21st century. As illustrated in the model (see Fig. 1), the development of new digital technologies and applications seems to be growing exponentially (e.g., the number of digital technologies steadily increases; every year the number of scientific publications increases worldwide, etc. [3]). This implies that a vast amount of potentially world-changing innovations is constantly being generated. As examples, the authors stated in [1] that in addition to improvements in smarter software, new hardware devices are being created, such as robots and 3D printers, equipping the computers with a kind of a body and reshaping the supply chains. In parallel, the cyber-world is evolving, increasingly encompassing technologies such as virtual and augmented reality (VR/AR), which provide the interface to a digital or cyber world. Many aspects of the world are now becoming “cyber”. For example, cryptocurrencies are evolving and expanding established forms of money; and enhancing humans with artificial robotic body parts is no longer a distant dream. Digital Innovation and Digital Business Transformation … 3 This book goes a step further to provide new examples of how digital technologies and/or business opportunities created by digitalization cause innovation and transformation. It therefore builds on the model with the four stages discussed in [2], but also proposes the two aspects “Digital Innovation” and “Digital Business Transformation”, which cross-sectionally cover the four stages. Here, digital innovation means the use of new or existing digital technologies (from another context) to support certain aspects of management or business. However, digital business transformation is—according to Gartner [4]—“the process of exploiting digital technologies and supporting capabilities to create a robust new digital business model.” To present and organize the subsequent chapters of this book, this introductory chapter briefly mentions the two aspects “digital innovation” and “digital business transformation”. The first aspect concerns a new technology (such as quantum computing) that is the starting point and enabler for new applications and business solutions. The second aspect concerns business/management ideas or concepts (such as digital eco-systems) that need digital technologies to support and implement them. As a guide for the readership of this book, all the ensuing chapters are briefly outlined and focus either on the first aspect “digital innovation” or the second “digital business transformation”. 2 The Technology of Quantum Computing as Digital Innovation 2.1 Identifying Promising Technologies In order to predict whether and how a particular new digital technology can become a promising technology in the sense of enabling one of the most important digital innovations of the future, the important technologies to come must be continuously identified. The question often is: What will come next? In the model [2] in Fig. 1, many readers may wonder what the question mark on the rising curve means. Albert Einstein answered this question very elegantly by saying: “I never think of the future—it comes soon enough.”1 According to this quote, the future seems to be approaching even faster today than in earlier times. A well-known source and reference for future technologies and trends is the Gartner hype cycle for emerging technologies [5]. It proposes and discusses future technologies and trends in four phases: 1. 2. 3. 4. Being a hype (peak of inflated expectations) Producing a disappointment (trough of disillusionment) Getting a realistic estimation (slope of enlightenment) Leading to common application (plateau of productivity). 1 https://quoteinvestigator.com/2013/07/23/future-soon/. 4 R. Dornberger and D. Schwaferts Once a new technology or trend appears on the hype cycle, it takes up a particular position on the cycle according to the estimation of its phase and in comparison to other technologies. Additionally, the predicted time until it becomes a common application is noted. 2.2 Quantum Computing as an Example for Digital Innovation One technology that has been on the rising slope of the Gartner hype cycle for years is “quantum computing”, which has appeared 11 times since 2000. In fact, Gartner predicted that quantum computing would reach its plateau of productivity in a maximum of five years or that over the next five to ten years it would be in general use. Thus, one can assume that the digital technology of quantum computing is an answer to the question mark in Fig. 1 and will become one of the next digital innovations. Why, however, will quantum computing become so important? Quantum computing [6] is a novel ground-breaking mechanism for computing/processing units, leading to a new computer generation with an immensely higher computing speed. Its principles are based on the phenomena of quantum physics that can be used within a new type of central processing units inside a computer. While the theory is well developed, its realization in new hardware and software to take advantage of its full potential remains difficult. Currently, there is a handful of highly advanced research labs, which possess more or less well-running prototypes of quantum computers. However, there is still a chance that the realization of application-ready quantum-computing devices will not be possible in this century. Assuming that ready-to-use quantum computing will be realized in the next years, one can say that it will change the world fundamentally. There will be a jump from today’s “nice computing power” to “unbelievable, ultimate computing power”, in which case the only limitations will be the computing memory and the available algorithms (software) to compute (almost) everything, known as Quantum Supremacy [7]. Some reflections below show why quantum computing is on the way to becoming such an important digital innovation—perhaps the most disruptive digital innovation of this century [8]. Bitcoin and similar cryptocurrencies will lose their motivation [9]. Theoretically, with quantum computing, all bitcoins will be instantly mined, because the computationally expensive computations for adding new blocks to the blockchain system will be carried out instantly. The miner (a specific role in the blockchain system) who gets first access to quantum computing will mine all bitcoins (perhaps introducing stupid transactions to generate new blocks with senseless back and forth transactions). Certainly, there are mechanisms, such as time delays that are integrated into the mining procedure; but today’s standard ideas of generating particular cryptocurrencies (through computationally expensive blockchain computations) will have to be completely rethought. Digital Innovation and Digital Business Transformation … 5 In cryptography, quantum computing will break through almost any current encryption in no time [10]. Certainly, a new discipline will come into play called “quantum cryptography” that adapts and replaces outdated encryption algorithms. Nevertheless, in every encryption and decryption process, quantum computers will always be superior to traditional computing devices, even to today’s supercomputers. For example, as cars exchange encrypted commands with their remote keys, it will allow hackers to use quantum computers to unlock not only one car, but also all other cars in the vicinity within a fraction of a millisecond. DeepMind Technologies is a subsidiary of Alphabet Inc., the mother company of Google. The research of DeepMind focuses on different aspects of extending and applying Artificial Intelligence (AI). A showcase is the gameplay of the ancient Asian board game “Go”. The projects AlphaGo, AlphaGo Zero, and Alpha Zero developed AI frameworks [11] that are (almost) always able to beat human world champions and other software in Go [12] and in various other kinds of single- and multi-player games [13]. DeepMind’s AI approaches are superior to most other (published) approaches, including other software solutions, as well as to human gameplay strategies in games with (almost) infinite possible moves. The algorithms are special heuristics. Some of them are trained with huge human-generated datasets; some of them use self-training by generating own dataset by applying the game rules. These powerful heuristics search limited parts of the entire search space, where promising next moves seem favorable. However, a quantum computer may be able to search the entire search space (not only limited parts) and discover unknown sequences of moves that could even beat the best powerful heuristics. In other words, the ultimate quantum-computing power has great potential to further enhance today’s best AI and decision-making algorithms—but also to make some AI approaches obsolete by instantly scanning the entire search space. Another major application field in which quantum computing will have a tremendous impact, is modeling and simulation to be known as Quantum Simulation [14]. For example, the knowledge of how the universe, the solar system, and earth evolved over billions of years is still incomplete. Many details are still insufficiently understood. Today, different physical astronomical models and approaches are available to describe the development of the universe. Unfortunately, these models are still partially incompatible, and even supercomputers take weeks to compute results in well-selected sub-cases. Here, quantum computing will be able to compute different models with increasing granularity in almost no time. Then it will be possible to easily compare the results of the different models, improve the models, and gain an understanding about which models are best suited for different cases and how they fit together. In addition to astronomy, which takes place on the large scales (i.e. space and time), quantum computing will also help to understand the effects on the small scales, where quantum physics explores them. Another discipline in which increased computing speed will have an extraordinary effect is engineering, which mostly uses search and optimization algorithms to computationally solve engineering problems [15]. In the last 40 years, numerical computations have already replaced various hardware tests and field experiments. Today, e.g., the most efficient engines, cars, and airplanes are developed and further 6 R. Dornberger and D. Schwaferts optimized on the computer; prototype testing is reduced to a minimum. Furthermore, with quantum computing, it will be possible to simultaneously compute and compare millions of design variants. Each product is designed and optimized on the computer until optimal properties are reached. Then, because of myriads of computational optimization loops, all products of various companies will be quite similar. For example, washing machines, smartphones, and airplanes will be almost identical independently of the producing company. On the other hand, the differentiation will lie in the personalization of engineered products [16]. Users will get an optimal, personalized product that is tailored to their requirements, e.g., a personalized car (shape and size, engine, color, interior, etc.). Hundreds of personal choice features, including an optimized personal product design, can be integrated during the online configuration. Hence, quantum computing will create a new “personalized product business” [17]. Extending the idea of being able to design whatever is technically possible, we will also be able to research and develop new ways to preserve nature [18]. Ideally, we will be able to model, simulate, and optimize countless variants of environmental technologies and sustainable, renewable energy solutions. We will be able to find ways to reduce carbon dioxide emissions and other pollution markers. In addition, we will be able to better understand society as a living organism, and, in turn, enhance the living conditions of billions of people, merely by discovering new technologies and new ways for these solutions to influence the quality of life. Certainly, all these examples of the benefits of quantum computing are very optimistic. However, as with any technology, quantum computing can also be used wrongfully, which we will not discuss here. In summary, when digital technologies find their way to common applications, they are adopted as digital innovations. In this chapter, the technology of quantum computing was chosen as an example to explain how a new technology might stimulate many new applications and business opportunities, which is consequently called digital innovation. Later in the book in the Digital Innovation part, several chapters are collected to present specific technologies or technological approaches, which stimulate new applications, businesses, processes, or even thinking. 3 Leveraging Digital Business Transformation Through Digital Technologies The section above describes the concept of digital innovation (used in this book to structure the upcoming chapters) using the example of quantum computing (i.e. how a digital technology might stimulate new applications). This section focuses on the possibility of realizing additional opportunities in business, management, learning, etc. by seeking and then exploiting new specific digital technologies. This is related to the already mentioned definition of digital business transformation [4] as “the process of exploiting digital technologies and supporting capabilities to create a Digital Innovation and Digital Business Transformation … 7 robust new digital business model.” However, the wording “business model” is used in a broader context, namely as “business, management or process opportunities”, which can be realized, supported, and/or improved by seeking and exploiting suitable digital technologies. In the next section, some reflections are made about rethinking managerial approaches for business transformation in the age of digital change. 3.1 Rethinking Managerial Approaches for Business Transformation in the Age of Digital Change When reflecting on the most important achievements of the industrial era around the 19th century, steam power, electrification, and the automobile could come to mind first. In parallel with the emergence of these new technologies, new management challenges also arose. According to the economic historian Pollard [19], current management approaches and topics have their origin in the emergence of largescale organizations during the industrial era, where large companies were explicitly confronted with challenges on another extended scale. E.g. human resource management for mass employment, leadership and management of a big entity, huge financial issues, procurement and sales on a broad level, etc. Following Pollard, the origin of today’s management approaches is strongly connected to the emergence of new technologies—in addition to various societal developments, ecological challenges and so forth. Today, a new type of technology, namely the digital technologies, such as the Internet of Things (IoT), AI, Big Data, Cloud Computing, robotics, etc. have an impact on life and work [20] (e.g., all the IoT technology to be applied in Industry 4.0). Such a transition to a digital society and economy leads to different expectations towards management by both customers and employees. The question could be which new managerial approaches are emerging for business transformation in the age of digital change—to be reflected on in the next section and in the upcoming book chapters under the umbrella term “Examples for an increasing need for Digital Business Transformation”. 3.2 About Examples in Digital Business Transformation The following are some reflections on the particular state of companies and new possibilities offered by digital business transformation (i.e. the exploitation of digital technologies and supporting capabilities assisting company needs). The question is how digital technologies influence companies in the age of digital change. Searching for management concepts and explanations of why companies in the industrial age are as they are, one might still refer to Coase [21], where he already asked in the 1930s, why and under what circumstances what kind of company would 8 R. Dornberger and D. Schwaferts emerge. He concluded that a meaningful enterprise size and a special state of the company are the result of the balance between disadvantages from transaction costs and disadvantages from overhead costs, imperfection in resource allocation, and inflexibility. In the 70s and 80s of the last century, Porter [22] defined the state of a company by the concept of the value chain, where the value added performed in an organization determines the size and the condition of the company. However, around the year 2000, Porter added the influence of digitalization to his model, where the value added is now divided among different partners in a networked structure, which he calls “the system of systems”. Examples in business are the automotive industry, which has long reduced its vertical integration, similar to early examples of virtual organizations, which have been building on this concept for more than 20 years [23]. In the age of digital change, the question arises: what exactly is the benefit of having the opportunity to exploit digital technologies to change the style of management and the state of companies? Can structures and processes be better defined to contribute more to the value creation? A particularly great opportunity that digital technologies offer today, is the simple division of tasks, the value chain, the management, the IT and so forth of every entity of the company (e.g., on the level of a project, a department or even the entire company). This means that many tasks—in terms of total value added—could now be divided and elaborated in different entities in parallel, fully supported by digital technologies. These entities could even lie outside the company, following the trend of focusing on one’s own unique abilities. Each entity makes its own contribution to value creation by being connected and working together via digital technologies. Here, Porter’s system of systems [22] with the flexible composition of services in the value chain and their orchestration or digital eco-systems [24] provide answers on how digitalization makes this possible [25]. Although these short reflections are only taken as an example in digital business transformation, it can be assumed that digital technologies have an important impact on management at various levels. Digital business transformation is strongly related, for example, to managerial approaches, the style of management, state of companies as well as to particular management topics (e.g. HR management or learning management) which are quickly changing nowadays. Therefore, it is indispensable to continuously investigate various upcoming management, business, societal demands, and further challenges in the search for digital technologies and solutions that support them. Selected topics are provided later in this book in the Digital Business Transformation part. 4 Organization of This Book As explained above, this book consists of two parts, which represent the abovementioned topic fields, namely Digital Innovation and Digital Business Transformation, and which, in turn, compile the more technology-based topics and the more business/application-relevant topics. In the following, a very short description and a brief summary of the book chapters is given. Digital Innovation and Digital Business Transformation … 9 (a) Examples of emerging Digital Innovation (according to chapter order later in the book) • Various digitization technologies have been researched for several years and optical character recognition (known as OCR) is nothing new. Nevertheless, an update with the latest machine-learning algorithms proves a new technological leap. “A Survey of State of the Art Methods Employed in the Offline Signature Verification Process” reviews the domain of offline signature verification and presents a comprehensive overview of methods typically employed in the general process of offline signature verification. • When it comes to innovative idea generation, Design Thinking is mentioned as one of the most widespread and effective methods. However, in addition to all attempts to improve the design thinking processes, a sensible technological implementation and related scenarios, as discussed in “Agile Visualization in Design Thinking”, promise a further innovative approach. • In general, enriching classical algorithms with machine-learning approaches further improves their potential. An example is given in the text mining case “Text Mining Innovation for Business”, reflecting on the business innovation enabled by developing text-mining solutions in response to the business needs of companies. • The sensible use of embedded digital technologies can lead to new applications and generate digital innovations as shown in “Using Mobile Sensing on Smartphones for the Management of Daily Life Tasks”, which explores the ability and the potential of the mobile-sensing technology to support daily life task management. • The use of an advanced software solution in university education provides an advanced tutoring system. “A Dialog-Based Tutoring System for Project-Based Learning in Information Systems Education” presents and discusses the design of an intelligent dialog-based tutoring system that is designed to support students during group projects, to maintain their motivation and give subtle hints for selfdirected discovery. • However, it is often difficult to teach technologically advanced topics, for example object-oriented programming. In order to identify and evaluate the students’ learning difficulties, a case of developing an assessment test is presented. “A Human Aptitude Test for Object-Oriented Programming in the Context of AI and Machine Learning” discusses the object-oriented programming paradigm and the potential future developments once AI and machine learning start to steadily increase their influence on the overall design of software systems. • Methods that belong to AI tend to be among the more challenging subjects for students. A case of evaluating different measures to improve a related university course over ten years is discussed in order to propose best practices. “Adapting the Teaching of Computational Intelligence Techniques for Improving the Learning Outcomes” shows the continuous efforts to integrate nature-inspired AI, and in particular, computational intelligence algorithms into today’s university teaching to increase the learning success of students. 10 R. Dornberger and D. Schwaferts • Some technologies force research to discuss them from the philosophical side. Maybe our view of technology is too strongly influenced by a too narrow perception of the world: We will see what we allow to be seen. “Automatic Programming of Cellular Automata and Artificial Neural Networks Guided by Philosophy” discusses an approach called “allagmatic method” that automatically programs and executes models with as little limitations as possible while maintaining human interpretability. • Some technologies themselves are “hardcore”, like robots. They are a hype; they will change the world and businesses. “Facial Recognition and Pathfinding on the Humanoid Robot Pepper as a Starting Point for Social Interaction” examines the facial recognition and navigation capabilities of the robot Pepper and relates them to a better human-robot interaction. (b) Examples for an increasing need to Digital Business Transformation (according to the chapter sequence later in the book) • However, robots are not only a new hardware device with extended software capabilities. Organizations might have special process needs which best can be suited by involving and integrating robots. “Social Robots in Organizational Contexts: The Role of Culture and Future Research Needs” describes the organizational contexts currently relevant for social robots addressing the respective cultural challenges. • Management must innovate and adapt in order to leverage the benefits of digitalization for a digitalized sustainability society. Here, “Digital Transformation for Sustainability: A Necessary Technical and Mental Revolution” outlines an integrated framework linking different levels of digitalness with necessary changes in managerial practice for organizational inquiry. • Keeping humans-in-the-loop provides many advantages. For example, “Visualization of Patterns for Hybrid Learning and Reasoning with Human Involvement” discusses the traditional roles of humans in hybrid systems and their influence on machine learning and/or knowledge engineering activities. • Learning can be improved by sensibly using selected technologies. Here, “Modeling the Instructional Design of a Language Training for Professional Purposes, Using Augmented Reality” presents a multi-layered instructional model for language training for professional adult learners. • The processes in e-commerce are continuously improved by steadily integrating and using new technologies. “Frictionless Commerce and Seamless Payment” discusses a potential solution for improving the user experience and conversion optimization in e-commerce and provides existing practical examples. • E-commerce has not necessarily led to disintermediation as predicted in the 1990s. “Direct to Consumer (D2C) E-Commerce: Goals and Strategies of Brand Manufacturers” suggests three strategies for brand manufacturers to mitigate conflicts with traditional sales partners while maintaining their presence on the Internet. Digital Innovation and Digital Business Transformation … 11 • Marketing is another management topic, where new technologies are highly appreciated to strengthen customer retention. “The Digital Marketing Toolkit: A Literature Review for the Identification of Digital Marketing Channels and Platforms” presents a study to support SMEs in their digital business transformation. • Supply chain management also quickly adsorbs new technologies to improve its processes. “How the Internet of Things Drives Innovation for the Logistics of the Future” discusses an innovative value chain framework relating technology advancement to business innovation in logistics. • Co-creation in networks is highly supported by digital technologies. As an example, “Recommendations for Conducting Service-Dominant Logic Research” addresses a research gap providing contextual guidance for applying servicedominant logic in research and a framework for innovation in business. • Many recent business models rely on data as the most important asset, where data privacy is always an issue. “Blockchain Technologies towards Data Privacy— Hyperledger Sawtooth as Unit of Analysis” presents a study that provides insights for selecting a suitable blockchain configuration that would comply with regulatory data privacy requirements. • “Leadership in the Age of Artificial Intelligence—Exploring Links and Implications in Internationally Operating Insurance Companies” describes a case that demonstrates the need (e.g., with the introduction of AI) for additional management approaches. 5 Conclusions Everywhere, examples and good practices in the field of digitalization appear again and again, which aim to promise a change of the world, stimulated partly by new technologies, partly by business/management/societal needs, often by both. A promising way to identify possible next disruptive leaps in today’s age of digital change is to continuously and profoundly explore a broad variety of examples. A small selection is shown in this book. This chapter is intended to introduce the central theme of the book for the arrangement of the upcoming chapters grouped to two parts. Therefore, the aspect of Digital Innovation is introduced using the example of the upcoming digital technology of quantum computing that will most likely change the world. Then, the aspect of Digital Business Transformation is proposed and reflected as a change of management/business to exploit digital technologies. Finally, these insights are briefly mapped to the following book chapters, which—as will become obvious during the course of the book—show various new trends and challenges in the age of digital change from different perspectives. Maybe one of them will lead to the next disruptive leap in digitalization. 12 R. Dornberger and D. Schwaferts References 1. Dornberger, R.: Business Information Systems and Technology 4.0: New Trends in the Age of Digital Change. Springer (2018) 2. Dornberger, R., Inglese, T., Korkut, S., Zhong, V.J.: Digitalization: yesterday, today and tomorrow. In: Dornberger, R. (ed.) Business Information Systems and Technology 4.0: New Trends in the Age of Digital Change, pp. 1–11. Springer International Publishing, Cham (2018) 3. Bornmann, L., Mutz, R.: Growth rates of modern science: a bibliometric analysis based on the number of publications and cited references. J. Assoc. Inf. Sci. Technol. 66, 2215–2222 (2015). https://doi.org/10.1002/asi.23329 4. Digital Business Transformation. https://www.gartner.com/en/information-technology/glo ssary/digital-business-transformation 5. Fenn, J., Blosch, M.: Understanding Gartner’s Hype Cycles. https://www.gartner.com/en/doc uments/3887767/understanding-gartner-s-hype-cycles 6. Farouk, A., Tarawneh, O., Elhoseny, M., Batle, J., Naseri, M., Hassanien, A.E., Abedl-Aty, M.: Quantum Computing and Cryptography: An Overview. In: Hassanien, A.E., Elhoseny, M., Kacprzyk, J. (eds.) Quantum Computing: An Environment for Intelligent Large Scale Real Application, pp. 63–100. Springer International Publishing, Cham (2018) 7. Harrow, A.W., Montanaro, A.: Quantum computational supremacy. Nature 549, 203–209 (2017). https://doi.org/10.1038/nature23458 8. Ferrie, C.: Why are scientists so excited about a recently claimed quantum computing milestone? http://theconversation.com/why-are-scientists-so-excited-about-a-rec ently-claimed-quantum-computing-milestone-124082 9. Hayes, A.S.: Cryptocurrency value formation: An empirical study leading to a cost of production model for valuing bitcoin. Telemat. Inform. 34, 1308–1321 (2017). https://doi.org/10. 1016/j.tele.2016.05.005 10. Hallgren, S., Vollmer, U.: Quantum computing. In: Bernstein, D.J., Buchmann, J., Dahmen, E. (eds.) Post-quantum cryptography, pp. 15–34. Springer, Berlin Heidelberg, Berlin, Heidelberg (2009) 11. Silver, D., Huang, A., Maddison, C.J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., Dieleman, S., Grewe, D., Nham, J., Kalchbrenner, N., Sutskever, I., Lillicrap, T., Leach, M., Kavukcuoglu, K., Graepel, T., Hassabis, D.: Mastering the game of Go with deep neural networks and tree search. Nature 529, 484–489 (2016). https://doi.org/10.1038/nature16961 12. Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., Hubert, T., Baker, L., Lai, M., Bolton, A., Chen, Y., Lillicrap, T., Hui, F., Sifre, L., van den Driessche, G., Graepel, T., Hassabis, D.: Mastering the game of Go without human knowledge. Nature 550, 354–359 (2017). https://doi.org/10.1038/nature24270 13. Lanctot, M., Lockhart, E., Lespiau, J.-B., Zambaldi, V., Upadhyay, S., Pérolat, J., Srinivasan, S., Timbers, F., Tuyls, K., Omidshafiei, S., Hennes, D., Morrill, D., Muller, P., Ewalds, T., Faulkner, R., Kramár, J., De Vylder, B., Saeta, B., Bradbury, J., Ding, D., Borgeaud, S., Lai, M., Schrittwieser, J., Anthony, T., Hughes, E., Danihelka, I., Ryan-Davis, J.: OpenSpiel: A Framework for Reinforcement Learning in Games. ArXiv190809453 Cs. (2019) 14. Georgescu, I.M., Ashhab, S., Nori, F.: Quantum simulation. Rev. Mod. Phys. 86, 153–185 (2014). https://doi.org/10.1103/RevModPhys.86.153 15. Williams, C.P.: Quantum search algorithms in science and engineering. Comput. Sci. Eng. 3, 44–51 (2001). https://doi.org/10.1109/5992.909001 16. Berry, C., Wang, H., Hu, S.J.: Product architecting for personalization. J. Manuf. Syst. 32, 404–411 (2013). https://doi.org/10.1016/j.jmsy.2013.04.012 17. Moon, J., Chadee, D., Tikoo, S.: Culture, product type, and price influences on consumer purchase intention to buy personalized products online. J. Bus. Res. 61, 31–39 (2008). https:// doi.org/10.1016/j.jbusres.2006.05.012 Digital Innovation and Digital Business Transformation … 13 18. Mohseni, M., Read, P., Neven, H., Boixo, S., Denchev, V., Babbush, R., Fowler, A., Smelyanskiy, V., Martinis, J.: Commercialize quantum technologies in five years. Nature 543, 171–174 (2017). https://doi.org/10.1038/543171a 19. Pollard, S.: The Genesis of Modern Management: a Study of the Industrial Revolution in Great Britain. Harvard University Press (1965) 20. Lasi, H., Fettke, P., Kemper, H.-G., Feld, T., Hoffmann, M.: Industry 4.0. Bus. Inf. Syst. Eng. 6, 239–242 (2014). https://doi.org/10.1007/s12599-014-0334-4 21. Coase, R.H.: The nature of the firm. Economica 4, 386–405 (1937). https://doi.org/10.1111/j. 1468-0335.1937.tb00002.x 22. Porter, M.E., Heppelmann, J.E.: How smart, connected products are transforming competition. Harv. Bus. Rev. 92, 64–88 (2014) 23. Porter, M.E.: Location, competition, and economic development: local clus-ters in a global economy. Econ. Dev. Q. 14(1), 15–34 (2000). https://doi.org/10.1177/089124240001400105 24. Hol, A., Ginige, A., Lawson, R.: System level analysis of how businesses adjust to changing environment in the digital eco-system. In: 2007 Inaugural IEEE-IES Digital EcoSystems and Technologies Conference, pp. 153–158 (2007) 25. Grieder, H., Schwaferts, D.: Digital ecosystem—how companies can achieve a sustainable competitiveness in the future and the digital age. In: Verkuil, A.H., Hinkelmann, K., Aeschbacher, M. (eds.) Digitalisierung und andere Innovationsformen im Management. Edition gesowip, Basel (2019) Digital Innovation A Survey of State of the Art Methods Employed in the Offline Signature Verification Process Michael Stauffer , Paul Maergner , Andreas Fischer , and Kaspar Riesen Abstract Handwritten signatures are of eminent importance in many business and legal activities around the world. That is, signatures have been used as authentication and verification measure for several centuries. However, the high relevance of signatures is accompanied with a certain risk of misuse. To mitigate this risk, automatic signature verification was proposed. Given a questioned signature, signature verification systems aim to distinguish between genuine and forged signatures. In the last decades, a large number of different signature verification frameworks have been proposed. Basically, these frameworks can be divided into online and offline approaches. In the case of online signature verification, temporal information about the writing process is available, while offline signature verification is limited to spatial information only. Hence, offline signature verification is generally regarded as the more challenging task. The present chapter reviews the field of offline signature verification and presents a comprehensive overview of methods typically employed in the general process of offline signature verification. Keywords Offline signature verification · Image preprocessing · Handwriting representation · Signature classification M. Stauffer · K. Riesen (B) University of Applied Sciences and Arts Northwestern Switzerland, Institute for Information Systems, Riggenbachstrasse 16, 4600 Olten, Switzerland e-mail: kaspar.riesen@fhnw.ch M. Stauffer e-mail: michael.stauffer@fhnw.ch P. Maergner · A. Fischer Department of Informatics, University of Fribourg, Boulevard de Pérolles 90, 1700 Fribourg, Switzerland e-mail: paul.maergner@unifr.ch A. Fischer e-mail: andreas.fischer@unifr.ch A. Fischer University of Applied Sciences and Arts Western Switzerland, Boulevard de Pérolles 80, 1700 Fribourg, Switzerland © Springer Nature Switzerland AG 2021 R. Dornberger (ed.), New Trends in Business Information Systems and Technology, Studies in Systems, Decision and Control 294, https://doi.org/10.1007/978-3-030-48332-6_2 17 18 M. Stauffer et al. 1 Introduction For several hundred years, handwritten signatures have been used as biometric authentication and verification measure in a broad range of business and legal transactions around the world. In fact, the first use of handwritten signatures can be traced back to the fourth century, where a signature was used to protect the Talmud (i.e. a central text in Rabbinic Judaism) against possible modifications. Later on, signatures (i.e. so-called subscripto) were used to authenticate documents in the Roman Empire during the period of Valentinian III [14]. More recently, several national and international digitization efforts for electronic signatures have been proposed. Yet, handwritten signatures remain very popular [8, 19, 20]. This is mostly due to the ubiquitous applicability as well as the overall high availability and acceptance of handwritten signatures when compared with electronic signatures. Due to the high commercial and legal relevance, handwritten signatures also expose a certain risk of misuse. To mitigate this risk, signature verification can be employed. That is, a questioned signature is compared with a set of reference signatures in order to distinguish between genuine and forged signatures [20]. Traditionally, this task has been conducted by human experts as part of graphology, i.e. the study of handwriting. However, due to increasing number of documents, automatic signature verification systems have been proposed already some decades ago [28]. Actually, manual signature verification has been continuously replaced since then. Today, automatic signature verification is still regarded as a challenging pattern recognition task, and thus, remains a very active research field [8]. In general, automatic signature verification approaches can be divided into online (also termed dynamic) and offline (also termed static) approaches [19]. In the case of online signature verification, signatures are acquired by means of an electronic input device, such as a digital pen, a digital tablet, or via input on a touch screen (e.g. smartphone or handheld device). Hence, dynamic temporal information about the signing process (e.g. acceleration, speed, pressure, and pen angle such as altitude and azimuth) can be recorded during the handwriting process. In contrast, offline acquired signatures are digitized by means of scanning. As a result, the verification task is solely based on the (x, y)-positions of the handwriting (i.e. strokes), and thus, offline signature verification is generally considered the more challenging case. Note that the present chapter focuses only on offline signature verification. Signature verification systems can also be distinguished with respect to the representation formalism that is actually used, viz. statistical (i.e. vectorial) and structural representations. The vast majority of signature verification approaches make use of statistical representations [8, 19]. That is, feature vectors or sequences of feature vectors are extracted from handwritten signature images. In early approaches, features are, for example, based on global handwriting characteristics like contour [6, 16], outline [10], projection profiles [30], or slant direction [2, 13]. However, more generic and local feature descriptors (i.e. descriptors that are not limited to handwriting images) were also used such as Histogram of Oriented Gradients (HoG) [41] and Local Binary Patterns (LBP) [12, 41]. In recent years, features have increasingly been A Survey of State of the Art Methods Employed … 19 extracted by means of a learned statistical model, viz. Deep Learning approaches like Convolutional Neural Networks (CNNs) [7, 18, 26]. Regardless of the actual type of feature, different statistical classifiers and/or matching schemes have been employed to distinguish genuine from forged signatures. Known approaches are for example Support Vector Machines (SVMs) [10, 12, 41], Dynamic Time Warping (DTW) [6, 30], and Hidden Markov Models (HMMs) [10, 13]. In contrast to statistical representation formalisms, structural (i.e. string-, tree-, or graph-based) approaches allow representing the inherent topological characteristics of a handwritten signature in a very natural and comprehensive way [27, 33]. Early structural approaches are, for example, based on the representation of stroke primitives [33], as well as critical contour points [4]. In a recent paper, graphs are used to represent characteristic points (so-called keypoints) as well as their structural relationships [27]. Hence, both the structure and the size of the graph is directly adapted to the size and complexity of the underlying signature. Moreover, these graphs are able to represent relationships that might exist between substructures of the handwriting. However, the power and flexibility of graphs is accompanied by a general increase of the computational complexity of many mathematical procedures. The computation of a similarity or dissimilarity of graphs, for instance, is of much higher complexity than computing a vector (dis)similarity. In order to address this challenge, a number of fast matching algorithms have been proposed in the last decade that allow comparing also larger graphs and/or larger amounts of graphs within reasonable time (e.g. [15, 32]). The goal of the present chapter is twofold. First, the chapter aims to provide a comprehensive literature review of traditional and recent research methods in the field of offline signature verification. Second, the chapter provides a profound analysis of the end-to-end process of signature verification. In particular, the chapter discusses the problems and solutions of data acquisition, preprocessing, feature extraction, and signature classification. Hence, this chapter serves as a literature survey as well as a possible starting point for researchers to build their own process pipe for signature verification. The remainder of this chapter is organized as follows. Section 2 presents a generic signature verification process. Subsequently, the most important processing steps are individually reviewed in the following sections. First, different publicly available offline signature datasets are compared in Sect. 2.1, while common image preprocessing methods are examined in Sect. 2.2. Next, Sect. 2.3 compares different statistical and structural representations that are eventually used for classification (reviewed in Sect. 2.4). Finally, Sect. 3 presents conclusions and possible trends in the field of signature verification. 20 M. Stauffer et al. 2 Signature Verification Process Most signature verification systems are based on four subsequent processing steps, as shown in Fig. 1 [8, 20]. First, handwritten signatures are digitized by means of scanning (in Sect. 2.1 this particular step is discussed in detail). In the second step, signatures are segmented and preprocessed to reduce variations (discussed and reviewed in Sect. 2.2). Next, preprocessed signature images are used to extract statistical (i.e. vectorial) or structural (i.e. string, tree, or graph) representations (discussed and reviewed in Sect. 2.3). Finally, a questioned signature q of claimed user u is compared with all reference signatures r ∈ Ru by means of a specific dissimilarity measure (discussed and reviewed in Sect. 2.4). Roughly speaking, if the minimal dissimilarity to all references is below a certain threshold, the unseen signature q is regarded as genuine, otherwise q is regarded as forged. In general, the larger the reference set R, the better the accuracy of the verification process. However, the acquisition of arbitrarily large reference sets is often not possible in practice, and thus, limited to a few reference signatures per user. In the following sections, each process step is reviewed separately. Data Acquisition Preprocessing Feature Extraction q = {x1 , . . . , xn } Questioned signature q of claimed user u Yes Classification min = 0.758 Below threshold? No r1 = {x1 , . . . , xm } Accept (Genuine) Reject (Forged) Reference signatures r ∈ Ru Data Acquisition Preprocessing Feature Extraction q= Questioned signature q of claimed user u Yes Classification min = 0.623 r1 = Below threshold? No Accept (Genuine) Reject (Forged) Reference signatures r ∈ Ru Fig. 1 Statistical (a) and structural (b) signature verification processes consist of four major steps: First, handwritten signatures are digitized by means of scanning. Second, noise and variations are reduced by means of different preprocessing algorithms. Next, preprocessed and filtered signature images are represented by either a statistical or a structural representation. In either case, a questioned signature q of claimed user u is compared with the corresponding set of reference signatures r ∈ Ru of user u. If the minimal dissimilarity between q and r ∈ Ru is below a certain threshold, q is regarded as genuine, otherwise q is regarded as forged A Survey of State of the Art Methods Employed … 21 2.1 Data Acquisition and Available Datasets During data acquisition, the available signature images are digitized into a machineprocessable format by means of scanning. In most cases, digitized signatures are based on greyscale images, commonly defined as [37] {S(x, y)}0≤x≤X,0≤y≤Y , where S(x, y) denotes the greyscale level at the (x, y)-position of the image and X , Y refer to the maximum values of the x- and y-axis of the image. Typically S(x, y) lies between 0 (= black) and 255 (= white). In the last decades, a number of different private and public datasets for offline signature has been proposed, see [8] for an exhaustive overview. These datasets can be characterized with respect to their writing style (e.g. Arabic, Bengali, Chinese, Persian, Western, etc.) as well as the number of genuine and forged signatures per user. Table 1 summarizes four widely used and publicly available datasets. In particular, the table shows summaries of three datasets with real world signatures (i.e. MCYT75 [13, 29], UTSig [34], and CEDAR [21]) as well as one synthetic dataset (i.e. GPDSsynthetic-offline [11]) for two different writing styles (i.e. Western and Persian). Note that GPDSsynthetic is replacing the widely used GPDS-960 dataset [40] that is no longer available due to new data protection regulations. 2.2 Preprocessing of Signature Images Once handwritten signatures are digitally available, preprocessing basically aims to reduce variations caused by the image itself (e.g. noisy background, skew scanning, segmentation from the background) as well as variations caused by inter- and intrapersonal variations in the handwriting (e.g. scaling and slant of handwriting). In most cases, preprocessing is based on several subsequent steps. First, signatures are extracted from the background (e.g. bank checks). Next, a certain filtering and binarization is employed before different normalization techniques are typically applied. Table 1 The writing style, the number of users, the number of genuine and forged signatures per user, and the resolution of the images in dpi of widely used datasets for offline signature verification Name Style Users Genuine Forgeries dpi MCYT-75 GPDSsyntheticoffline UTSig CEDAR Western Western 75 4000 15 24 15 30 600 600 Persian Western 115 55 27 24 42 25 600 300 22 M. Stauffer et al. (a) Original (b) Segmentation (c) Filtering (d) Binarization (e) Skeletonization Fig. 2 Common image preprocessing steps for GPDSsynthetic-Offline [11]: a bank check with signature,4 b segmented signature image, c filtering by means of Difference of Gaussian, d binarization by means of thresholding, e skeletonization by means of morphological operators Note, however, that the process sequence is not given a priori. The following paragraphs give an overview of some common methods and techniques that are widely accepted as standard preprocessing steps. These steps can be divided into four main phases (see also Fig. 2): 1. 2. 3. 4. Segmentation Filtering Binarization Skeletonization For an exhaustive overview on preprocessing techniques, see [8, 20]. 1. Segmentation As illustrated in Fig. 2a, the signatures are in many cases part of some larger document (e.g. bank checks or forms), and thus, need to be extracted first. Especially bank checks are regarded as a challenging segmentation problem due to colored background images, logos and preprinted lines. To address this issue, different segmentation approaches have been proposed in the literature [2, 12, 13, 33]. In [9], for example, the segmentation is based on subtracting a blank check from the signed artifact. More generic approaches, are based on Speeded Up Robust Features (SURF) features in combination with an Euclidean distance metric [1]. Recently, different Deep Learning techniques based on CNNs have been proposed for this task [22, 34]. In general, the importance of signature segmentation is decreasing due to two reasons: First, the increasing use of blank signature fields and second, the declining importance of bank checks. A Survey of State of the Art Methods Employed … 23 2. Filtering In many cases, scanned handwritten signatures are affected by noisy background, i.e. small pixel fragments that do not belong to the actual signature. The reasons for that can be filthy scanners and papers, smearing as well as non-smooth writing surfaces, to mention just a few. To reduce noise, different filter methods are often employed such as low pass filter [4, 12, 24], median filter [3], morphological filter [2, 6, 13, 16] as well as Difference of Gaussian filter [27]. 3. Binarization To clearly distinguish between foreground and background pixels in an image, scanned signature artifacts are often binarized [6, 27]. That is, each pixel of a greyscale image is either assigned black (i.e. ink of signature) or white (i.e. background). In many cases, binarization leads to better classification results as well as overall lower processing times. A number of different binarization approaches have been proposed for this particular task such as Niblack’s algorithm [4], Otsu’s method [12, 16–18, 23, 24] as well as thresholding [2, 4, 13, 27, 35]. 4. Skeletonization Further image preprocessing approaches are mainly aimed at the reduction of intrapersonal variations in the handwriting. In [2, 3, 13], for example, the handwriting scaling is corrected, while skew and slant are corrected in [3, 23, 24, 30]. Intrapersonal variations can also be reduced by means of skeletonization that is also known as thinning [3, 4, 23, 24, 27]. Note that a number of Deep Learning approaches for signature verification have been proposed recently [8]. In these cases, the neural network is often restricted to a fixed size input, and thus, a certain resizing is necessary [7, 18]. Moreover, the size of training data is often rather small in the case of signature verification which might be a problem in the paradigm of Deep Learning. Hence, data augmentation is often employed in this case to artificially increase the number of available training data [17, 18]. 2.3 Feature Extraction Based on segmented and preprocessed signature images, different types of statistical and structural representations can be extracted. The following two subsections give a brief overview of both representation types. 2.3.1 Statistical Representations of Signatures Statistical representations are extracted from handwritten signature images either on a global or local scale [37]. In the former case (also known as holistic approach), 24 M. Stauffer et al. features are extracted on complete signature images, and thus, represented by a single (typically high-dimensional) feature vector x ∈ Rn . In the case of local representations, subparts of the handwritten signature images are independently represented by local feature vectors. For example, features are extracted by means of a sliding window that moves from left to right over the whole signature image. Then, at each window position a set of different features is extracted. As a result, a linear sequence of m local feature vectors is acquired for every signature image, i.e. x1 , . . . , xm with xi ∈ Rn . In early approaches, features are often used to represent different global handwriting characteristics [37]. For instance, features that are based on the direction of the contour, as shown in Fig. 3a [6, 16]. Similarly, the handwriting outline has also been used as representation formalism in [10]. Moreover, the direction of slant, i.e. the inclination of the handwriting, has been found to be useful [2, 13]. Last but not least, projection profiles, i.e. the number of foreground pixels in the horizontal or vertical direction, have been employed as global features in [30]. In contrast, features can also be extracted from a local perspective on the signature [37]. For example, different morphological features have been proposed in [23]. In case of Gradient, Structural and Concavity (GSC) features, three types of features are extracted by means of a grid-wise segmentation and concatenated to form a feature vector [35]. Global features (e.g. the signature height, projection profiles, slant angle, etc.) can also be combined with different local features (e.g. grid and texture features), as shown in [3]. Further local features are often based on more generic texture descriptors, i.e. features that can be used to represent arbitrary patterns [37]. In [41], for example, HoG have been employed, as illustrated in Fig. 3b. That is, the direction of gradients is first extracted at different local cell positions and then concatenated to form a single histogram. Likewise, LBP are used to describe circular structures of an image at different neighborhood levels [12, 41], while so-called surroundedness features are extracted in [24]. Finally, it is worth mentioning that a number of Deep Learning approaches for feature extraction have been proposed in the last years [18, 26, 36]. That is, rather than extracting handcrafted features by means of a certain predefined procedure, this type of feature representation is learned in an unsupervised manner by means of a CNN. In particular, different CNN architectures have been proposed for this task such as spatial pyramid pooling [17], triplet network [26] as well as siamese network [7]. 2.3.2 Structural Representations of Signatures Structural representations are based on powerful data structures such as strings, trees, or graphs [4, 27, 33, 37]. In general, a graph is defined as a four-tuple g = (V, E, μ, ν) where V and E are finite sets of nodes and edges, μ : V → L V and ν : E → L E are labeling functions for nodes and edges, respectively [31]. In general, the graph size is not fixed and thus graphs are—in contrast to global fea- A Survey of State of the Art Methods Employed … 25 (a) Contour (b) HoG (c) Keypoint Graph (d) Inkball Model Fig. 3 Visualization of contour features (see Fig. 3a), HoG features (see Fig. 3b), keypoint graph (see Fig. 3c), and inkball model (see Fig. 3d) of one signature from GPDSsyntheticOffline [11] ture vectors—flexible enough to adapt their size to the size and complexity of the underlying handwritten signature image. Moreover by means of edges, graphs are capable of representing binary relationships that might exist in different subparts of the underlying signature. A vast amount of signature verification approaches still makes use of statistical rather than structural approaches [8, 19, 37]. This is rather surprising as the inherent representational properties of structural formalisms would be well-suited to represent handwriting. An early structural signature verification approach is based on the representation of signatures by means of stroke primitives [33]. Another approach is based on the detection of characteristics points—so-called keypoints—on contour images [4]. Following this line of research, a recent approach is based on the detection of keypoints in skeletonized signature images, as illustrated in Fig. 3c [27, 37]. In this particular scenario, nodes are used to represent keypoints (e.g. start, end, and intersection points of the handwritten strokes), while edges are used to represent strokes between these keypoints. Other graph-like representations, that have become more popular in recent years are, for example, inkball models, as illustrated in Fig. 3d [25]. Similar to keypoint graphs, nodes in inkball models are based on characteristic points of the skeleton (i.e. junction and end points). Note, however, that inkball models are in contrast with keypoint graphs based on rooted trees rather than general graphs. 2.4 Classification of Signatures Finally (and regardless the actual representation formalism used), the signature verification is based on a classification task. That is, a questioned signature q of claimed user u is compared with all reference signatures r ∈ Ru of registered user u (see 26 M. Stauffer et al. Fig. 1). However, the actual method that determines whether q is regarded as genuine or forged depends on the chosen representation formalism, i.e. whether signatures are represented by means of a statistical or structural formalism. The following two subsections give an overview of both approaches. 2.4.1 Statistical Approaches for Signature Verification In case of sequences of feature vectors (i.e. the signature is represented by x1 , . . . , xm ), the classification task is often based on a dynamic programming approaches such as, for example, DTW [6, 30]. That is, two sequences of feature vectors (i.e. one sequence for q and one sequence for r ) are optimally aligned along one common time axis. The minimal sum of alignment costs can then be used as dissimilarity measure between q and r . In many cases, however, signatures are represented by holistic and often high-dimensional feature vectors (i.e. the signature is represented by x ∈ Rn ). Consequently, feature vectors are compared by means of different distance measures like for example Euclidean distance [7, 10, 26], χ 2 distance [12, 16], or Mahalanobis distance [2, 13]. In contrast to these so-called learning-free methods, a number of learning-based approaches have been proposed as well [12, 13, 17]. That is, a statistical model is trained a priori on a (relatively large) training set of reference signatures. Different statistical models have been employed for this task for example SVMs [10, 12, 17, 18, 23, 24, 41], HMMs [10, 13] as well as Neural Networks [3, 23, 24]. Generally, learning-based approaches lead to lower error rates when compared with learning-free methods. However, this advantage is accompanied by a loss of flexibility and generalizability, which is due to the need to learn a statistical model on some training data. In fact, the accuracy of many learning-based approaches is crucially depending on the size and quality of the labelled training data. However, in case of signature verification the acquisition of training data is often expensive and/or otherwise limited. 2.4.2 Structural Approaches for Signature Verification Structural representations, in particular graph-based representations, offer some inherent representational advantages as discussed before. However, this advantage is accompanied with an increased complexity with respect to basic dissimilarity measures. In fact, the exact computation of a graph dissimilarity measure on general graphs is known to be NP-complete, and thus several fast approximations for this task have been proposed in the last decades [31]. In an early work, stroke primitives are first merged locally and then compared by means of a global dissimilarity measure [33]. Moreover, keypoints of contour images are matched by means of a point matching algorithm in [4]. However, the actual structural relationships of the handwriting are not considered during these first structural matching procedures. A Survey of State of the Art Methods Employed … 27 More recent structural signature verification approaches make use of fast approximations for Graph Edit Distance (GED) [15, 32]. GED measures the minimum amount of distortion (i.e. insertion, deletion, substitution of both nodes and edges) needed to transform one graph into another. In the case of tree-based inkball models, the classification is based on an energy function that measure the amount of deformation, similar to GED [25]. Recently, different approaches have been proposed that allow the combination of statistical and structural classification approaches [26, 38, 39]. In [38], for example, subgraphs are first matched by means of graph matching and then optimally aligned by means of DTW. Another approach is based on an ensemble method that allows the combination of a Deep Learning approach (i.e. triplet network) with a fast GED approximation [26]. Finally, prototype graph embedding allows to embed graphs into a vectorial representation [39]. This approach combines the representational power of graphs and the mathematically well-founded vector space for dissimilarity computation. 3 Conclusion and Outlook Handwritten signatures have been an important verification measure for many business and legal transactions for several centuries [37]. The popularity of handwritten signatures remains high, even though several international initiatives for electronic signature have been announced. That is, handwritten signatures offer some inherent advantages such as ubiquitous applicability as well as the overall high acceptance and availability. However, handwritten signatures expose also a certain risk of misuse. For this reason, automatic signature verification has been proposed [8, 19, 20]. Most signature verification frameworks are based on four processing steps. First, signatures are digitized by scanning. Second, scanned handwritten signature images are typically preprocessed in order to reduce variations caused by both the artifact (e.g. noisy background, skew scanning) and the handwriting (e.g. scaling of the handwriting). Next, preprocessed signature images are represented either by a statistical (i.e. vectorial) or structural (e.g. graph-based) formalism. Finally, a questioned signature is compared with a set of reference signatures in order to distinguish between genuine and forged signature. The present chapter provides a comprehensive overview of the state of the art in signature verification based on offline data. This chapter reviews in particular traditional and very recent methods used in the various steps in a generic process of offline signature verification. This analysis can be used either as a literature review or as a starting point for developing an own signature verification engine. In general, structural approaches offer some inherent representational advantages when compared with statistical formalisms. That is, graphs are able to adapt both their size and structure to the underlying pattern. Moreover, graphs are able to represent a binary relationship that could exist between subparts of the handwriting. However, most signature verification approaches still make use of a statistical rather 28 M. Stauffer et al. than structural representation. This is mainly due to the fact that algorithms and methods that need graphs as input have a substantially greater complexity than their statistical counterparts. Yet, in recent years, several powerful and efficient methods for graph processing have been proposed [31]. These methods enable the use of graphs in the field of signature verification. In the future, great potential for further structural signature verification approaches can be expected, especially with the rise of Deep Learning methods for graphs, the so-called Geometric Deep Learning [5]. Acknowledgements This work has been supported by the Swiss National Science Foundation project 200021_162852. References 1. Ahmed, S., Malik, M.I., Liwicki, M., Dengel, A.: Signature segmentation from document images. In: 2012 International Conference on Frontiers in Handwriting Recognition, pp. 425– 429 (2012) 2. Alonso-Fernandez, F., Fairhurst, M., Fierrez, J., Ortega-Garcia, J.: Automatic measures for predicting performance in off-line signature. In: IEEE International Conference on Image Processing, pp. I–369–I–372. IEEE (2007) 3. Baltzakis, H., Papamarkos, N.: A new signature verification technique based on a two-stage neural network classifier. Eng. Appl. Artif. Intell. 14(1), 95–103 (2001) 4. Bansal, A., Nemmikanti, P., Kumar, P.: Offline signature verification using critical region matching. In: International Conference on Future Generation Communication and Networking Symposia, pp. 115–120. IEEE (2008) 5. Bronstein, M.M., Bruna, J., Lecun, Y., Szlam, A., Vandergheynst, P.: Geometric deep learning: going beyond euclidean data. Signal Process. Mag. 34(4), 18–42 (2017) 6. Deng, P.S., Liao, H.Y.M., Ho, C.W., Tyan, H.R.: Wavelet-based off-line handwritten signature verification. Comput. Vis. Image Underst. 76(3), 173–190 (1999) 7. Dey, S., Dutta, A., Toledo, J.I., Ghosh, S.K., Llados, J., Pal, U.: SigNet: convolutional siamese network for writer independent offline signature verification (2017) 8. Diaz, M., Ferrer, M.A., Impedovo, D., Malik, M.I., Pirlo, G., Plamondon, R.: A perspective analysis of handwritten signature technology. ACM Comput. Surv. 51(6), 1–39 (2019) 9. Djeziri, S., Nouboud, F., Plamondon, R.: Extraction of items from checks. Proc. Fourth Int. Conf. Doc. Anal. Recognit. 2, 10–13 (1997) 10. Ferrer, M.A., Alonso, J., Travieso, C.: Offline geometric parameters for automatic signature verification using fixed-point arithmetic. IEEE Trans. Pattern Anal. Mach. Intell. 27(6), 993– 997 (2005) 11. Ferrer, M.A., Diaz-Cabrera, M., Morales, A.: Static signature synthesis: a neuromotor inspired approach for biometrics. Trans. Pattern Anal. Mach. Intell. 37(3), 667–680 (2015) 12. Ferrer, M.A., Vargas, J.F., Morales, A., Ordonez, A.: Robustness of offline signature verification based on gray level features. IEEE Trans. Inf. Forensics Secur. 7(3), 966–977 (2012) 13. Fierrez-Aguilar, J., Alonso-Hermira, N., Moreno-Marquez, G., Ortega-Garcia, J.: An off-line signature verification system based on fusion of local and global information. In: International Workshop on Biometric Authentication, pp. 295–306. Springer, Berlin (2004) 14. Fillingham, D.: A Comparison of Digital and Handwritten Signatures. Technical report, Massachusetts Institute of Technology (1997) 15. Fischer, A., Suen, C.Y., Frinken, V., Riesen, K., Bunke, H.: Approximation of graph edit distance based on Hausdorff matching. Pattern Recognit. 48(2), 331–343 (2015) A Survey of State of the Art Methods Employed … 29 16. Gilperez, A., Alonso-Fernandez, F., Pecharroman, S., Fierrez, J., Ortega-Garcia, J.: Off-line signature verification using contour features. In: International Conference on Frontiers in Handwriting Recognition. Concordia University (2008) 17. Hafemann, L.G., Oliveira, L.S., Sabourin, R.: Fixed-sized representation learning from offline handwritten signatures of different sizes. Int. J. Doc. Anal. Recognit. 21(3), 219–232 (2018) 18. Hafemann, L.G., Sabourin, R., Oliveira, L.S.: Learning features for offline handwritten signature verification using deep convolutional neural networks. Pattern Recognit. 70, 163–176 (2017) 19. Hafemann, L.G., Sabourin, R., Oliveira, L.S.: Offline handwritten signature verification— literature review. In: International Conference on Image Processing Theory, Tools and Applications, pp. 1–8. IEEE (2017) 20. Impedovo, D., Pirlo, G.: Automatic signature verification: the state of the art. Appl. Rev., IEEE Trans. Syst., Man Cybern. Part C (2008) 21. Kalera, M.K., Sargur, S., Xu, A.: Offline signature verification and identification using distance statistics. Int. J. Pattern Recognit. Artif. Intell. 18(07), 1339–1360 (2004) 22. Kleber Santos Leite Melo, V., Byron Leite Dantas, B.: A Fully convolutional network for signature segmentation from document images. In: International Conference on Frontiers in Handwriting Recognition, pp. 540–545. IEEE (2018) 23. Kumar, R., Kundu, L., Chanda, B., Sharma, J.D.: A writer-independent off-line signature verification system based on signature morphology. In: First International Conference on Intelligent Interactive Technologies and Multimedia, pp. 261–265. ACM Press (2010) 24. Kumar, R., Sharma, J., Chanda, B.: Writer-independent off-line signature verification using surroundedness feature. Pattern Recognit. Lett. 33(3), 301–308 (2012) 25. Maergner, P., Howe, N., Riesen, K., Ingold, R., Fischer, A.: Offline signature verification via structural methods: graph edit distance and inkball models. In: International Conference on Frontiers in Handwriting Recognition, pp. 163–168. IEEE (2018) 26. Maergner, P., Pondenkandath, V., Alberti, M., Liwicki, M., Riesen, K., Ingold, R., Fischer, A.: Offline signature verification by combining graph edit distance and triplet networks. International Workshop on Structural. Syntactic, and Statistical Pattern Recognition, pp. 470–480. Springer, Berlin (2018) 27. Maergner, P., Riesen, K., Ingold, R., Fischer, A.: A structural approach to offline signature verification using graph edit distance. In: International Conference on Document Analysis and Recognition, pp. 1216–1222. IEEE (2017) 28. Nagel, R., Rosenfeld, A.: Computer detection of freehand forgeries. IEEE Trans. Comput. C-26(9), 895–905 (1977) 29. Ortega-Garcia, J., Fierrez-Aguilar, J., Simon, D., Gonzalez, J., Faundez-Zanuy, M., Espinosa, V., Satue, A., Hernaez, I., Igarza, J.J., Vivaracho, C., Escudero, D., Moro, Q.I.: MCYT baseline corpus: a bimodal biometric database. Vis., Image, Signal Process. 150(6), 395 (2003) 30. Piyush Shanker, A., Rajagopalan, A.: Off-line signature verification using DTW. Pattern Recognit. Lett. 28(12), 1407–1414 (2007) 31. Riesen, K.: Structural pattern recognition with graph edit distance. In: Advances in Computer Vision and Pattern Recognition. Springer, Berlin (2015) 32. Riesen, K., Bunke, H.: Approximate graph edit distance computation by means of bipartite graph matching. Image Vis. Comput. 27(7), 950–959 (2009) 33. Sabourin, R., Beaumier, L.: Structural interpretation of handwritten signature images (1994) 34. Sharma, N., Mandal, R., Sharma, R., Pal, U., Blumenstein, M.: Signature and logo detection using deep CNN for document image retrieval. In: International Conference on Frontiers in Handwriting Recognition, pp. 416–422. IEEE (2018) 35. Siyuan Chen, Srihari, S.: A new off-line signature verification method based on graph. In: International Conference on Pattern Recognition, pp. 869–872. IEEE (2006) 36. Soleimani, A., Araabi, B.N., Fouladi, K.: Deep multitask metric learning for offline signature verification. Pattern Recognit. Lett. 80, 84–90 (2016) 37. Stauffer, M.: From signatures to graphs. Master’s thesis, University of Applied Sciences and Arts Northwestern Switzerland (2015) 30 M. Stauffer et al. 38. Stauffer, M., Maergner, P., Fischer, A., Ingold, R., Riesen, K.: Off-line signature verification using structural dynamic time warping. In: International Conference on Document Analysis and Recognition (2019) 39. Stauffer, M., Maergner, P., Fischer, A., Riesen, K.: Graph embedding for offline handwritten signature verification. In: International Conference on Biometric Engineering and Applications (2019) 40. Vargas, F., Ferrer, M., Travieso, C., Alonso, J.: Off-line handwritten signature GPDS-960 corpus. In: International Conference on Document Analysis and Recognition, pp. 764–768. IEEE (2007) 41. Yilmaz, M.B., Yanikoglu, B., Tirkaz, C., Kholmatov, A.: Offline signature verification using classifier combination of HOG and LBP features. In: International Joint Conference on Biometrics, pp. 1–7. IEEE (2011) Agile Visualization in Design Thinking Emanuele Laurenzi , Knut Hinkelmann , Devid Montecchiari , and Mini Goel Abstract This chapter presents an agile visualization approach that supports one of the most widespread innovation processes: Design Thinking. The approach integrates the pre-defined graphical elements of SAP Scenes to sketch digital scenes for storyboards. Unforeseen scenarios can be created by accommodating new graphical elements and related domain-specific aspects on-the-fly. This fosters problem understanding and ideation, which otherwise would be hindered by the lack of elements. The symbolic artificial intelligence (AI)-based approach ensures the machineinterpretability of the sketched scenes. In turn, the plausibility check of the scenes is automated to help designers creating meaningful storyboards. The plausibility check includes the use of a domain ontology, which is supplied with semantic constraints. The approach is implemented in the prototype AOAME4Scenes, which is used for evaluation. Keywords Symbolic AI · Design thinking · Agile and ontology-aided meta-modeling · AOAME4Scenes · Innovation processes E. Laurenzi (B) · K. Hinkelmann · D. Montecchiari · M. Goel Intelligent Information Systems Research Group, FHNW University of Applied Sciences and Arts Northwestern Switzerland, Riggenbachstrasse 16, Windisch, Switzerland e-mail: emanuele.laurenzi@fhnw.ch K. Hinkelmann e-mail: knut.hinkelmann@fhnw.ch D. Montecchiari e-mail: devid.montecchiari@fhnw.ch M. Goel e-mail: mini.goel@students.fhnw.ch E. Laurenzi · K. Hinkelmann Department of Informatics, University of Pretoria, Lynnwood Rd, Pretoria 0083, South Africa D. Montecchiari University of Camerino, Camerino, Italy © Springer Nature Switzerland AG 2021 R. Dornberger (ed.), New Trends in Business Information Systems and Technology, Studies in Systems, Decision and Control 294, https://doi.org/10.1007/978-3-030-48332-6_3 31 32 E. Laurenzi et al. 1 Introduction Visual sketches are an efficient and effective way to communicate ideas [2, 33]. They empower designers to explore the creative design space [36]. Sketching is a traditional and conventional technique used to draw new ideas. Its usage has seen a massive growth after the Renaissance as designed objects have become more complex and more novel [3]. A pioneer on this subject was Leonardo da Vinci. Many of his sketches expressed the ability to communicate with someone else about (1) which artifact should be built, (2) how it should be built, (3) how it should work, as well as (4) the use of thinking and reasoning to support the solution design. Sketches are not limited to the use as a means of communication but can also assist the cognitive process of conceiving a new solution design. These characteristics make sketching a prominent technique in todays’ most widespread innovation processes [45]. In the Google Sprint [14], a whole day is dedicated to the sketch of possible solutions. In Design Thinking [28] sketches are used from an early phase to increase the understanding of the problem, and later to ideate the new solution designs [13, 29]. There are already several approaches in which the sketches are drawn with free hands, e.g. Crazy 8s [23, 43], Collaborative Sketching [8], Sketch Storming [38], CollabSketch [25], SketchStorm [19]. However, freehand sketching can be problematic for the smooth progress of innovation processes. Some people are not capable, hesitant, or even refuse to draw sketches, while others feel uncomfortable when being asked to draw [6]. De Vere et al. [40] associate this discomfort with the lack of wisdom in expressing ideas with sketches. Another problem is the comprehensibility of the drawn sketch. Since ideas are mainly sketched on tangible means (e.g., sticky notes), Hohmann [11] experienced that when any changes are made to sketches, they tend to become messy, and thus incomprehensible. Such problems motivated the emergence of approaches that use a predefined set of elements such as SAP Scenes [31]. SAP Scenes consists of a collection of predefined graphical elements, the combination of which allows you to sketch storyboards about products or services. Predefined graphical elements help users create visual stories quickly and collaboratively. Therefore, drawing skills are not necessary, and users can focus directly on shaping ideas and scenarios in the form of illustrative storyboards. However, the practice of sketching storyboards with predefined graphical elements has its limitations (see Sect. 2). Such limitations hinder innovation processes, which is also why digital tools are difficult to use in practice. In this chapter, we aim to show how a new agile and ontology-aided meta- modeling approach can overcome these limitations in order to promote the understanding of problems and ideation in innovation processes. The approach has already been validated in other application domains [15, 17]. The rest of the chapter is structured as follows: Sect. 3 discusses the adopted methodology. Section 4 introduces the state of the art of Design Thinking and visualization tools that allow the creation of sketches. In Sect. 5 , we set the requirements Agile Visualization in Design Thinking 33 for the proposed artifact for the agile and ontology-aided modeling approach. The approach is then introduced in Sect. 6. The evaluation of the approach is described in Sect. 7. Finally, the conclusion and future work are discussed in Sect. 8. 2 Limitations that Hinder the Innovation Processes and Related Work The first limitation is about the inability of the pre-defined elements to model any underlying reality. SAP Scenes provides “Scenes add-ons” [31] with some additional topic- and industry-specific elements, but they are not sufficient to cover all application scenarios. A real-world example for such limitation is the Air as a Service storyboard,1 which was realized in the context of the EU-funded project DIGITRANS [7]. Figure 1 shows a part of the final storyboard. The storyboard supported the transformation of the business model of a German company from the production of ventilation systems to the innovative concept of Air as a Service. The new value proposition of the business model becomes a continuous supply of clean air. In the Air as a Service use case, ventilation of large buildings like parking houses were identified as possible application scenarios. However, SAP Scenes did not foresee ventilators or parking lots. The newly needed graphical elements had to be physically created (see Fig. 1). For physical SAP Scenes, this limitation can be easily overcome by manually drawing new elements. This is different, when the Scenes are modeled with a digital tool. Scene2Model [21] contains the digital representation of SAP Scenes with which users can design digital storyboards. The tool is based on the meta-modeling tool ADOxx [27]. Adding new modeling elements follows the current sequential process of meta-modeling, which reverts the waterfall-like cycle [15] (1) adapt the metamodel; (2) deploy the adapted meta-model in a new tool instantiation; and finally (3) test the new version of the modeling language. Workshops with innovation experts (see Sect. 5) have confirmed that the sequential engineering process can lead to delays in innovation processes. An agile metamodeling approach for the quick integration and adaptation of new elements would promote the adoption of digital tools among Design Thinking practitioners. Another limitation that hinders the practice of Design Thinking is the lack of support from digital solutions in the elaboration of storyboards, especially in the ideation phase. In this phase, a set of storyboards are elaborated and should be available to provide meaningful input to the next phase [1], i.e., the prototype phase. There exist digital tools (e.g. [34, 35]) for online collaboration between participants to work on an idea. However, their use assumes a high level of expertise in innovation processes. In cases where designers (especially the less experienced ones) have a low domain expertise, relevant aspects in the storyboards may be overlooked. A possible 1 The storyboard was realized in the innovation room of the Herman Hollerith Zentrum (HHZ), University of Reutlingen. 34 E. Laurenzi et al. Fig. 1 Air-as-a-Service storyboard created with an extension of SAP scenes consequence would be the numerous back and forth loops between the ideation, the prototype and the test phases until a new solution design is refined enough for the Design Thinking process to end [5]. In the worst case, the new solution design will lack realistic foundations that need to be taken into account. This lack of support can best be addressed by an approach that allows automation of the knowledge stored by SAP Scenes. The objective, in this case, would be to automatically support the creation of plausible and meaningful storyboards. The automation of knowledge can be achieved with the use of symbolic artificial intelligence, i.e., ontologies. The latter have already been used in storyboards of Scene2Model [21, 22] with the purpose of increasing information transparency and clarity of the meaning of storyboards. In order to address the identified need for agility and automation to support the creation of storyboards, we propose to adopt the agile and ontology-aided metamodeling approach already conceived by Laurenzi et al. [17]. The approach allows for the integration of new modeling elements and the adaptation of the existing ones on the fly, as well as the seamless alignment of graphical elements to ontologies. 3 Methodology In this work, the chosen methodology is the Design Science Research (DSR) [10, 39], which aims to build an artifact by going through the following phases: awareness of problem, suggestion, development, and evaluation of the artifact and conclusion. In order to increase the awareness of problem, insights were derived from the following three different research activities: • Research Activity 1 (RA1): Analyzing theories and existing approaches relevant to this research work (reported in Sects. 2 and 4). • Research Activity 2 (RA2): Modeling a use case by adopting a modeling approach, which makes use of SAP Scenes to create digital storyboards in innovation pro- Agile Visualization in Design Thinking 35 cesses (see Sect. 5). During this activity, we confirmed our hypothesis about the need to deal with unforeseen scenarios and derived a list of requirements. • Research Activity 3 (RA3): Two workshops were held with innovation experts (see Sect. 5) to increase understanding of whether and how the agile and ontology-aided modeling approach can support an innovation process. • Research Activity 4 (RA4): Next, the conceptualized artifact was implemented in a variation of AOAME [17], which we called AOAME4Scenes. It solves the research problem and satisfies the requirements that were derived in the first three research activities. The research activities cover the Design Science Research Cycle as proposed by Hevner [10]: RA4 corresponds to the design of the artifact, the rigor cycle (RA1) provides theoretical foundation to the designed artifact and the relevance cycle (RA2 and RA3) ensures the appropriateness of the application context. 4 State of the Art This section first explains what Design Thinking consists of. It then describes existing visualization tools that support innovation processes by sketching. 4.1 Design Thinking Design thinking is an iterative process tha...

Critically analyze the paper.

Which one of the text mining scenarios is more of a challenge to business?

1. Anonymization
2. Quality of statistical machine learning models
3. Being on the top of new methods or open-source software
4. etc.

Provide your reasoning and rationale behind your choice.

Professor's Guide:

As you read and assimilate this new trend of managing part of a business, take a close look at the required automation in the extraction of structured information from pdf files. Try to critically analyze the reasoning behind the standard path to innovation in such projects, including business process modeling and the need to implement novel technological solutions.

NOTE:
1. Post your 500-700 word answers
2. Offer at least two 300-400 word comments (replies) to posts from your peers’ discussions