This is a machine learning subject question

This is a machine learning subject question.For this problem, imagine that you have graduated and are working on an email product for a new start-up company. You have been asked to develop a Spam Filter, i.e., a tool that will identify incoming email as unwanted and not personally relevant to the recipient, i.e., Spam. You have been provided with a large number of emails that have been classified by users as Spam, as well as a large number of emails that have not been marked as Spam, so you have labelled data of both positive and negative examples of Spam.

1. (a) Name the machine learning concept, and (b) describe the basic steps you would follow to approach this task. Be sure to describe the full process beginning with the data all the way to a working model. You do not need to specify the details of data processing or feature engineering, the focus should be on the high-level goal of building a working machine learning model. 
2. After a few weeks your boss wants to know how well your system works. He is a critical person, and asks for a convincing justification for your evaluation approach. Convince your boss of the optimal evaluation strategy by comparing it with a less promising strategy -- i.e., compare and contrast two evaluation strategies. 
3. Assume that so far you used a set of indicator features where each feature corresponds to a word type in your training data -- i.e., you have a vast feature set comprising many English words. (I) Define the concept of "Feature Selection" and (II) describe two approaches for feature selection in the context of this task and your feature set. (III) Do you think Feature Selection will be beneficial for this application? Justify your answer.