Implementing a recommender system for Bangladeshi faculty search using machine learning

Abstract

Machine learning is a one of the popular fields in Computer Science. In my thesis research the focus is to implement a recommender system for Bangladeshi faculty search. Selecting a appropriate faculty or thesis supervisor is a very important part in a student’s life. Even choosing right academy is also an important part in their study life. This research paper presents a faculty recommender system to assist students in making these choices. Here the main focus is to cover our own country, Bangladesh, to help the students of our country to pursue their own interest. I proposed this recommender system by using collaborative filtering algorithm. I used a very popular machine learning algorithm, K-Nearest Neighbor algorithm with cosine similarity to predict faculty members. It works on a vast database and being analyzed by different criteria. It applies multiple filtering conditions to retrieve relevant supervisor or faculty member based on the research interest or preferences. The preference field of the faculties based on preferred research area, making part of the decision specific. This system helps a user finding faculty or supervisor according to own individual interests. It contains information about faculties around Bangladesh from different institutions. A classification accuracy of 76.0 % for the predicted results ac hived by the proposed model.