Multi-Criteria Decision Support System for Ph.D. Supervisor Selection

Selection of a suitable Ph.D. supervisor is a very important step in a student's career. This dissertation presents a multi-criteria decision support system to assist students in making this choice. The system employs a hybrid method that first utilizes a fuzzy analytic hierarchy process to extract the relative importance of the identified criteria and sub-criteria to consider when selecting a supervisor. Then, it applies an information retrieval-based similarity algorithm (TF/IDF or Okapi BM25) to retrieve relevant candidate supervisor profiles based on the student's research interest. The selected profiles are then re-ranked based on other relevant factors chosen by the user, such as publication record, research grant record, and collaboration record. The ranking method evaluates the potential supervisors objectively based on various metrics that are defined in terms of detailed domain-specific knowledge, automating part of the decision making process. In contrast with other existing works, this system does not require the professor's involvement and no subjective measures are employed.