The Pitt Grapevine: An Advisor-in-the-Loop Academic Recommender System

Abstract

Grapevine is a recommender system for helping Pitt students learn about and connect with the vast field of people, experiences, and programs that are available at Pitt. It is data-driven and intended to improve graduates’ satisfaction with the degrees and programs they went into, to increase student involvement in research and projects, increase students’ cross-disciplinary exploration, and to create new knowledge by supporting students’ and faculties’ network-building and exploration. Grapevine asks users to “Tell us about you,” and users simply enter a text description —from a paragraph about themselves to a list of topics. Grapevine then provides a list of faculty persons and the interests that overlap with the student’s and seem to be a good match. Preliminary research demonstrated its feasibility for representing undergraduate students and the faculty. We also found that while student users may not know the best ways to describe and interpret academic advice, advisors do and are eager to use analytic information about people and projects. Thus our proposal is to build out Grapevine as a hybrid recommender system in which the recommendations come from two critical sources: (1) Grapevine’s analyses of Pitt data and (2) advisors’ professional knowledge and social networks. The system will target both student and educator/advisor populations with different functions and interfaces. With the advisor, or at some point the educator, in-the-loop, student privacy will be maintained on the advisor side. The system will give educators new conversational support during advising sessions, the powerful analytics about social “knowledge” networks, and add their insights and expertise to the system’s database of recommendations.