Employing open-web for Contextual Suggestion using tag-tag similarity

Abstract

The TREC 2016 Contextual Suggestion task aims at providing recommendations on points of attraction for different kind of users and a varying context. Our group DPLAB IITBHU tries to recommend relevant point-of-interests to a user based on the information provided on the candidate attractions and her past preferences. We employ open-web information in a novel way to capture the best possible setting for a user’s tastes and distastes in terms of tag scores. The scores are then ranked using a heuristic to suggest the most pertinent attraction to the user. One of our methods exceed the TREC-CS 2016 median of the standard evaluation scores of all participant runs, which reflects the effectiveness of our approach. © 2016 25th Text REtrieval Conference, TREC 2016 - Proceedings. All Rights Reserved.