Skip to content Skip to navigation
Qualifying Exam
9/26/2016 03:30 pm
CoRE B (305)

Towards a More Generative Model of Query Recommendation

Matthew Mitsui, Dept. of Computer Science

Examination Committee: Prof. Chirag Shah (Chair), Prof. Amélie Marian, Prof. Pranjal Awasthi, Prof. Kostas Bekris

Abstract

Loosely stated, the goal of query recommendation is to recommend queries that obtain desired information from a search engine, based on the searcher's intentions.  Yet largely overlooked are direct solutions to the following problems: 1) Recommending "information" and 2) Discovering a searcher's "intentions".  We show how previous solutions in query recommendation and intention modeling fall short.  We outline our work towards a more general query recommender, taking a generative approach that contrasts with discriminative approaches.  We draw relationships between query logs and intentions and argue for unified recommendation approach incorporating a context of intentions and a context of information.