Karthik Raman, Paul N. Bennett, and Kevyn Collins-Thompson
Current research on web search has focused on optimizing and evaluating single queries. However, a significant fraction of user queries are part of more complex tasks which span multiple queries across one or more search sessions. An ideal search engine would not only retrieve relevant results for a user's particular query but also be able to identify when the user is engaged in a more complex task and aid the user in completing that task. Toward optimizing whole-session or task relevance, we characterize and address the problem of intrinsic diversity (ID) in retrieval, a type of complex task that requires multiple interactions with current search engines. Unlike existing work on extrinsic diversity that deals with ambiguity in intent across multiple users, ID queries often have little ambiguity in intent but seek content covering a variety of aspects on a shared theme. In such scenarios, the underlying needs are typically exploratory, comparative, or breadth-oriented in nature. We identify and address three key problems for ID retrieval: identifying authentic examples of ID tasks from post-hoc analysis of behavioral signals in search logs; learning to identify initiator queries that mark the start of an ID search task; and given an initiator query, predicting which content to prefetch and rank.
In Proceedings of the 36th Annual ACM SIGIR Conference (SIGIR 2013)