Exploring Combinations of Sources for Interaction Features for Document Re-ranking
Emanuelle Di Buccio, Massimo Melucci, Dawei Song · Open Research Online (The Open University) · 2010
The behavior of the user when interacting with a result page or the corresponding landing documents is a possible source of evidence that Information Retrieval (IR) systems can ex- ploit to assist the user when searching for information. In- teraction features can be adopted as evidence to model the user behavior, thus making it usable to assist relevance pre- diction. One issue when dealing with interaction features is the selection of the sources from which these features are distilled. Individual users and group of users which perform a similar task or look for information matching the same query are possible sources. This paper will focus on these two sources, particularly investigating group of users searching for the same topic as source for interaction features to be used as an alternative to, or in combination with, indi- vidual users. The objective of this work is to investigate the impact of diverse combinations of these sources on the retrieval effectiveness, specifically when interaction features are used as evidence to support document re-ranking.