Investigating Multiple periodicities in User’s Web Search Behaviour to Improve Personalization of Information Retrieval Systems

Web users generally get irrelevant results when they submit an ambiguous and/or under-specified query. Search engines are not able to distinguish the different intents associated with a query. The results provided are generally biased towards the popular intent. Every user has a distinct context associated with a query. The context of a user can be inferred from his/her past searches. If that context is considered, the query can be disambiguated.

This study aims to explore time-sensitive search patterns in a user’s search behavior, using the query logs. The time-sensitive search patterns would help in revealing the various topics of interest of a user and also show how the dominance of these interests depends on time. These patterns will be leveraged
to disambiguate the context of a user query according to the time
at which the search is made and thereby improve the relevance of
ranked results.