Mining Time-Sensitive Queries and their Sensitivities from Query Log

In this paper, we present a new concept called time-sensitive query. A query is time-sensitive if the set of relevant pages changes when the query is expanded with different temporal terms. Knowing which queries are time-sensitive and what sensitivities they have can be very useful in automatic query expansion for improving search effectiveness and user experience. Existing methods related to time-sensitive query detection are more sudden event oriented and have limitations in detecting other kinds of time-sensitive queries and in determining the time sensitivity. We propose an approach to detecting time-sensitive queries and determining their time sensitivities. Our experiment shows that the approach works well. The accuracy of time-sensitive query detection can reach 92% and the accuracy of time sensitivity calculation can reach 93.12%.