Modelling Users Search and Browsing Behaviour for Relevance

PhD Qualifying Examination


Title: "Modelling Users Search and Browsing Behaviour for Relevance"

by

Mr. Weizhu Chen


Abstract:

When a user is seeking information in Internet, most of her behaviour will
be recorded into click-through logs of a search engine. Analyses of user
behaviour logs can benefit many Internet applications, such as search
relevance learning, Ads click through rate prediction, and user
satisfaction estimation. However, the information encoded in the log data
is implicit which pose challenges to uncover its characteristic. Many
research works have proposed to use generative models to better interpret
user behaviour data and learn a user-perceived relevance, and one of the
major techniques is the \emph{click model}, which is a recently developed
attractive technique for estimating an unbiased user-perceived relevance
for query-document pairs. In this survey, I will start with the recent
advances in user behaviour analysis in both organic search and sponsor
advertisement. I will then explore the state-of-the-art research results
in interpreting user click behaviour with generative models and how
researchers have applied this technique for search $\&$ Ads relevance.
Finally, I will provide a study of user behaviour beyond search engine,
such as incorporating user browsing behaviour in popular websites to
improve relevance, and then present the findings of some of my recent
works.


Date:                   Thursday, 26 May 2011

Time:                   2:00pm - 4:00pm

Venue:                  Room 1504
                         lifts 25/26

Committee Members:	Prof. Qiang Yang (Supervisor)
                         Prof. Dit-Yan Yeung (Chairperson)
 			Dr. Lei Chen
 			Prof. Dik-Lun Lee


**** ALL are Welcome ****