- Query auto completion in information retrieval
- Award date
- 14 June 2016
- Number of pages
- Document type
- PhD thesis
- Faculty of Science (FNWI)
- Informatics Institute (IVI)
Query auto completion is an important feature embedded into today's search engines. It can help users formulate queries which other people have searched for when he/she finishes typing the query prefix. Today's most sophisticated query auto completion approaches are based on the collected query logs to provide the best possible queries for each searcher's input.
In this thesis, we develop new query auto completion methods for information retrieval. First, we consider the information of both time and user to propose a time-sensitive personalized query auto completion approach. In previous work, these two sources of information have been developed separately. We bring them together and pay special attention to long-tail prefixes. Second, based on a learning-to-rank framework, we propose to extract features originating from so-called homologous queries and from the semantic similarity of terms, which allow the contributions from similar queries and from semantic relatedness to be used for query auto completion.
In addition, we study the problem of query auto completion diversification, where we aim to diversify aspect-level query intents of query completions. This task has not been studied before. Given that only a limited number of query completions can be returned to users of a search engine, it is important to remove redundant queries and improve user satisfaction by finding an acceptable query. Finally, we conduct an investigation on when to personalize query auto completion by proposing a selectively personalizing query auto completion approach, where the weight of personalization in a query auto completion model is selectively assigned based on the search context in session.
The experimental results in this thesis indicate that our proposed query auto completion approaches can improve the ranking performance of query completions in terms of well-known metrics, like Mean Reciprocal Rank. The unique insights and interesting findings in this thesis may be used to help search engine designers to improve the satisfaction of search engine user by providing high quality query completions.
- Research conducted at: Universiteit van Amsterdam
Series: SIKS dissertation series 2016-23
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library, or send a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.