This document proposes a system to organize user search histories into query groups for web personalization. The system has four main modules: 1) a query group module that computes query groups from search histories, 2) a search history module that stores user search queries and clicks over time, 3) a query relevance module that calculates relevance between queries, and 4) a dynamic query grouping module that uses a similarity function to dynamically group queries. The goal is to better understand users' search contexts and tailor their search experiences.
4. Modules
The proposed system has the following
modules:
•Query Group
•Search History
•Query Relevance
•Dynamic Query Grouping
5. Input Dataset
Query Time Query ClickURL
2008-11-13
00:01:30
kitchen counter
in new or leans
http://www.su
perpages.com
2008-11-13
00:01:33
photo example
quarter
doubled die
coin
http://www.coi
nresource.com
2008-11-13
00:01:39
plays Perry cox
wife scrubs
http://www.ref
erence.com
6. Query Group Module
• This module is responsible for computing groups.
• First and foremost, query grouping allows the
search engine to better understand a user’s
session and potentially tailor that user’s search
experience according to her needs.
• Once query groups have been identified, search
engines can have a good representation of the
search context behind the current query using
queries and clicks in the corresponding query
group.
7. Search History
• This module is responsible for storing the
search history of the user.
• User’s search history consists of the Query,
URL with the corresponding time and date.
• User’s search history is stored in the database
which is used for organizing according to the
group.
8.
9.
10.
11. Query Relevance Module
• This module is responsible to compute query
relevance between two queries using QFG.
• The edges in Query Fusion Graph correspond
to pairs of relevant queries extracted from the
query logs and the click logs.
• Query Fusion Graph merges the information
of both Query Reformulation Graph and
Query Click Graph.
12. • This module calculates the query relevance by
performing random walks over the query
fusion graph.
13. Dynamic Query Grouping Module
• This module is responsible to group queries
dynamically.
• The proposed similarity function is used to
find the similarity of queries while grouping
them.
14. References
• Organizing User Search Histories Heasoo Hwang, Hady W. Lauw, Lise
Getoor, and Alexandros Ntoulas IEEE TRANSACTIONS ON KNOWLEDGE
AND DATA ENGINEERING, VOL. 24, NO. 5, MAY 2012
• Agglomerative clustering of a search engine query log Doug Beeferman
Lycos Inc. 4002 Totten Pond Road Waltham, MA 02451