Detection of User Interesting Fields and Personalization of Search Results using
User Search History Information
Heejun Han, Heeseok Choi, Jaesoo Kim
Department of NTIS, Korea Institute of Science and Technology Information
245 Daehak-ro, Yuseong-gu, Daejeon, South Korea
hhj@kisti.re.kr, choihs@kisti.re.kr, jaesoo@kisti.re.kr
ABSTRACT
When a user inputs a search query to find a
document or information, the search service
provides a list of search results. To get more
information, the user explores the provided list and
selects one of the results to move to the page with
detailed views. Most search services provide the
same results for the same search queries. But
providing the same results for the same search
queries each time cannot satisfy the ever so
different users‟ needs. In this article, we propose a
method to analyze the user search query, search
result set, and the user history for detailed views to
decide an individual's field of interest, and reflect
this information to individualize the search results.
To obtain classification information for
comprehending the users‟ fields of interest, the
Dewey Decimal Classification (DDC) codes pre-
assigned to the published journal document records
are used, and this information is applied to re-rank
the search results individually, using the boosting
method. Using this method, different search results
individualized to each user‟s preference are
suggested for the same search queries by the users.
KEYWORDS
Search personalization; search history; user
preference detection; information retrieval; boosting
1 INTRODUCTION
Most internet users perform web searches to
find wanted information. But most web search
services provide the same search results for a
search query regardless of individual interests
[1-4]. This is problematic for users wanting to
easily access information of interest and
preference. It is most important to provide the
desired information of the users at the top of the
search results. If two users entered the same
search term „virus‟ into a search service, the
results desired by a computer engineer and a
biologist may be different. For the computer
engineer, information regarding computer
viruses should be at the top of the search
results, and for the biologist, information
regarding disease related viruses is more
appropriate. In this article, the users‟ search
queries, search result set created by the search
query, and users' history on provided detailed
views are analyzed to determine individual
fields of interest. Every time an individual
performs a search, the weighted information on
the user's fields of interest is recalculated and
re-determined, and this information is reflected
on the search results again to provide the
information of preference at the top of the
search results. The published journal database
serviced by KISTI (Korea Institute of Science
and Technology Information) was used as a
search subject for this experiment, and the DDC
codes were used to determine individual fields
of interest. In section 2, we explain some
related works about search technologies such as
content personalization by user profile, content-
based recommendation system, and a
personalization retrieval system based on a
classification and user query. In section 3, the
user search history information used for
determination of individual fields of interest
will be defined and its storage structure is
designed. And a method for determining
individual fields of interest by analyzing search
ISBN:978-0-9891305-8-5 ©2014 SDIWC 34