|
Title:
|
ON CONTEXT-DRIVEN ONLINE SEARCH-PHRASE SUGGESTERS FOR LARGE TEXTUAL DOCUMENT REPOSITORIES |
|
Author(s):
|
Sulieman Bani-ahmad |
|
ISBN:
|
978-972-8924-79-9 |
|
Editors:
|
Miguel Baptista Nunes, Pedro Isaías and Philip Powell |
|
Year:
|
2009 |
|
Edition:
|
Single |
|
Keywords:
|
Search-phrase suggester, Information Retrieval, Textual-Document Digital Libraries |
|
Type:
|
Full Paper |
|
First Page:
|
151 |
|
Last Page:
|
160 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Guiding users choice of search-terms and search-phrases is useful as it allows for successful search. Studies show that
user spend considerable amount of time choosing search terms and, probably modifying their chosen terms in order to
focus search outputs. This means that each search-session is composed of multiple, probably unsuccessful search queries.
This in turns degrades scalability of online textual document repositories.
In this paper we propose SISSTOR, a prototype of Scalable Inline Search-phrase Suggestor for Textual-document Online
Repositories. We achieve scalability through making suggestions from contexts (topics) of interest of the current user.
Our proposal has two major offline steps; the pre-analysis step in which the documents at hand are (i) processed to
extract frequent search phrases from text, and (ii) are put into contexts such that each context represents a topic. At
search time, topics of interest are gradually identified as the user enter more keywords and even parts-of-words.
Suggestions are made out of the identified topics.
We experimentally evaluate SISSTOR against its basic design principles. For that we use a repository of more than
14,891 publications from the computer science domain. Our proposal promises high-quality suggestions with no startup
requirements as the case in Google Suggest which requires having a search history to make suggestions from. |
|
|
|
|
|
|
