ON CITATION-BEHAVIOR-GUIDED SEARCHPHRASE SUGGESTERS FOR ONLINE DIGITAL
Author(s):
Sulieman Bani-ahmad
Paper abstract: A content-driven search-keyword (SK-) Suggester for keyword-based search in digital libraries is
proposed. Suggesting search terms while the user is entering search terms is helpful for constructing
correctly-typed and focused search terms for digital library queries. The proposed SK-Suggester is based
on pre-analyzing step of the publication collection to be searched. The pre-analysis step consists of the
following. (i) We parse the document collection using a Link-Grammar parser, a syntactic parser of
English, next, (ii) we group publications based on their research topics, (iii) after that, the parser output is
used to build a hierarchical structure of simple and compound tokens to be used to suggest search terms.
In order to sort the suggested terms, we use the TextRank algorithm, a text summarization tool, to assign
topic-sensitive scores to the simple and compound tokens. The identified research topics are used to help
user entering focused search terms prior to the actual search query execution. The topic-sensitive
TextRank scores are further refined to incorporate the user’s citation behavior model proposed in [Bani-
Ahmad, S., Ozsoyoglu, T. 2009].
We experimentally show that the proposed framework promises a more scalable, high quality, and userfriendly
SK-Suggester when compared to its competitors. We validate our proposal experimentally using
a subset of the ACM SIGMOD Anthology digital library as a testbed, and by employing the researchpyramid
model to identify the research topics.
Keywords:
Online digital libraries, Search keyword suggesters, The research-pyramid model.
