International Journal of Scientific & Engineering Research Volume 8, Issue 7, July-2017 2218
ISSN 2229-5518
IJSER © 2017
http://www.ijser.org
Projecting the Future Direction of
Publication PatternsUsing Text Mining
Adebola K. Ojo
Department of Computer Science, University of Ibadan, Nigeria
adebola_ojo@yahoo.co.uk
Abstract - In this study, text mining techniques were used to identify various research trends in academic journal
publications. These techniques were applied to figure out trends in research patterns related to various specialisation
areas in Computer Science academic journal articles within a period of two decades. The corpus mined were crawled
online, pre-processed and transformed into structured data using filtering and stemming algorithms. The data were
grouped into series of word features based on bag of words document representation. Abstracts and the keywords of
the articles selected from these journal articles were used as the dataset. It was discovered that the publication trends
have changed tremendously from communications and security to artificial intelligence over time.
Index Terms– Computer Science, Filtering and Stemming Algorithms, Journals, Trends
1 INTRODUCTION
Articles published in peer reviewed journals
are likely to remain a very important means
of distributing research findings for the
foreseeable future [1].It is a mathematical
technique that uses historical results to
predict future outcome. During the
publication of research articles in academic
journals, it is necessary to identify trends.
The process of identifying trends is called
the trend analysis.
Trend Analysis is a mathematical
scientific approach that eliminates potential
error by utilizing precise calculations in
order to provide the utmost accuracy. It is
the most dependable and efficient method
for anticipating possible future behavior and
desired outcome of a specific journal
publication.It is a quantitative review of
what happens over a period of time.
Article abstracts provide a
comprehensive yet concise overview of an
article[2][3]. Abstracts are much shorter than
the full text, which minimises the influence
of data noise. Therefore, this study focused
on prediction of abstracts of a journal article.
The data used for this study were retrieved
from the Institute of Electrical and
Electronics Engineers (IEEE) Transactions on
Computers, a monthly publication with a
wide distribution to researchers, developers,
technical managers, and educators in the
computer field. It publishes papers on
research in areas of current interest to the
readers. Journal of Institute of Electrical and
Electronics Engineers (IEEE) Transactions on
Computers was chosen because it is one of
the highly rated Computer Science journals
with ISI indexed ranking and impact
factors[4][5][6][7].
This study focused on the mining of
trends of journal publications using the
abstracts, keywords and authors’
bibliometric information of these journal
articles. It was based on trend analysis using
text mining techniques. Text mining is a
multidisciplinary field, concerning retrieval
of information, analysis of text, extraction of
information, categorization, clustering,
visualization, mining of data, and
machine learning [8]. As the core of the
knowledge management systems, text
mining is a cross between information
retrieval (IR) and artificial intelligence (AI). It
is estimated that 90% of the world’s online
content is based on text (Oracle
Corporation). An effective means to deal
with structured, numeric content has been
developed via database management
systems (DBMS), but text processing and
analysis is significantly more difficult. The
status of knowledge management systems is
much like that of DBMS twenty years ago.
The real challenges, and the potential payoffs
for an effective, universal text solution, are
equally appealing.
Text mining predictive methods
help organizations enhance the value of
unstructured information by deploying
insight from text analysis in software
applications and business processes. Once
textual information is transformed into a set
IJSER