You are on page 1of 24

Profile Analysis of Graduate Theses

Conducted on Computer Science and


Engineering in Turkey

Volkan TUNALI & T. Tugay BİLGİN


Maltepe University
Content
 Introduction
 Motivation
 Aim and Scope
 Data Acquisition Method
 Thesis Distribution by Years
 Number of Theses by University
 Change in the Themes of Theses
 Evaluation
 Conclusion and Future Work

October 24-25, 2013 / Kuşadası ISCSE 2013 2/24


Introduction
 Graduate theses have significant value in the
development of a discipline.
 Analysis of theses conducted on a discipline can
 provide valuable information about coverage and depth
of the field,
 present the general view of the field, revealing the trends
and also the topics saturated over the years.

October 24-25, 2013 / Kuşadası ISCSE 2013 3/24


Motivation
 There is no comprehensive and up-to-date
study that examines and analyzes the theses
conducted specifically on computer science
and engineering in Turkey.

October 24-25, 2013 / Kuşadası ISCSE 2013 4/24


Aim and Scope
 Our aim is
 to investigate the profile of the graduate theses conducted
on computer science and engineering in Turkey,
 to understand the research interests and trends over the
last three decades in this field from the theses.
 We examined 6,307 master’s and PhD theses
 5,595 MSc & 712 PhD theses,
 submitted to the national thesis database of The Council
of Higher Education (Yükseköğretim Kurulu – YÖK),
 between 1983 and 2012.

October 24-25, 2013 / Kuşadası ISCSE 2013 5/24


Data Acquisition Method 1/5
 YÖK thesis database is publicly available via
the YÖK thesis web portal
 Unlimited search by several criteria,
 Unlimited abstract access (Turkish & English),
 Limited full text download.

October 24-25, 2013 / Kuşadası ISCSE 2013 6/24


Data Acquisition Method 2/5

October 24-25, 2013 / Kuşadası ISCSE 2013 7/24


Data Acquisition Method 3/5
 We needed thesis abstracts for an extensive
text mining study.
 There were over 300,000 theses in the
database as of July 2012.
 We developed a web crawling and processing
tool in C# programming language.

October 24-25, 2013 / Kuşadası ISCSE 2013 8/24


Data Acquisition Method 4/5
 Downloaded
 308,870 HTML files for thesis registration data,
 308,870 HTML files for thesis abstracts.
 Performed cleaning, parsing, and information
extraction from these HTML files.
 Recorded the extracted data into a relational
database (for further easy querying and
processing).

October 24-25, 2013 / Kuşadası ISCSE 2013 9/24


Data Acquisition Method 5/5
 We selected 6,307 theses for processing
 whose at least one subject topic field contains the
text “Bilgisayar Mühendisliği Bilimleri” (computer
engineering sciences).
 We made use of the index terms (keywords)
of the theses.

October 24-25, 2013 / Kuşadası ISCSE 2013 10/24


Thesis Distribution by Years

October 24-25, 2013 / Kuşadası ISCSE 2013 11/24


Thesis Distribution by Universities
University MSc PhD Total %
Middle East Technical University 910 141 1,051 16.7
Boğaziçi University 508 60 568 9.0
İstanbul Technical University 473 59 532 8.4
Ege University 322 78 400 6.3
İhsan Doğramacı Bilkent University 328 33 361 5.7
Gazi University 266 30 296 4.7
Marmara University 230 36 266 4.2
Yıldız Technical University 157 31 188 3.0
Dokuz Eylül University 137 32 169 2.7
Selçuk University 120 22 142 2.3

October 24-25, 2013 / Kuşadası ISCSE 2013 12/24


Change in the Themes of Theses
 We analyzed the themes of theses in 5 periods of
about 5 years as
 1983-1990,
 1991-1995,
 1996-2000,
 2001-2005,
 2006-2012.
 Most popular research topics of these periods are
presented.

October 24-25, 2013 / Kuşadası ISCSE 2013 13/24


1983-1990 Period
Research Topic Count %
Software 27 7.12
Computers 14 3.69
Database 8 2.11
Expert systems 7 1.85
Education 6 1.58
Computer aided design 5 1.32
Microcomputers 5 1.32
Artificial intelligence 4 1.06
Computer assisted education 4 1.06
Computer graphics 4 1.06
Computer networks 4 1.06
Information systems 4 1.06
System analysis 4 1.06
Computer assisted instruction 3 0.79
Computer software 3 0.79
October 24-25, 2013 / Kuşadası ISCSE 2013 14/24
1991-1995 Period
Research Topic Count %
Database 30 2.26
Design 30 2.26
Database management system 24 1.81
Software 23 1.73
Artificial neural networks 20 1.50
Expert systems 20 1.50
Computer networks 18 1.35
Algorithms 15 1.13
Artificial intelligence 15 1.13
Programming languages 14 1.05
Computer aided control 13 0.98
Object oriented database 11 0.83
Operating systems 11 0.83
Computer assisted education 9 0.68
Data communication 9 0.68
October 24-25, 2013 / Kuşadası ISCSE 2013 15/24
1996-2000 Period
Research Topic Count %
Database 49 2.07
Internet 47 1.99
Artificial neural networks 41 1.73
Computer networks 40 1.69
Fuzzy logic 29 1.23
Algorithms 25 1.06
Distributed systems 25 1.06
Software 24 1.01
Simulation 22 0.93
Information systems 21 0.89
Java 19 0.80
Computer assisted education 14 0.59
Object oriented database 13 0.55
Robots 13 0.55
Artificial intelligence 12 0.51
October 24-25, 2013 / Kuşadası ISCSE 2013 16/24
2001-2005 Period
Research Topic Count %
Internet 51 2.31
Information systems 46 2.09
Simulation 30 1.36
Software 27 1.22
Database 25 1.13
Data mining 23 1.04
Computer networks 22 1.00
Artificial neural networks 21 0.95
Distance education 21 0.95
Electronic commerce 17 0.77
Distributed systems 15 0.68
Genetic algorithms 15 0.68
Java 13 0.59
Artificial intelligence 12 0.54
Natural language processing 11 0.50
October 24-25, 2013 / Kuşadası ISCSE 2013 17/24
2006-2012 Period
Research Topic Count %
Data mining 103 2.04
Artificial neural networks 102 2.02
Wireless networks 78 1.54
Image processing 60 1.19
Artificial intelligence 57 1.13
Fuzzy logic 47 0.93
Genetic algorithms 38 0.75
Software engineering 37 0.73
Simulation 33 0.65
Computer networks 30 0.59
Ontology 29 0.57
Natural language processing 28 0.55
Software 26 0.51
Computer vision 25 0.49
Optimization 24 0.47
October 24-25, 2013 / Kuşadası ISCSE 2013 18/24
Evaluation 1/4
 Artificial neural networks have almost always been
a hot research topic.
 Artificial intelligence has been gaining its old
popularity again.
 Genetic algorithms is also highly popular in the last
decade.
 Fuzzy logic
 was very popular during 1996-2000,
 losts interest during 2001-2005,
 gains its popularity again during 2006-2012.

October 24-25, 2013 / Kuşadası ISCSE 2013 19/24


Evaluation 2/4
 Database was one of the most popular
subjects until the period of 2001-2005.
 There is a paradigm shift in the academia
towards data mining and related techniques.
 Natural language processing has been gaining
interest.

October 24-25, 2013 / Kuşadası ISCSE 2013 20/24


Evaluation 3/4
 Internet and related technologies and
applications were highly popular through
1996-2005.
 Recent communication-related studies are
mostly on wireless networks and
communication systems.

October 24-25, 2013 / Kuşadası ISCSE 2013 21/24


Evaluation 4/4
 Software engineering has been drawing
attention and it has been becoming a major
research area.
 Java was highly popular in academic studies
during the 1996-2000 period, but it is now a
well-understood development tool rather
than a special research concept.

October 24-25, 2013 / Kuşadası ISCSE 2013 22/24


Conclusion and Future Work
 We examined 6,307 master’s and PhD theses conducted on
computer science and engineering between 1983 and 2012
in Turkey.
 We presented the profile analysis of the theses, and analyzed
the popularity of research areas over the years.
 Our analysis of theme change is based on index terms
supplied by thesis authors.
 As a future work, we plan to extend and improve our analysis
with automatic keyword and index term extraction from the
thesis abstracts using text mining and natural language
processing techniques.

October 24-25, 2013 / Kuşadası ISCSE 2013 23/24


Thank You!
 Any questions are welcome!

October 24-25, 2013 / Kuşadası ISCSE 2013 24/24

You might also like