=Paper=
{{Paper
|id=Vol-1312/swcib2014_paper3
|storemode=property
|title=Researcher Profiling for Researcher Analysis Service
|pdfUrl=https://ceur-ws.org/Vol-1312/swcib2014_paper3.pdf
|volume=Vol-1312
|dblpUrl=https://dblp.org/rec/conf/jist/LeeCJJ14
}}
==Researcher Profiling for Researcher Analysis Service==
https://ceur-ws.org/Vol-1312/swcib2014_paper3.pdf
Researcher Profiling for Researcher Analysis Service
Mikyoung Lee1, Minhee Cho11, Changhoo Jeong1, Hanmin Jung1
1
Korea Institute of Science and Technology Information (KISTI),
245 Daehak-ro, Yuseong-gu, Daejeon, South Korea
{jerryis, mini, chjeong, jhm}@kisti.re.kr
Abstract. This study examines a method of generating comprehensive profiling
information for a researcher analysis service. In addition to basic and
performance-based information about researchers necessary to generate
profiling information, we introduce researcher performance index models for
researcher analysis service. The models can that measure qualitative and
quantitative performance, researcher influence, and growth potential, which are
necessary to analyze the skills of a researcher from multiple perspectives. We
measure the qualitative performance index of researchers by using the citation
index of the papers they have published. The quantitative performance index
can be measured based on a researcher’s published papers. The Influence index
measures the social influence of researchers according to their academic work.
The growth potential index measures the speed at which a researcher improves
research performance. We expect to develop a researcher analysis service that
can evaluate a researcher’s performance and enhance his or her research
abilities.
Keywords: Researcher Profiling, Research Performance analysis, Researcher
Analysis service
1 Introduction
Many services exist that provide researchers with a wealth of academic
information such as scholarly literature, including Google Scholar, MS Academic
Searchand Elsevier SciVal. However, services are lacking that help improve the
research abilities of researchers by analyzing their research skills. In Informetrics,
various studies have represented researcher research skills in the form of objective
numbers by examining various bibliometric indicators for research policy [1].
Currently, the h-index, which uses a citation index, is widely used to measure
researcher skills. However, the h-index fails to produce objective assessments of the
abilities of individual researchers. This is because it reflects only the combined
achievements of co-researchers and is limited in the manner in which it comparatively
analyzes them with the achievements of researchers in other fields. In addition, the h-
index does not reflect the assessments of competent new researchers. Numerous
studies have been conducted to compensate for the existence of a index, which is
insufficient for analyzing the comprehensive research skills of researchers.
1 Corresponding author
This study describes researcher profiling information used in a researcher analysis
service for improving research skills.
2 Related Work
InSciTe Advisory(http://inscite-advisory.kisti.re.kr/search) deals with various
textual big data including papers, patents, Web, social data, and linked data. It
analyzes researcher's competitiveness and recommends attainable strategy and plan
based on prescriptive analytics as well as descriptive analytics. It consists of two main
analytics: descriptive analytics and prescriptive analytics. The former includes both
activity history analysis and research power analysis for a selected researcher [2][3].
- Commerciality: Ability to produce practical products and profits
- Scholarity: Ability to produce new knowledge and academic outputs
- Influentiality: Ability to spread leverage to other researchers
- Diversity: Ability to extend research scope and degree of variation of research
field
- Durability: Ability to keep research consistently in some research field
- Technology emergability: Degree of emerging about research area performed
by researcher
- Partner Trend: Status of research area changing about partner researchers
- Market Trend: Status of market size changing about researcher area
-
Fig. 1. Descriptive analytics of InSciTe Advisory
ArnetMiner(http://www.arnetminer.org) aims to provide comprehensive search and
mining services for researcher social networks. In this system, we focus on: creating a
semantic-based profile for each researcher by extracting information from the
distributed Web; integrating academic data from multiple sources; accurately
searching the heterogeneous network; analyzing and discovering interesting patterns
from the built researcher social network [4].
ArnetMiner offers insights into the capabilities of researchers by using the seven
indices of activity, papers, citation, h-index, g-index, sociability, and diversity. As Fig.
2 shows, users can judge a researcher’s ability by examining the graph related to each
index.
- Citation: The number of citations of all publications by an expert.
- Papers: The number of all publications by an expert.
- H-index: An expert has index h if h of his or her N papers have at least h
citations each, and the other papers have at most h citations each.
- Activity: People's activity is simply defined based on one's papers published
in the last years.
- Diversity: An expert's research may include several different research fields.
Diversity is defined to quantitatively reflect the degree
- Sociability: The score of sociability is basically defined based on how many
coauthors an expert has.
Fig. 2. Researcher Information on ArnetMiner
3 Researcher Performance Index Model
This chapter discusses the researcher performance index model used for researcher
profiling. This index model collects the bibliography and citation information of
researchers and compares them numerically to measure their research skills. We
define four indices based on qualitative and quantitative performance, influence, and
growth potential.
3.1 Qualitative Performance Index
We measure the qualitative performance index of researchers by using the citation
index of the papers they have published. The qualitative performance index is a
measurement of the influence of published papers and, thus, an indicator of their
research quality. In general, indices such as citation index, h-index [5] and g-index [6]
are used to measure researcher performance. A citation index can be used because the
h-index and g-index for all authors in our collection are unavailable.
3.2 Quantitative Performance Index
The quantitative performance index can be measured based on a researcher’s
published papers. The more active the researcher has been producing recent studies,
the more weight is given to the annual performance index.
The qualitative performance index is not proportional to the quantitative
performance index because the more papers that are published, the higher is the
researcher’s score on the quantitative index, thus indicating active research. However,
the higher the quality of the papers, the higher is the researcher’s quantitative
performance index because the rank of the journals and conferences are reflected in
the calculation of the quantitative performance index. If researchers being considered
are the main or corresponding authors, they are more likely to score high on the
quantitative index when they are actively involved in research. Using (1), the
quantitative performance index can be calculated based on the rank of the journals
and conferences, the weight of the author order.
𝑄𝑃 = ∑𝑛1 𝐽𝑜𝑢𝑟𝑛𝑎𝑙𝑟𝑎𝑛𝑘 × 𝑎𝑢𝑡ℎ𝑜𝑟𝑜𝑟𝑑𝑒𝑟 + ∑𝑚
1 𝑃𝑟𝑜𝑐𝑒𝑒𝑑𝑖𝑛𝑔𝑟𝑎𝑛𝑘 × 𝑎𝑢𝑡ℎ𝑜𝑟𝑜𝑟𝑑𝑒𝑟 (1)
The rank of journals can be calculated by the journal’s impact factor. The journals
from each field are ranked according to the impact factor. Regarding conference
proceedings, the rankings of the conferences in each field are used. With respect to
authors, main and corresponding authors who write the majority of a paper can earn
one point, whereas co-authors receive points equal to 1/nth, where n is the total
number of authors.
3.3 Influence Index
Using a co-author network, the influence index measures the social influence of
researchers according to their academic work. The influence index is an application of
Google’s PageRank [7] based on the co-author network and assumes that co-authors
of influential research also have high influence.
𝐼𝐼(𝐴) = (1 − 𝑑)/𝑛 + 𝑑(∑𝑛1 𝐼𝐼( 𝑇𝑛)/𝐶(𝑇𝑛)) (2)
In (2), II(A) refers to the value of researcher A’s influence, II(T1) refers to the
value of researcher T1’s influence, and C(T1) represents the number of co-authors
with researcher T1. In addition, d equals 0.85, as in the PageRank algorithm.
3.4 Growth Potential Index
The growth potential index measures the speed at which a researcher improves his
or her research performance. We measure the growth potential index based on the
amount of time a researcher spends reaching his or her current research performance,
thus enabling to the system to predict how quickly the researcher can become a leader.
𝐺𝑃 = 𝑄𝑃/𝑛 (3)
QP represents the quantitative performance index and n denotes the elapsed time
from the beginning of the study to the present.
4 Researcher Profiling
4.1 Design
During the development of our researcher analysis service, we define research
profiling, which includes all analytical information about the researcher. Researcher
profiling consists of a list of basic information about researchers such as names and
affiliations. In addition, it identifies essential information derived from their research
such as main research skills (as shown in Fig. 3), research area (Fig. 3), and the co-
author network. Moreover, it includes indices provided by research performance
models.
Fig. 3 shows an example of researcher profiling designed in this study. The profiling
extracts basic information from bibliographies and citations such as names,
affiliations, beginning years of research, paper types, author order, number of
citations, and major technology used. Information such as the level of technology and
co-author network is obtained from analysis models. All acquired information is used
to calculate the four indices of researcher performance index models. Most profiling
preferences include time information, which facilitates the observation.
Fig. 3. Design of Researcher Profiling
4.2 Applications
Multiple aspects of research performance can be evaluated using the four indices of
qualitative and quantitative performance, influence, and growth potential provided by
research performance index models. Each index can be used to evaluate researcher
tendencies and characteristics. For example, researchers can be divided into various
types such as competent researchers who produce high quality research results, those
who produce highly quantitative outcomes, and those who have immense research
potential. In addition, yearly trends related to researcher profiling information can
confirm changes in a researcher’s performance based on flow patterns of falling,
rising, and stagnancy. Moreover, the service can also be used for prescriptive
analytics to reinforce researchers’ strengths, correct research weaknesses, and
compare researchers of similar research capabilities and patterns. Finally, it can be
used to analyze research performance patterns and to determine researcher’s type.
5 Conclusion
This study described a researcher profiling method necessary to develop a
researcher analysis service. Research profiling information consists of meta-
information, performance information, and information of researchers acquired by
analytical model. We evaluated researchers from multiple perspectives. We used a
qualitative performance index, which indicates the quality of papers; a quantitative
performance index, which indicates the number of papers published; an influence
index, which refers to the amount of influence a researcher has among his or her peer
researchers; and a growth potential performance index, which indicates degrees of
change (i.e., levels of improvement) in research performance over time.
Our researcher profiling service represents an aggregation of all information that
can be used to evaluate researcher and their research characteristics and patterns. This
service is necessary to develop a service to assist researchers in improving their
performance.
Acknowledgements
This work was supported by the IT R&D program of MSIP/KEIT. [2014-044-024-
002, Developing On-line Open Platform to Provide Local-business Strategy Analysis
and User-targeting Visual Advertisement Materials for Micro-enterprise Managers].
References
1. Rodrigo, C., Maria, B.: The h-index: Advantages, limitations and its relation with other
bibliometric indicators at the micro level. In: Journal of Informetrics, vol 1 issue 3, pp193-
203, Elsvier (2007)
2. S. Song, D. Jeong, J. Kim, M. Hwang, J. Gim, H. Jung, :Research Advising System based
on Prescriptive Analytics, In Proc. of International Workshop on Data-Intensive Knowledge
and Intelligence in conjunction with the 9th FTRA International Conference on Future
Information Technology. (2014)
3. Jinhyung Kim, Myunggwon Hwang, Janwon Gim, Sa-Kwang Song, Do-Heon
Jeong,Seungwoo Lee, Hanmin Jung.: Intelligent Analysis and Prediction Model of
Researcher’s Capability for Prescriptive Analytics. In: AIM 2014 (2014)
4. Jie, T., Jing, Z., Limin, Y., Juanzi, L., Li, Z., Zhong. S.: ArnetMiner: extraction and mining
of academic social networks. The 14th ACM SIGKDD international conference on
Knowledge discovery and data mining, pp 990-998, ACM, New York, NY, USA (2008)
5. H-index, http://en.wikipedia.org/wiki/H-index
6. G-index, http://en.wikipedia.org/wiki/G-index
7. Page, L., Brin, S., Motwani, R., Winodgrad, T.: The pageRank Citation Ranking: Bringing
Order to the Web, Technical Report, Standford InfoLab. (1999)