A Scientometric Analysis of Scientific Productivity of Artificial Intelligence Research in India

The study presents a scientometric analsyis of publications related to ‘Artificial Intelligence’ research in India during 2009-2018. In today’s ICT driven world, artificial intelligence has taken up some tasks of our daily life to make it easier. As a consequence, extensive research is going on “Artificial Intelligence” to find out it’s potential in knowledge development. The paper analyses the bibliographic data retrived from Scopus database extracted with a suitable search query. The study was conducted taking the chronological growth of research publications, relative growth rate, doubling time, scientometric profile of authors, document type of publications, source profile, keyword analysis, institution wise distribution of publications, funding agency wise distribution. The analysis was conducted using the MS-Excel. The study reveals that a maximum number of publications are in the form of conference procedings and articles. Artificial Intelligence, Learning system, algorithms, data mining are the keywords with maximum number of occurences. The findings of the study implies India need become more competitive with the world leaders in artificial intelligence research. To get more return from AI applications, the stakeholders are required to play a catalytic role to build and strengthen research capacity in the nation by paving quality research environment, adequate funding, research incentives, and development of IT infrastructure.


INTRODUCTION
Artificial Intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to suppose like humans and mimic their actions. The term may additionally be applied to any machine that exhibits traits related to a person's mind like learning and problem-solving. The characteristic of AI is its ability to rationalize and take actions that have the most effective probability of achieving a particular goal. AI is a wide-ranging branch of engineering science involved with building sensible machines capable of playing tasks that generally need human intelligence. AI is a knowledge base science with multiple approaches, however, advancements in machine learning and deep learning area unit making a paradigm shift in just about every sector of the technical school business. Artificial intelligence could be a science and technology-supported disciplines like technology, Biology, Psychology, Linguistics, arithmetic, and Engineering. A serious thrust of AI is within the development of computer functions related to human intelligence like reasoning, learning, and problem-solving. Out of the subsequent areas, one or multiple areas will contribute to making an intelligent system. India, being the fastest growing economy with the second largest population in the world, has a significant stake in the AI revolution. Recognising AI's potential to transform economies and the need for India to strategise its approach, Hon'ble Finance Minister, in his budget speech for 2018 -2019, mandated NITI Aayog to establish the National Program on AI, with a view to guiding the research and development in new and emerging technologies. In pursuance of the above, NITI Aayog has adopted a three-pronged approachundertaking exploratory proof-of-concept AI projects in various areas, crafting a national strategy for building a vibrant AI ecosystem in India and collaborating with various experts and stakeholders. [1] Since the start of this year, NITI Aayog has partnered with several leading AI technology players to implement AI projects in critical areas such as agriculture and health.
Artificial intelligence has been predominant in different fields, for example, gaming, natural language processing, master frameworks, vision frameworks, speech recognition, handwriting recognition, and astute robots. The main aims of artificial intelligence are to create an expert system, to implement human intelligence in machines.
carried out a scientometric analysis on artificial intelligence research in India from 1968 to 2014 and found that a total of 6529 records which has grown considerably since 2004, and the last 2 years have witnessed a large publication output in the field of AI. The average citation recorded per paper is 3.06. The average number of authors per paper is three. "Artificial Intelligence" was found to be the most popular keyword, followed by "Algorithms." A total of 12.64% of the papers have been published with international collaboration. The maximum number of publications have given by Anna University. It was found that the IITs played a major role in the field of artificial intelligence research in India.

METHODOLOGY
One of the most famous products provided by Elsevier in the Scopus database, it is a very comprehensive multidisciplinary citation database indexing a large number of science and technology research papers globally, provided by subscription basis. The following search string to be used to collect the data: (TITLE-ABS-KEY ("Artificial Intelligence") AND (LIMIT-TO (PUBYEAR, 2018) TO (PUBYEAR, 2009)) AND (LIMIT-TO (AFFILCOUNTRY, "India")) AND (LIMIT-TO (PUBSTAGE, "final"))). Worldwide a total of 191362 research papers were found in which 14108 publications contributed by Indian authors, which were available on Scopus databases from the marked period of study. These research papers along with full bibliographical details like the growth of publications, relative growth rate and doubling time of publications, author's profiles, document types, source title, significant keywords, affiliations name, funding agencies have been extracted from the Scopus database. The raw data was downloaded on 13/02/2019 to conduct the study and the data were analyzed and tabulated in MS-Excel applications software.

Scope of the Study
The scope of the study is limited to the "Artificial Intelligence" research publication and the period is limited to a 2009-2018 year. The study was also limited to the publications which were contributed by Indian authors indexed in the Scopus database.

Background study of Artificial Intelligence
In the past years, a few selected studies were undertaken on quantitative analysis of worldwide output on artificial intelligence research covering various periods and various aspects of the subject in their research analysis. Bhattacharjee (2019) [2] conducted a scientometric study on AI in one decade i.e., 2009 to 2018. After the analysis, it has been found that AI as an academic subject has gained tremendous momentum in the recent past which is reflected through the annual research publications in the particular field. The beginning year 13936 while in the ending year 27061 publications were recorded across the countries during the study period. He also mentinede that India has been taken the fifth position after the US, China, UK and Germany in the last 20 years with 16903 publications. He further mentioned that if the same search keyword use for the last ten years then it was found that India secured 3 rd position with 13923 publications just after the US and China. This substantial growth clearly shows the popularities of AI research in India in recent years. Gupta and Dhawan (2018) [3] conducted a study a scientometric assessment on artificial intelligence research publications in India from (2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) in which a total of 9730 research papers were published by Indian authors and it has gained tremendous momentum in the recent past which is reflected through the annual research productivity of the particular field of study. The most productive organization was Anna University with 294 publications, followed by Jadavpur University 270 publications and the most prolific authors was S. Das from ISI Kolkata with 36 publications, followed by B. K. Panigrahi with 32 contributions. The significant keywords were Artificial Intelligence with 9496 frequency. The most productive journals in artificial research in India during the period 2007-2016 was Applied Soft Computing Journal. Niu et al. (2016) [4] carried out a bibliometric analysis on global research on artificial intelligence publications during the period (1990-2014) in which a total 20715 articles were published and the maximum 344 articles were published in the Journal of Expert Systems with Applications, followed by Engineering Applications of Artificial Intelligence with 161 publications. The most productive authors were J. Neves from University Minho with 35 contributions, followed by G. Klopman from Case Western Reserve University with 30 contributions. A maximum 123 articles were contributed by Chinese Academy Sciences organization while the highest 5841 frequency is an Artificial Intelligence keyword. Shrivastava and Mahajan (2016) [5] to evaluate the yearly trends in research productivity (Kumar and Kaliyaperumal, 2015). [6] Relative Growth Rate and Doubling time of the publications Figure 2 shows the relative growth rate and doubling time of the research paper from the marked period of study. On the observation of Figure 2, it has been shown that the relative growth rate was in decreasing order in the year 2009-2014 i.e., (0.97 to 0.31) while the next year i.e., 2015 it was increased with (0.33) and then after in 2016-2018, it was again in decreasing trend with (0.28 to 0.16). The mathematical representation of the mean relative growth rate of articles over a specific period is derived from the following formula given by (Mahapatra, 1985). [7] RGR W W T T = − −  [8] found that doubling time is increased every year. The following formula is used to analyze the doubling time.
Scientometric profile of authors publications

Document type publications in Artificial Intelligence research
The productivity of researchers on Artificial Intelligence spreads over ten types of publication media, such that journal articles, review, conference paper, book chapters, letters, editorials, notes, short surveys, books, and erratum. Figure 3 depicts the document type publications in artificial intelligence research publications in India for 10 years i.e., (2009-2018).
On the observations of a particular Figure, it has been shown that the maximum 11326 (80.28%) of records were published in Conference paper, followed by Article with 2379 (16.86%) of publications were published while the minimum 2 (0.01%) of records were published in Erratum. The overall data of document type publications in artificial intelligence research was shown in Figure 3.   Profile of source title in Artificial Intelligence research  Figure 4 depicts the overall source title in increasing order of the publications from the marked period of study. Institution-wise distribution of publications in Artificial Intelligence research Figure 6 shows that a scientometric profile of the top 10 most productive institute in artificial intelligence from India depicts that their publications varied from 158 to 357 publications. Around 160 organizations have been identified from the literature, the top 10 organizations were listed in Figure 6 and it was found that the maximum 357 publications were published in the Jadavpur University and Anna University, followed by 335 research papers were published by the Vellore Institute of   Profile of funding agencies.

DISCUSSION
Bibliometric analysis has turn out to be an effective tool for summarizing the current status and predicting the future development trends in any knowledge domain which differentiate it from review works. [9,10]  For authors, number of citations and h-index are used as indicators to analyse their producitivity. It is calculated on the basis of number of times an article is cited, according to Scopus. This is a preliminary indicator of the impact of research by the researchers. [11] h-index is a rigourious metric that has been used for quantifying a researcher's scholarly output. Morover, a number of previous studies also used it for evaluating the productivity and impact of a researcher, country, institution, or journal. [12] Our study showed that A. Abraham of Mir Labs, Aburn is the most impactful author with highest number of citations and h-index (15994, 60).

For source of publication, Advances in Intelligent Systems and Computing, Lecture Notes in Computer Science Including Subseries
Lecture Notes in Artificial Intelligence and Communications in Computer and Information Science published maximum articles on Artificial intelligence. These journals are of high impact factor and attract papers of high quality and publication of high quality articles in turn increases the impact of these journals in their respective fields. It is evident that in future also research and development works are more likely to get published in this sources.
In bibliometrics, analysis of frequently appearing keywords can also reveal the hot spot catgories and the development of a research area. [13] According to the analysis of keywords, artificial intelligence, learning systems, algorithms, data mining, learning algorithms are mostly occurring keywords. Occurrence of keywords like neural network, feature extraction implies diversification in research on artificial intelligence.

CONCLUSION
The present study aims to analyze the artificial intelligence research papers published in the Scopus database for ten years i.e., (2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018). Using publications data from particular database, this study provides a quantitative and qualitative description of artificial intelligence research, after the observation, it has been found that a maximum of 2384 publications was contributed by the authors in the year 2016. The maximum (164.21) annual growth rate was recorded in the beginning year of the study i.e., 2009. The relative growth rate was recorded between (0.97 to 0.16) during the period of study this shows that the RGR was in decreasing trend however, the doubling time was shown in increasing trend with (0.71 to 4.26). It is also observed that a maximum of 11326 (80.28%) artificial intelligence research papers were published in conference, followed by journal article with 2379 (16.86%) publications. A maximum of 1224 articles was published in the Journal of Advances in Intelligent Systems and Computing and the highest 357 each publication was contributed by Jadavpur University and Anna University. The most productive funding agency's name was the University Grants Commission (UGC) with 49 publications. The most prolific author's name was B.K Panigrahi from New Delhi with 41 publications, followed by S. Das (ISI, Kolkata) with 39 research papers published in artificial intelligence research. The study finds that India's qualitative performance has not been as stunning as its quantitative performance in research. It is important that stakeholders pay special attention to academic and research organizations in the country encouraging them to undertake more and more new research projects, programmes in collaboration with international hubs in artificial intelligence research.