Abstract
Computer Science degrees are very popular currently among institutions worldwide. The proliferation of these programs in different universities has led to the creation of rankings for classifying programs according to their prestige and quality. However, these rankings do not specify the quality of research. This study develops a bibliometric overview of all the journals that are currently indexed in the Web of Science (WoS) database in any of the seven categories connected to Computer Science research. These categories include Artificial Intelligence, Cybernetics, Hardware and Architecture, Information Systems, Interdisciplinary Applications, Software Engineering and Theory and Methods. This study aims to identify the leading institutions over the last 25 years (1991–2015) in each area selected according to a wide range of bibliometric indicators. The results indicate that American universities are the most influential in Computer Science research. This study concludes that Computer Science traverses many institutions.
Introduction
During recent years, research on the productivity of institutions has proliferated. The ranks elaborated influence an institution’s reputation [23] and affect its ability to raise funds [2, 22]. These rankings are based on a multitude of factors, such as peer-review surveys, research, teaching, the size of the institution, the grants, etc. [24]. However, these rankings are criticized because they lack accuracy and only measure results, which implies that many indicators are left behind when ranking institutions [8]. This is why one must also take into account the article count method and the quality-weighted method. It counts citations as one of the main quality indicators of publications, although there are other indicators, such as the h-index and impact index, that can rank institutions [15].
Some of the bibliometric studies that perform ranks usually include incomparable methods [20, 30], as some of them are based on article counts instead of publication quality. Others use citations but ignore article counts.
Bibliometric studies have been carried out in a large number of fields. Among them, we can highlight the analysis of journals [6, 29], countries [3, 33], research topics [32] and institutions [4, 39]. In Computer Science, there are many bibliometric studies concerning a wide range of issues, including journals [9, 37], countries [13, 38] and institutions [36].
In a recent paper, the authors focused on analyzing the 100 most productive institutions in India, comparing the results obtained by these institutions with major world institutions. This analysis was based on research output data indexed in Scopus between 1989 and 2013.
However, no research has applied a bibliometric overview to all journals currently indexed in the WoS database in any of the seven categories connected to Computer Science research. Additionally, there are no existing tools for ranking universities according to their production of Computer Science research. As a consequence, no ranking includes other institutions such as companies and research centers. Attempting to fill this research gap, the bibliometric analysis in this paper is particularly useful since it identifies the main institutions that have published in the Computer Science field and enables the evaluation of output performed over the years by these institutions.
This paper aims to develop an analysis that identifies the most productive and influential organizations in Computer Science research. The specific objectives of this study are (1) to create a ranking of Computer Science organizations according to bibliometric indicators; (2) to create rankings of Computer Science organizations for each category of Computer Science; (3) to identify the most productive and influence institutions in Computer Science research; and (4) to identify the number of research institutions that are universities, companies and research centers.
The rest of the study is organized as follows. First, we present the methodology used in this article. Next, the results and discussion analyse the results of the paper. Finally, conclusions summarize the main findings, limitations and suggestions for future research.
Methods
There are different approaches to classifying the bibliographic material, with bibliometrics being one of the most used approaches. Bibliometrics is a research field that quantitatively studies bibliographic material by analyzing a research area and identifying its leading trends [26].
The source used for the selection of the panel of journals for this study was the WoS, which is considered the most influential database in the world since it collects high-quality articles recognized by the scientific community (This is without prejudice to other databases such as Scopus or Google Scholar.). The WoS includes information from more than 15,000 periodicals and 50,000,000 articles, which are ranked in 251 categories and 151 research areas [27].
By searching for keywords, WoS shows bibliometric indicators for the selected papers. This work searches for publication names in a period of time of 25 years (1991–2015). This time period is selected because it is representative of the latest developments in the field. Data collection was performed during the first half of 2016.
The WoS categorizes Computer Science into seven categories: Artificial Intelligence (106 journals), Cybernetics (21 journals), Hardware and Architecture (34 journals), Information Systems (106 journals), Interdisciplinary Applications (90 journals), Software Engineering (65 journals), and Theory and Methods (58 journals). In each case, the analysis considers a Global Ranking where all the institutions are ranked together. This allows comparisons between subfields when needed.
For each Computer Science category in the WoS, we search for the 30 organizations with the highest number of articles. Next, some bibliometric indicators are selected for each organization. Several indicators are considered, including the total number of papers, the total number of citations, the ratio of total citations/total paper, the h- index [16] and the number of papers that an institution has among the most cited ones for a specific field. The most cited papers are found in the WoS by ordering the results of a search from the most cited papers to the least cited [29].
In addition to the above indicators, two more indicators have been added that refer to quality indicators for universities, such as the Shanghai Academic Ranking of World Universities (ARWU) and Quacquarelli Symonds (QS). Although both the ARWU and QS have several critiques, including the difficulty of defining and measuring quality and their lack of objectivity and precision [23], they are useful for providing a reference on the global position of a university.
We have utilized several indicators in this paper, since there is no general agreement on the optimal method for measuring research. In fact, the current literature addresses many indicators that have appeared during the recent years [1, 29].
Results and discussion
This section presents the main bibliometric results found in the WoS for all Computer Science journals from 1991 to 2015. First, the paper analyzes the global results obtained in the Computer Science field. It then analyzes the individual results of each category that constitute the Computer Science field.
Global results
Figure 1 shows the typology of the different research organizations that have published in the Computer Science field. For a better understanding, we have classified these institutions into three categories: universities, companies and research centers. The difference between the last two categories is that research centers have the purpose of generating knowledge, while companies aim to commercialize goods and services.

Types of research organizations in Computer Science field.
As shown, universities are the institutions with a greater presence in each area that forms the Computer Science field. In fact, universities represent more than 80% of the most prolific institutions in each of the 7 areas that constitute the field. Therefore, it can be asserted that universities have a strong research presence in the Computer Science field. However, research centers and companies have a more residual presence, although these institutions also promote Computer Science research. This is due to the various idiosyncrasies of universities. Each year they welcome new students that contribute fresh perspectives to research. With respect to research centers and companies, it can be seen how their weight in the TOP 50 is quite similar, except in the Interdisciplinary Applications field, where there are no company contributions.
Table 1 shows the global results.
Global ranking
Source: Own elaboration. Abbreviations: (1) Total Papers (TP), which indicates the numbers of papers associated for each organization; (2) Total Citations (TC), that means the sum of citations obtained by the selected institution; (3) Total Citations/Total Papers (TC/TP) it is a ratio for the average number of citations for each published article; (4) h-index, which indicates X number of articles that have at least X number of citations [16]; (5) ≥250, that indicates the papers that have at least 250 citations; (6) ≥100, which specifies the number of papers that have at least 100 citations; (7) ≥50, that identifies the number of papers that have at least 50 citations.
Table 1 presents the 30 most productive institutions in Computer Science research. The ranking is listed in a descending order according to the number of citations. The top 5 institutions in the ranking constitute approximately 32% of all citations received by the TOP 30. Of these 5 institutions, there are three universities, one company and one research center, which shows the relevance of companies and research centers in Computer Science research. In this respect, these 5 institutions are very influential because they account for 22% of all publications made by the TOP 30.
The most productive institution is the Center National de la Recherche Scientifique CNRS, an institution that has 43% more publications than the institution that occupies the second place, which is the American International Business Machines (IBM). The University of California Berkeley has the best ratio of total citations to total papers presented. Massachusetts Institute of Technology (MIT) is the institution with the highest h-index and the highest number of publications that received more than 250, 100 and 50 appointments, respectively. All these institutions are part of the TOP 5 of institutions, ordered by the number of citations received.
In general, the institutions that constitute the TOP 30 worldwide by number of citations belong to business areas such as telecommunications and software development. The research that these institutions perform are usually improvements to the goods that they commercialize. In addition, these institutions have a wide presence in the ARWU and QS ranks, which implies that highly cited organizations are not necessary the most prestigious. Finally, the ranking presented shows a high dispersion between different institutions, indicating that Computer Science research transverses many organizations.
Next, we discuss the individual results for each category of the Computer Science field.
According to the definition of the WoS, “Computer Science, Artificial Intelligence” covers resources that focus on research and techniques to create machines that attempt to efficiently reason, problem-solve, use knowledge representation, and perform analysis of contradictory or ambiguous information. This category includes resources on artificial intelligence technologies such as expert systems, fuzzy systems, natural language processing, speech recognition, pattern recognition, computer vision, decision-support systems, knowledge bases, and neural networks.
Artificial intelligence research has experienced a rapid growth over recent decades [31], both in terms of the number of papers and the number of citations, with an increasing number of articles.
Table 2 shows the most influential institutions in this category.
Artificial intelligence
Artificial intelligence
Source: Own elaboration. Abbreviations are available in Table 1.
Artificial Intelligence is the most influential area of Computer Science research, since it is the one that has received the greatest number of citations. It is also the second area with the greatest number of publications, just behind Interdisciplinary Applications. For this reason, the results for Artificial Intelligence do not differ from Global Results.
The institution that has received more citations and the highest h-index in this field is the Massachusetts Institute of Technology (MIT) with 81,953 citations and an h-index of 128, followed by the University of California Berkeley with 70,459 citations. Moreover, this university had the highest average citation rate per paper. The second-best h-index is achieved by Carnegie Mellon University (118), which is also the institution that is third in the number of citations. These three institutions dominate the number of citations in the field and comprise approximately 22% of citations. It is worth mentioning that the first five institutions constitute 32% of all Artificial Intelligence citations. Note that American institutions have more citations from published articles than other organizations from other countries.
Analyzing the T50, T100 and T250 ranking results, the Massachusetts Institute of Technology (MIT) is in the top positions with 253 papers in the T50 ranking of the category, followed by Carnegie Mellon University with 235 of the papers in the T50. The same institutions have 154 and 131 papers in the T100 ranking, respectively, and 63 and 51 papers in the T250 ranking, respectively. Note that all the institutions of the category have an h-index bigger than 52, reflecting its influence and productivity.
According to the definition of the WoS, “Computer Science, Cybernetics” includes resources that focus on the control and information flows within and between artificial (machine) and biological systems. Resources in this category draw from the fields of artificial intelligence, automatic control, and robotics. Cybernetics is the area with the fewest number of journals in the Computer Science field. In fact, the number of research papers published has experienced a slight increment between the time span of 15 years (1997–2011), presenting some ups and downs throughout the said time period [34].
Table 3 shows the most influential institutions among this category.
Cybernetics
Cybernetics
Source: Own elaboration. Abbreviations are available in Table 1.
The institution that has received more citations and the highest h-index in this field, as in the previous category, is the Massachusetts Institute of Technology (MIT) with 6,272 citations and an h-index of 42, followed by Stanford University with 5,219 citations. Moreover, this university had the highest average citation rate per paper. The second-best h-index is achieved by the Max Planck Society (36), an institution that is third in the number of citations. The top five institutions comprise of 32.7% of all citations in this field. Though Asian countries have led Cybernetics research [5], the most influent and productive organizations are American and British, which combine for 65% of the total number of citations received by the TOP 30 (American institutions accounts for 54.6% of all citations, while British institutions account for the remaining 10.4%). The remaining 9 countries with a presence in TOP 30 received a very small percentage of citations.
Analyzing the T50, T100 and T250 ranking results, the Massachusetts Institute of Technology (MIT) is in the top position with 34 papers in the T50 ranking of the category, followed by the Max Planck Society with 26 of the papers in the T50. The same institutions have 16 and 11 papers, respectively, in the T100 ranking, and 4 and 1 papers, respectively, in the T250 ranking. Note that all the institutions of the category have an h-index higher than 15, the lowest of all categories analyzed, reflecting the small influence that this area has in the Computer Science field.
According to the definition of WoS, “Computer Science, Hardware & Architecture” covers resources on the physical components of a computer system: main and logic boards, internal buses and interfaces, static and dynamic memory, storage devices and storage media, power supplies, input and output devices, networking interfaces, and networking hardware such as routers and bridges. Resources in this category also cover the architecture of computing devices, such as SPARC, RISC, and CISC designs, as well as scalable, parallel, and multi-processor computing architectures. The number of papers published in this area has experienced uneven growth. In fact, according to [35], it appears to be an area of increasing irregularity.
Table 4 shows the most influential institutions among this category.
Hardware and architecture
Hardware and architecture
Source: Own elaboration. Abbreviations are available in Table 1.
The institution that has received the most citations is AT&T. This institution, together with the University of California Berkeley, are the ones with the highest h-index (77). The University of California Berkeley also had the highest average citation rate per paper. The top five institutions comprise 35% of all Hardware & Architecture citations. The United States has produced the highest number of papers [17]. In all, 80% of organizations in this ranking are American universities. These universities constitute the first 200 positions of both the ARWU and QS rankings, meaning that highly prestigious American universities are the most influential in this area.
Analyzing the T50, T100 and T250 ranking results, AT&T is in the top position with 126 papers in the T50 ranking of this category, followed by International Business Machines (IBM) with 124. In contrast, the University of California Berkeley is the institution with the largest number of papers in the T100 and T250, with 57 and 18 papers, respectively. Regarding the T250, the University of California Berkeley shares the lead with the Massachusetts Institute of Technology (MIT).
All the institutions of this category have an h-index higher than 33.
Information Systems is third most important area in Computer Science research by number of citations. According the WoS, “Computer Science, Information Systems” covers resources that focus on the acquisition, processing, storage, management, and dissemination of electronic information that can be read by humans, machines, or both. This category also includes resources for telecommunications systems and discipline-specific subjects such as medical informatics, chemical information processing systems, geographical information systems, and some library science.
Table 5 shows the most influential institutions in Information Systems field.
Information systems
Information systems
Source: Own elaboration. Abbreviations are available in Table 1.
The institution that has received the most citations and the highest h-index in this field is Stanford University with 52,764 citations and an h-index of 95, followed by the Massachusetts Institute of Technology (MIT) with 48,630 citations and an h-index of 91. The University of California Berkeley had the highest average citation rate per paper (52.99), followed very closely by Stanford University and the Massachusetts Institute of Technology (MIT) with 51.83 and 50.34, respectively. These three institutions are the only ones that exceed a ratio of 50 citations per paper. The first five institutions dominate the number of citations in the field with almost 34% of all citations.
American institutions dominate both the number of publications and citations received in this field. These institutions have produced 72% of all publications and have received almost 81% of all citations received in the TOP 30. The other countries that participate in this ranking have a rather testimonial presence. In fact, only 6 countries, in addition to the United States, have a presence in the TOP 30.
Analyzing the T50, T100 and T250 ranking results, Stanford University is in the top position in the T50 and T100 rankings, with 188 and 89 papers, respectively, followed by the Massachusetts Institute of Technology (MIT) with 161 and 84 papers, respectively. This last institution leads the T250 ranking with 38 publications. All the institutions of the category have an h-index higher than 40. Additionally, the universities that are in this classification are prestigious, according to the ARWU and QS rankings. It worth noting that the most productive organizations are not necessarily the most influential, such as with the Centre National de la Recherche Scientifique (CNRS). This means that the articles published in Information Systems by the different organizations are highly cited.
According to the WoS, “Computer Science, Interdisciplinary Applications” includes resources concerned with the application of computer technology and methodology to other disciplines, such as information management, engineering, biology, medicine, environmental studies, geosciences, arts and humanities, agriculture, chemistry, and physics.
This area is the most productive in Computer Science research, since it is an interdisciplinary area that embraces articles from many knowledge areas.
Table 6 shows the most influential institutions among this category.
Interdisciplinary applications
Interdisciplinary applications
Source: Own elaboration. Abbreviations are available in Table 1.
This area is the only one that does not have any companies among the 30 most productive organizations. One of the reasons for this is that journals of this category are multidisciplinary, so companies like International Business Machines (IBM), Microsoft or AT&T publish articles in more specialized areas of Computer Science.
The institution that has received more citations is the United States Department of Energy (DOE). Additionally, this institution had the highest h-index (123), with values much higher than the rest of universities, and leads the T50, T100 and T250 ranking results with 355, 160 and 58 papers, respectively. The University of California Los Angeles had the highest average citation rate per paper.
The top five institutions constitute 34% of all citations. The United States is the most productive country and the one that receives the most citations. Next are France and the UK. In this regard, it should be noted that this category has the largest number of countries within the TOP 30. Specifically, universities from 11 countries are part of the TOP 30.
Note that all the institutions of the category have an h-index higher than 49.
The objectives of this research area are software principles and techniques [12]. According to the WoS, “Computer Science, Software Engineering” includes resources that are concerned with the programs, routines, and symbolic languages that control the functioning of the hardware and direct its operation.
Table 7 shows the most influential institutions among this category.
Software engineering
Software engineering
Source: Own elaboration. Abbreviations are available in Table 1.
The institution that has received more citations and the highest h-index in this field is International Business Machines (IBM) with 84,602 citations and an h-index of 133, followed by the Massachusetts Institute of Technology (MIT) with 48,240 citations and an h-index of 99. It can be seen how the first institution practically duplicates the number of citations of the second. Stanford University had the highest average citation rate per paper (36.29), followed very closely by the Massachusetts Institute of Technology (MIT) (35.84). The first five institutions dominate the number of citations in the field with 39% of all citations.
American institutions dominate research in this field. In fact, 24 of the 30 institutions that appear in the TOP 30 are from the United States. In addition, American institutions that occupy the TOP 10 receive 57% of all TOP 30 appointments. Despite this, [11] indicated that over a 40-year period (1968–2009), approximately 60% of the Software Engineering literature has been contributed by only 7% of all countries, and the share of American contributions to this discipline has declined from 71.43% (in 1980) to 14.90% (in 2008). Meanwhile, China is the country with the biggest share of growth in the number of publications (from 0.82% of the entire Software Engineering publications in 1991 to 13.82% in 2009).
Analyzing the T50, T100 and T250 ranking results, International Business Machines (IBM) is in the top position in the T50, T100 and T250 rankings, with 367, 179 and 44 papers, respectively, followed by the Massachusetts Institute of Technology (MIT) with 207, 98 and 34 papers, respectively. All institutions of this category have an h-index larger than 44 It is also relevant to note that only the first 10 organizations of the ranking have more than 10 articles with more than 250 citations.
According to the WoS, “Computer Science, Theory & Methods” includes resources that emphasize experimental computer processing methods or programming techniques such as parallel computing, distributed computing, logic programming, object-oriented programming, high-speed computing, and supercomputing.
Table 8 shows the most influential institutions among this category.
Theory and methods
Theory and methods
Source: Own elaboration. Abbreviations are available in Table 1.
The institution that has received more citations is the Centre National de la Recherche Scientifique (CNRS) with 30,400 citations, followed by International Business Machines (IBM) with 26,056 citations. With respect to the h-index, the two previous institutions also lead the rankings but exchange their positions. Stanford University had the highest average citation rate per paper (37.73), followed by the University of California Berkeley (29.22). The first five institutions dominate the number of citations in the field with 33% of all citations.
Although American institutions dominate this field, the presence of European institutions is noteworthy. This implies that Asian countries are less influential and productive in this area.
Analyzing the T50, T100 and T250 ranking results, International Business Machines (IBM) is in the top positions of the T50 and T100 (with 120 and 44 papers, respectively), while the University of California Berkeley leads the T250 ranking (with 16 papers).
Although all the institutions of this category have an h-index higher than 25, this research area is the one with the lowest average of citations per published article.
This study aims to characterize Computer Science research institutions indexed in the WoS database between 1991 and 2015. This analysis illustrates the results obtained under a wide range of bibliometric indicators. It is very useful to see the general results from a broader perspective than the Journal Citation Reports of the WoS. Particularly, this is very useful for PhD students and newcomers in the field in order to obtain a general orientation of the leading institutions in Computer Science.
The results provide a general picture of the current position of the leading institutions in this field for the seven categories analyzed. In all, 78 institutions have a presence in the elaborated ranks. Of these, only 5 institutions (Massachusetts Institute of Technology (MIT), University of California Berkeley, Stanford University, Centre National de la Recherche Scientifique (CNRS) and Carnegie Mellon University) appear in the eight elaborated ranks. Of these institutions, 4 are from the United States and 1 from France. This last institution is a research center, while the first four institutions are universities.
Another 6 institutions have a presence in seven of the eight ranks analyzed. Of these institutions, 2 were not ranked among the most influential for Interdisciplinary Applications (International Business Machines (IBM), University of Maryland College Park), two in the Cybernetics (University of Texas Austin and University of Illinois Urbana Champaign), and two in information system (INRIA and Purdue University). Of these institutions, four are universities, one is a research center and the other one is a company.
The 11 institutions mentioned above are present in the overall ranking. Four of these institutions hold the top five positions in the ranking, while 7 of them are in the Top 10. Another 3 institutions are located between positions 11 and 20 of that ranking, and the last institution are locate in the 28th position.
The Massachusetts Institute of Technology (MIT), University of California Berkeley, Stanford University and International Business Machines (IBM) have the highest number of citations, which shows the largest absolute influence. However, some other institutions achieve better results when looking to other indicators. For example, according to the number of papers, the leading institution is the Centre National de la Recherche Scientifique (CNRS). This indicates that the institutions considered in the study have different profiles with a wide variety of objectives.
The results also indicate Artificial Intelligence is the most influential category, although Interdisciplinary Applications is the largest. Another important fact is that 31 institutions only have a presence in a single ranking of those presented. Of these, 35% are only present in the Cybernetics Theory and Methods category, while the Information Systems category is the one with the smallest ratio (3%) of institutions not present in other ranks.
American universities are the most influential of those investigated, but there is a wide range of institutions that published in this area. Companies and research centers from different countries like France, Germany, China, Japan or South Korea are important in research. However, there are no South American or African countries in the rankings.
Regarding the three continents represented in the different rankings elaborated, Centre National de la Recherche Scientifique (CNRS) is the most important institution from Europe, Technion Israel Institute of Technology from Asia and Massachusetts Institute of Technology (MIT).
The global rankings also show that research centers and companies have little published presence in the Computer Science field, but they are in high positions of the presented rankings. In fact, universities represent 63% of all published papers, while research centers and companies account for 31% and 6%, respectively.
In conclusion, the findings and suggestions of this study can help scientific researchers understand the performance and trends of Computer Science research. With the help of these findings, researchers can make informed decisions regarding their research directions in terms of identifying top disciplinary institutions and choosing research exchange platforms.
Some of the limitations of our paper are similar to those found in other papers [29]. First, this study only accounted for journals indexed in the WoS without considering other databases. However, the material published in the WoS may be sufficiently representative of state-of-the-art research for the analyzed fields.
Despite these limitations, the paper provides a starting point for future bibliometric studies in these fields. Future research could include bibliometric studies focused on the identification of the main authors, countries and published documents.
