The collaboration network of the Brazilian Symposium on Databases

Abstract

The Brazilian Symposium on Databases (SBBD) celebrated its 30th edition in October 2015. As the database community has evolved over the years, so has the data analysis area. To celebrate such accomplishments, this article goes over the SBBD history from distinct social perspectives. Overall, we investigate the complete SBBD co-authorship network built from bibliographic data of SBBD’s 30 editions, from 1986 to 2015, and analyze several network metrics, considering the network evolution over the three decades. In particular, we analyze the progress of the most engaged SBBD authors, the number of distinct authors, institutions, and published papers, and the evolution of some of the most frequent terms presented in the titles of the papers, as well as the influence and impact of the most prominent SBBD authors.

Introduction

The Brazilian Symposium on Databases (SBBD) celebrated its 30th edition in October 2015. Besides specific scientific meetings held during the Annual Congress of the Brazilian Computer Society (CSBC), SBBD is only the second national scientific event promoted by SBC to accomplish 30 editions (the Brazilian Symposium on Computer Networks and Distributed Systems – SBRC, celebrated its 30th anniversary in 2012 [22]). As the database community has evolved over the years, so has the data analysis area. Most prominently, the area of social networks analysis (SNA) has recently bloomed by automatically applying a number of specific metrics over big volumes of data.

In order to celebrate such accomplishments and taking advantage of recent advancements in SNA, this article goes over the SBBD history from distinct social perspectives. Specifically, the main objective is to present a deep analysis of the Brazilian database community based on the publications at the SBBD proceedings and its associated journal, the Journal of Information and Data Management (JIDM). We achieve this by investigating the complete SBBD co-authorship network built from bibliographic data of SBBD’s 30 editions, from 1986 to 2015, and analyzing several network metrics, such as degree, density and assortativity, as well as considering the network evolution over the three decades. In particular, we analyze the most engaged SBBD authors, the number of distinct authors, institutions, and published papers, and the evolution of some of the most frequent terms presented in the titles of the papers. Finally, we discuss the productivity, influence, and impact of SBBD authors through centrality measures and show that the SBBD network follows a phenomenon typical in social networks known as small-world [35].

The remaining sections of this article are organized as follows. The “Related work” section covers related work on academic collaboration. The “Background informationBackgroundinformation” section provides background information on data acquisition, network modeling, and key metrics considered in our study. The results of our analyses are presented from the “Basic statistics” section to the “Homophily” section, covering (respectively) statistics of the network and authors, collaborations and newcomers, communities, influential authors, and homophily. Finally, the “Conclusions” section summarizes the main conclusions and possible future directions.

Related work

Collaboration networks analyses have been well explored to reveal interesting features of academic communities. Newman presents the first studies in this area [27, 28]. He presents distributions of collaborators and their clusters and characterizes different patterns of collaboration between distinct fields. Then, Newman [29] answers a broad variety of questions about collaboration patterns, presenting several structural and topological features from bibliographic databases in biology, physics, and mathematics. Likewise, other studies build and analyze academic networks by countries [16, 24], universities [7, 16, 21], researchers [3, 19], subareas [18, 19], and venues [2, 5, 22, 26, 31, 32].

There are also studies that regard Computer Science (CS) as a specific area. For example, Freire and Figueiredo [10] characterize the structural properties of the collaboration network in CS by exploring external collaborations of groups and individuals. Lima et al. [18] propose ranking authors across multiple research areas by characterizing the profile of top Brazilian researchers. Mena-Chalco et al. [23] study the Brazilian co-authorship network by exploring topological metrics.

In this article, we perform a deep analysis of the Brazilian database research community by evaluating the collaboration network of the Brazilian Symposium on Databases on the occasion of its 30th edition. Considering other specific database conferences, Nascimento et al. [26] analyze the co-authorship network of the ACM SIGMOD International Conference on Management of Data, whereas Ameloot et al. [2] go over the 30 years of history of the ACM Symposium on Principles of Database Systems (PODS).

Regarding other CS areas, Duarte et al. [8] analyze the works presented at the the Brazilian Symposium on Multimedia and the Web in the period 1995–2012 in order to provide an overview of its community and show how its research topics have evolved over time. Likewise, Neto et al. [11] investigate the impact of international research in the Brazilian Symposium on Software Engineering (SBES) by analyzing its first 24 editions. Moreover, Smeaton et al. [33] analyze the co-authorship network and research topics at the 25-year celebration of the International ACM SIGIR Conference on Research and Development in Information Retrieval. Liu et al. [20] focus on the first decade of the Digital Libraries community by analyzing the coauthorship network of past ACM, IEEE, and joint ACM/IEEE digital library conferences. Similarly, Maia et al. [22] study the collaboration network of the Brazilian Symposium on Computer Networks and Distributed Systems (SBRC) over its 30 editions. Besides analyzing its co-authorship network, they also consider several network properties to describe the kinds of collaborations found and identified the main communities within SBRC.

Similar to the aforementioned works, here, we analyze SBBD’s co-authorship network statistics, including average of articles by author, articles by edition, and coauthors by article. We also evaluate several structural and temporal characteristics, such as the main type of co-author relationship among authors, the most prominent communities within SBBD, the collaborations and newcomers on the network, and the homophily of SBBD from two perspectives: affiliation and gender.

The SBBD community was briefly studied by Procópio et al. in a short paper [31] on the occasion of its 25th anniversary, with an analysis of both its structural characteristics and temporal evolution. Besides including the data of the last 5 years and contributing to a more thorough analysis of the SBBD network, we provide a deeper investigation underlying social perspectives such as collaborations between authors in this network, communities assembled inside SBBD, and authors with central roles in the network. Hence, we considerably expand the previous work, not only in the time interval considered but also in the many analytical dimensions added.

Background information

This section describes background information on data acquisition, network modeling, and key metrics considered in our study. Given the importance of data acquisition in this kind of analysis, we start by describing our dataset building process.

Dataset

For a more complete study, we consider both publication statistics and co-authorship social perspectives. Therefore, our dataset comprises bibliographic data of SBBD’s 30 editions from 1986 to 2015, which includes for each paper: its title, year of publication, list of authors with their respective affiliations, and the language the paper was written. As each SBBD edition is unique in terms of its program and aiming at a more robust analysis, we consider only full papers actually presented at the symposium (i.e., our dataset disregards short papers, demos and tools, tutorials and keynotes, and workshop papers).

To ensure continuity, the dataset analyzed by Procópio et al. [31], comprising SBBD’s first 25 years, was extended with data from the remaining 5 years (2011–2015) as collected from BDBComp¹, the Brazilian Digital Library of Computing [15]. For consolidating our dataset and keeping data consistency, we have also performed name disambiguation to avoid splitting the publications of one person under two different author names.

Note that, from 2010 to 2014, SBBD full papers were published as articles in the Journal of Information and Data Management - JIDM², whereas the SBBD proceedings included short papers, demos and workshop papers. Then, the 2015 edition published full papers in the proceedings (mostly in Portuguese) and articles in JIDM as well. JIDM has also published full versions of invited SBBD short papers, as well as invited papers from other conferences such as the Brazilian Symposium on Geoinformatics (GeoInfo), the Symposium on Knowledge Discovery, Mining and Learning (KDMiLe), and the Brazilian Symposium on Multimedia and the Web (WebMedia). Thus, seeking a more round analysis, for the 2010–2015 period, we consider only full papers from SBBD proceedings and all SBBD-related papers from JIDM (both regular articles and invited full versions of short papers). Selected papers from the other conferences are not considered, as they do not reflect publications from the SBBD community, the focus of our study.

Network model

Following Maia et al. [22], in this study, we also represent the SBBD network as a temporal graph G _y =(V _y ,E _y ), where V _y is the set of vertices, E _y is the set of edges, and y is the year a network refers to. The graph G _y =(V _y ,E _y ) is an undirected weighted graph, where the vertices represent authors, and the edges indicate that two authors have published together in or before the year y. In addition, each edge has a corresponding weight based on the weighted collaboration network proposed by Newman [27]: the weight for each edge is discounted by the size of the collaboration, according to the formula $w = \sum _{p} \frac {1}{N_{p}-1}$, where N _p is the number of authors of paper p. For instance, given a,b,c,d∈V _y and assuming that a and b wrote a single paper with no other co-author, and b and c have a joint co-author d. Then, the weight of the edge (a,b) is 1.0, while the weight of the edges (b,c), (c,d), and (b,d) is 0.5, as the weights among nodes are calculated by the number of authors for each paper. Figure 1 shows the complete SBBD network as viewed in 2015, representing 30 years of history.

Fig. 1

The complete SBBD undirected co-authorship network

The complete SBBD collaboration network, built from all papers published in its 30 editions, has a total of 1034 authors (vertices) and 2299 collaborations (edges), comprising a total of 674 papers. The largest connected component (LCC) (the largest subgraph in which any pair of nodes is connected by paths) has 781 nodes, representing 75.53% of the whole community. Then, 162 nodes compose smaller components containing three to seven authors. Finally, there are 80 nodes that form pairs of authors, and 11 nodes that correspond to sole authors.

The average number of papers per year is 22.47 (with a standard deviation of 4.86), and the average number of authors per year is 60.53 (with a standard deviation of 17.53). Finally, the average number of papers per author is 1.98 (with a standard deviation of 3.21), while the average number of authors per paper is 3.04 (with a standard deviation of 1.47).

Network metrics

Several metrics characterize and enable to investigate collaboration networks as the one studied here. This section summarizes the definition and applicability of the metrics used in this work.

The degree of a vertex is the number of its adjacent edges. Hence, authors who have high number of papers with different co-authors also have high degree. The degree distribution is the probability distribution of these degrees over the whole network. The network density is calculated by dividing the number of edges by the number of nodes present in the graph. The assortativity measures the similarity of connections in the graph with respect to the node degree. High coefficient means that authors of high degree tend to connect to authors of high degree (assortative network).

A connected component of an undirected graph is a subgraph in which any two vertices are connected to each other by paths. The size of such a component is given by dividing the number of its nodes by the total number of vertices of the graph. The average path length is the average number of steps along the shortest paths for all possible pairs of network nodes. The graph diameter is the length of the longest shortest path between all pair of vertices.

Betweenness and closeness regard the centrality of nodes in the network. The former is equal to the number of shortest paths from all vertices to all others that pass through a given node, whereas the latter measures how close a vertex is to all other vertices in the graph. The clustering coefficient CC(x) of a vertex x is the ratio between the number of edges among the neighbor-set of x and the total possible number. The clustering coefficient of the whole network is the average CC(x) over all x. Finally, homophily is the tendency of authors to interact with others with similar features. We refer to Albert and Barabási [1] and Costa et al. [6] for further information and complete definitions of all these network metrics.

Basic statistics

We start our study by discussing the SBBD growth. Figure 2 presents the evolution of the number of distinct authors (Fig. 2 a), institutions (Fig. 2 b), and published papers (Fig. 2 c) per year. The regression line (which includes a 95% confidence region) of Fig. 2 a shows that the number of authors has increased over the years, which suggests that SBBD has been attracting the contribution of new authors. The Pearson’s correlation coefficient³ between the number of authors and institutions per year indicates a high correlation (0.75, p value = 1.493e-06). Thus, we can go further and say that, besides attracting the participation of new authors, SBBD is also attracting authors from new institutions, which contributes largely to the symposium scientific strength and reachability.

Fig. 2

Evolution of the number of distinct (a) Authors, (b) Institutions, and (c) Papers

On the other hand, the sharp decrease in the number of authors and papers at the most recent editions might be due to different reasons. Most of all, if a researcher wants to publish in a conference, there are many options for possible venues of interest. Specifically in Brazil, there are at least two clear examples of other conferences whose topics of interest intersect with SBBD. One is WebMedia, which was initially restricted to multimedia and hypermedia systems (in Portuguese: Simpósio Brasileiro de Sistemas Multimídia e Hipermídia) and then evolved to Multimedia and the Web in 2003. As databases have also evolved to the Web, now, SBBD and WebMedia have common topics of interest such as Semantic Web, Linked Data, and Ontologies. Another is KDMiLe, with its first edition in 2013, specialized in topics related to data mining and knowledge discovery that were mostly covered by SBBD. Nonetheless, we can only speculate, as a true answer could only come from interviewing the authors of papers whose topics are related to SBBD but are published elsewhere.

Figure 3 presents the SBBD network density over the years. The network density is calculated by dividing the number of edges by the number of vertices present in the graph. It achieves its highest value in the first years of the symposium. This can be explained by an unusual paper with 18 authors published in 1989 [34]. After that, as the number of authors increases, the density starts reducing, but starts growing again from 1996 on. The figure also shows the network density without considering the paper with 18 authors, which in fact represents a collaboration pattern outlier. The density presents some oscillation in the first decade, but with a clear increasing tendency later, possibly explained by the arrival of new authors that collaborate with others who are already in the network.

Fig. 3

Network density over time

Tables 1 and 2 display the distributions of authors per paper and papers per author. Most SBBD authors (71.76%) have only one paper published throughout the years, and about 28% have two or more papers. Single-paper authors comprise students who most likely published with their supervisors, authors who have published just once, and employees from companies, for instance. Despite the high number of single-paper authors, some of the authors are really engaged (70 authors published five or more papers) and have published consecutively over the years. Currently, the record holder is Marcos André Gonçalves with a streak (uninterrupted period of publishing) of 11 years (2004–2014). He is followed by three authors with streaks of 9 years: Rubens Nascimento Melo (1992–2000), Alberto H. F. Laender (1994–2002), and Caetano Traina Jr. (2003–2011). This distribution shows that only 13.25% of the authors have been continuously responsible for the core publications.

Table 1 Distribution of authors per papers published

Table 2 Distribution of papers per number of authors

Using the number of publications, Table 3 shows the top 20 authors considering the 30 years of the symposium, which is also illustrated in Fig. 4 by the authors’ name cloud. Then, Fig. 5 illustrates the most engaged SBBD authors by the number of publications separated by decades. Some authors who had a high number of publications during the first years do not have the same participation in the last years. For example, Rubens Nascimento Melo is one of the founders of the SBBD community who is already retired.

Fig. 4

Name cloud of SBBD authors

Fig. 5

Evolution of the most enganged SBBD authors over the three decades (a) 1986-1995, (b) 1996-2005, and (c) 2006-2015

Table 3 Top twenty most prolific SBBD authors ranked by number of papers

SBBD is a national symposium targeted at the Brazilian database research community. Despite that, more than half of its papers (50.37%) were published in English in its 30-year history. Figure 6 shows the number of papers published in English over the years, compared with the number of papers published in Portuguese. Note that, over the period 2010–2013, all SBBD papers were published in English. This period corresponds to the first 4 years of JIDM, when all SBBD full papers were submitted in English and the accepted ones published in this journal. This was an important effort of the SBBD community to make its results more internationally visible. With an international Editorial Board and a fast track submission scheme, the JIDM effort produced very good results in terms of the quality of the accepted papers, but considerably reduced the number of submissions to SBBD, particularly those by younger authors. Thus, in 2014, SBBD resumed its traditional submission scheme, but including in its technical program all JIDM database-related articles published in the same year.

Fig. 6

Number of papers published in English and in Portuguese over time

Regarding the main topics of interest covered by SBBD, the study reported by Kauer and Moreira [14] presents the evolution of research keywords extracted from both title and abstract of SBBD papers in the first 25 editions. For the sake of completeness, we have analyzed the terms present in SBBD paper titles for all 30 editions. Specifically, Fig. 7 illustrates the main words found in the titles of the papers in Portuguese (Fig. 7 a) and in English (Fig. 7 b). Most words, in both languages, are typical database-related terms, which is expected since SBBD is the official database event of the Brazilian Computer Society. In order to further investigate the evolution of these terms and visualize how Brazilian data-related research has evolved over the last 30 years, Fig. 8 highlights the amount of papers per year associated with 15 of the most frequent words and terms.

Fig. 7

Word clouds from the paper titles (a) in Portuguese, and (b) in English

Fig. 8

Evolution of some of the most frequent terms over time, as collected from titles of papers both in Portuguese and English

Some terms, such as databases and data, received a high number of mentions over all the years. Particularly, database studies usually focus on different aspects of the data life cycle, such as modeling, designing, and managing spatial, object, distributed, and relational databases. The appearance of specific terms such as XML, mining, and Web from 2000 onwards corroborates the rise of these topics in Computer Science. Figure 8 also shows that some related terms do not keep high growing rates. For example, the term object-oriented achieves its highest number of titles in mid 1990s, decreasing its importance afterwards most probably because the idea of an object-oriented database was replaced by the more realistic solutions over object-relational systems.

Collaborations and newcomers

In this section, we discuss the SBBD authors’ collaboration networks from two points of view: existing collaborations and newcomers (SBBD first time authors).

Figure 9 shows the values for mean, skewness, and variance of the degree distribution over the years. Both skewness and mean tend to stabilize, with similar behavior from 2000 on. Note that, skewness and variance values are significantly large, indicating that some authors possess a high degree. We note that such trends also happen in SBRC [22].

Fig. 9

Three moments of the degree values over the years

Using Newman’s relation weight scheme, the range of a single collaboration in all editions of SBBD varies from 0.06 to 8.72. The former is due to (again) the unusual paper with 18 authors published in 1989, whereas the latter is product of a long-term collaboration between Agma J. M. Traina and Caetano Traina Jr., who coauthor 26 papers.

At SBBD, papers are mostly published by two authors (38.87%). Publications from single authors took place from 1986 to 2003, comprising only 4.90% of all papers. Moreover, the number of single-author papers (Fig. 10) has decreased over the years, thus suggesting that SBBD authors are collaborating more. Indeed, 53.33% of the authors who published a single-author paper have published at least one joint paper after it.

Fig. 10

Number of single-author papers by year

The average degree of the network is 4.45, i.e., on average, an SBBD author collaborates with many distinct coauthors. The degree distribution (Fig. 11) follows a power-law, i.e., there are few authors with high degree (the highest being Marcos A. Gonçalves with 68), and most of the authors have only few collaborations. This is a common feature of many complex networks, closely related to the rich-gets-richer effect [4].

Fig. 11

Degree frequency

A further investigation into collaborations shows that the SBBD network is non-assortative, with a value close to 0 (-0.06). Figure 12 shows the time evolution of assortativity and how predominant this kind of collaboration is becoming. This indicates that nodes with high degree tend to connect to low degree nodes, i.e., new authors with few collaborations are increasingly tending to connect to authors with higher degree. This kind of behavior is usually observed between students and their supervisors.

Fig. 12

Assortativity over time

We also notice that, at early years, the number of newcomers connected to the largest connected component was close to the number of newcomers connected outside it, probably because this component was not large enough. Figure 13 shows that SBBD initially received several authors with no link to its largest connected component, but recently, it has become more common to join the symposium collaborating with someone already connected to it.

Fig. 13

Number of newcomers connected to the largest connected component and outside it with at least one reappearance

An important question is whether a newcomer who joins SBBD as a part of the LCC has more chance of reappearing than an author who joins it outside this component. Table 4 shows the number of newcomers with at least one reappearance in the symposium in the first and in the last 15 years. Most of the authors who have joined the symposium in the first 15 years and reappeared at least once did not join it directly into the LCC (around 81%). However, this has changed in recent years, as authors who have more than one SBBD paper usually contribute to a paper with a well-connected SBBD author first (62.18% of reappearances from 2001 to 2015 are from authors who joined the SBBD community in collaboration with an author already in the LCC). In addition, the Pearson’s correlation coefficient between the number of newcomers connected to the LCC and the density of the graph (0.46, p value = 0.009) indicates a slight positive correlation. Such a result suggests that the network is becoming more dense as the number of newcomers connected to the LCC increases. This reinforces the increasing tendency of the density values observed from 2000 onward in Fig. 3.

Table 4 Number of newcomers connected to the LCC and connected to nodes outside the largest connected component

Analysis of the collaboration groups

We now take a further look into the SBBD collaboration groups by using social network metrics and the k-clique algorithm [30].

Clustering coefficient

The SBBD network has a high global clustering coefficient of 65.6%, indicating dense groups of authors who publish together (PODS, for example, achieved only 35% in its 30th edition in 2011 [2]), indicating that there is a stronger triadic closure effect [12] in SBBD. Even though the number of authors in the network has substantially increased, new authors tend to collaborate with coauthors that maintain previous collaborations, establishing triangle network motifs. In other words, previous collaborations increase the chance of a pair of authors establishing a collaboration with a newcomer in the network⁴.

Figure 14 shows the evolution of the average shortest path length and the clustering coefficient of SBBD, as well as their equivalent random networks. A high clustering coefficient (compared to its equivalent random network) and a small average shortest path (as low as its equivalent random network) characterize the SBBD network as a small-world network [25]. This phenomenon has long been subject of scientific studies and is typical in social networks [35].

Fig. 14

Comparison to random networks for (a) average shortest path length and (b) clustering coefficient

Table 5 shows the size (number of vertices) of the two SBBD largest connected components over the years, and Fig. 15 the evolution of their relative sizes. The LCC for the 30 editions has 2004 edges, with a relative size of 87.17%. It grows with an average of approximately 26 new nodes yearly over the 30 years, 32 over the last 20 years, and 41 over the last 10 years, reaching 75.53% of all nodes in 2015 (13.06% in 1995 and 36.07% in 2005). These values confirm that most of the SBBD authors are connected through a single component, with the second largest connected component (SLCC) having only 22 edges, with a relative size of 0.96%. The remaining cases correspond to smaller components and authors who did not collaborate with other authors. There are only 11 cases of sole authors (see Fig. 1), whose papers were published between 1986 and 2003 (see Fig. 10). It is worth noticing that most of such papers are short communications describing projects developed by large technology companies (e.g., Telebrás and Embratel) or present results of their authors’ thesis or dissertation.

Fig. 15

Relative size of the largest connected component (LCC) and second largest connected component (SLCC)

Table 5 Sizes of the largest and second largest connected components over the years

Formation of collaboration groups

Decomposing a complex network into groups (sets of highly connected nodes) is very important, as it may help to understand a-priori unknown features and properties of the network. In this section, we focus on discovering and analyzing collaboration groups inside SBBD.

Specifically, we use the k-clique algorithm [30] that relaxes the notion of clique and has shown great success in detecting clusters on a large scale. A collaboration group is then defined as the maximal union of k-cliques that can be reached from each other by a series of adjacent k-cliques (they are adjacent if they share k-1 nodes). The k-clique algorithm returns a large number of groups inside SBBD, thus showing that collaboration has greatly increased over the years. According to Fig. 16, there are currently 148 (with k = 3), 116 (with k = 4), and 75 (with k = 5) groups formed.

Fig. 16

Number of 3-cliques, 4-cliques, and 5-cliques over the years

Despite the high number of groups, they do not include many authors. In fact, most of them are really small groups: 51.4% of the 3-cliques have only three authors. The largest 3-clique has 230 nodes and contains the biggest 4-clique inside it (61 nodes). Considering 5-cliques, the number of authors in the largest, second largest, and third largest groups are respectively 18, 12, and 12.

Using the top three groups found by the k-clique algorithm with k = 4, which only uses topological features of the graph, there are clusters in which most authors belong to one or at most two distinct institutions, as shown in Table 6. The largest group found is composed by authors from the states of Minas Gerais and Amazonas (due to a longstanding collaboration between Alberto H. F. Laender from UFMG and Altigran S. da Silva from UFAM). On the other hand, the second and third largest groups are mainly composed by authors from the states of São Paulo and Rio de Janeiro, respectively. This indicates that the geographical location and the affiliation of an author is a strong factor to determine which group this author belongs to in the SBBD community.

Table 6 Predominant affiliations of k = 4 groups

Influential authors

In this section, we analyze the productivity, influence, and impact of SBBD authors. We start by ranking them based on the structural information of the network. Tables 7 and 8 show authors ranked by the number of collaborators and Newman’s weight [27]. Even though the rankings differ when considering only the last 10 years, the tables include widely known and prolific authors in the community. The size of the intersection between these ranks is high, as there are eight authors who appear in both tables. Since prolific authors tend to have a higher number of collaborators, such a correlation between the metrics is expected.

Table 7 Top ten degree

Table 8 Top ten Newman’s weight

We can also use centrality metrics to discover whether the location of a node in the social network is strategic or not. Here, we consider both betweenness and closeness [1]. Tables 9 and 10 display the top 10 authors according to such metrics. The betweenness of a node measures the percentage of all shortest paths in a network that passes through that node. Hence, the idea is that there is a higher probability of information being disseminated through nodes that lie on more shortest paths than those that do not. On the other hand, closeness measures the inverse of the sum of distances to all other nodes, which allows to estimate how close a node is to all others.

Table 9 Top ten betweenness

Table 10 Top ten closeness

Both tables show prolific and well-known authors in the SBBD community. However, some prolific authors are not listed in any of the two rankings (e.g., Agma J. M. Traina who appears in Tables 7 and 8). Thus, we should evaluate if these metrics are indeed accurate for finding influential authors. For instance, Fernanda Lima has only two publications in SBBD, but she appears in Table 9. An attentive analysis explains such a phenomenon, since both of her publications are joint papers with influential authors in the network such as Marta L. de Queirós Mattoso and Rubens Nascimento Melo. For this reason, her overall distance in the network has decreased after collaborating with these two strategically positioned authors. This shows how centrality measures are positively influenced when an author collaborates with other central authors.

Furthermore, only few authors are key nodes in the network, as shown in Table 11. Although most of these authors maintained a high betweenness centrality coefficient over the years, some of them had their coefficient decreased over time. For instance, in 1997, about 19% of all shortest paths went through Rubens Nacimento Melo, while in 2015 this measure reduced to 14%. Even being retired, he still is in the center of the network for historical reasons and because some of his collaborators are still publishing at SBBD, for example Sergio Lifschitz.

Table 11 Most central authors per year

Even though these centrality measures are helpful at finding important nodes, they are not the absolute answer to this question. Specifically, we show that these metrics fail to include some prolific and outstanding authors, and they might also rank authors who have not published for a long time in SBBD (e.g., Décio Fonseca and José Luís Braga).

Therefore, as an effort to capture another facet of the authors’ influence, we list in Table 12 the top SBBD authors ranked according to their h-index⁵ as collected from Google Scholar⁶ in June 2017. Notice that for some authors, the total number of SBBD publications is not proportional to their h-index, which is expected as other publications (besides their SBBD papers) contribute to their h-index value.

Table 12 h-index and number of publications of the authors who appear in at least one of those top ten ranks: Newman’s weight, betweenness, closeness, and degree

Finally, Fig. 17 a, b shows the first three moments of the betweenness and closeness distributions, respectively. Regarding the betweenness distribution, high positive values for skewness indicate that SBBD network has a small set of influential nodes. These nodes act as bridges by connecting different parts of the graph, spreading information and new research trends. On the other hand, for the closeness distribution, skewness values go from positive to negative, indicating that, over time, the network density is growing and its nodes are closer to each other. We highlight that similar results were found on the SBRC collaboration network [22], probably indicating that this behavior could be inherent to this type of network.

Fig. 17

Three moments of the (a) betweenness and (b) closeness values

Homophily

Homophily is the tendency of people to connect with similar ones. In this section, we investigate this phenomenon from the perspective of affiliation. We also overview data regarding gender. Such analyses can shed light on how the surrounding contexts of a network (affiliation and gender of an author in our case) can drive the formation of its links [9].

Table 13 shows the top 10 institutions ranked by the number of affiliated authors who have contributed to a paper presented at SBBD over the years. We can see a high predominance of institutions from the southeast of Brazil followed by some from the northeast, with only two exceptions (UFRGS and UFAM). The number of publications are primarily concentrated into the following five states: Minas Gerais, Rio Grande do Sul, Rio de Janeiro, Pernambuco, and São Paulo. Despite geographical distance, collaborations between regions occur. For instance, UFMG and UFAM, which are from Southeast and North respectively, have many affiliated authors who publish together in SBBD.

Table 13 Top-10 institutions considering the number of distinct authors

In order to understand how collaborations are influenced by similar author characteristics, Fig. 18 shows the homophily of SBBD by affiliation and gender. There is an increasing tendency of connections between different affiliations (e.g., two different universities), resulting in the decrease of the affiliation homophily during the years. Moreover, gender homophily was high in initial years due to the very small percentage of women among the authors (4 of 22 authors in 1986 for example) and has oscillated over the years.

Fig. 18

Homophily over time

Even though collaborations are getting more diverse in SBBD, there is still a predominance of authors publishing with similar authors in terms of affiliation and gender. As we can see from Fig. 18, these two types of homophily have evolved to 0.625 and 0.658, respectively, indicating that similar characteristics are observed between collaborators more often than dissimilar ones.

Finally, Fig. 19 shows the distribution of authors’ gender by each year, in which the female distribution has oscillated between 16.44 and 38.46%. However, it is worth noticing that, with few exceptions (1986, 1996, 1997, 2003, 2008, 2012, and 2013), the female collaboration over the years was above 25%.

Fig. 19

Authors’s gender distribution over time

Conclusions

In this article, we went over the SBBD history from distinct social perspectives. Specifically, we presented a deep analysis of the Brazilian database community based on the publications included in the SBBD proceedings and its associated journal JIDM. We achieved this by investigating the complete SBBD co-authorship network built from bibliographic data of SBBD’s 30 editions and analyzing several network metrics (e.g., degree, density, and assortativity) considering the network evolution over three decades. In particular, we analyzed the involvement of the most engaged SBBD authors, the number of distinct authors, institutions, and published papers, and the evolution of some of the most frequent terms presented in the titles of the papers. Finally, we discussed the productivity, influence, and impact of SBBD authors given by centrality measures and showed that the SBBD network follows a phenomenon typical in social networks known as small-world.

Among our main findings, we provide evidence that the SBBD community is becoming more collaborative over the years. Moreover, the increasing number of newcomers is followed by an increasing number of new institutions, which contribute to the symposium scientific strength and reachability. In spite of some authors being really engaged and publishing consecutively for several years, only few authors are key nodes in the network and have been continuously responsible for the core publications. Besides, one common feature of many complex networks, closely related to the rich-gets-richer effect [4], can also be observed in the SBBD community, as most authors have only few collaborations while there are few that correspond to high degree nodes.

Our analysis demonstrates that even though the number of authors increased, new authors tend to publish papers with coauthors that maintain previous collaborations, which can be associated with the high number of collaboration groups with few authors found in the network. We showed that geographical location and the affiliation of an author are strong factors to determine which group this author belongs to in the SBBD community, although collaborations between regions such as Southeast and North, for instance, have many affiliated authors who publish together in SBBD. Furthermore, the connection of high degree nodes to low degree ones (usually observed between students and their supervisors) is another factor that leads to collaborations between authors.

Despite the deep analyses of the Brazilian Database community based on its publications, our work could be improved by further studying the geographic location of the collaborations. It would also be interesting to consider analyses based on the collaboration network of members of SBBD program committees, equally studying the existing members and appearance of newcomers through the years. Finally, we could also investigate how SBBD researchers contribute to other communities, similar to the broader study of Silva et al. [32].

Endnotes

¹ BDBComp: http://www.lbd.dcc.ufmg.br/bdbcomp

² JIDM: https://seer.ufmg.br/index.php/jidm/index

³ Pearson’s correlation coefficient provides a measure of the strength of a linear association between two variables [17].

⁴ A similar case of the “friend of my friend is also my friend” phenomenon found in social networks [28].

⁵ The h-index [13] of an author is the highest number of her publications with at least that many citations, e.g., an author with 10 papers with at least 10 citations each has h-index of 10.

⁶ Google Scholar: https://scholar.google.com.br