Identi ng Social and Technical Aspects in the Startup Ecosystem in Rio de Janeiro State

Abstract

Startups ecosystems are important drivers for innovation, responsible for generating jobs and revenue in urban centers. They promote technological development through collaborative networks of entrepreneurs, startups builders and investor groups. The complex relationships formed in these communities are essential to ensure access to resources that enable the execution of projects, such as technologies, know-how, infrastructure and nancing. However, understanding how these partnerships are formed and maintained is not a trivial task, because they depend on several regional factors. This paper presents a mapping of the startup ecosystem of the state of Rio de Janeiro. In our approach, we use data of di erent sources to de ne technical and social aspects of the entrepreneurial community. Then, social network analysis are used to characterize predominant sectors, competencies, interests and relevance of each group of entrepreneurs.

Keywords

• Startups ecosystems Urban Centers Entrepreneurship Big Social Data Social Network Analysis Rio de Janeiro State

-

Startup ecosystems are creative workspaces in which entrepreneurs seek to validate innovative ideas in a short period, converting them in disruptive business with low costs [ 9 ]. Besides acting as innovation drivers, these entrepreneurial communities have great potential in job creation and income. They are vital for the development and reinforcement of economic activities in urban centers [ 21 ].

These communities are composed by entrepreneurs, institutions and processes. They are situated in a given geographic location, where the actors interact through both formal and informal connections [ 2 ]. The distinct nature of these actors reveals di erent purposes in the network. Entrepreneurs pursue new business creation and technological development. While ecosystem builders, such as incubators and accelerators play the role of facilitators in the enterprise development process by providing infrastructure and administrative support. In turn, investor groups provide nancial support for scalability [ 11 ]. The interactions among these entities aid the development of new companies and the community as a whole [ 1 ].

The size of a startup ecosystem can be determined by its reach and geographic location [ 24 ]. A local ecosystem is restricted to the community formed around an entity of interest, university or research center. In a regional startup ecosystem there is a large number of participants, increasing the possibilities of partnerships and the availability of resources. The pluralism in the network improves the innovation process, once the collaboration among di erent participants brings new perspectives to the ventures [ 25 ]. However, the identi cation of suitable partners is a challenge. Ensuring the convergence of interests among various agents through such arrangements is not trivial. The di erent objectives, besides the circumstances, can make the conjugation less harmonious and complex [ 8, 26 ].

In this context, it is important to know the inherent characteristics of the community to facilitate integration and improve the e ciency of network interactions. According to Audretsh and Belitski [ 5 ], the complex nature of relationships in ecosystems is due to an unique combination of regional factors. They are cultural, social and material elements that in uence the discovery and exploration of opportunities. The present work presents the mapping of the regional ecosystem of the state of Rio de Janeiro, tracing a technical and social pro le of this entrepreneurial community.

This paper is strutured as follows. Section 2 brings an overview of startup ecosystem. Section 3 describes the main concepts of social network analysis. Section 4 presents our approach to plan, collect and analyze data. Section 5 shows the research ndings and limitations of the experiment. Finally, Section 6 presents the conclusions and suggests future works. 2

Startup Ecosystems

Startups are endeavors that search for a scalable and repeatable business, despite uncertain conditions, little experience and limited resources [ 9 ]. These ventures have greate potential to launch innovations [ 10 ]. However, to overcome their constraints, startups make use of entrepreneurial communities. It is imperative to have partners that o er infrastructure, administrative services and networking to facilitate access to suppliers, technological assets and funding [ 27, 19 ].

An entrepreneurial community behaves like a biological ecosystem - a system of di erent species living in the same habitat. The business activities are expressed by relations of interdependency and coevolution. This metaphor is useful to analyze the interrelationships existents in entrepreneurial environments. The interdependence denotes the complex nature of relationships among its participants, who compete for resources and collaborate for the common bene ts, in a relationship called coevolution [ 20 ].

A startup ecosystem can be de ned as a set of di erent agents that promote the entrepreneurial spirit. They follow and support the startup development process, stimulating entrepreneurship, generating innovation and economic growth [ 28, 26 ]. They are formed by actors with di erent roles and interests. Understanding the nature of the relationships among these di erent participants is imperative for their success as a whole. According to Agueda [ 2 ], these agents can be grouped into three categories: { Entrepreneurs: people who are searching for some business opportunities to start a deal. Torres and Souza [ 28 ] emphasized that in developing countries many people undertake for lack of good jobs. They are called necessity entrepreneurs. On the other hand, there are those who are looking for new challenges, the so-called serial entrepreneurs. { Ecosystem Builders: they are support institutions that act in the development, support and encouragement of entrepreneurial actions. They represent bridges between ecosystem participants and ensure that the entrepreneurs have all the necessary resources to increase the chance of success of the ventures [ 2, 6 ] { Investor Groups: they are responsible for funding high growth startups. In Brazil, the initial investments in the entrepreneurial communities have been carried out by government agencies [ 4 ]. The mission of development agencies is related to the public policies of technological development adopted in the ecosystem region. There are also other nancial entities that have realized excellent business opportunities by o ering credit to startups. These are venture capital funds or even experienced entrepreneurs who have decided to support new investments in order to get nancial pro ts.

The size of a startup ecosystem can be determined by its reach and geographic location. This dimension is a key factor for the development of innovations, bringing direct in uences to collaborative activities of creation and di usion of knowledge, capacity development, resource sharing and networking [ 24 ]. A greater diversity in the community impacts on the creative process, because the collaboration with di erent participants brings new perspectives to the ventures [ 25 ]. Pombo-Juarez et al. [ 24 ] establish four levels of entrepreneurial community coverage: local, regional, national and international. A local ecosystem is limited to participants in a university or research center. A regional ecosystem is a bit more comprehensive than the local, with more participants and resources. A national ecosystem involves institutions and entrepreneurs from a whole country. An international ecosystem involves several countries, such as initiatives by companies or groups of countries interested in developing entrepreneurial policies. 3

Social Network Analysis

The behavior of certain elements can not be studied separately due to the inuences produced by the environment. In such cases, it must to study how connections are formed and what their relevance is to the problem in question. In the case of startup ecosystems, partnerships provide members with a range of resources that they would otherwise not have access to [ 18 ]. Considering that an entrepreneurial community is a set of interdependent organizations, the study of its dynamics can be facilitated by the use of social network analysis [ 12, 22 ].

A social network is an abstraction that allows to codify relationships between pairs of individuals, such as ties of friendship, a nity, common interests or commercial relations [ 14 ]. There are a number of phenomena occurring in networks that depend fundamentally on their structure. Therefore, the study of the properties of networks can reveal patterns of interaction. The social networks analysis can assess the level of coordination of partnerships, the intensity of interactions, the emergence of communities, the level of connectivity, the relevance of participants, the in uence of groups and patterns of group behavior [ 3 ]. 3.1

Network Topological Characteristics

The structural aspects of the network can reveal important information about relationships in communities. The social network analysis aids to identify critical points in the community's performance [ 15 ]. The networks are normally represented using graphs: the actors are the nodes (or vertices) and the bonds are the edges (or links) of the graphs. Nodes and edges can receive di erentiated weights to represent the number of node connections or frequency of interactions. Thus, it is possible to represent di erent characteristics of a social network. Edges can use di erent weights to indicate intensity, number of occurrences, or probability of relationships [ 12, 14 ].

The number of nodes and edges of a network de ne its density. The network density represents the ratio between the links in the graph and the total number of edges that the graph could have. In turn, the density of a node is the ratio between the number of neighbors of the node and the number of possible neighbors. This measure indicates how well connected a node is in the network [ 14 ].

In the study of organizational networks, an important issue refers to the concept of centrality. It determines the extent to which a speci c node is connected to the others in the network. In general, the degree centrality of a node is determined by its number of edges. A high degree centrality implies a greater number of relationships and better opportunities because they have choices [ 15 ]. In a network of partnerships, the degree centrality points the relevance of the participant in the community. Thus, incubators that have a great importance for the enterprises have a high centrality [ 25, 19 ].

There is a set of speci c centrality metrics that can be applied in speci c cases. The simpler measures consider only the presence or not of an edge, however, more sophisticated metrics can take into account the weight of the edges [ 14 ]. The closeness centrality metric is based on the total distance between a particular node and all others and the total number of other nodes accessible from the observed vertex. Nodes with high values for closeness centrality have great importance in the dissemination of information in the network [ 12 ]. Betweeness centrality measures the frequency at which a node is used as a bridge between two others. The intermediary has the power to interrupt relations and isolate actors, preventing contact between them [ 12 ]. In order to analyze the importance of nodes with low degree centrality, the eigenvector centrality can be applied. This metric checks the impact of a node's relationships through a score assigned to all nodes in the network. If an actor has few relationships, but with other nodes of great relevance, their importance will also be considered [ 15 ].

In a network of entrepreneurs, individuals with more relationships will have more access to resources. The role played by the actor will determine which centrality metric should be applied. As reported by Grassi et al. [ 15 ], there are four strategic positions: 1. central individuals: they have many connections with others, that represents great opportunities for interactions. They can mobilize more resources and in uence partners to achieve results. In the entrepreneurial communities, it is the role played by incubators, accelerators and technology parks. They are hubs for entrepreneurs, the diversity of contacts is fundamental to expand business opportunities [ 6, 13 ]. In this case, degree centrality is the most appropriate tool. 2. brokers: individuals who act in the community controlling the ow of information. They do not have a large number of connections, but they have a betweenness centrality. The startups behave in this way [ 21 ]. 3. boundary spanners: they maintains relationships with individuals from outside their community, seeking new opportunities. Companies usually establish these relationships in entrepreneurial communities [ 6 ]. The best tool here is eigenvector centrality [ 15 ]. 4. boundary specialists: they have high level of technical skills or speci c information, and they establish in the border of the network. They have a low centrality measure [ 15 ]. 3.2

Network Behaviors in Startup Ecosystems

The complex nature of ecosystem relationships is the result of a unique combination of environmental aspects. They are sociocultural and material factors that in uence the discovery and exploitation of opportunities [ 17, 5 ]. Cultural aspects are based on implicit beliefs and norms that shape the perception of ecosystem members in relation to entrepreneurship. A friendly culture is concerned with establishing the environmental conditions necessary to stimulate entrepreneurial activity, through a climate of greater acceptance of risks. According to Audretsch and Belitski [ 5 ], tolerance and openness to diversity establishes the conditions for testing new possibilities, assuming the chances of failure and making the environment richer by tolerating di erent ideas and ways of thinking, ethnicities and cultures. The culture also appears related to the sense of con dence and security necessary to establish activities of collaboration in the community [ 26 ]. The in uence of family and friends is also mentioned as a factor that can a ect the actions of entrepreneurs [ 28 ].

The social factor, the so-called social capital, refers to the bene ts obtained or acquired through the social network of the community. The importance of this mechanism has been widely discussed in Jha [ 18 ]. It has a fundamental role in discovering new knowledge about opportunities and technologies, helping new ventures to obtain funding and in uencing new perspectives and entrepreneurial skills. Social capital depends on the stablished connections and culture existent in the network. These aspects create an atmosphere of trust among the agents, that is a basic condition to encourage the sharing of scarce resources among entrepreneurs, investors and other entities [ 26 ].

The material aspects are related to the physical conditions necessary for the establishment of the startups ecosystem. According to Audretsch and Belitski [ 5 ], the infrastructure of the region can in uence connectivity and the recognition of opportunities. The facilities o ered by the region can make it more attractive to a greater number of entrepreneurs, local and regional authorities, researchers and academics, educational institutes and other supportive agents promoting community development [ 27 ].

The universities and research centers in the region act as providers of new technologies and catalysts of market opportunities. They form human capital and are responsible for the development of new academic ventures and spin o s [ 6 ]. In turn, the companies can establish partnerships with universities, absorb the skilled workforce or seek solutions collaborating with startups. The formed partnerships promote the monitoring of new technologies, facilitating the absorption of knowledge and the generation of competitive di erentials [ 10 ].

The existence of formal support institutions helps regulate the governance model for the operation of the ecosystem. The government also has a relevant role in the community. Their actions establish important incentives for the emergence of new businesses through measures that can reduce bureaucracy, provide e cient administrative services, and prioritize resource allocation and nancial support [ 17, 6, 26 ].

The availability of investment funds is determinant for the development of entrepreneurial communities [ 17 ]. These kinds of nancing include public funds, venture capital, angel investors, family, banks, self- nancing, friends and incubators [ 28 ]. Another important startup ecosystems' requirement is the existence of a consolidated market with speci c needs. The perceived demands on interactions with potential clients facilitate the identi cation of opportunities and the perception of value creation. The target audience creates an early validation mechanism that reduces the costs of launching new products and boosts business growth on scale [ 5, 26 ]. 4

Methodological Approach

This paper presents a mapping of startup ecosystem of state of Rio de Janeiro, de ning the technical and social pro le of its participants and their relationships. As presented, such aspects are crucial for understanding network behaviors [ 5 ]. Our approach was divided into 4 stages: (1) Data collection; (2) Classi cation and clustering; (3) Building of graphs; and (4) Analysis and visualization.

The rst phase corresponds to the extraction, structuring and storage of the raw data. The data about the enterprises and support institutions were collected from the website of the ReINC4. ReINC is a non-governmental organization in support of entrepreneurship that aims to leverage the economy through incentives for innovation. Considering the development of social computing, much of social interaction is nowadays mediated by information technology [ 23 ]. So, the entrepreneurs' pro les from LinkedIn5 were used to extract social data of the startup ecosystem. We chose the LinkedIn platform because it is a business online social media that connects professionals from all over the world, providing relevant information that allows a view of the pro le of its members.

From database of ReINC, it is possible to identify the location, sectors and development stage of the ventures. It also has details about theirs products or projects and general data about their responsible. While LinkedIn data reveals social information from entrepreneurs, such as: their academic and experience background, explicited interests, and recommendations for users' skills. The recommendations provide some evidences of user engagement and reputation in the network, whilst expressed interests help characterize the pro le of the entrepreneur [ 13 ].

During the second phase, the data collected and structured in the previous step are submitted to procedures for eliminating redundancies and disambiguating terms. We used the taxonomy of knowledge areas provided by CAPES6 and the taxonomy of productive niches de ned by REINC itself. We used also a non-supervised k-means clustering algorithm to aid in the categorization of terms. To increase the reliability of the results, the base was inspected by pairs of researchers.

In the third phase, the relationships between startup ecosystem participants are mapped using graphs. Several aspects are represented by graphs, such as a liations and geolocation, interests, and competencies. The a liation and geolocation graph allows to identify startups linked to the same incubator or technology park. In this way, we were able to analyze projects that share the same culture and norms and visualize their location. It is also possible verify the relevance and power of incubators for di erents sectors. This graph points the sectors that attract most attention in the community. As ReINC represents a regional ecosystem, through this map, it is possible to determine degrees of distance between actors in the network, helping in the identi cation of potential partnerships.

The interests graph maps the subjects that generate greater a nity for the entrepreneurs. The topics may be related to companies, groups, educational and research institutions or even personalities.

Finally, the competencies graph represents skills and endorsements about entrepreneurs. The identi cation of the skills and recommendations about entrepreneurs as well as the capacities necessary for the development of the products and services o ered represent an important source of information about 4 ReINC { Network of Promoters of Innovative Enterprises Agency. Available in http://reinc.org.br 5 LinkedIn. Available in http://linkedin.com 6 CAPES { Brazilian Foundation for Coordination for the Improvement of Higher

Education Personnel. Available in http://capes.gov.br the members of the entrepreneurial ecosystem. These aspects of participants are related to their prestige and reputation. The identi cation of a complementary competence can help in the formation of partnerships or broaden synergies.

The analysis of the formed networks is performed in phase four. The concepts and metrics associated with generated graphs were de ned in order to support this step. The social network analysis metrics are used to know the basic characteristics of the network topology to infer the following characteristics: local productive vocation, in uence of location, pro le of entrepreneurs, reputation, competence and expertises in the network. 4.1

Related Work

The mapping of startup ecosystems was carried out in Arruda et al.[ 4 ]. However, its goal was to detail the structural characteristics of the startup ecosystems: the basic conditions for the success of the communities. They did not discuss the in uence of relationships in the community. The description of the communities was also carried out in Isenberg [ 17 ].

The analysis of relationships in business ecosystems was discussed in Basole et al.[ 7 ]. The authors identi ed the segment, total number of partner collaboration, number of collaborations, specialties, and trust in a business ecosystem. The graph visualization model supports an intelligent management of partnerships and decision support. They did not address the speci cities of startup ecosystems, but the issues raised in these studies elucidate important points about the relationships and the impact of interactions. 5

Results Obtained

The present study used as object the entrepreneurial community of the state of Rio de Janeiro. It represents an important entity of the federation, being the second richest and most populous state in Brazil[ 16 ], ilustrated in Figure 1. The data collection process was carried out between February and March 2018. The last update identi ed in the ReINC database was in August 2017. It is important to consider the dynamic nature of the entrepreneurial community, so that the results obtained represent a snapshot this period. It was identi ed 18 incubators and 7 technology parks, 132 startups, as well as 227 graduated ventures and 27 associated companies. They act in 14 di erent sectors: agribusiness, biotechnology, design/creative economy (CE), drugs & health, education, energy, oil & gas (O&G), engineering and robotics, environment, food & beverage, information and communication technologies (ICT), industrial technology, logistics, mining & earth sciences, and solidarity economy, as shown in Table 1.

Most of the observed startups are involved in knowledge intensive ventures. This is related to the pro le of the entrepreneurs, because the diversity of entrepreneurs leads to more creative spaces [ 25 ]. Professionals with a high level of education and experience are keys factors for the success of business. In communities in which projects of greater complexity are developed, there is a great concentration of masters and doctors. In the COPPE/UFRJ, which is maintained by the largest federal university in Brazil (UFRJ7), it has the highest percentage of masters and doctors among its entrepreneurs: 76.4%. The incubator of the UFF8 has 58.3% and the incubator Instituto G^enesis of Puc-Rio9, has 50% of scientists managing projects.

With respect to the productive sector, there is a great concentration in ICT, creative economy and environment areas. There are ICT startups in almost every technology incubator because they do not require so much infrastructure resources. Merely the thematic incubators do not have ventures of this nature: Bio Rio is focused on biotechnology; Rio Criativo/SEC incubator just supports creative economy projects; INEAGRO incubator aids agribusiness; and social incubators (ITCP, ITECS, ITESS) work speci cally with solidarity economy.

The incubators and technology parks have predominant sectors, as shown in Figure 2. The ecosystem builders with the greatest number of enterprises have greater relevance in the network. However, the diversi cation of areas also impacts on the importance of the institution. For instance, the UFF incubator, which has 31 projects, has a greater betweeness centrality than the COPPE/UFRJ incubator, which has 87 projects. On the other hand, Rio Criativo/SEC incubator, despite having the same number of projects, because it is sectorized, its betweeness centrality is zero. The technology parks have attracted the attention of large companies from several sectors, specially on the Oil and Gas area and the development of Medicines, next to UFRJ. 7 UFRJ { Federal University of Rio de Janeiro 8 UFF { Federal Fluminense University 9 Puc-Rio { Ponti cal Catholic University of Rio de Janeiro

The recomendation of skills found in Linkedin demonstrate the social ability of their users. The most frequent and online user commonly have more recommendations. Therefore, this indicator provides information about its competencies and the recognition of the professional in the community in which it participates. The skills were grouped into classes adapted from the knowledge taxonomy of CAPES. Figure 3 illustrates the most frequent recommendations per incubator. Observing the eigenvector centrality, it is possible to perceive which nodes have the greatest impact on the network. Table 2 details the 5 incubators in which their competencies generate greater in uence in the community and the most important terms cited. Despite having more recommendations than Rio Criativo/SEC, IETEC/CEFET-RJ is less relevance in this aspect. This is because Rio Criativo/SEC behaves as a boundary specialist.

The graph of geolocation links the ecosystem builders that have common sectors and that are within a certain radius of distance, represented by Figure 4. The parameter was adjusted by the maximum distance between entities belonging to the same region of the state, the metropolitan region. It was used 30 kilometers, which is the distance between the BIO RIO incubator, located in the capital and INMETRO incubator, located in an adjacent municipality. Proximity between hubs can stimulate informal relationships among their participants [ 5 ]. The centrality degree of this graph determines the incubators or technology parks with the best possibilities of establishing relationships in the network due to their location. According to the weighted degree calculated, the ve best located institutions are: COPPE/UFRJ, Instituto G^enesis/Puc-Rio, IETEC/CEFET-RJ, UFF and UFRJ Technology Park.

The graph of interests analyzes which subjects are tracked by the entrepreneurs on LinkedIn. In this network, the nodes represent entrepreneurs and the edges represent the a nities between them. A nity was established as a certain number of interests in common. We used 5 interests as parameter. The resulting graph, illustrated in Figure 5, shows the predominance of a nities among entrepreneurs linked to the same incubator. The culture of the incubator may have an in uence on this indicator. However, there are similarities between entrepreneurs of distinct and distant incubators. In these cases, either the academic or experience background of the entrepreneurs were similar.

Regarding the limitations, this study, like any other, has several limitations and threats that may a ect the validity of the results. Concerning to the construction validate, the approach relies on the participating of entrepreneurs in some social platform. Despite being a widely used nowadays, it generates an important bias. Moreover, this experiment there was no statistically established population. Because the entrepreneurial community is dynamic, new participants may have been added over time, making it di cult to accurately represent the ecosystem. Entrepreneurial communities are habitats where di erent actors coexist and interact by seeking resources and partnerships to develop ventures. This diversity can give rise to more innovative businesses [ 25, 1 ]. However, managing resources in the network may not be a trivial task. There are a number of sociocultural and material aspects that must be considered to enhance integration and provide greater network e ciency. Therefore, it is necessary to have mechanisms that help in understanding network behaviors.

We presented a mapping of the startup ecosystem of Rio de Janeiro state using an approach based on social network analysis. The pro le of entrepreneurs, incubators and technology parks were identi ed according to technical and social aspects. The use of social network analisys was a good solution to broaden understanding about some implicit aspects of entrepreneurial communities. The main contribution of this paper is building of a process of mapping, that can be used in others startup ecosystems.

In future work, we intend to increase the understanding about interactions in startup ecosystems through a more detailed investigative analysis. We are currently working on the mapping model, creating di erent views of the ecosystem according to the actor's pro le - entrepreneur, ecosystem builders or investors. It is also necessary to improve the metrics used in this article and develop tools that facilitate the execution of the method. The aim is to give participants the entrepreneurial community a broader view of the possible partnerships and networking management mechanisms of cooperation.

Acknowledgements

Kesia Mamede was partially sponsored by a grant from Capes. Rafael Escalfoni is supported by CEFET/RJ, according to the act #122 of January 17th, 2018.

References

1. Aarikka-Stenroosa , L. , Ritala , P. : Network Management in the Era of Ecosystems: Systematic Review and Management Framework . Industrial Marketing Management 67 ( 1 ), 23 { 36 ( 2017 ). https://doi.org/10.1016/j.indmarman. 2017 . 08 .010

2. Agueda , A. F. P. : Interconnectivity between Ecosystem Builders and Investor Groups in European Startup Ecosystems . Dissertation | Faculdade de Ci^encias e Tecnologia da Universidade Nova de Lisboa, Lisbon, Mar 2016 .

3. Albalooshi , N. , Mavridis , N. , Al-Qirim , N.: A Survey on Social Networks and Organization Development . In: Collaboration Technologies and Systems (CTS) , pp. 539 { 545 . IEEE, Denver ( 2012 ). https://doi.org/10.1109/CTS. 2012 .6261102

4. Arruda , C. , Nogueira , V. S. , Costa , V. : The Brazilian Entrepreneurial Ecosystem of Startups: an Analysis of Entrepreneurship Determinants in Brazil as seen from the OECD Pillars . Journal of Entrepreneurship and Innovation Management 2 ( 3 ), 17 { 57 ( 2013 ). https://doi.org/10.1007/978-3- 319 -11412-5 2

5. Audretsch , D. B. , Belitski , M. : Entrepreneurial Ecosystems in Cities: Establishing the Framework Conditions . Journal of Technol. Transf . 42 ( 5 ), 1030 { 1051 ( 2016 ). https://doi.org/10.1007/s10961-016-9473-8

6. Bandera , C. , Thomas , E. : Startup Incubators and the Role of Social Capital . In: IEEE Technology & Engineering Management Conference (TEMSCON) , pp. 142 { 147 . IEEE, Kansas ( 2017 ). https://doi.org/10.1109/TEMSCON. 2017 .7998368

7. Basole , R. , Huhtamaki, J., Still , K. , Russelld , M. : Visual decision support for business ecosystem analysis . Expert Systems With Applications 65 , pp. 271 { 282 ( 2016 ). https://doi.org/10.1016/j.eswa. 2016 . 08 .041

8. Bengtsson , M. ; Raza-Ullah , T. : A Systematic Review of Research on Coopetition: Toward a Multilevel Understanding . Industrial Marketing Management 57 , 23 { 39 ( 2016 ). https://doi.org/10.1016/j.indmarman. 2016 . 05 .003

9. Blank , S. G. , Dorf , B. : The Startup Owners Manual: the step-by-step guide for building a great company. K&K Ranch Publishing Division , Pescadero ( 2012 ).

10. Chesbrough , H.: Inovaca~o Aberta: como Criar e Lucrar com a Tecnologia . Bookman, Porto Alegre ( 2012 ).

11. Dedehayir , O. , Makinen , S. J. , Ortt , J. R. : Roles During Innovation Ecosystem Genesis: a Literature Review . Technological Forecasting & Social Change , 1 { 12 ( 2016 ). https://doi.org/10.1016/j.techfore. 2016 . 11 .028

12. Easley , D. , Kleinberg , J.: Networks, Crowds, and Markets: Reasoning about a Highly Connected World . Cambridge University Press, Cambridge ( 2010 ). http://www.cs.cornell.edu/home/kleinber/networks-book. Last access in May 29th , 2018 .

13. Escalfoni , R. , Franca , T. , Irineu , M. , Vivacqua , A. , Oliveira , J.: A Method to Support the Identi cation of Interests in Startup Ecosystems . In: XIV Brazilian Symposium on Information Systems (SBSI) , ACM, Caxias do Sul ( 2018 ).

14. Franca , T. , Faria , F. F. , Miceli , C. , Rangel , F. , Oliveira , J.: Big Social Data: Princ pios sobre Coleta , Tratamento e Analise de Dados Sociais. In: Proceedings of Brazilian Symposium on Database (SBBD) , pp. 1 { 40 . SBC, Porto Alegre ( 2014 ).

15. Grassi , R. , Stefani , S. , Torriero , A. : Centrality in organizational networks . International journal of intelligent systems 25(3) , 253 { 265 ( 2010 ). https://doi.org/10.1002/int.20400

16. IBGE. Censo 2010 , https://censo2010.ibge.gov.br. Last acessed in May 29th , 2018 .

17. Isenberg , D. J.: The Entrepreneurship Ecosystem Strategy as a New Paradigm for Economic Policy: Principles for Cultivating Entrepreneurship . In: Presentation at the Institute of International and European A airs . Babson Global , Dublin ( 2011 ).

18. Jha , S. K. : Social Network Perspective on Innovation: a Review . In: Internet Multimedia Systems Architecture and Application (IMSAA) , pp. 1 { 6 . IEEE, Washington DC ( 2011 ). https://doi.org/10.1109/IMSAA. 2011 .6156356

19. Mian , S. , Lamine , W. , Fayolle , A. : Technology Business Incubation: an Overview of the State of Knowledge . Technovation 50 , 1 { 50 ( 2016 ). https://doi.org/10.1016/j.technovation. 2016 . 02 .005i

20. Moore , J. F. : Predators and Prey: the New Ecology of Competition . Harvard Business Review 71 ( 3 ), 75 { 83 ( 1993 ).

21. Motoyama , Y. , Waltins , K. Examining the Connections within the Startup Ecosystem: a Case Study of St . Louis. Entrepreneurship Research Journal 7 ( 1 ), ( 2017 ). https://doi.org/10.1515/erj-2016 -0011

22. Neumeyer , X. , Santos , S. C. : Sustainable Business Models, Venture Typologies, and Entrepreneurial Ecosystems: a Social Network Perspective . Journal of Cleaner Production 172 , 4565 { 4579 ( 2018 ). https://doi.org/10.1016/j.jclepro. 2017 . 08 .216

23. Olshannikova , E. , Olsson , T. , Huhtamaki, J., Karkkainen, H.: Conceptualizing Big Social Data . Journal of Big Data 4 ( 3 ), 1 { 19 ( 2017 ). https://doi.org/10.1186/s40537- 017-0063-x

24. Pombo-Juarez , L. , Konnola , T. , Miles , I. , Saritas , O. , Schartinger , D. , Amanatidou , E. , Giesecke , S. : Wiring up Multiple Layers of Innovation Ecosystems: Contemplations from Personal Health Systems Foresight . Technological Forecasting & Social Change 115 , 278 { 288 ( 2017 ). https://doi.org/10.1016/j.techfore. 2016 . 04 .018

25. Scillitoe , J. L. ; Chakrabarti , A. K. : The Role of Incubator Interactions in Assisting New Ventures . Technovation 30 ( 3 ), 155 { 167 ( 2010 ). https://doi.org/10.1016/j.technovation. 2009 . 12 .002

26. Spigel , B. : The Relational Organization of Entrepreneurial Ecosystems . Entrepreneurship Theory and Practice 41 ( 1 ), 49 { 72 ( 2017 ). https://doi.org/10.1111/etap.12167

27. Stam , E.: Entrepreneurial Ecosystems and Regional Policy: a Sympathetic Critique . European Planning Studies 23 ( 9 ), 1759 { 1769 ( 2015 ). https://doi.org/10.1080/09654313. 2015 .1061484

28. Torres , N. N. J. , Souza , C. R. B. : Uma Revisa~o da Literatura sobre Ecossistemas de Startups de Tecnologia. In: XII Brazilian Symposium on Information Systems (SBSI) , pp. 385 { 392 . ACM, Florianopolis ( 2016 ).