Social network analysis (SNA) provides tools to examine relationships between people. Text mining (TM) allows capturing the text they produce in Web 2.0 applications, for example, however it neglects their social structure. This paper applies an approach to combine the two methods named "content-based SNA" (CB-SNA). Using the R mailing lists, R-help and R-devel, we show how this combination can be used to describe people's interests and to find out if authors who have similar interests actually communicate. We find that the expected positive relationship between sharing interests and communicating gets stronger as the centrality scores of authors in the communication networks increase. / Series: Research Report Series / Department of Statistics and Mathematics

Submitted by Adriana Alves Rodrigues (aalves@espm.br) on 2017-11-22T15:02:28Z No. of bitstreams: 1 ANDRE SILVA DE CARVALHO.pdf: 3017440 bytes, checksum: 72f0dd79324eb16e16c0fca2fea756db (MD5) / Approved for entry into archive by Adriana Alves Rodrigues (aalves@espm.br) on 2017-11-22T15:02:51Z (GMT) No. of bitstreams: 1 ANDRE SILVA DE CARVALHO.pdf: 3017440 bytes, checksum: 72f0dd79324eb16e16c0fca2fea756db (MD5) / Approved for entry into archive by Ana Cristina Ropero (ana@espm.br) on 2017-11-23T10:56:03Z (GMT) No. of bitstreams: 1 ANDRE SILVA DE CARVALHO.pdf: 3017440 bytes, checksum: 72f0dd79324eb16e16c0fca2fea756db (MD5) / Made available in DSpace on 2017-11-23T10:56:31Z (GMT). No. of bitstreams: 1 ANDRE SILVA DE CARVALHO.pdf: 3017440 bytes, checksum: 72f0dd79324eb16e16c0fca2fea756db (MD5) Previous issue date: 2017-03-24 / Being innovative in a more and more competitive market can be anything but trivial. There is a complex variables system to be taken into account throughout an innovation process, and hardly ever will there be enough data to support a research or decision. It is always possible to turn to human inference, or cognitive bias, when enough data is not available, or when time for decision-making is scarce. Consumer Insight technique has been used for this research purpose and aimed at lowering cognitive bias, seeking to find out what are consumers' wishes and needs so that decision-making or innovation could be supported. This paper proposes to mitigate the influence of cognitive bias, by means of data analysis techniques, in search for patterns which can identify opportunities to give both decision-making and search for innovation some support. In order to achive this purpose, unstructured data from 26.514 telephone talks had in a big financial market company between 01.12.2016 e 31.12.2016 have been used. Analysis has been carried out with the transcript from voice into text concomitantly with Text Mining and Social Network analysis. The results have led us to identify main client demands from a sales perspective, cancellation resquest, as well as the reason for inefficiency in offering new products from elements of higher centrality identified in the word association networks. It is implied that the combined use of analytical techniques applied to unstructured data may give rise to findings in which cognitive bias is lower. / Em um mercado cada vez mais competitivo, ser inovador pode ser um diferencial, porém não é uma atividade trivial. Existe um sistema de variáveis complexas que deve ser considerado ao longo de um processo de inovação e nem sempre há dados suficientes que suportem uma pesquisa ou decisão. A inferência humana, ou viés cognitivo, pode ser uma alternativa quando não existem dados suficientes ou quando o tempo para a tomada de decisão é menor que o necessário. A técnica de Consumer Insight foi utilizada nesta pesquisa com o objetivo de diminuir o viés cognitivo, buscando descobrir os anseios e necessidades do consumidor, para suportar o processo de tomada de decisão ou inovação. Este estudo apresenta uma proposta para mitigar a influência do viés cognitivo, a partir de técnicas de análise de dados, em busca de padrões que possam identificar as oportunidades para suportar o processo decisório ou a busca pela inovação. Neste trabalho foram utilizados dados não estruturados de 26.514 conversas telefônicas realizadas no período de 01/12/2016 a 31/12/2016, provenientes de uma empresa do mercado financeiro. A metodologia analítica consistiu na transcrição de voz para texto e no uso associado de técnicas de Text Mining e Análise de Redes Sociais. Os resultados obtidos permitiram identificar as principais demandas dos clientes na perspectiva de vendas, pedido de cancelamento e a razão da ineficiência das ofertas de novos produtos, a partir dos elementos de maior centralidade identificados nas redes de associação de palavras. Implica-se que o uso combinado de técnicas analíticas em dados não estruturados pode permitir a obtenção de achados com menor influência do viés cognitivo.

De plus en plus, les entreprises incitent aujourd’hui leurs employés à adopter des modes de travail agiles et flexibles. Cette transformation est outillée par de nouvelles plateformes informatiques intitulées "plateformes de média sociaux d’entreprise" (de l’anglais "Enterprise Social Media platforms – ESM platforms"). Ces plateformes portent la promesse d’améliorer la communication des employés dans leur environnement de travail tout en outillant leurs liens sociaux. Néanmoins, le bénéfice de l’usage de ces plateformes par leurs employés demeure vague et peu mesurable pour les entreprises, dont les dirigeants ont pourtant besoin d’évaluer cet usage et de mesurer ainsi tangiblement le succès de leurs investissements. De plus, le fait d’introduire des plateformes ESM dans les environnements de travail impacte la façon dont les outils existants, tels que la messagerie électronique, sont considérées et utilisés. La question qui se pose est alors de savoir si ces nouvelles plateformes pourraient remplacer l’usage des outils existants. Dans ce contexte, cette thèse adresse la problématique de la place des plateformes ESM dans les entreprises d’aujourd’hui. Elle examine l’usage réel de ces plateformes en offrant une analyse quantitative des réseaux sociaux, établie à partir de l’usage actuel de ces plateformes. En s’appuyant sur un modèle de recherche conçu pour caractériser l’activité, la thèse applique des tests d’hypothèse qui permettent de comparer l’usage prévu de ces plateformes à leur usage réel, et ainsi de mesurer le succès du déploiement de ces plateformes. De plus, cette thèse étudie, à travers une approche déductive et expérimentale, la capacité des plateformes ESM à remplacer l’usage des outils existants dans les environnements de travail. Ces plateformes fonctionnent aujourd’hui plutôt en tandem avec d’autres outils tels que la messagerie électronique. Cette thèse propose ainsi d’améliorer la messagerie par une fonctionnalité de tagging social afin que l’utilisateur puisse bénéficier de son réseau social sous-jacent. Ce service de tagging est ensuite testé et validé via un prototype. Pour conclure, cette thèse propose des scenarios pour les futurs systèmes sociaux, englobant la totalité des outils de l’environnement de travail au sein des entreprises / Today’s organizations are increasingly promoting a shift of employees’ ways of working towards more agility and flexibility. This shift is empowered by new IT tools commonly known as “Enterprise Social Media” (ESM). Taking mainly the form of integrated platforms, these tools are deployed with the promise of enhancing employees’ communicative activities at the workplace while capitalizing on their social connections. Yet, it remains unclear for companies how their employees benefit from the use of these new tools (i.e., ESM platforms). Managers are unable to evaluate the use of these platforms and thus, tangibly measure the success of their investments. Moreover, introducing ESM platforms to the workplace impact the way the existing tools, such as email, are regarded and used. This leads to the question of whether these new platforms may be substitutes for the use of the existing tools. This thesis tackles the mentioned polemical questions about the place of ESM platforms in today’s enterprises. Carried out jointly with Orange labs, the thesis assesses the role ESM platforms could play in supporting organizations in reaching their objectives and hence sustaining their businesses in an extreme competitive context. It explores the real use of these platforms by offering a quantitative social network analysis and data mining of an established ESM platform currently in use. While relying on a proposed research model that characterizes the activity, the applied hypothesis tests allow determining the correct use of ESM platforms, which in turn indicates the success of their deployments. Moreover, the thesis argues through a reasoning approach that the ESM platforms are not capable of substituting the use of existing workplace tools. Instead, they work in tandem with certain tools such as email. Based on the latter argument, the email tagging service is introduced as an example of how existing workplace tools may be enriched with social features to derive increased benefit for the workforce. The thesis finally validates the service through implementation and testing. To conclude, it proposes and discusses scenarios of future social systems that involve the overall set of workplace tools in enterprises