Carlos Eduardo R. de Mello

GCC: a knowledge management environment for research centers and universities

Research centers and universities are knowledge-intensive institutions, where the knowledge creation and distribution are constant – and this knowledge should be managed. In spite of it, scientific work had been known for being solitary work, in which human interaction happened only in small groups within a research domain. Nowadays, due to technology improvements, scientific data from different sources is available, communication between researchers is facilitated and scientific information creation and exchange is faster than in the past. However, the focus on information exchange is too limited to create systems that enable true cooperation and knowledge management in scientific environments. To facilitate a more expressive exchanging, sharing and dissemination of knowledge and its management, we create a scientific knowledge management environment in which researchers may share their data, experiences, ideas, process definition and execution, and obtain all the necessary information to execute their tasks, make decisions, learn and disseminate knowledge.

Mining and Analyzing Organizational Social Networks Using Minimum Spanning Tree

This work focuses on using data mining techniques to identify intra and inter organization groups of people with similar profiles and that could have relationships among them. Our clustering method identifies clusters with a link mining-based technique that uses the minimum spanning tree to construct group hierarchies. In this paper we analyze the scientific scenario in Computing Science in Brazil, assessing how researchers in the best universities and research centers collaborate and relate to each other.

Mining and analyzing organizational social networks for collaborative design

A social network is a social structure consisting of individuals or organizations usually represented by nodes tied by one or more types of relationships. The resulting structures are complex and analyzing them enables us to detect several inter and intra connection problems. This work focuses on using data mining techniques to identify intra and inter organization groups of people with similar profiles that could have relationships among them. The clusters identified allow us to identify the collaboration in the Brazilian scientific scenario of Computing Science, assessing how researchers in the best universities and research centres collaborate and relate to each other. This kind of approach can help in the identification and improvement of collaboration on innovative and multidisciplinary teams, especially on manufacturing and design scenarios.

Active Learning Applied to Rating Elicitation for Incentive Purposes

Active Learning (AL) has been applied to Recommender Systems so as to elicit ratings from new users, namely Rating Elicitation for Cold Start Purposes. In most e-commerce systems, it is common to have the purchase information, but not the preference information, i.e., users rarely evaluate the items they purchased. In order to acquire these ratings, the e-commerce usually sends annoying notifications asking users to evaluate their purchases. The system assumes that every rating has the same impact on its overall performance and, therefore, every rating is worth the same effort to acquire. However, this might not be true and, in that case, some ratings are worth more effort than others. For instance, if the e-commerce knew beforehand which ratings will result in the greatest improvement of the overall system’s performance, it would be probably willing to reward users in exchange for these ratings. In other words, rating elicitation can go together with incentive mechanisms, namely Rating Elicitation for Incentive Purposes. Like in cold start cases, AL strategies could be easily applied to Rating Elicitation for Incentive Purposes in order to select items for evaluation. Therefore, in this work, we conduct a extensive benchmark, concerning incentives, with the main AL strategies in the literature, comparing them with respect to the overall system’s performance (MAE). Furthermore, we propose a novel AL strategy that creates a k-dimensional vector space, called item space, and selects items according to the density in this space. The density-based strategy has outperformed all others while making weak assumptions about the data set, which indicates that it can be an efficient default strategy for real applications.

Acompanhamento de Campanha Eleitoral pelo Twitter

O processo eleitoral vem tomando proporções cada vez maiores, de maneira que a ciência do marketing é, hoje, uma das principais preocupações das equipes responsáveis pelas campanhas eleitorais. O crescente uso das redes sociais criou um novo canal entre candidatos e eleitores. Contudo, é possı́vel afirmar que este canal tem sido pouco explorado. É um hábito para muitos usuários divulgar nas redes sociais suas opiniões a respeito dos candidatos e partidos polı́ticos. Se as equipes de marketing eleitoral tivessem acesso direto a tais informações, poderiam utilizá-las como “termômetro” em suas campanhas. Entretanto, filtrar as opiniões dos usuários não é uma tarefa trivial devido à enorme quantidade de informações fúteis sendo compartilhada nas redes. Portanto, este trabalho propõe-se a extrair do Twitter informações relevantes, i.e., as opiniões dos usuários a respeito de um determinado candidato. Como estudo de caso, tweets referentes a um dos candidatos à prefeitura da cidade do Rio de Janeiro (RJ) foram coletados, durante a campanha eleitoral de 2012, e um filtro foi aplicado para diferenciar os tweets que transmitem a opinião dos usuários (relevantes) dos demais (irrelevantes). Técnicas de Aprendizado de Máquina foram empregadas com o intuito de efetuar tal filtragem, entre as quais estão Naive Bayes, Árvore de Decisão e Máxima Entropia. Em nossos resultados, onde atingimos acurácia média de até 75%, tiramos importantes conclusões a respeito dos fatores que interferem no desempenho de tais técnicas. Keywords—Marketing Eleitoral, Aprendizado de Máquina, Processamento de Linguagem Natural, Mineração de Redes Sociais.

Group Recommender Systems: Exploring Underlying Information of the User Space

This work proposes a new methodology for the Group Recommendation problem. In this approach we choose the Most Representative User (MRU) as the group medoid in a user space projection, and then generate the recommendation list based on his preferences. We evaluate our proposal by using the well-known dataset Movie lens. We have taken two different measures so as to evaluate the group recommender strategies. The obtained results seem promising and our strategy has shown an empirical robustness compared with the baselines in the literature.

Analysis and visualization of the geographical distribution of atlantic forest bromeliads species

This work presents a spatial distribution analysis of Brazilian Atlantic Forest Bromeliad species catalogued by the Rio de Janeiro Botanical Gardens Research Institute. Our analysis aims at identifying probable endangered species and conservation areas (environmental reservations) for specific Bromeliad species. For that, we propose a solution using Data Mining techniques and the data visualization through the TerraView tool. Thus, it was possible to define where are typical species, allowing more effective field tours.

Temporal Profiling for Opportunistic Partnership Recommendation

Locating experts or work partners in large communities can sometimes be hard. The most common way of accomplishing this task is through recommendations from known acquaintances. This networked search for others who fit required profiles is a form of social navigation. However, needs, interests and expertise change rapidly, so time is an important factor in this type of situation, and recommendations need to be time-sensitive. This paper presents a peer-to-peer system for social network navigation, which builds user profiles through a temporal analysis of ongoing activities and matches these to find opportunities for collaboration.