Wladmir Cardoso Brandão

Assessing the Efficiency of Suffix Stripping Approaches for Portuguese Stemming

Stemming is the process of reducing inflected words to their root form, the stem. Search engines use stemming algorithms to conflate words in the same stem, reducing index size and improving recall. Suffix stripping is a strategy used by stemming algorithms to reduce words to stems by processing suffix rules suitable to address the constraints of each language. For Portuguese stemming, the RSLP was the first suffix stripping algorithm proposed in literature, and it is still widely used in commercial and open source search engines. Typically, the RSLP algorithm uses a list-based approach to process rules for suffix stripping. In this article, we introduce two suffix stripping approaches for Portuguese stemming. Particularly, we propose the hash-based and the automata-based approach, and we assess their efficiency by contrasting them with the state-of-the-art list-based approach. Complexity analysis shows that the automata-based approach is more efficient in time. In addition, experiments on two datasets attest the efficiency of our approaches. In particular, the hash-based and the automata-based approaches outperform the list-based approach, with reduction of up to 65.28% and 86.48% in stemming time, respectively.

Formal Concept Analysis Applied to Professional Social Networks Analysis.

From the recent proliferation of online social networks, a set of specific type of social network is attracting more and more interest from people all around the world. It is professional social networks, where the users’ interest is oriented to business. The behavior analysis of this type of user can generate knowledge about competences that people have been developed in their professional career. In this scenario, and considering the available amount of information in professional social networks, it has been fundamental the adoption of effective computational methods to analyze these networks. The formal concept analysis (FCA) has been a effective technique tosocial network analysis (SNA) , because it allows identify conceptual structures in data sets, through conceptual lattice and implication rules. Particularly, a specific set of implications rules, know as proper implications, can represent the minimum set of conditions to reach a specific goal. In this work, we proposed a FCA-based approach to identify relations among professional competences through proper implications. The experimental results, with professional profiles from LinkedIn and proper implications extracted fromPropImalgorithm, shows the minimum sets of skills that is necessary to reach job positions.

SSV: An Interactive Visualization Approach for Social Media Stock-related Content Analysis

Users interactions in social media have proven to be highly correlated with changes in the Stock Market, and the large volume of data generated every day in this market makes the manual analytical processing impractical. Data visualization tools are powerful to enable this analysis, generating insights to support decisions. In this article we present SSV, our data visualization approach to analyze social media stock-related content. In particular, we present the SSV architecture, as well as the techniques used by it to provide data visualization. Additionally, we show that the visualizations displayed by SSV are not disposed arbitrarily, by contrary, it uses a ranking system based on visualization entropy. Moreover, we perform experiments to evaluate the ranking system and the results show that SSV is effective to rank data visualizations. We also conducted a case study with finance specialists to capture the usefulness of our proposed approach, which points out room

Predicting the Success of NFL Teams using Complex Network Analysis.

The NFL (National Football League) is the most popular sports league in the United States and has the highest average attendance of any professional sports league in the world, moving billions of dollars annually through licensing agreements, sponsorships, television deals, ticket and product sales. In addition, it moves a billionaire betting market, which heavily consumes statistical data on games to produce forecasts. Moreover, game statistics are also used to characterize players performance, dictating their salaries. Thus, the discovery of implicit knowledge in the NFL statistics becomes a challenging problem. In this article, we model the behavior of NFL players and teams using complex network analysis. In particular, we represent quarterbacks and teams as nodes in a graph and labor relationships among them as edges to compute metrics from the graph, using them to discover implicit properties of the NFL social network and predict team success. Experimental results show that this social network is a scale-free and small-world network. Furthermore, node degree and clustering coefficient can be effectively used to predict team success, outperforming the usual passer rating

Exploiting linkedin to predict employee resignation likelihood

Turnover is the organizational movement of hiring and dismissing employees motivated by different reasons, such as termination, retirement, and resignations. High turnover rates can be harmful to organizational productivity, potentially disrupting investments in human resources, generating loss of tacit knowledge and non-scheduled costs with staff replacement. Usually, organizations are unprepared for premature employee resignation, and the problems arising from turnover are even more harmful in this case. Estimating the employees inclination to resignation is paramount to reduce turnover, thereby reducing its negative impact on organizational performance. In this article, we exploit LinkedIn to predict employee resignation likelihood. Particularly, we introduce the ERP approach, which collect professional profiles from LinkedIn and use them as a source of features about employees inclination to resignation. Additionally, we evaluate different algorithms used by ERP to classify employees, considering their resignation likelihood. Experimental results show that the decision tree is the most effective algorithm, classifying correctly more than 88% of the employees. Furthermore the kappa measure show a substantial agreement between the decision tree and an optimal classifier.

Professional Competence Identification Through Formal Concept Analysis

As the job market has become increasingly competitive, people who are looking for a job placement have needed help to increase their competence to achieve a job position. The competence is defined by the set of skills that is necessary to execute an organizational function. In this case, it would be helpful to identify the sets of skills which is necessary to reach job positions. Currently, the on-line professional social networks are attracting the interest from people all around the world, whose their goals are oriented to business relationships. Through the available amount of information in this kind of networks it is possible to apply techniques to identify the competencies that people have developed in their career. In this scenario it has been fundamental the adoption of computational methods to solve this problem. The formal concept analysis (FCA) has been a effective technique for data analysis area, because it allows to identify conceptual structures in data sets, through conceptual lattice and implications. A specific set of implications, know as proper implications, represent the set of conditions to reach a specific goal. So, in this work, we proposed a FCA-based approach to identify and analyze the professional competence through proper implications.

Automatic formal specification generation of APIs by mining unit tests

Specification mining is a research field which combines the power of rigorous mathematical methods and data mining to automatically produce reliable formal specifications. In this article, we introduce MUTE to address the problem of generating a formal specification for API clients by using unit tests of the consumed library as the only input. We evaluate the proposed approach using the popular JDK 6 library and investigate how generated and handcrafted unit tests contribute to the learning process based on an extended framework developed to support automatic accuracy assessment. With precision values always above 83%, we demonstrate how both handcrafted and randomly generated unit tests can be harnessed to learn valid usage scenarios of an API.