Edilson Ferneda

Emerging Technologies of Text Mining: Techniques and Applications

Massive amounts of textual data make up most organizations stored information. Therefore, there is increasingly high demand for a comprehensive resource providing practical hands-on knowledge for real-world applications. Emerging Technologies of Text Mining: Techniques and Applications provides the most recent technical information related to the computational models of the text mining process, discussing techniques within the realms of classification, association analysis, information extraction, and clustering. Offering an innovative approach to the utilization of textual information mining to maximize competitive advantage, Emerging Technologies of Text Mining: Techniques and Applications will provide libraries with the defining reference on this topic.

On the Effectiveness of Candlestick Chart Analysis for the Brazilian Stock Market

Abstract Several techniques have been developed in pursuit of understanding the behavior of the financial market, in an attempt to predict the asset pricing behavior. The candlestick chart created in the 18th century is one of these techniques. In 2006, Greg Morris conducted a study on the effectiveness of this technique for the U.S. capital market. However, no similar work was done for the Brazilian market. In this paper, the behavior of part of the Brazilian capital market was studied using sixteen candlestick patterns. We considered the data series of ten stocks between 2005 and 2009, totaling approximately 40% of Ibovespa (Sao Paulo Stock Exchange Index) turnover. The frequency of confirmation of each pattern was measured along seven exchange sessions after occurrence of such pattern, and results were compared to those presented by Morris. Additionally, adjustments of the observed proportions of hits were tested for their statistical significance. Results found in the frequential analysis showed a discrepancy in relation to Morriss study. Likewise, in statistical analysis few patterns have confirmed the behavior expected of them. In at least one case the trend expressed by data, although significant, was contrary to the original interpretation of the pattern. Therefore, direct application of patterns developed for other markets, times or actions is not recommended. Such results do not allow for an affirmation that candlestick patterns have the power to predict future behavior of stocks traded in the Ibovespa stock market. However, we found statistically significant evidence of the predictive ability of some patterns, which may indicate that the technique must be adapted to the market where it is intended to be used. The main contributions of this paper are a partial replication of Morris’ study for a set of stocks traded in the Brazilian market, and a statistical analysis of the effectiveness of candlestick patterns as predictors of the behavior of those stocks.Click here and insert your abstract text.

Improving image mining through Geoprocessing

Geoprocessing Information Systems (GIS) deal with structured information concerned some geographical localization. So one uses three-dimensional image representation systems in a huge database, where it is possible to insert many data about some interest domain, say, agriculture, economics, industry, demographics and so on. Images are powerful information sources that can soundly support decision making processes. An image can be seen as a set of elements with spatial localization and color. To interpret an image includes deriving clusters and relations between such elements. This article proposes an integration of Geoprocessing and Image Mining to support image based decisions in several domains such as healthcare.

Basic ideas to support pressure ulcers treatment by means of image mining

A Pressure ulcer (PU) occurs in a significant amount of patients that must remain in bed without movements for long periods. To improve its treatment in order to increase patient life quality and to reduce medical healthcare is yet a challenge. It seems to be helpful to search for medical decision making information in both PU images features and in data concerning patient life so far. Once suitably stored such information sets data and image mining procedures are supposed to be able to allow inferences and finding of patient clusters. Is seems to be also possible computationally to generate a PU grade inference that will help medical experts to accomplish therapeutic procedures. Current research efforts with such purposes are here presented.

Text mining: crossing the chasm between the academy and the industry

The existence of a chasm between the development phase and the adoption of new technologies has been widely recognized. Some reasons that make hard the transition academy-industry for new technology are: (a) the weak usability comonly presented by emergent technology in regard to the required ease of ordinary users; (b) few successful experiences reported; and (c) the lack of an adequate methodology to new tools. In this paper we argue that text mining technology is exactly in the chasm point and study the hypothesis (c) mentioned above. The start point of our argumentation is the contradiction posed by the extraordinary amount of information in text form - about 800/0 of all existing information in a company - while the amount of text mining/web mining applications does not go beyond ‘7°/0. At the same time, we observe that the available technological alternatives present an excellent level of maturity, with many functions and adequate interfaces for the common user. The research was carried out by means of a case study in which we used texts issued by a journalistic agency. In order to explore our hypothesis, we applied the CRISP-DM method that was originally conceived for data mining. The contribution of this work includes the examination of the methodological hypothesis for the lack of text mining applications, an experience report in which we describe the steps carried out to apply CRISP-DM to text mining, and the findings in the target domain.

Mining high performance managers based on the results of psychological tests

Selecting high performance managers represents a risky task mainly due to the costs involved in a wrong choice. This fact led to the development of many approaches to select the candidates that best fit into the requirements of a certain position. However, defining what are the most important features that condition a good personnel performance is still a problem. In this paper, we discuss an approach, based on data mining techniques, to help managers in this process. We built a classifier, based in the Combinatorial Neural Model (CNM), taking as dependent variable the performance of managers as observed along their careers. As independent variables, we considered the results of wellknown psychological tests (MBTI and DISC). The rules generated by CNM enabled the arising of interesting relations between the psychological profile of managers in their start point in the company and the quality of their work after some years in action. These rules are expected to support the improvement of the selection process by driving the choice of candidates to those with a best prospective. Also, the adequate allocation of people - the right professional in the right place - shall be improved.

Extracting definitions from brazilian legal texts

In order to avoid ambiguity and to ensure, as far as possible, a strict interpretation of law, legal texts usually define the specific lexical terms used within their discourse by means of normative rules. With an often large amount of rules in effect in a given domain, extracting these definitions manually would be a costly undertaking. This paper presents an approach to cope with this problem based in a variation of an automated technique of natural language processing of Brazilian Portuguese texts. For the sake of generality, the proposed solution was developed to address the more general problem of building a glossary from domain specific texts that contain definitions amongst their content. This solution was applied to a corpus of texts on the telecommunications regulations domain and the results are reported. The usual pipeline of natural language processing has been followed: preprocessing, segmentation, and part-of-speech tagging. A set of feature extraction functions is specified and used along with reference glossary information on whether or not a text fragment is a definition, to train a SVM classifier. At last, the definitions are extracted from the texts and evaluated upon a testing corpus, which also contains the reference glossary annotations on definitions. The results are then discussed in light of other definition extraction techniques.

Digital certification in the Brazilian e-government

This article aims to analyze the role of digital certification in the development process of the electronic government actions in Brazil, as well to provide future perspectives. Qualitative research was done and the data gathering performed through semi-structured interviews. Ten nationally recognized stakeholders were interviewed either for the conceiving of public policies regarding electronic government as well for the implementation and inspection of related projects. The study demonstrates that digital certification is straightly connected with information security. Indirectly, the security that digital certification provides is the major condition for the development of the electronic government, once it can provide the bases for the improvement of internal process in Public Administration and the increase of public services and better quality for the interface between the State and the citizen.

Using data mining techniques for time estimation in software maintenance

Measuring and estimating are fundamental activities for the success of any project. In the software maintenance realm the lack of maturity, or even a low level of interest in adopting effective maintenance techniques and related metrics, has been pointed out as an important cause for the high costs involved. In this paper, data mining techniques are applied to provide a sound estimation for the time required to accomplish a maintenance task. Based on real-world data regarding maintenance requests, some regression models are built to predict the time required for each maintenance. Data on the team skill and the maintenance characteristics are mapped into values that predict better time estimations in comparison to the one predicted by the human expert. A particular finding from this research is that the time prediction provided by a human expert works as an inductive bias that improves the overall prediction accuracy of the models.

Reasoning about External Environment from Web Sources

Most organizations approach internal and external challenges with a varied degree of effectiveness. One of their biggest challenges is the ability to identify and respond appropriately to changes in their external environments. These changes affect not only their technological choices, but also their internal structures and cultures. In this context, we have seen an increasing demand for computational tools capable not only to support information storage but also to help in reasoning about the organizational environment. In particular, it is observed that the availability of a huge set of information in the Web offers a new opportunity to learn and reason about the organizational context. In this paper we present an empirical model to proceed the knowledge extraction from Web sources and support the reasoning process in the Competitive Intelligence domain.