Search

The analysis of crimes represents a great challenge to law enforcement agencies and organizations that collect information to analyze accurately and effectively the distribution of the behavior of criminal events, considering that the sources to be used in the process of generating intelligence are diverse in content and/or structure. Tools that have contributed to the analysis of fields in which massive amounts of data are available is "data mining", a powerful tool that can be used effectively in the analysis of large amounts of data and in the subsequent derivation of important analytical results. This paper presents a methodology of analysis of crime facts from online newspapers, identifying the different communes where the greatest number of criminal events occur, which gives an idea of potentially more dangerous places, through the detection and geographical mapping of critical points, or the analysis of the nature of the crime through the extraction of entities. Statistics that measure the predictive capacity of the model indicate that the methodology is robust to recognize crime events within the body of the news. [ABSTRACT FROM AUTHOR]

The evolution of web technologies brought a new view over Web published data, making the information available to both humans and machines. However, as societies are built over citizens' needs, it is difficult to determine how governments are performing on addressing these needs. Social networks are an interesting tool for people to express their aspirations, but monitoring peoples' activities to understand their behavior is a difficult task. This paper presents an application of the semantic uplift technique on social networks to store violence and criminality data. The procedure uses the output of natural language processing techniques to classify criminal activity in a general taxonomy, so data can be compared to worldwide statistics. [ABSTRACT FROM AUTHOR]