For decision making in government, it is necessary to have well-structured sources of information. In several countries, it is difficult to access government data as the information are dispersed, disconnected, and poorly structured. For this reason, this work presents a framework to gather, unify, and enrich missing person data from distributed web sources. The framework allows inserting new tasks specific to the user’s domain to improve data quality. In this study, Brazilian missing person data from non-governmental organizations (NGOs) and governmental (...) websites were collected and semantically enriched. To enhance the understanding of the gathered missing people cases, we create interpretive models using machine learning techniques to extract knowledge and to encourage the use of standards for publishing the data that are frequently ignored by organizations, hindering analysis and decision-making on data. After the collection and semantic enrichment process, there was an increase of approximately 11% in the data present in the base. Also, the mining process evidenced the disappearance and reappearance of a person in Brazil according to several factors such as age, state initiatives, skin tone, hair colors, etc. (shrink)

Sentiment analysis is the field of natural language processing to analyze opinionated data, for the purpose of decision making. An opinion is a statement about a subject which expresses the sentiments as well as the emotions of the opinion makers on the topic. In this paper, we develop a sentiment analysis tool namely SENTI-METER. This tool estimates the success rate of social campaigns based on the algorithms we developed that analyze the sentiment of word as well as blog. Social campaigns (...) have a huge impact on the mindset of people. One such campaign was launched in India on October 2, 2014, named Swachh Bharat Abhiyan. Our tool computes an elaborated analysis of Swachh Bharat Abhiyan, which examines the success rate of this social campaign. Here, we performed the location-wise analysis of the campaign and predict the degree of polarity of tweets along with the monthly and weekly analysis of the tweets. The experiments were conducted in five phases namely extraction and preprocessing of tweets, tokenization, sentiment evaluation of a line, sentiment evaluation of a blog and analysis. Our tool is also capable of handling transliterated words. Unbiased tweets were extracted from Twitter related to this specific campaign, and on comparing with manual tagging we were able to achieve 84.47 % accuracy using unigram machine learning approach. This approach helps the government to implement the social campaigns effectively for the betterment of the society. (shrink)