Historical data provide valuable information for the understanding of human interactions through time. However, mining this data is challenging as the available records are generally noise digitized handwritten, typewritten or press printed documents. In this research proposal, we plan to develop tools and techniques for pre-processing and extracting information from documents of the military dictatorship period that ruled Brazil from 1964 to 1985. The data to be analyzed consists of digitized

more »

... mages of records from DEOPS/SP (São Paulo State Department of Political and Social Order), an emblematic police agency which have monitored (and in some cases, harassed and tortured) hundreds of thousands Brazilian citizens during that period. The idea is to use state-of-the-art powerful artificial intelligence algorithms in conjunction with crowdsourcing techniques to preprocess and extract information from this important period of the Brazilian History.