Developing a software for police of the city of Amsterdam to help in identification of human traffic subjects.
One of the most interesting studies in which we participated, was a joint project of police of the city of Amsterdam and Moscow National Research University «Higher School of Economics».
As part of this project was the task of semi-automatic identification of actors involved in human trafficking (human traffic). This area of criminal activity is one of the fastest growing. According to studies, up to two million women a year are involved in this type of illegal business.
Dutch police officers on the required daily reporting of any, even the most inconspicuous events and observations. On the basis of this information is stored database of hundreds of thousands of reports from which potentially can be learned a lot of useful information not apparent contact between the suspects and the social graph of interactions, temporal relations and so on.
The big problem is the fact that the accumulated information is stored in unstructured form, preventing analysis. As a result, one of the main tasks is to extract information from the data set, i.e., the task of data mining.
To solve the problem of semantics allocation was used indexer Lucene in conjunction with a built tezaurusum, including both individual terms found in the reports, as both semantic clusters.
Visualization and data exploration is carried out using formal concept lattice analysis, one of the latest methods of data analysis.
As a result, the program has been successfully implemented in the work of police of the city of Amsterdam. Due to the interaction of experts to analyze data for this product have been successfully detected crimes in the area of human trafficking and provide invaluable assistance in the fight against human trafficking.