Using Text Mining in Finding Criminal Data on the Web

摘要

With the rapid growth of cyber crime, law enforcement agencies are trying rigorously to find new methods to cope with the problem. Most cyber crimes seldom expose themselves explicitly, making the search of criminal information a difficult task. Fortunately, based on our experience, each type of crime has certain properties that can be recognized by human experts or by computers. In this paper we illustrate the use of text mining to find criminal information by an example — cautioning drugs on the Web. The underlying technique of text mining is the application of a special-purpose search engine, the eDetective system, constructed according to Facet Analysis Method (FAM). To formulate our search target, we organize the keywords of sample pages provided by law enforcement agencies in four facets, namely article, form & effect, exchange and marketing, and to represent the concept of cautioning drugs on the Web. A Web page will be compared with the search target and the similarity between them will be determined. Experimental results reveal that the average search precision is 81.70% and the highest F measure is 0.74510, if the exponential model of FAM is used in the Web page comparison. The result shows the applicability of our method in carrying out difficult and complex search tasks.