基于主题预测与信息指纹的文章抄袭判断

In this paper,for a given number of articles, our goal is to determine which of these articles are plagiarized from each other and the likelihood of plagiarism.This article mainly the following several aspects in detail in this paper, the thought and technology theory: first of all, to deal with each article, main purpose is to grab the theme of each article content (that is, for each article keywords), use of technology is the TF - IDF, namely quantitative calculating the weights of every word, that is to get the word to predict the size of the article theme\'s contribution.If a word\'s ability to predict a topic is stronger, the weight is higher, and vice versa.The next main work is to classify these articles according to the degree of similarity of the topic. The more similar the topic is, the more likely it is to be classified into the same category.The technique used is cosine similarity, and then a bottom-up merging method is used to categorize all articles, so that articles with similar topics can be grouped into the same category.Finally, we use the information fingerprint to judge the probability of plagiarism between two articles with similar topics. We process the articles with the same topic respectively to get the information fingerprint of each article, and then make pair-to-pair comparison.