Research on construction of semantic dictionary in the football field
Jiguang Wu, Ying Li · 2017
This paper is mainly through the following research process to show: first of all, making a pretreatment to news text corpus in football field, including word segmentation, de-stop words and POS tagging, and then extract the candidate concept set in the processed corpus, And then use the Word2Vec tool to train the corpus to get the multidimensional vector model file for all the words in the corpus, after that combining the statistical algorithm TF-IDF and word embedding to determine the domain concepts. Finally, we calculate all cosine values between any a concept in domain concepts and any a word in model file, and extract three words with higher cosine value as synonyms for a input word to construct semantic dictionary after those words are sorted by value size.