Abstract: Disclosed is a similarity mining method, comprising: acquiring user behavior data and brand word data, wherein the user behavior data includes user search word data and user comment data; acquiring search brand words according to the user search word data and pre-stored brand word data; constructing a corpus important vocabulary library relating to the search brand words according to the user behavior data; using the corpus important vocabulary library as an input of a word vector tool to perform word vector model training to acquire word vectors of the search brand words; calculating a distance between the search brand words according to the word vectors of the search brand words. The invention further provides a similarity mining device, which calculates similarities of the brand words using a clustering algorithm (e.g., word2vector) according to the user search word data and the user post-purchase comment data.

Abstract: Disclosed is a-similarity mining method, comprising: acquiring user behavior data and brand word data, wherein the user behavior data includes user search word data and user comment data; acquiring search brand words according to the user search word data and pre-stored brand word data; constructing a corpus important vocabulary library relating to the search brand words according to the user behavior data; using the corpus important vocabulary library as an input of a word vector tool to perform word vector model training to acquire word vectors of the search brand words; calculating a distance between the search brand words according to the word vectors of the search brand words. The invention further provides a similarity mining device, which calculates similarities of the brand words using a clustering algorithm (e.g., word2vector) according to the user search word data and the user post-purchase comment data.