摘要 |
PROBLEM TO BE SOLVED: To convert a Word Embedding of a word which is not present in learning data into a Word Embedding which enables information related to class to be estimated.SOLUTION: A classification model learning system 100 includes a WE acquisition unit 130 and a conversion function leaning unit 140. The WE acquisition unit 130 acquires, for each of plural pieces of text, information related to a class to which text belongs, and a first Word Embedding which is a vector indicating information related to each word included in the text. The conversion function leaning unit 140 learns a conversion function for converting the first Word Embedding into a second Word Embedding which enables estimation of a class to which a text including words having the second Word Embedding belongs, based on information related to the class to which the text belongs, and the first Word Embedding of each word included in the text.SELECTED DRAWING: Figure 1 |