Research on Chinese Short Text Classification of Bidding Project Names with Fusion Feature Item Category Distribution

Abstract

Within the framework of word2vec, aiming at the feature of Chinese bidding project names, this paper proposes a TF-IDF-CDW weighted word2vec model, which combines the Category Distribution Weight (CDW) of feature items to generate short text vectors of project names. The short text vector is constructed in three ways, namely the mean word2vec model, the TF-IDF weighted word2vec model, and the TF-IDF-CDW weighted word2vec model. Finally, the three models are applied to the text classification of bidding project names. The experimental results are compared to verify the effectiveness of the new method.