A Semi-Supervised E-commerce Review Sentiment Analysis Method Based on Tripartite Graph and Cluster Analysis

Abstract

The invention discloses a semi-supervised e-commerce comment sentiment analysis method based on tripartite graphs and cluster analysis. The method comprises the following steps of calculating word similarity based on word vectors in combination with an emotion dictionary and part-of-speech information; introducing a word group mode to add context information, and eliminating the influence of a one-word multi-meaning phenomenon; establishment a word-document-word group tripartite graph taking the document as the center, calculating the similarity among the documents; based on the sample clustering hypothesis, mining cluster structure distribution in the corpus to obtain global information of the corpus; carrying out weighted fusion on the global information of the corpus and the similarityinformation in the three graphs to obtain a relation graph of the final sample; and executing a label propagation algorithm according to the relational graph, and propagating the labels with the labeled samples to the unlabeled samples to realize the emotion classification of the unlabeled samples. According to the method, the global information and the similarity information in the tripartite graph are subjected to weighted fusion, a high-quality sample relation graph model is obtained on the basis of combining the characteristics of the comment corpus, and a better emotion classification effect can be achieved.

Description

technical field [0001] The invention relates to the field of document classification of natural language processing technology, in particular to a semi-supervised method based on a tripartite graph and a cluster analysis method for sentiment classification technology of e-commerce review documents. Background technique [0002] With the rapid development of the Internet, a large number of user comments have been generated on online platforms such as e-commerce websites, and the emotional information contained in the comments can not only help other users make better purchasing decisions, but also facilitate merchants to track and manage Consumer Feedback Information. Therefore, how to automatically classify the sentiment of user review documents has gradually become a research topic that has attracted more and more attention in the field of natural language processing. [0003] Document sentiment classification methods can be mainly divided into unsupervised learning, super...

AI Technical Summary

Problems solved by technology

However, they still lack thinking about the problem of limited labeled data. Li et al. [Two-View Label Propagation to Semi-supervised Reader Emotion Classification] proposed a semi-supervised label propagation algorithm based on two views of news documents and review documents in 2016. Sentiment classification, taking into account the modeling of labeled data and data relationships

Images