Method of Extracting Sentiment Labels From Online Agricultural Product Reviews

This paper takes the online consumer reviews of agricultural products in the e-commerce platform as the research object and studies the extraction method of the sentiment labels. Firstly, we use Chinese character segmentation software to segment and label the reviews. Then we use the feature extraction method named TF-IDF to filter the keywords of comment object. After that, we calculate the coherence of each related word by using PMI. We further judge the sentiment word tendencies according to the rule based on user ratings and thus get emotional label collection. Finally, we use the agricultural product review corpus which is captured on the Internet as test data to test the sentiment label set. Higher extraction accuracy and recall rate indicate that this method has a good adaptability in various fields.