Core demands
1. Ranking of comments: It is hoped that the comments with high scores and poor comments (the product scores are high, but the comment text is poor) and low scores and good comments (the product scores are low, but the comment text is good) can be distinguished and sorted in descending or ascending order. Promote the conversion rate by ascending the real "praise".
2. Comment tag: the comment module is too simple, so we hope that we can extract and analyze the comment content, define comment tags for each product, so that users can intuitively see other users' views on the product, and guide users to purchase.
Solution
1. Calculate the emotional tendency and emotional value of user comments through Baidu's emotional tendency analysis interface, and sort them according to the emotional value. Really ascending the "praise" to promote the conversion rate.
2. Automatically analyze users' comment concerns and comments through Baidu's comment opinion extraction interface, and output comment opinion tags and comment opinion polarity to gather comments under the commodity. It can support the opinion extraction of user comments on 13 types of products, including food, hotels, cars, scenic spots, etc., help the fresh food store to analyze products, and assist users to make consumption decisions.
Old version comment module: comments are sorted according to submission time, and comment tags are single and fixed
New version comment module: comments are sorted according to the sorting value, and new product evaluation tags are added to gather comments under the product
