A recommendation method and device based on article comments
The invention provides a recommendation method and device based on article comments, belonging to the computer technical field. The method comprises the following steps of: collecting user comments; dividing the comments into N categories by unsupervised clustering; finding out the most representati...
Saved in:
| Main Authors | , |
|---|---|
| Format | Patent |
| Language | Chinese English |
| Published |
11.01.2019
|
| Subjects | |
| Online Access | Get full text |
Cover
Loading…
| Summary: | The invention provides a recommendation method and device based on article comments, belonging to the computer technical field. The method comprises the following steps of: collecting user comments; dividing the comments into N categories by unsupervised clustering; finding out the most representative keywords in the N categories by means of information gain and chi-square test; expanding the keywords by a shallow neural network to generate a review thesaurus of a plurality of review categories; determining whether a comment to be processed matches the comment thesaurus; if the comment to be processed matches any of the comment categories in the comment thesaurus, recommending the article corresponding to the comment category to the user. Therefore, more dimension feature tags can be extended to improve the accuracy of the recommendation of the article so as to overcome the technical problems existing in the prior art that cannot be precisely recommended.
本发明实施例提供的种基于文章评论的推荐方法及装置，属于计算机技术领域。该方法包括：采集用户评论；通过无监督聚 |
|---|---|
| Bibliography: | Application Number: CN201811084474 |