Project Recommendation Methods and Devices

Abstract

The invention discloses an item recommendation method. The item recommendation method includes: acquiring text information and score information of items; determining semantic feature vectors on the basis of the topic of the items according to the text information of the items; determining the scores of the items to be history feature vectors of the items, determining the feature vectors of the items according to the history feature vectors and the semantic feature vectors and determining prediction scores corresponding to to-be-scored items of a user according to the feature vectors; determining a preset number of the to-be-scored items to be the items recommended to the user. The invention further discloses an item recommendation device. By the technical scheme, similarity among the items can be accurately determined, accuracy of the prediction scores of the to-be-scored items can be improved, and the items newly generated and high in time efficiency requirements can be timely recommended to the user.

Description

technical field [0001] The present invention relates to item recommendation technology, in particular to an item recommendation method and device. Background technique [0002] With the development and popularization of Internet technology, the Internet has become an important way to obtain information. The recommendation device can obtain items that meet user needs from the massive data information on the Internet and recommend them to users. Therefore, it has increasingly become the focus of technology. The items here are recommended to users. objects such as books, news, and applications. Existing recommendation devices usually use a combination of collaborative filtering recommendation, content-based recommendation, and knowledge-based recommendation for item recommendation. The main problems are as follows: [0003] The recommendation device based on collaborative filtering uses the similarity between the item to be rated and the historical item that the user has used ...

AI Technical Summary

Problems solved by technology

[0003] The recommendation device based on collaborative filtering uses the similarity between the item to be rated and the historical item that the user has used and the user's rating of the historical item used to predict the user's rating of the item to be rated, and recommends the item with high rating to the user. The recommendation device Use the text information of the project to extract keywords to calculate the similarity. On the one hand, in order to maintain the semantic information contained in the original text information, the dimension of the extracted keywords is relatively high, resulting in the similarity between different items to be rated and user history items. On the other hand, the recommendation device directly uses the extracted keywords to calculate the similarity, so it can only rely on the same keywords that appear in different items to indicate the similarity between items, and cannot identify synonyms of the same keywords in different items or synonyms, resulting in inaccurate calculation of the similarity between items, resulting in similar similarities between different items to be rated and the user's historical items, and it is impossible to perform differentiated scoring on the items to be rated;

[0004] Existing recommendation devices treat items with high timeliness requirements in the same way as items with low timeliness requirements, so that newly generated items with high timeliness requirements cannot be recommended to users in a timely manner, and even outdated or obsolete items will be recommended

Images