Knowledge recommendation method and equipment based on similarity

Abstract

The invention discloses a knowledge recommendation method and equipment based on similarity. The method comprises the following steps: obtaining one piece of knowledge which comprises a title part and a content part; independently carrying out word segmentation processing on the content part and the title part, determining the weight of each segmented word, and independently determining the characteristic words of the content part and the title part according to the weight; according to the characteristic words of the two parts, determining a characteristic word set of the obtained knowledge; forming a new characteristic word set with the obtained characteristic word set and a characteristic word set of another piece of knowledge stored in a repository, independently determining the frequency of each characteristic word in the two pieces of knowledge in the formed characteristic word set to obtain the word frequency vectors of the two pieces of knowledge; on the basis of the word frequency vectors of the two pieces of knowledge, determining the similarity between the two pieces of knowledge; and under a situation that the similarity is greater than or equal to a threshold value, recommending the other piece of knowledge to the user. Therefore, the accuracy and the authenticity of the similarity can be improved, and the knowledge with the highest similarity is recommended to the user.

Description

technical field [0001] The present invention relates to the field of information retrieval, in particular to a similarity-based knowledge recommendation method and equipment. Background technique [0002] The knowledge recommendation method usually refers to finding the knowledge matching the knowledge input by the user from the existing knowledge stored in the knowledge base according to the current knowledge input by the user, and recommending it to the user. [0003] In the existing knowledge recommendation methods, some methods perform knowledge matching based on knowledge categories. This method only uses categories as the only matching standard for knowledge similarity, which may lead to low correlation of knowledge content and poor matching accuracy. Most of the knowledge recommended to users by this method is not the knowledge that users are really interested in. [0004] There are also some methods for knowledge matching based on knowledge content. This method us...

AI Technical Summary

Problems solved by technology

This may lead to the fact that the final extracted keywords do not include the keywords in the title, resulting in inaccurate selection of keywords (also called "feature words"), which may not be truly representative, and may eventually lead to failure to recommend for users. best matching knowledge

Images