People interest label extraction method based on social network

Abstract

The invention discloses a people interest label extraction method based on a social network. The method includes the steps: A data preprocessing: cleaning, screening and replacing social network dataof people to form a set comprising a plurality of words; B candidate label deriving: sequentially reading in and judging the words in the set to form a candidate label set comprising topic labels, word candidate labels and word group candidate labels; C interest label extraction: determining a candidate label TF value, calculating a candidate label IDF value, performing ordering according to a candidate label TFIDF value, leading out parts of topic labels into the candidate label set, calculating weight values between the candidate labels, calculating scores of the candidate labels, and acquiring an interest label set. The method at least has the advantages that frequency of an interest label in a document library and a document is considered, influence of a document structure on the interest label is further considered, and more accurate effects can be acquired.

Description

technical field [0001] The invention relates to the technical field of tag extraction, in particular to a method for extracting tags of people's interests based on social networks. Background technique [0002] With the rapid development of Internet applications, social networks have an increasing influence on users. People are increasingly relying on social networks for information exchange and sharing, which has brought about an explosive growth of Internet data. At the same time, users' demand for personalization is becoming stronger and stronger, such as recommending users' favorite products, games, music, movies or News and more. Character interest tags are usually used to describe the identity attributes and interest attributes of characters, which are very helpful for character retrieval and recommendation, character behavior analysis, discovery of character hobbies and character portrait models. [0003] Commonly used interest tag extraction technologies include TF...

AI Technical Summary

Problems solved by technology

[0005] Most of the existing interest tag extraction algorithms use single words as interest tags, while ignoring phrases and unique topic tags in social networks

In addition, the TFIDF algorithm mentioned above only considers the frequency of words in documents and document libraries, but does not consider the text structure of documents.

On the contrary, the TextRank algorithm only considers the role of candidate words in the document structure, but ignores the representation of candidate words in the entire corpus, which is easily affected by meaningless words (such as stop words, etc.)

Images