Keyword extraction method, keyword extraction device, and computer-readable storage medium

Abstract

The present application relates to a keyword extraction method, a keyword extraction device and a computer-readable storage medium. The keyword extraction method includes: separately calculating the correlation between the text vector of the target text and each candidate word in the candidate vocabulary; extracting K candidate words from the N candidate words whose correlation is greater than the correlation threshold, and generating the target text The candidate word joint vector, N and K are both natural numbers greater than 1; respectively calculate the first similarity and second similarity; and K candidate words corresponding to the joint vectors of candidate words whose first similarity is greater than the second similarity are used as keywords of the target text. By comparing the consistency between the title of the article and the extracted keywords, and introducing auxiliary titles for semantic discrimination, the unsupervised problem is turned into a supervised problem, and the accuracy of keyword extraction is improved.

Description

technical field [0001] The present application belongs to the field of computer software applications, especially a keyword extraction method, a keyword extraction device and a computer-readable storage medium. Background technique [0002] When performing natural language processing or news recommendation, it is usually necessary to extract keywords for articles, or to make personalized recommendations based on keywords, so different keyword extraction algorithms are used. [0003] Most of the existing keyword extraction algorithms are algorithms based on word frequency statistics, such as keyword extraction algorithms based on TF-IDF or Textrank, and keyword extraction algorithms based on TF-IDF generally endow TF (word frequency) and IDF (inverse document Term Frequency) with different weights, the product of TF and IDF is TF-IDF (Term Term Frequency Inverse Document Term Frequency Feature), and the words with the highest TF-IDF value are selected as keywords. The algori...

AI Technical Summary

Problems solved by technology

[0004] The keywords extracted by these two algorithms are all high-frequency words, but because important words may not appear many times, it is not comprehensive enough to measure the importance of a word simply by "word frequency". In addition, the above methods may make an article Several synonyms that appear frequently in the text are all selected as keywords, resulting in the repetition of keywords, and only the words contained in the article can be obtained, and the semantic abstraction cannot be carried out, so that the extraction results of keywords are limited to the article and not precise

Images