Real-time keyword extraction method and device in text streaming environment

摘要

The present invention relates to a method and apparatus for extracting real-time keywords using a micro-batch processing-based TextRank algorithm. A real-time keyword extraction apparatus according to an embodiment of the present invention includes: a data receiving unit for receiving word data of a first sentence input in a text streaming environment; a storage unit for calculating the input word data of the first sentence, generating a micro table in which an operation value of the word data of the first sentence is stored, and storing the operation value in the generated micro table; a word weight calculator for calculating word weights of words included in the word data using a TF-IDF (Term Frequency-Inverse Document Frequency) algorithm based on the calculation values stored in the micro table; a word node graph generating unit that generates a word node graph based on the calculated word weight; an importance value calculator for calculating importance values of words included in the word data using a PageRank algorithm based on the word weight and the number of adjacent word nodes connected in the word node graph; and a keyword extraction unit for extracting keywords according to the calculated importance value.