Title high-frequency segmentation-based news hotspot phrase extraction method

Abstract

The invention provides a title high-frequency segmentation-based news hotspot phrase extraction method. The method comprises the following steps of extracting a news title for each hotspot topic class; performing word segmentation on the news title, performing statistics on a word frequency of each segmented word, and screening out first n segmented words with the highest word frequencies to serveas a high-frequency word set; searching for a high-frequency segmentation boundary of the news title by using the high-frequency word set, and according to the segmentation boundary, performing segmentation on the news title to obtain candidate phrases; obtaining a candidate phrase set; and performing evaluation on each candidate phrase in the candidate phrase set, and performing screening to obtain the candidate phrases with the highest evaluation indexes, thereby serving as optimal phrases. The method has the advantages that one hotspot phrase which describes topic contents accurately in asimplified way can be extracted for each hotspot topic; a solution is provided for quick summarization and effective display of the hotspot topic contents of current news; and the information displayefficiency and the information obtaining efficiency of a user are improved.

Description

technical field [0001] The invention belongs to the technical field of news text data mining, in particular to a method for extracting news hot phrases based on high-frequency segmentation of titles. Background technique [0002] With the explosive growth of Internet news data, how to identify current real-time news hotspots and present them to users has become more and more important. Based on this purpose, the detection technology of news hot topics was developed. However, the amount of news detected in the same hot topic may still be huge for users and cannot be directly presented to users. How to briefly and accurately summarize hot topics according to the news in the topic has become a key issue. Users only need to By reading the summarized topic description, you can quickly understand the current hot events happening in real time. [0003] At present, the main methods of generalizing topic content can be roughly divided into: abstract type, tag type, keyword type, an...

AI Technical Summary

Problems solved by technology

Among them, the abstract method is to extract multi-document abstracts, which has the following disadvantages: because the length of the multi-document abstract itself is too long, and the splicing order of the summary sentences of each document cannot be determined, and the abstract may have incorrect semantics and references. So this method cannot be effectively applied to this scenario

It has the following disadvantages: due to limitations in the description of words, complete information cannot be accurately expressed like sentence phrases, and there are different syntactic combinations for multiple words. The same keywords or tags may be different due to different order of user understanding produce a completely different meaning

It has the following disadvantages: most of the phrases are too short and can only be regarded as phrases to a certain extent, and in order to describe a topic completely, multiple phrases are still needed, so fragmented descriptions still cannot meet the user's reading habits

[0004] It can be seen that the current various ways of summarizing topic content have certain limitations, and cannot provide a brief and accurate summary of hot topics, and cannot meet the needs of users to quickly understand current hot events happening in real time

Images