Method and system for automatically obtaining news headline, computer equipment and storage medium

Abstract

The invention discloses a method and system for automatically obtaining news headlines, computer equipment and a storage medium, and the method comprises the steps: obtaining the coordinate information of each single-row textbox of a single-frame picture in a news video and the text information in the textbox through OCR, and determining a to-be-selected headline through employing a textbox clustering and character similarity comparison method; using BERT, LSTM and CRF jointly to extract entities of text information, screening out non-headline information through entity recognition results, and determining finally news headlines according to in-out point information of a single piece of news. The method has a good effect on extracting different types of news headlines; meanwhile, manual auxiliary operations such as manual marking and template making are not needed, so that the labor cost can be greatly saved, and the method has profound significance in news headline extraction work.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a method, system, computer equipment and storage medium for automatically obtaining news headlines. Background technique [0002] In recent years, TV news programs have developed and expanded rapidly, and with the popularity of TV, TV news has gradually replaced paper news as the first way for people to obtain news. Among them, news headlines, as a high-level summary of news and topic essence, can be used as content identification and indexing of video clips, which is of great significance for understanding the content of news. However, manual identification of news headlines is time-consuming and laborious. Therefore, automatic positioning, extraction and identification of news headlines are the corresponding It provides a practical and effective way to perform advanced semantic annotation, video database establishment and intelligent retrieval on video streams. It has ...

AI Technical Summary

Problems solved by technology

[0006] However, in practical engineering applications, just getting the OCR recognition results in the news scene cannot accurately distinguish the headlines of the news.

The main reason is that there are scrolling subtitles and news titles in the news, which cannot be achieved by using location information and text information alone. The title can be extracted well. If the template and location information are roughly used to determine the title , then changing a piece of news requires changing the template and position threshold. This method increases the labor cost and is still not advisable.

Images