Social network event detection method based on kleinberg online state machine

Abstract

The invention discloses a social network event detection method based on a Kleinberg online state machine, comprising the following steps: S1, acquiring tweet data of a social network, and preprocessing the acquired tweet data; S2, adding tweet text Quantitative clustering, dividing the text according to the similarity of the text; S3, using the Kleinberg state machine to establish a burst detection model for the generation time interval sequence of the word-related text, and identifying the burst structure of the word; S4, judging the burst event. The invention adopts a continuous time model, which can identify the burst structure information of word features in a fine-grained manner, which helps to alleviate the early detection of social network events; it can relatively comprehensively detect the word burst features of events, and is suitable for streaming data , using the emergent structure relationship and co-occurrence relationship of events, the accuracy of social network event detection can be improved.

Description

technical field [0001] The invention relates to a social network event detection method based on a Kleinberg online state machine. Background technique [0002] The rise and development of social networks have brought great convenience and changes to people, and social networks have gradually become an important platform for social media at home and abroad. For example, Twitter is one of the most popular social networking platforms in the world, and more and more users express their views on popular events through the Twitter platform. With its refined content and rapid dissemination characteristics, the Twitter platform generates a large amount of data information reflecting current social emergencies every day. Compared with traditional media, the data information on the Twitter platform can provide researchers with a more comprehensive research perspective . [0003] Events refer to things that happen and have an impact at a specific time and place. Due to the generatio...

AI Technical Summary

Problems solved by technology

However, in the early days of the event, the event has not yet become a popular event, and its related text data is relatively scarce, which is not enough to ensure that the event detection has a good enough effect

Secondly, the massive data flow caused by the flood information dissemination of social networks brings new challenges to real-time event detection

On the one hand, the emergent events in massive data have different scales. Traditional burst detection methods are often related to fixed thresholds, which cannot detect events of different scales under the condition of ensuring the effect of event detection. On the other hand, huge data The scale itself has strict requirements on the computational efficiency and real-time performance of the event detection model

Images