A topic recognition method for social media events based on keywords

Abstract

The invention discloses a topic recognition method for social media events based on keywords, comprising the following steps: S1, performing noise reduction processing on social media texts; S2, calculating distribution vectors of words and events on topics; S3, extracting social media texts Subject terms of the event; S4. Construct a Bayesian inference model based on subject terms. The event topic word extraction algorithm of the present invention utilizes the difference between words and events about topic distribution to calculate and extract topic words. The event topic classification algorithm based on Bayesian inference can efficiently classify event topics, has high classification accuracy and practicability, and the algorithm proposed by the present invention does not rely on feature engineering, and replaces the training set for different topic classification tasks and retrains It can be used again and has high portability.

Description

technical field [0001] The invention relates to a method for identifying topics of social media events based on keywords. Background technique [0002] Events in social media refer to important things that actually happen at a specific time and place in the real world, which are stored and disseminated in social networks in the form of text and other data. Social media event topic recognition is a new research direction that has emerged with the development of the Internet and the emergence of social media in recent years. Its research purpose is to classify the topic of event texts flooded in social media. Against the backdrop of the rapid development of the Internet and the popularity of various network applications, social networking, as one of the most popular applications in network applications, has a large number of users worldwide. Due to the characteristics of fast and timely information release and wide spread of social networks, many events will be sent by users ...

AI Technical Summary

Problems solved by technology

However, the above method does not make a targeted selection of vocabulary, resulting in too high dimensionality of vocabulary vector representation and sparse feature representation.

At the same time, this method relies on feature engineering to construct features. When migrating this method to other classification tasks, it needs to rebuild the features, and the construction of features depends on the user's experience, resulting in poor portability of the method. When the test set changes, the classification low accuracy

Images