A Text Classification Method Based on Naive Bayes

Abstract

The invention discloses a text classification method based on naive Bayesian, comprising: step 1: using a word segmentation tool to form a feature vector of the text to be classified, comparing the feature vector with common words, and removing meaningless words in the text to be classified ;Treat each word s that appears in the classification text i weight w i set; get P(w 1 ,...,w n ) in the training text set D i The probability set Q(w 1 ,...,w n ); put Q(w 1 ,...,w n ) attributes are multiplied to get P(w 1 ,...,w n ) in the training text set D i The prior probability P(w|D i ); Step 3: training text set D i Divide the number of files in the entire training text set by the total number of training texts to get the prior probability P(D i ), P(D i )*P(x|D i ) to get P(w 1 ,...,w n ) in the training text set D i The posterior probability P(D i |w), step 4: repeat steps 2 and 3 to calculate all posterior probabilities; step 5: compare the results of step 4 with the largest posterior probability P(D i ), D i class is P(w 1 ,...,w n ) belongs to the category, the performance of this method is better, and it has a good practical application value in the text classification of power user appeals.

Description

technical field [0001] The invention relates to the field of railway catenary detection, in particular to a text classification method based on naive Bayesian. Background technique [0002] The power customer service department has to face a large number of user appeal information every day. In the traditional mode, the operator classifies the user's appeal information through subjective judgment, and then delivers it to the corresponding department for processing. This method requires manual checking and confirmation one by one, and the informatization and intelligence are seriously insufficient. [0003] The content of text classification of power user appeals is very rich, and these contents are often found in various international conferences and related journals or magazines such as information retrieval, machine learning, knowledge mining and discovery, pattern recognition, smart grid, power science and application, etc. Representative review articles include "Machin...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a text classification method based on naive Bayesian, which solves the technical problem that the existing text classification algorithm is not ideal. Assuming this deficiency independently, the performance of the method is better, and it has a good practical application value in the power user appeal text classification problem

Images