Online public opinion text information sentiment polarity classification processing system and method

Abstract

The invention belongs to the technical field of computer science, and discloses an online public opinion text information emotion polarity classification processing system and method, the online public opinion text emotion polarity is widely applied to a public opinion monitoring system, however, a feature engineering extraction module of a traditional machine learning method is large in text information loss, and the accuracy of a classification model is not high enough. The method comprises the steps of preprocessing data; the method comprises the following steps of: constructing a word vector in a way of pre-training a model fin-tuning through BERT; the BERT model calculates the correlation between the characters in the sentence and each of the other characters; the constructed word vector can better solve the problems of'one-word polysemy 'and'synonym' of Chinese; the loss of word vector representation is greatly reduced; in the classification model, firstly Bi-LSTM is used for effectively learning context information, then Attention is used for capturing main semantic information and effectively filtering valuable public opinion information, finally softmax classification is used, and the performance of an obtained public opinion text emotion polarity classification result is better than that of a current mainstream algorithm.

Description

technical field [0001] The invention belongs to the technical field of computer natural language information processing, and in particular relates to a system and method for classifying and processing emotional polarity of network public opinion text information. Background technique [0002] With the development of the Internet, the Internet has become an important platform for people to exchange ideas and express opinions. When hot spots or focal issues in the economy and society are spread on the Internet, tendentious views will be formed and social public opinion with strong influence will be formed. It has a non-negligible impact on social stability and national security. Therefore, in order to meet the needs of network public opinion analysis, it is urgent to propose an emotional tendency classification algorithm for short texts on the network to help relevant organizations mine effective public opinion information from massive text data to assist decision-making. Th...

AI Technical Summary

Problems solved by technology

Word vector construction is mostly based on the word2vec method. The idea of ​​word2vec is to use the relationship between a word and its contextual words in the text, use high-dimensional word vectors to represent words, try to put words with similar meanings in similar positions, but pay attention to The length of the context is fixed

It is not conducive to texts with large changes in length and short sentences

Although the deep learning algorithm has optimized the feature extraction process to a certain extent, there are still deficiencies. For example, CNN cannot recognize the contextual order relationship of processing text. RNN can use internal memory to process input sequences of any time sequence, but it can only perform partial sequences. Memory, which performs poorly on long sequences, LSTM is a special structure type of RNN model, which adds three control units of input gate, output gate, and forget gate, which can better solve the problem of long sequence dependence in neural networks, but Only the following dependencies can be obtained in one direction, but the above information cannot be obtained

[0005] To sum up, the problems existing in the existing technology are: the feature engineering module of the traditional machine learning method has a large loss of text information, and the accuracy of the classification model results is not high enough

Although the deep learning model has been optimized to a certain extent, the word vector representation module is still not accurate enough

The feature extraction process of deep learning cannot handle context dependencies well, and there is a large loss in feature extraction

[0006] The main difficulty in solving the above technical problems lies in the more precise expression from text information to word vectors, on the one hand, and the inability of existing deep learning classification algorithms to accurately identify contextual relationships and extract the most important features of texts, resulting in emotional polarity. Analytical model performance needs to be improved

Images