Multiple features fused bidirectional recurrent neural network fine granularity opinion mining method

Abstract

The invention discloses a multiple features fused bidirectional recurrent neural network fine granularity opinion mining method. The method comprises the following steps of: capturing comment data of a specific website through internet and carrying out labelling and preprocessing on the comment data to obtain a training sample set; carrying out training by using a Word2Vec or Glove model algorithm to obtain word vectors of the comment data; carrying out vectorization after carrying out part of speech labeling, dependence relationship labeling and the like; and inputting the vectors into a bidirectional concurrent neural network to construct a bidirectional recurrent neural network fine granularity opinion mining model. According to the method, attribute words in fine granularity opinion mining is extracted and emotional polarity judgement is carried out through the training of a model, so that plenty of model training time is further saved and the training efficiency is improved; no professionals are required to carry out manual extraction on the attribute words, so that a lot of manpower cost is saved; and moreover, the model can be trained by using a plurality of data sources, so that cross-field fine granularity opinion analysis can be completed, thereby solving the problem of long-distance emotional element dependency.

Description

technical field [0001] The invention relates to the technical field of natural language processing and neural network, in particular to a fine-grained opinion mining method of bidirectional recurrent neural network integrating multiple features. Background technique [0002] At present, with the continuous increase of text data on the Internet, it is very important for data mining and analysis tasks. For the field of text mining and opinion analysis, traditional methods are based on dictionaries, artificial feature templates, and frequent methods for pattern mining. This type of method mainly completes two tasks, one is attribute extraction and entity recognition, and the other is sentiment analysis and polarity analysis based on attribute words. Related research on opinion mining mainly focuses on sentiment classification at the sentence or chapter level. Users are more looking forward to fine-grained opinion mining results. Among the existing mainstream methods of opinion...

AI Technical Summary

Problems solved by technology

This type of method mainly completes two tasks, one is attribute extraction and entity recognition, and the other is sentiment analysis and polarity analysis based on attribute words. Related research on opinion mining mainly focuses on sentiment classification at the sentence or chapter level. Users are more looking forward to fine-grained opinion mining results. Among the existing mainstream methods of opinion mining, the flexibility and scalability of extraction methods using rules need to be improved, while attribute extraction based on hidden Markov models or conditional random fields (CRF) The method cannot deal with the problem of long-distance emotional factor dependence

[0003] Most of the current research work is on opinion analysis and sentiment classification under specific conditions, such as given a comment text and a target word, the sentiment polarity of the word segmentation target word in the current sentence, or based on the given comment text The different attributes and entity words that appear, judge the emotional polarity of each entity, and the premise of being able to complete this task is to have a large amount of labeled data, and mark the target word or attribute word for each piece of data, and mark the emotional polarity at the same time. The general attribute word extraction model also needs the process of manually selecting features and formulating templates. These processes need to be completed by talents with relevant professional backgrounds, which consume a lot of manpower, and the efficiency of manually labeling corpus is low, and mainly uses neural networks and The methods of deep learning are all based on word representation learning as feature input, so traditional linguistic features such as part of speech and dependency relationship will be lost

Images