Research On Chinese Short Text Sentiment Analysis Algorithm Based On BLSTM

Sentiment analysis is the basic task of natural language processing and one of the important research directions in the field of artificial intelligence.It has attracted extensive attention of researchers.Sentiment analysis algorithm mainly calculates the current text's emotional tendencies,such as praise and derogation,according to the characteristics provided by the text data,and provides an effective basis for decision-making.As one of the key technologies of human-computer interaction and artificial intelligence,sentiment analysis is widely used in the fields of national defense construction,government management,public opinion analysis,medical and health,commerce and so on.Through large data text analysis,it guides the formulation of national policies,social reform,economic operation of enterprises,daily work and life of individuals.At present,although the existing emotional analysis algorithms have achieved some results,there are still some problems and challenges.For example,the text representation method of word vector has the phenomenon of ambiguity in word segmentation and can't express polysemy.In addition,the common neural network can't better recognize the more important part of sparse features of short text,and can't make full use of the syntactic structure information.In view of the above problems,the main research works of the thesis are as follows: 1.Aiming at the problem that the word vector text representation in Chinese information processing field requires high accuracy of word segmentation and cannot deal with the problem of word segmentation ambiguity and polysemy,an improved BLSTM algorithm based on character vector text representation is proposed.The character vector performs fine-grained representation of Chinese short text,and the context semantic capture is performed by BLSTM,which effectively improves the algorithm's emotion recognition accuracy.2.In view of the problem that the common neural network algorithm cannot pay more attention to the local key features in Chinese short text and the weak ability to fit global information,a sentiment analysis algorithm based on Self-Attention and BLSTM is proposed.After BLSTM encodes the text sequence,Self-Attention is adopted.Attention performs dynamic weight adjustment,and comprehensively considers global semantic training to obtain key features representations,and the performance of the model is greatly improved.3.Aiming at the problem that the existing deep learning algorithms do not consider the hierarchical structure and grammatical information of sentences,an emotional analysis algorithm based on Self-Attention and BLSTM's dividing strategy is proposed.With the idea of dividing and conquering,the final sentence representation is obtained by the bottom-up merging of character vectors.At the same time,the structural information of sentences is introduced.The experimental results show that the algorithm is effective and feasible.