Research On Sentiment Analysis Of Commodity Short Text Evaluation Based On Improved Adaboost Model

With the advent of the Internet era,online shopping has gradually become a part of people's daily life.The installed capacity of Taobao,Jingdong,Suning e-commerce and other software is increasing.According to the official data of Taobao,only on November 11,2019,the daily activity of Taobao was close to 500 million,and the final turnover reached 268.4 billion,followed by consumers' purchase of their own Product evaluation feedback,how to extract important information quickly and accurately from these data which contain a lot of interference and invalidity,and then feed the extracted information back to consumers and businesses is becoming more and more important,among which analyzing the emotional tendency of online shopping evaluation has become a hot topic for researchers.This paper makes the following research:(1)Aiming at the shortcoming that bat optimization algorithm is easy to fall into local extremum,a bat optimization algorithm based on cosine control factor and iterative local search(CILSBA)is proposed.Firstly,the nonlinear inertia weight based on cosine control factor is added to enhance the accuracy and stability of the algorithm.Secondly,before the end of each iteration,the iterative local search strategy is introduced to disturb the local optimal solution and search the global optimal solution again,which improves the ability of the algorithm to obtain the global optimal solution.The simulation results show that CILSBA can also obtain the optimal solution in high dimension,and the convergence speed of the function is higher than the basic bat algorithm,and the optimal solution,the worst solution and the average value are better than the basic bat algorithm.(2)Combined with the idea of integrated learning,aiming at the defects of weight updating in Ada Boost algorithm,the paper proposes weight threshold and new adaptive weight updating,which greatly reduces the possibility of over fitting phenomenon and improves the accuracy of the model.Through the data collection of online shopping website for network evaluation,and then comparative analysis of the experimental results,the improved Ada Boost algorithm model proposed in this paper improves the accuracy and accuracy of the traditional support vector machine by 2% and 6% respectively.(3)Based on the CILSBA algorithm and the improved Ada Boost algorithm model,an online shopping sentiment analysis system is developed.The system collects the comments of a certain kind of goods through the crawler orientation,realizes and integrates the database and the model designed,and finally achieves a system with comprehensive functions and good effects.In summary,through the real-time classification of the network comment data,it is verified that the method in this paper can get more accurate classification results,and the information in the network evaluation can be presented more clearly and intuitively,which not only improves the efficiency of users to obtain effective information,but also improves the efficiency of merchants to quickly recognize their shortcomings and make improvements.