Design Of Online Shopping Commodity Evaluation System Based On Deep Semantics

With both the rapid growth of the amount of information on the Internet and the diversification of information categories,it becomes more and more difficult to obtain the content of effective information quickly and accurately.It has been an urgent problem how to achieve the semantic analysis of text content.On the other hand,with the quick development of online e-commerce in the era of big data,studies on how to use the semantic analysis technology to dig out the potential emotional attitude of text will promote the development of online public opinion supervision and automatic after-sale selection,which is of very important application value.Therefore,this dissertation first studies the semantic meaning of words.Second,the semantic analysis of sentence and short article is discussed.Finally,one two-way long and short term memory neural network is developed to achieve the emotional analysis of texts? semantic content,after combining the convolutional neural network with the attention mechanism.The current work is not only helpful for promoting the application of deep learning network in natural language processing,but also provides specific reference for the accurate recommendation of network media.The main work and achievements are summarized as follows:A.Firstly,the web crawler technology is used to construct a corpus by pre-processing common Chinese words and phrases.Secondly,on the basis of studying the generative question and answer response model,the seq2 seq neural network is used to build an efficient Chinese word definition model for the semantic analysis of word,after combining the attention mechanism with the dropout strategy.The effectiveness of the model is verified by experimentally comparative analysis based on different models.B.Usually,the performance of the TextRank algorithm greatly depends on the initial node?s weight setting,while the acquired text by text extraction involves in redundant information.Therefore,an improved TextRank approach,which includes interdependence syntactic analysis and four factors,is developed to extract the subject-predicate sentence structure of a text and achieve the semantic analysis of sentence level.The four factors consist of the location information of sentence,the closeness degree between sentence and text,the effective term of sentence,and the weight of summative word.The numerical experiments show that the improved algorithm can behave well with the aspects of the accuracy of text abstract extraction and the closeness degree between the acquired text and the text topic.C.After intensively analyzing the basic principles and model structures of the convolutional neural network and the long and short-term memory network,we design an attention mechanism-based bidirectional long and short term memory network and also a text emotion analysis model,since the long and short-term memory network can only learn the historical information of text.The model is combined with the Python lightweight Web framework in order to obtain one commodity evaluation system which can execute online shopping commodity information.The experimental results show that,compared with the traditional long-term and short-term memory network,the acquired model outperforms the compared models based on the specific indices of recall rate and F-value,while the attention mechanism can improve the effect of the model?s emotion feature extraction.