Research On Information Extraction Technology For Commodity Information

Information extraction is a text processing technology that extracts the specified type of entity,relationship,event and other factual information from natural language text and forms structured data output.As far as we know,most of the current information extraction focuses on specific fields such as health care,finance,commodity reviews and so on,while the research on information extraction for commodities is less.Therefore,this paper studies the information extraction of commodities,including subject information,entity and entity attribute information,and keyword information.The research contents are as follows:(1)The topic information is extracted by sentence classification.A topic information extraction model based on convolutional neural network and attention mechanism is proposed.The experimental results show that the accuracy of SACNN model is 0.4%-1.4%higher than other CNN models on the benchmark dataset,and the accuracy of SACNN model for extracting subject information of commodities is equivalent to that of benchmark dataset SST-1.The model is theoretical and feasible.(2)This paper proposes a named entity recognition method based on dependency parsing and rules.The extracted information includes entity information and the attribute information of the entity.Firstly,the entity information is obtained by matching the syntactic relations and extraction rules obtained by dependency parsing.Secondly,the entity attribute information is obtained by matching the dictionary with the fixed syntax pattern.The experimental results show that the accuracy rate,recall rate and F1 value of entity and entity attribute extraction are 86%,79%and 82%,respectively.This method is effective and feasible.(3)This paper proposes a keyword information extraction method based on word vector clustering,which realizes key information extraction of commodities.The experimental results show that the F1 values of the experiment are 78.75 and 81.43 respectively when the data set size is 7000 and 700.This method can get better effect of automatic keyword extraction.