| In order to strengthen the quality and safety management of import and export textile raw materials,Shandong inspection and Quarantine Technology Center needed to obtain the quality and safety information of import and export textile materials from the media,forums,blogs,micro-blog and other websites,and made statistics,analyzed,evaluated,and reported on the information.In view of the actual needs of the quality and safety management of import and export textile raw materials,this thesis analyzed,designed and implemented the "automatic import and export textile material summarization system",the main research work was as follows:(1)Aimed at the low accuracy of ambiguity segmentation in existing Chinese word segmentation,a new word segmentation algorithm,TD-MD algorithm,was proposed.This algorithm divides the ambiguity based on the statistical mutual information and the t-information difference linear superposition in order to improve the accuracy of the ambiguity segmentation.Experiments on two large corpus PDA9801 and PDA9802 of the people's daily are carried out by programming.The experimental results showed that the accuracy of ambiguity segmentation of TD-MD algorithm was 2.72%and 5.95%higher than that of mutual information algorithm and t-information difference algorithm.(2)Aimed at the problem of low quality of abstracts extracted by LDA theme model,a new algorithm of abstracting abstracts,HTMM algorithm,based on HMM model was proposed.The algorithm used relative entropy to calculate the similarity between the sentence theme distribution and the document theme distribution to select the selected text sentences,and combined the multiple features of the document content to choose the candidate sentences.The HTMM algorithm compared with the LDA theme model by quality evaluation tools for abstracts,ROUGE.The results showed that the key information coverage evaluation index was improved by 0.14166,and the evaluation index of the abstract readability was improved by 0.02619.(3)On the basis of algorithm research,OO and UML technology were used to analyze,design and implement the "Automatic summarization system for quality and safety of import and export textile raw materials".The system has realized the functions of data crawled,data preprocessed,data uploaded,summary query and summary analysis and evaluation for the quality and safety information of import and exported textile raw materials.The TD-MD algorithm and HTMM algorithm studied in this paper have improved the quality of abstracts,and had certain reference and reference value for the research and implementation of other Web text automatic summarization algorithms.The system developed meets the quality and safety management of import and export textile raw materials.The actual demand had improved the efficiency and quality of quality and safety management. |
