Key Technologies For Information Aggregation Analysis Of E-commerce Based On Meta-synthesis

In recent years,with the vigorous development of information technologies,online shopping market has been developing rapidly and e-commerce has become one of the leading drivers of the economic growth in China.However,with the continuous accumulation of information,it has been more and more difficult for users to obtain an accurate and comprehensive understanding of products,sellers and brands.The dual-structural network,which consists of the Internet architecture and a broadcast-storage secondary structural network,can use the Uniform Content Label(UCL)to achieve effective governance of massive and disordered content data.Although traditional methods for e-commerce information aggregation can analyze the review text qualitatively and quantitatively,there are still many problems that need to be solved.On the one hand,traditional methods for sentiment analysis of review text is coarse-grained.On the other hand,traditional methods for e-commerce information aggregation lack an effective information organizational structure,which can lead to that those methods can neither process e-commerce descriptive information nor take all related review texts as a whole to conduct quantitative analysis.Aiming at solving the problems above,this dissertation puts forward an e-commerce information hall for workshop(ECIHWS),which draws on the principle of Meta-Synthesis,in the dual-structural network.According to the characteristics of ECIHWS,ECALSA(E-Commerce Aspect-Level Sentiment Analysis)and ECIA(E-Commerce Information Aggregation),based on the ECIL_TLI(E-Commerce Information Library with Two-Layer Infrastructure),are proposed in this dissertation.The main work of this dissertation is as follows:1)Concentrating on solving the problem of the coarse sentiment analysis on e-commerce review texts,this dissertation proposes an algorithm named ECALSA.Firstly,ECALSA draws on the PageRank algorithm to extract comment targets based on three features.Secondly,ECALSA adopts a dynamic sliding windows scheme for extracting comment aspects based on the LDA(Latent Dirichlet Allocation)model.Finally,ECALSA expands synonyms and frequently co-occurring field adjectives or adverbs to the basic sentiment dictionary and conducts sentiment analysis on e-commerce review texts with the expanded sentiment dictionary and negative word dictionary.2)Aiming at solving the problems of lacking effective information organizational structure in traditional methods for e-commerce information aggregation,an e-commerce information library named ECIL_TLI and an algorithm named ECIA is proposed in this dissertation.ECIL_TLI is a dynamic e-commerce information library that needs to be continuously updated.On the one hand,ECIA completes the sentiment computing of related review texts within the range of an ECIL_TLI lower-library.On the other one hand,ECIA achieves the integration of e-commerce descriptive information and review texts in the ECIL_TLI to meet the needs of users' divergent thinking.3)Based on the principle of Meta-Synthesis,a prototype system of ECIHWS was designed and implemented in the dual-structural network.Based on the prototype system,ECALSA and ECIA algorithmswere experimented and analyzed.The experimental results verified the feasibility of the above algorithms,which shows that ECALSA can complete the fine-grained sentiment analysis on e-commerce review texts,e-commerce descriptive information and review texts can be effectively organized in ECIL_TLI,and ECIA can achieves the integration of e-commerce descriptive information and review texts in the ECIL_TLI.