基于多Agent的电子商务个性化推荐系统研究-管理科学与工程专业论文.docxVIP

AbstractElectronic Abstract Electronic business has been developed tremendously in just few years．The trade volume for global E-business transactions has reached approximately 13 trillion US dollars in 2006，which accounted for 20％of worldwide business transactions．The scope，therefore，makes it more and more difficult for users tO glean rapidly the agcurale information the IntemeL T0 set up well·eonnocted recommendation platform for E-business,therefore，will benefit the nsers for locating quickly the interesting goods among many E-business websites，OR one hand,and increase the influence ofthese websites，securing and promoting users’fidelity and carrying out interactive sales,on the other hand． It seems that the mass—public recommendation pattern is hardly satisfied with the need for personal individuality．The recommendation services of E-business have been transformed from the initial browsing ofclassification ofcommodity catalogs to the recommendation for peculiarity by delving intO the potential interests ofusers． There m four approaches for individual recommendation aimed at v撕ous consuming-patterns of users：on users-visiting pattern,on users·appraising pattern，on the peculiarity of texts and on credit-assessing of users．The operational mechanisms and procedures of these strategies will be analyzed in details tO dapiet thoroughly the patterns of data and algorithm needed by every strategy． The originality ofprogram using Agent method has been initiated from the perspective the society．The sottware system Can embody the social intelligence ofhuman being,making it to be more flexible and adaptable．Base the E-business peculiarities of Agent,the main researching works for recommendation systems Can be as follows：analyzing the structure relationship of the system to construct its structure in frame，setting up the structure of model Agent based on VarioIls strategies,describing thoroughly the communication protOcols and connection patterns．and explaining the operational