Research On The Precision Service Recommendation Based On Twice Matching

Service industry has become an important development area for the nationaleconomy, and the modern service industry increasingly catch the populargovernments’ great attention. Overall, China’s industrial activity unit is moving in theservices sector, especially in the emerging service industry direction. IT applicationand emerging service industry promote with each other, modern information networktechnology has spawned the development of new service. With the development ofinformation network technology, service recommendation technology has graduallybecome a common research direction by the business community and academia.Service recommendation technology has been applied from web servicerecommendation gradually to the library recommendation services, cloudrecommendation services, personalized medical care recommendation, financialinformation recommendation services and other fields. At present, the service levelhas been greatly improved as the information technology applied in the field ofservice, but the customer has increasingly strong demand for personalization anddifferentiation, so we still need to constantly absorb advanced technology and ideas tofurther enhance service quality to meet customers’ need. Meanwhile, there are manycurrently existing theories and algorithms about service recommendation which arebased on data on a small scale. Along with the era of big data, we more and moreurgently need service recommended method which adapt to massive data.The dissertation uses a variety of techniques, including data mining, customerrelationship management, intelligent decision-making, knowledge management andother IT technology and management theory and method, and put forward a newservice recommendation theory and method which emphasis on active servicesystematically from the point of view of the supporting technology, knowledgestructure, behaviour analysis and matching process. Explore the customer base andindividual customer management technology, and research the group-level mappingand personalized mappings relationships when customer characteristics, customerneeds and customer service are multidimensional attribute. According to thequalitative feature, determine the "quality" difference among customer base. Based onthe individual characteristic difference, adjust the individual’s rough service to obtainaccurate service content. Ultimately realize the customer service recommendation from customer base grade level to the individual prospective.Firstly, build customer-service knowledge structure model. Build customergroups, individual customers, meta-service and secondary meta-service domainontology model based on Ontology, and clarify the relationship between the variousontology, building customer-service knowledge structure model, obtaining "customerset, customer groups, individual customers" and "meta-service, secondarymeta-service, accurate service " hierarchy. Secondly, accurately measure the similaritybetween customer groups and the difference of individual characteristic based onN-dimensional cloud model. Basing on the existing cloud model, propose the conceptof N-dimensional cloud model, and build a variety of N-dimensional cloud modelgenerators and the N-dimensional cloud model service inference engine.On this basis,propose precise measurement algorithm for multi-attribute customer groups similarityand individual client characteristic difference from the quantitative point of view. Andthen implement the twice matching of service rule. The first match is between thecustomer base and meta-service, exploring the content and temporal association rulesbetween customer base and meta-service to identify the services collections whichmeet certain customer group from the meta-service. The second match is between theindividual customer and secondary meta-service, exploring the content and temporalassociation rules between customer’s individual characteristics and secondarymeta-service, thus reasoning precise service content that meet customer’s personalitycharacteristics. Finally, the instance data simulations show that the proposed twicematching algorithm can improve accuracy and efficiency of customer service.The main innovations of the dissertation are: Firstly, Construction a model ofN-dimensional cloud. Based on the existing one-dimensional, two-dimensional andthree-dimensional cloud model, popularize the common cloud model to anydimension cloud model, propose the definition and algorithm of an N-dimensionalcloud model, Making the cloud model can be more widely used in the managementfield; Secondly, Propose a similarity measure algorithm between customer groups.Based on existing research results, propose a new similarity measure algorithm formulti-attribute groups based on high-dimensional cloud model, which overcomes theshortcomings of traditional methods to meet the nature of the similarity measure. Andverify the validity of the method through empirical analysis; Thirdly, Propose measurealgorithm for individual client’s personality difference. Propose two measure methodsfor individual client’s personality difference, one is based on membership degree, and the other is based on standard customer of customer groups, which can measure thedifference between an individual customer characteristic and their common feature;Firstly, propose the idea of twice matching in the field of service management. Designservice matching inference, from the level of customer groups, to achieve thematching between customer groups and meta-service based on commoncharacteristics, and the match between individual customer and service projects basedon individual characteristics.