【作者】 蒲国林；

【导师】 邱玉辉；

【作者基本信息】 西南大学 ， 基础心理学， 2009， 博士

【摘要】 Web服务作为一种自治的、开放的以及与平台无关的网络化构件,可使分布式应用具有更好的复用性、灵活性和可增长性。Web服务是独立的、模块化的应用,能够通过互联网来描述、发布、定位及调用。而服务质量是判断用户是否获得满意服务的主要因素。Christian Gronroos根据认知心理学的基本理论,提出了顾客感知服务质量的概念。他认为感知服务质量是顾客对服务期望与实际感知的服务绩效之间的比较,即当实际服务绩效大于服务期望时,顾客感知服务质量是好的,反之亦然。Christian Gronroos还指出,服务质量由功能质量和技术质量组成。用户在提出请求服务时,不但对服务功能提出了要求,而且对服务质量也有一定的期望。好的服务质量能提升用户的满意度,不低于用户对服务功能和质量期望的服务都是用户满意的服务。服务质量将是判断服务提供者是否成功的重要因素,它将成为服务的重要卖点和区分点。如何提供可区分的服务,如何为用户提供有质量保障的服务,以及如何让用户获得满意的服务都是服务计算的重要课题。根据Gronroos感知服务质量理论,满意的服务不只是一个技术问题,更是一个心理学问题。因此服务质量直接关系到用户是否获得了满意的服务。虽然现有的服务发现、选择、组合、推荐和主动服务等技术也重视服务质量的研究,但还存在以下一些不足:一是服务质量有哪些维度的问题;二是对服务质量各维度研究不深入;三是在服务质量维度中侧重于技术层面因素而忽略了心理层面的因素。本文在感知服务质量的启发下,尝试在服务质量中引入心理学要素,研究如何为用户提供满意的服务。主要工作包括:1.提出了一种基于QoS的多Agent协同服务模型现有的服务模型以满足用户的功能需求为主,而忽视了用户的服务质量需求。基于该现状,提出了一种基于QoS的多Agent协同服务模型。在模型中主要突出了:基于用户兴趣的个性化推荐、QoS匹配和管理Agent。它们都以服务质量为中心,确保用户获得满意的服务。在模型中,还重点刻画了服务功能匹配的语义,作为后继研究的基础。2.提出了一种Web服务的感知服务质量计算方法。传统的服务质量计算主要是从技术角度进行分析,没有考虑用户的心理因素。而本文提出的Web服务的感知服务质量计算方法,参考了Gronroos感知服务质量模型,使QoS中包含了心理学要素和市场经济学要素。在计算QoS时考虑了三个方面的因素:技术方面、服务提供者方面和服务请求者的心理因素。主要研究了QoS的时间、安全性、可靠性、信任和价格等维度。其中,时间、安全性和可靠性等由网络状态决定;价格则是最能体现市场经济的因素;而信任具有强烈的心理学色彩,带有主观性。最后给出了基于认知心理学的Web服务感知服务质量的加权平均计算方法。感知服务质量是服务发现、服务推荐和服务选择等的直接依据,是判断用户是否获取满意服务的主要标准。3.给出了一种QoS驱动的服务组合化简算法。随着用户需求的不断增长,单个服务已经不能满足用户更深层次的复杂需求。将多个服务组合成一个组合服务,可以满足用户更多更复杂的服务需求,同时也为服务提供者创造了新的商机。复杂服务的服务组合一般很复杂,怎样有质量保障地进行服务组合则是一个关键问题,它直接涉及到该组合服务是否是用户满意的服务。传统的服务组合化简方法在考虑时间复杂度的同时忽略了服务质量。本文在研究了服务组合时的逻辑关系和数据关系的基础上,给出了服务组合的逻辑语义和服务组合的数据语义。在数据关系中,一个重要的数据关系是服务组合时的QoS数据关系,即服务组合时的服务质量约束满足条件。最终提出了一种QoS驱动的服务组合化简方法。该方法以QoS值为参考,有质量保障地对组合服务进行剪枝,无需对复杂路径组合服务进行化简和合并,因而具有较好的时间效率。4.提出了一种基于用户兴趣的个性化推荐方法。传统的服务推荐算法带有盲目性,没有针对用户的个性化特征。不管用户是否喜欢,无论用户是否感兴趣,都以广告的形式进行推荐,命中率低,服务质量差。Mizzaro认为,个性化服务是系统收集和存储用户的使用信息,并分析这些信息得到用户的特定兴趣和需求,然后在合适的时间向每一位访问者发送正确的信息。个性化主要体现在系统针对不同的用户提供不同的服务方式和自动搜索并主动向用户推荐用户最需要和最感兴趣的服务。个性化服务系统的目的是利用用户兴趣模型提供更有针对性的个性化服务,产生最符合用户需求的服务列表。本文提出了一种基于用户兴趣的个性化推荐方法,根据同模式用户兴趣模型进行推荐,因为具有相同兴趣模式的用户对同一资源感兴趣的可能性比较大。在对用户进行同模式判定时,主要考虑了权重、个人行为指数和遗忘因子。实例证明,基于用户兴趣的个性化推荐可以帮助用户发现可能感兴趣的新资源,并培养和发展用户新的兴趣主题,而不局限于用户的现有兴趣和已经感兴趣的资源主题。因此,用户兴趣模型是服务推荐的前提,是个性化服务的基础,直接涉及到服务的质量。总之,本文在感知服务质量模型的基础上,构建了服务模型,并对感知服务质量计算、服务组合化简和个性化推荐方法进行了研究,同时提出了Web服务的感知服务质量计算方法、QoS驱动的服务组合化简算法和基于用户兴趣的个性化推荐方法。本文将感知服务质量应用于Web服务,探索服务过程中的服务质量问题,研究如何为用户提供满意的服务,具有一定的理论和实践意义。但是本论文的研究只是个开端,作者将以此为基础做进一步的深入研究。更多还原

【Abstract】 As an autonomous, open and platform-independent component of the network, Web Services will enable distributed applications with better reusability, flexibility and growth. Web Services are independent, modular applications that can be described, published, located and called via the Internet.The Quality of Service (QoS in short) is a main factor to determine whether users were satisfied with the service. Christian Gronroos in accordance with the basic theory of cognitive psychology presented the customer perceived service quality concept. He believes that the customer perceived service quality is customer expectations and actual perceived service performance comparison between, that is, when actual perceived service performance is greater than the expectations, customer perceived service quality is good, and vice versa. Gronroos also pointed out that the quality of service is composed of functional quality and technical quality. User is not only the service function of the demands, but also has a certain QoS expectations. Therefore the quality of service is not only a technical level, is also a psychological level, it is directly related to user satisfaction with services received. QoS will be an important factor to determine the service provider whether or not success; it will become an important selling point of service.How to provide distinguishable services, how to provide users with guaranteed quality of service, as well as how to provide users with satisfied services are important subjects in services computing? In accordance with Gronroos perceived service quality theory, satisfied service is not just a technical issue, it is a psychological problem. Therefore quality of service is directly about whether users are satisfied with the services or not. Although the existing service discovery, selection, composition, recommendation and proactive services technology also attaches great importance to study the quality of service, but still less than the following: First, the issue of service quality dimensions, and the other is the dimension of QoS studies depth, three dimensions in the QoS focused on the technical level of the factors to the neglect of psychological factors. In this paper, under the Gronroos Perceived Service Quality Model, we presented a service model, also studied three methods in the course of the service: Perceived Service Quality methods for Web Service, QoS-driven service composition simplification algorithm, personalization recommended method based on user interest. Main tasks are:1. A QoS-based Multi-Agent Collaborative Service Model.Existing services Model meet the needs of the user’s functional requirements and have ignored the needs of the user’s service quality. So presents a QoS-based Multi-Agent Collaborative Model for Services. In the model, the main highlights: personalized recommendations based on user interest, QoS matching and management Agent. They are centered on quality of service to ensure that users gain satisfied services. Also we focused on the semantics of the service functions matching.2. Presents a Perceived Service Quality methods for Web ServiceTraditional quality of service is calculated from a technical point of view, does not take the user’s psychological factors. In this paper, we present a calculating method of perceived quality of service for Web service, referenced Gronroos Perceived Service Quality Model, so that QoS factors include psychological factors, and market economics factors. In QoS we take into account three factors: technology, service providers, and service requester’s psychological factor. Main studied time, security, reliability, trust, and price dimensions. Among them, the time, safety and reliability of the decision are controlled by the network status; prices are the best way to reflect the market economy; and trust is with a strong psychology color and with subjectivity. And gives the calculate method of customer cognitive quality of service.3. This paper presents a QoS-driven service composition simplification algorithm.With the growing of user’s demands, individual service can no longer meet user’s complexity needs. Composing multiple services into a composed service, it can satisfy users with more complex services, as well as it provide the new business opportunities. It is complex for composed complex services in general. It is a key issue how to service with high QoS. These reflect user’s satisfaction of services. Traditional composed services simple methods are good of time complexity, and are bad of quality of service. We studied the composte logic relationships and the composite data relationships. Composite logical relationship of services include: the serial logic, circle logic, selection logic, and the concurrency logic. These logical relationships Describes the logic semantics of composite service; service composite data relationships presents the necessary conditions for services composition, it describes the data semantics of composite service. QoS data relationship is an important data relationship in service composite data relationships. And we gave a QoS-driven services compositin simplification method.4. Presents personalized recommendation method based on user interests.Traditional recommended algorithm for the service with blindness, there is no response to the user’s personality characteristics. Whether users like it or not, whether users are interested or not, services are advertised for every one. Its hit rate is low and with poor quality of service. Mizzaro believe that personalized service is that ,system collect and store the user’s information, analysis these and gain the user’s specific interests and needs, and then send the right message to every visitor at the right time. Personalization is mainly reflected in the system for different users with different modes of service delivery and automatic search the users most interested and needs services and recommended him. Personalized service is aimed at users interested in using a more targeted model of personalized service, resulting in the best list of service user needs. The realization of personalized service is based on the real relationship between services, the user preferences and other factors to select and recommend services. Therefore, the user interest model is the premise of the recommended services is the basis for personalized service, directly related to the quality of service.In short, this article constructed a service model based on the perceived service quality model, and studied Perceived Service Quality methods for Web Service, simplify services composition and personalized recommended approach. In this paper, perceived quality of service will be applied to Web Services, and exploring the quality of service issues in the process, has a certain theoretical and practical significance, but the study was only a start, we will study in-depth on this basis.更多还原