中文摘要：

信任是决定实体双方在互联网上成功交易的一个重要原因，而商家的信誉又是客户选择交易对象的关键因素，传统的信任评估方法对评估对象提出种种假设．不能分辨恶意客户的虚假推荐．因而影响了评估结果的客观性和可信性，针对上述问题，提出了以灰色系统理论为基础、以灰色聚类评估算法为主要内容的信誉报告机制方案，同时证明了实体的灰关联度在t时刻存在的充分必要性、方案首先采用味集群方式采集原始数据，以被评估的实体为聚类实体、以客户对实体关键属性的评分为评估依据，运用灰关联分析得到实体的评估向量，按事先约定的灰类归纳整理，判断聚类实体的灰类，得到相应实体的信任度，方案的评估依据是客户对实体关键属性的实际评分，并且允许用贴近实际的灰数表达形式，避免了种种假设数据：味集群方法采集原始数据，能够有效地避免恶意推荐．实例说明，实体的信任可以被量化，实体间的信任可以具有可比性，方案具有评价可靠、可操作性强、适合软件自动处理等特点，是网络环境中具有实用价值的一种对实体信任度评估的新方法．

英文摘要：

Trust is an important causation between seller and customer in the successful network business, and the prestige of merchant is a vital factor that customers take into account when they select bargainer. In the traditional trust evaluation methods, various hypotheses are put forward for the evaluated entities, and the inveracious recommendation of malicious customers can not be avoided, so the objectivity and creditability of the evaluation result is affected. To the above questions, a mechanism scheme based on the gray system theory is put forward in this paper. It is proved that in the t time, the existence of the entity＇ gray relation degree is sufficient and necessary. In this scheme, original data is firstly collected through using taste concourse method, the evaluated entities are regarded as clustering entities, and the grade point value of their key attributes given by customers are regarded as the basis of valuation. Using a gray relation analytical method, the evaluated vectors of the entities can be gotten. Through inducing and adjusting according to the arranged gray level, gray level of the clustering entity is judged, and trust degree of the entity can be obtained. The actual mark of the entity＇ key attributes given by customers are regarded as the basis of valuation in this scheme. Gray number close to fact is adopted as the expressive mode so that the various hypotheses are avoided. Original data is collected through using taste concourse method and it can effectively avoid malicious recommendation. In the example, it is explained that entity＇s trust is quantified and trust among entities can be compared. The scheme has some advantages, such as credible evaluation, strong maneuverability, and suiting for the automatic processing of software, etc .It is a practical and valuable new method to evaluate entity＇s trust degree in network circumstance.