With the rapid development of cloud computing, various types of cloud services emerge. However, it remains a big challenge for cloud users to find suitable and cost-effective services for two major reasons. One is that it can be a possibility that providers dont provide services in accordance with their declared Service Level Agreements(SLA). The other is that it is difficult for customers to describe their requirements accurately. In order to help users select cloud services efficiently, this paper presents a Trust enabled Self-Learning Agent Model for service Matching (TSLAM). In the model, Agents represent cloud entities to take actions and the learning module embedded in broker agents help to capture the implicit demands and find the real service preferences of users. Along with the transactions, under the effect of the learning mechanism, services are classified into different types, and each type is managed by different broker agents, due to the randomness of customers needs faced by different brokers. Moreover, trust is introduced into the process to improve the security of cloud market against malicious behaviors. Extensive experiments prove that TSLAM is able to optimize the service market so that services can be matched to customers requirements efficiently.