Traditional crowdsourcing based on cen-tralized management platform is vulnerable to Distributed denial of service (DDoS) attack and single point of failure. Combining blockchain technology with crowdsourcing can well solve the above problems, enabling users to realize peer-to-peer transactions and collaboration based on decentralized trust in distributed systems where nodes do not need to trust each other. Although the current methods have solved the above problems, task publishers select workers based on their reputation values, which has two disadvantages: subjectivity and difficulty in initial value setting. Due to the complexity of crowdsourcing network, there will be malicious users in the network. The requirement for anonymity protects both legitimate and malicious users. In order to solve these problems, we propose an attribute-based worker selection scheme using the private set intersection technology. Our scheme also realizes the malicious user identity disclosure function. A concrete example of our scheme is given.