This paper presents a scalable approach for managing trust in peer-to-peer (P2P) environments, focusing on reputation-based trust management at both the data and semantic levels. Traditional trust management methods rely on centralized databases or global knowledge, which are not scalable. The proposed method uses decentralized data structures and algorithms to assess trust based on agents' past interactions, enabling efficient trust management in large P2P systems. The approach is implemented in a P2P environment and does not require centralized services for trust assessment or data management. The paper discusses the challenges of managing trust in decentralized systems, including the need for reliable data collection and the potential for malicious behavior. It introduces a global trust model based on binary trust, where an agent is either trustworthy or not. The trustworthiness of an agent is determined by the number of complaints it has received and filed. The method uses a decentralized storage structure called P-Grid to store and retrieve complaint data, allowing for efficient and scalable trust management. The paper also presents algorithms for computing trust, including a simple majority decision strategy and a more sophisticated approach that recursively checks the trustworthiness of witnesses. The results of simulations show that the method effectively identifies cheaters in P2P environments, even when the number of cheaters is large. The method is robust and can handle varying levels of cheating probability, making it suitable for large-scale P2P systems. The paper concludes that scalable trust management is essential for P2P systems to support more complex applications beyond simple file sharing.This paper presents a scalable approach for managing trust in peer-to-peer (P2P) environments, focusing on reputation-based trust management at both the data and semantic levels. Traditional trust management methods rely on centralized databases or global knowledge, which are not scalable. The proposed method uses decentralized data structures and algorithms to assess trust based on agents' past interactions, enabling efficient trust management in large P2P systems. The approach is implemented in a P2P environment and does not require centralized services for trust assessment or data management. The paper discusses the challenges of managing trust in decentralized systems, including the need for reliable data collection and the potential for malicious behavior. It introduces a global trust model based on binary trust, where an agent is either trustworthy or not. The trustworthiness of an agent is determined by the number of complaints it has received and filed. The method uses a decentralized storage structure called P-Grid to store and retrieve complaint data, allowing for efficient and scalable trust management. The paper also presents algorithms for computing trust, including a simple majority decision strategy and a more sophisticated approach that recursively checks the trustworthiness of witnesses. The results of simulations show that the method effectively identifies cheaters in P2P environments, even when the number of cheaters is large. The method is robust and can handle varying levels of cheating probability, making it suitable for large-scale P2P systems. The paper concludes that scalable trust management is essential for P2P systems to support more complex applications beyond simple file sharing.