Design and implementation of a semantic search engine for Solidity smart contracts using embedding and vector space techniques

The rapid growth of blockchain technology, smart contracts, and decentralized applications (DApps) has led to a significant increase in the volume of data generated by the Web۳ ecosystem. In such a situation, fast and accurate retrieval of smart contracts has become one of the main challenges for developers and researchers. Conventional search methods that mainly rely on keyword matching face limitations in understanding semantic relationships and retrieving related contracts. In this research, a semantic search engine for Web۳ smart contracts is designed and implemented, which enables fast and accurate retrieval of smart contracts by utilizing the BAAI/bge-small-en-v۱.۵ embedding model, FAISS vector database, and FastAPI-based modular architecture. In the proposed method, after extracting the structural and semantic features of the contracts, their vector representation is created and indexed in FAISS. Then, the user query is mapped to the vector space and the closest matches are retrieved and ranked based on semantic similarity. The performance of the system was evaluated using the Precision@۱۰, Recall@۱۰, NDCG@۱۰, MAP, MRR, and average response time metrics. The evaluation results showed that the proposed system provides better retrieval accuracy, ranking quality, response speed, and scalability than keyword-based methods, and provides a practical framework for intelligent search engines in the Web۳ ecosystem.

کلیدواژه ها:

Smart Contract ، Semantic Search ، FAISS ، Embedding ، Sentence Transformer