This paper aims to address the issues of insufficient professionalism and precision in information retrieval within the fisheries domain by designing and implementing a vertical search engine specifically for fisheries. The system employs a specialized structure, integrating a fisheries terminology-based word segmentation algorithm, a query expansion mechanism based on knowledge graphs, and a multi-source fisheries data collection and processing workflow. It establishes a domain knowledge framework that includes various entities such as fishing gear, fish species, fishing grounds, and legal regulations. The system enhances text representation and retrieval relevance by using word segmentation techniques that combine mutual information and left-right entropy, as well as TF-IDF weighting and the vector space model. Experiments show that the system’s response time for retrieving information from fisheries-specific databases is within 1 second, which is a significant improvement compared to traditional search engines. The system demonstrates good domain adaptability and practical value.