Abstract: This paper aims to solve the pain points of low information retrieval efficiency and low accuracy of retrieval results in the data asset management system, and integrates an intelligent retrieval system based on the ranking learning algorithm to improve the relevance of retrieval results and user requests. The theory of ranking learning algorithm is studied, the commonly used ranking learning algorithms are optimized, the classification problem is extended to the text ranking problem, the related objective function and loss function are defined, and the machine learning method is used to improve the accuracy of the retrieval results. The intelligent retrieval system built in vertical distributed search engine technology and ranking learning algorithm improves the efficiency of retrieval request conversion through correlation engineering. Experiments show that this system can enhance the relevance between retrieval sentences and returned results on the basis of optimizing retrieval rate and polish up the accuracy of retrieval.

Key words: computer application, information retrieval, relevant search, learning to rank, Elasticsearch