3results about How to "Improve search quality" patented technology

The application discloses a search quality real-time evaluation method and device, equipment and a storage medium, and relates to the technical field of Internet, and comprises the following steps: a sample pool construction module is used for online search based on each preset search vocabulary and corresponding prediction category and entity recognition result to determine a corresponding number of positive and negative samples to obtain a target commodity sample pool; a vocabulary routing module is used for judging whether a target search vocabulary currently input by a user belongs to a preset search vocabulary, and if yes, the target search vocabulary is routed to the target commodity sample pool to perform a search operation; a data uploading module is used for storing detailed data of each search stage into a preset persistent database based on a preset burying point when the search operation is performed; and a quality evaluation module is used for determining a quality evaluation result based on the detailed data and each preset quality index formula. The application determines the search quality by evaluating the detailed data of each search stage when the preset search vocabulary is input, and effectively guarantees the real-time performance of the evaluation.

The application discloses a cross-document type retrieval method and system based on differential nested coding. The method includes six steps: document preprocessing, document type identification, differential coding, index construction and optimization, online retrieval and result fusion. First, the document is standardized and preprocessed to extract structured information. Second, the BERT classifier is used to analyze the document features and identify the document type. Then, according to the document type, the corresponding encoder is selected to generate the original coding vector. Next, the differential nested coding index is constructed and optimized. In the online retrieval stage, the user query is preprocessed, the relevant document type is predicted, and the approximate nearest neighbor retrieval is performed. Finally, the retrieval results are sorted by a stepwise weight fusion strategy, and the top-100 retrieval results are returned. Through differential coding and index optimization, the application realizes efficient retrieval of different types of documents, and has significant practical value and application prospect.