A word library-based matching recommendation method, an apparatus, a device, and a storage medium. The method comprises: reading an object name and object attribute data from a preset database, and converting the read object name and object attribute data into a vocabulary list and adding same into a predetermined word library; performing word segmentation processing on the read object name, adding each word after word segmentation to a word set matched in the word library, and establishing a multi-level index table for the words matched with the word set; receiving an object name to be matched input by a user, and then performing word segmentation processing to obtain an initial word set; matching the initial word set with the multi-level index table to obtain a target object set matching the words in the initial word set; and screening objects from the target object set on the basis of a plurality of preset screening conditions, and recommending same to the user. A memory space required by the matching process can be reduced, the calculation time is shortened, and fast fuzzy matching under a bulk data scene is realized.
展开▼
