Fuzzy query method and device and query system

Abstract

The embodiment of the invention discloses a fuzzy query method and device and a query system. The method comprises the steps of providing a single-character inverted index, wherein the single-character inverted index is constructed and generated by storing single characters into an inverted index after dividing a document into the single characters; obtaining a query character string, dividing the query character string into the single characters and inquiring a phrase matching mode in the single-character inverted index with the single characters being phrases. Through the adoption of the fuzzing query method and device and the query system, the achieving difficulty and the complexity of fuzzy query can be lowered, the query speed can be increased, original IR system invasion is reduced, and the system performance is improved.

Description

technical field [0001] The application belongs to the technical field of computer data query, and in particular relates to a fuzzy data query method, device and query system. Background technique [0002] Word segmentation search is an information retrieval method commonly used in existing information retrieval. It mainly divides a piece of text into a string of words according to certain rules, and each word is used as the smallest index unit. With the continuous development of information retrieval technology, the industry is no longer satisfied with word segmentation search, but also expects fuzzy query (or fuzzy search) under massive data. [0003] The fuzzy query usually refers to a fuzzy search with keywords, such as querying the route name "Suzhou-Hangzhou Three-Day Special Tour", the keywords can be "Suzhou-Hangzhou", "Suzhou", "Hangzhou", "Rite Special", " Color tour" and so on. From the 1970s to the present, there have been various fuzzy query methods. Most of th...

AI Technical Summary

Problems solved by technology

However, the method of using k-gram is limited by the selection of k. If k is too small, the post-filtering complexity will be too high. If k is too large, the query granularity will be too large, and accurate query will not be guaranteed.

The selection of k is often difficult to determine a reasonable value because of its own algorithm, and usually requires a large number of value attempts or rich experience, which is difficult to achieve

However, using a round index is easy to cause index expansion, because a word with a length of l will generate l rounds, and using a b-tree to match prefixes is too inefficient

Moreover, both of the above two methods require additional major modifications to the IR (Information Retrieval) system to adapt to complex post-check or prefix matching algorithms

[0004] Existing fuzzy search methods similar to the above require additional large-scale modification of the IR system, which is difficult and complex to implement, and the additional processing workload is large, and the query speed and accuracy of the algorithm itself are difficult to guarantee.

Images