Real-time search method, device and system

Abstract

The invention discloses a real-time search method. The method comprises the following steps: S1, setting interest point data specified by a system; S2, according to the interest point data, capturing associative data from a target web site into the system; S3, according to a preset data acquisition cycle, traversing the target web site; S4, judging whether an updated target web site exists in the target web site, wherein the updated items include newly-presented web pages and changed web pages; if no updated target web site exists in the target web site, back to the step 2; otherwise turning to a step S5; and S5, capturing associative data on the updated target web site into the system and updating the target web site so as to achieve synchronous acquisition. The invention also discloses a real-time search device and a real-time search system. By using the real-time search method, the real-time search device and the real-time search system disclosed by the invention, instant information can be searched in real time at high speed with small resource occupation.

Description

technical field [0001] The invention relates to the field of network search, in particular to a method, device and system. Background technique [0002] Information is an important factor most closely related to people's work and life, ranging from the whole world to every enterprise, business, and even a family and individual. Although the technology of search engines has become more and more advanced in these years, there is still a big problem no matter whether the information search on the Internet is successful or not. Anyone who has used a search engine has experienced this feeling: sometimes you will not find the results you want, on the contrary, sometimes you will find millions of unnecessary results. In fact, the second result is the most troublesome and difficult to deal with. If you want to find the information you really need from these millions of search results, it is like looking for a needle in a haystack. [0003] Suppose the Internet is a giant library,...

AI Technical Summary

Problems solved by technology

[0010] 2. The merging mechanism of incremental index and full index database is complicated and difficult to control

If a single incremental index and a single full index are used, the merge process will be extremely slow due to the full index becoming too large in the long-term operation, which will also affect the retrieval performance

If the method of multi-level incremental index and multi-level full index library is adopted, the update and delete operations of existing data contained in the incremental index will be distributed in multiple full index libraries, and additional management is required when merging Agencies assist in processing, which greatly increases the complexity of the system, and is also prone to data inconsistencies

[0011] 3. Traditional indexes usually create an index for a specific application. Each specific index and its supporting resources (such as tokenizers, similarity calculators, etc.) are independent, and supporting resources between multiple indexes cannot be shared

For example, the thesaurus of the tokenizer will take up a lot of memory. If multiple indexes are deployed on the same server, each index must load a thesaurus independently, resulting in a lot of memory waste.

Images