The Dictionary of information retrieval and its fault-tolerant retrieval

1, the main content: there are spelling errors in the query or there are different spellings of the robustness of the spelling correction technology ①, supporting the dictionary to quickly find multiple data structures; ②, wildcard query, ③, misspelled query, automatic correction technology (independent correction for individual words; for the entire query string Overall correction technology); 2. The data structure of the dictionary search: determine if each query is in the glossary, see: 1, data structures and algorithmic principles behind the MySQL index, 2. From B-Tree, ④, b* tree to R-Tree; 3. In-depth study of B-Tree index ①, ha Method of ②, search tree; There is a way to allow the sub-tree of the internal performance point to change in a fixed interval to achieve the tree insert delete operation, B-Tree: B-tree is characterized by the two-tree multi-layered "flattening" to a layer and get the tree structure, in memory space is not enough to This is very effective when you have to save all the dictionaries and you have to have a subset of the dictionaries resident in the disk, because in this case "flattening" allows for a pre-read of the next-known test when the dictionary is transferred into memory. 3, wildcard query: [K-gram index] 4, spelling correction: see: ①, English version: How to write a spelling Corrector②, Chinese version: How to write a spelling checker ③, Bayesian push Break and its Internet application (iii): Spell check ④, spell correction 5, pronunciation-based correction technology [SOUDEX algorithm] basic idea: For each word item, a speech hash operation, the pronunciation of similar terms are mapped to the same value.