A Levenshtein Automaton implemented in C#, for fuzzy searching.
Given a libary of words, and an original word W, Levenshtein Automaton can find all the words similar to W from the libary. The similarity is specified by Levenshtein distance.
The word library is processed and used to construct a TRIE for searching.
When given a word to search, Levenshtein Automaton will first created a NFA (nondeterministic finite automata) based on given word and Max levenshtein distance, then transform the NFA to DFA (deterministic finite automata). After getting DFA, do depth-first-search in this DFA and use TRIE node to validate the jump between states in DFA duing the DFS. When reaching final states in DFA, the word we reach in TRIE at the same time would be one of the fuzzy search results.
For each word in test case or libarary, non-alphabet character is removed and capital letters are transformed to lowercase in pre-processing.
To view more details about Levenshtein Automaton, please refer to http://blog.notdot.net/2010/07/Damn-Cool-Algorithms-Levenshtein-Automata.
The codes transforming from NFA to DFA refers related part of Regex Engine, codes can be viewed in link http://www.leniel.net/2009/05/regex-engine-in-csharp-match-strings.html#sthash.YON5CVU0.dpbs
The project contains implementation of Levenshtein Automaton, and compares its performance when traditional way of fuzzy search (check all words in libraray and calculate Levenshtein distance from given word by Dynamica programming).
"Data" folder/ WordLib.txt: serves as the libary. Three files as test cases, containing words to search.
Program.cs Main entrance to run the project.
Following is result of running test case 2 (TestCase_2_WordsToSearch.txt), with given Max Levenshtein distance value = 1
----Automaton way----
results size for "gay": 1
results size for "add": 1
results size for "he": 2
results size for "it": 2
results size for "share": 1
results size for "do": 2
results size for "ye": 1
results size for "so": 2
results size for "to": 3
results size for "she": 1
results size for "get": 1
results size for "she": 1
Total hits: 0; Max distance : 1; Total time consumed(milisec): 166
----Traditional way----
results size for "gay": 1
results size for "add": 1
results size for "he": 2
results size for "it": 2
results size for "share": 1
results size for "do": 2
results size for "ye": 1
results size for "so": 2
results size for "to": 3
results size for "she": 1
results size for "get": 1
results size for "she": 1
Total hits: 0; Max distance : 1; Total time consumed (milisec): 4894