private IEnumerable <LevensteinInfo> FindClosestWords(string word, int editDistance)
{
var wordTrigrams = TrigramIndex.GetTrigramsFrom(word);
return(trigramIndex.GetWordListUnion(wordTrigrams).Select(
dictionaryWord => new LevensteinInfo(dictionaryWord, word)).Where(info => info.GetDistance() <= editDistance));
}
public void TestIndexCreation()
{
var index = new TrigramIndex(new List <string>());
Console.WriteLine(String.Join("\n",
index.Trigrams.OrderByDescending(t => t.Value.Count).Select(
t => t.Key + "\t" + t.Value.Count)));
}
public Mispellings(IEnumerable <string> someWords, IEnumerable <string> correctWords)
{
trigramIndex = new TrigramIndex(correctWords);
WordFrequencies = someWords
.GroupBy(w => w, (w, ws) => Tuple.Create(w, ws.Count()))
.ToDictionary(it => it.Item1, it => it.Item2);
var unknownWords = WordFrequencies.Keys.Where(w => !trigramIndex.ContainsWord(w));
var levensteinInfos = RetrieveLevensteinInfos(unknownWords).ToList();
fuzzyDictionary = GetFuzzyDictionary(levensteinInfos);
}
