What is the best Fuzzy Matching Algorithm (Fuzzy Logic, N-Gram, Levenstein, Soundex ....,) to process more than 100000 records in less time?
I suggest you read the articles by Navarro mentioned in the Refences section of the Wikipedia article titled Approximate string matching. Making your decision based on actual research is always better than on suggestions by random strangers.. Especially if performance on a known set of records is important to you.