A novel similarity algorithm for fixing erroneous turkish text and detection of roots

Finding roots of words is widely used in document classification and text mining. Computational methods of text similarity are intensely utilized on the English words and successful outcomes are obtained. On the other hand, applying the aforementioned methods on the Turkish words did not give the similar success. In this study, a novel similarity computation algorithm is developed. By using this algorithm it is aimed to find correct words or advice possible alternatives from the written erroneous Turkish words as a highest accuracy rate.