IJSRP, Volume 6, Issue 1, January 2016 Edition [ISSN 2250-3153]
Vimal P.Parmar, Dr. CK Kumbharana
Abstract:
Neural networks are widely used in computer science in various research applications specifically for pattern recognition. Neural network can solve complex problems by nature and implements solution as being similar to be performed by human being. It is possible to incorporate such a neural network concept with phonetic algorithms that try to identify phonetic similarity among the words that will result in improvement of overall result. Phonetic algorithm based programs are used to identify phonetic similarity in which pronunciation of the different words with different meaning in English language are similar. There exists number of such algorithms with advantages and disadvantages which are introduced here and then after author has derived, designed and developed new algorithm. Each algorithm provides certain result at some extent but by integrating these algorithms in form of neural network results in improvement over any individual algorithm implementation. Neural networks are familiar as a solution for pattern recognition and here efforts are being made to identify similar patterns based on pronunciation of the English words to correct the misspelled English word as well as to suggest number of alternative English words as a result which have similar phonetic identity.