6

I have constructed an algorithm to transliterate from English to multiple languages, Since we should show them appropriate suggestion for the words they have entered, I have made logic to search in that language dictionary.

Logic I have implemented to search in that language dictionary :

  1. Difference in the last typed vowel and find the words. { Ex : re — > r* }
  2. Replacing all the vowels in all possible combinations . { Ex : test — > [ tAst*, tEst*, tIst*, tOst*, tUst* ] }
  3. Least Distance possible words from dictionary. ( Levenshtein Distance algorithm )
  4. Finding phonetically similar words in dictionary. { Ex : tast —> [ tEst*, tEAst*, .. ] }
  5. Stressing the consonants between vowels and searching in dictionary. { Ex : posible —> [ poSSible* ] }

Is there any standard Algorithms to implement Transliteration and Transliteration suggestions for the above logic ?

Sab
  • 61
  • 4
  • do you really mean transliteration? https://en.wikipedia.org/wiki/Transliteration – Walter Tross May 24 '19 at 10:46
  • 1
    I think the data structure to use would be a [trie](https://en.wikipedia.org/wiki/Trie), but I can't give you precise links to algorithms that can do just what you want. – m.raynal May 24 '19 at 11:04
  • @WalterTross - Yes exactly. The above examples that I have provided are in English so that every one could understand the logic behind it. – Sab May 24 '19 at 11:29

1 Answers1

0

Recurrent neural networks such as LSTM can be used for predicting the next set of words or suggestions. A training set of English language sentences and supposedly their translated versions with all the possible combination of the words in that language like word vectors can be made.

Stardust
  • 1
  • Thanks for your answer: Its not about predicting the next set of words.
I wish to clarify that the doubt is in Transliteration
(https://en.wikipedia.org/wiki/Transliteration). I have tried to achieve transliteration through keyboard substitution for each language.In this case: 1. There can be similar pronounced consonants or vowels in each language. 2. The way people try to type word may vary.For “Peter”- (“Piter”, “Peeter”,..) but the result should be the proper data of that language pronouncing “Peter”. Hence we have to find the nearest dictionary word of that language. – Sab May 27 '19 at 07:19