Word predicting application
Stefan Loska
4/24/2015
Processing the data
- n-gram sets obtained from http://www.ngrams.info/
- for 4-grams: grouped by the initial 3-gram and sorted by frequency
- top <=5 words collected, dictionary: 3-gram | word set
- analogous procedure for 3- and 2-grams
Algorithm
- dictionary (dict) searched for the 3-gram
- if >=5 words found (res_3), return; else search dict for the 2-gram
- combine all search results and again check if have >=5 words; if not, keep searching
- once >=5 collected, return top 5 words
