# \\\|/// # \\ ~ ~ // # ( @ @ ) #**********************************-oOOo-(_)-oOOo-************************************ # # Natural Language Processing # # CS8761 - Dr. Ted Pedersen # # FINAL PROJECT # # An approach to sentiment classification using machine readable dictionaries # and World Wide Web # # # Melgar FBC - Prashant Jain , Anand Takale , Nan Zhang ,Sumalatha Kothadi # # #***************************************-OoooO-*************************************** # .oooO ( ) # ( ) ) / # \ ( (_/ # \_) Algorithm to classify a review as positive or negative using Machine Readable Dictionaries and World Wide Web **Start ** 1. Initialize BigMac and LDOCE interfaces. 2. Get all the words containing "good" in their definition and classify them as good words and make a good word hash. (This would be done for other good words such as excellent, nice etc. which indicate a positive context) 3. Get all the words containing "bad" in their definition and classify them as bad words and make a bad word hash. (This would be done for other bad words which indicate a negative context) 4. Get all adjectives from LDOCE and BigMac and create a hash of that. 5. Collect all the adjectives from a particular review. 6. Using the WebReader module classify the adjectives as good or bad and get a score for an adjective using t-score, odd's ratio and dice coefficient. 7. For every adjective occuring in the review find whether the adjective is present in the good word hash or the bad word hash and assign score to the adjective accordingly.If it occurs in good word hash assign a score of +1 and if it occurs in the bad word hash assign a score of -1 to the adjective. 8. Classify every adjective according to its score. The adjective is classified as an adjective reflecting positive sentiment if it has a positive score. The adjective is classified as an adjective reflecting negative sentiment if it has a negative score. Adjective having a score of 0 is classified as positive adjective. 9. Compute the total scorefor the review according to the score of adjectives. 10. If the score is greater than 0 then it is a positive review and if it is less than 0 then it is a negative review and if it is 0 then the review is classified as Can't Say. 11. Compute the precision, recall and f-measure for the data-set under consideration ** End of the Algorithm **