Sentiment Analysis Using Telugu
SentiWordNet
In
this paper using SentiWordNet author is detecting positive or negative
sentences from Telugu sentences, this detection consists of two parts in which
using first part we can detect objective or subjective from sentences and if
objective words appear in the neutral list of SentiWordNet then that sentence
will be consider as Neutral, if words not appear in SentiWordNet Neutral list
then sentence words will check inside positive and negative list of SentiWordNet,
if sentence words found in positive list then sentence will be consider as
positive otherwise negative, if sentences contains words from both positive and
negative list then we take ratio of both positive and negative words and if
positive ratio higher then sentence will be consider as positive else negative.
Formula for ratio
Positive_ratio =
Total_no_postitive_words/total_no_of_words_in_sentence
Negative_ratio =
Total_no_negative_words/total_no_of_words_in_sentence
Based on that above score precision,
recall and fscore will be calculated
Above sentiment detection will be run
inside two algorithms
First algorithm will check objective
words of sentences in SentiWordNet Neutral list
Second Algorithm will check subjective
words of sentence from positive and negative list of SentiWordNet.
In
recent times, sentiment analysis in low resourcedlanguages and regional
languages has become emerging areasin natural language processing. Researchers
have shown greaterinterest towards analyzing sentiment in Indian languages
suchas Hindi, Telugu, Tamil, Bengali, Malayalam, etc. In best ofour knowledge,
microscopic work has been reported till datetowards Indian languages due to
lack of annotated data set.In this paper, we proposed a two-phase sentiment
analysis forTelugu news sentences using Telugu SentiWordNet. Initially,
itidentifies subjectivity classification where sentences are classifiedas
subjective or objective. Objective sentences are treated asneutral sentiment as
they don’t carry any sentiment value. Next,Sentiment Classification has been
done where the subjectivesentences are further classified into positive and
negative sentences.With the existing Telugu SentiWordNet, our proposedsystem
attains an accuracy of 74% and 81% for subjectivity andsentiment classification
respectively.
We
downloaded BOOK REVIEWS sentences from internet to detect sentiment and
downloaded SentiWordNet list also.
BOOK
REVIEWS sentences store inside ‘Book_Reviews_Sentences’ folder
SentiWordNet
list saved inside ‘Telugu_SentiWordNet’ folder
thank you for your comment
pls call me on 8125424511