Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Sentiment analysis of twitter data


Published on

Comparison of semantic analysis approache

Published in: Engineering
  • Login to see the comments

Sentiment analysis of twitter data

  1. 1. Sentiment Analysis of Twitter Data Presented By Team 5 Bhagyashree Deokar (bdeokar) Milinda Sreenath (mrsreena) Rahul Singhal (rsingha2) Rohit Sharma (rsharma9) Yogesh Birla (ydbirla)
  2. 2. Purpose of sentiment analysis Why Twitter Data Challenges of Using Twitter Data Introduction
  3. 3. Simplest Probabilistic Classifier Based on Bayes Theorem Strong(naïve) independence assumption between words in document Considers the frequency of each term in document Multinomial Naïve Bayes Classifier
  4. 4. Based on Recursive Neural Tensor Network Uses Stanford Sentiment Bank Example: “I love this movie.” Recursive Deep Model
  5. 5. Influence of special characters like “@”, “!” eliminated Intelligence added for not recognizing single sentence as multiple sentences Mapping of new words to closest existing words in tree bank Our Contribution - Improvements in Recursive Deep Model
  6. 6. Data Collection using Twitter API Data Preprocessing Execution of Algorithm on 1400 classified tweets Our Work
  7. 7. Parameter/ Algorithm Multinomial Naive Bayes Recursive Deep Model Accuracy 77.03 % 81.6 % Time of Execution 0.06 sec 45.96 sec Result Accuracy Time of Execution Simplicity Ease of Model Learning Multinomial Naïve Bayes Classifier Recursive Deep Model
  8. 8. Considering logical relation between words, Recursive Deep Model provides better accuracy than Multinomial Naïve Bayes Classifier Multinomial Naïve Bayes is simple, easy to train and has less execution time Recursive Deep Model can be enhanced to provide multilingual support Conclusion & Future Direction
  9. 9. Thank You!