Twitter Sentiment Impact on Bitcoin Pricing

Natural Language Processing ¹ (NLP) used in conjunction with a Support Vector Machine ² (SVM) to classify the sentiment of tweets to determine a correlation with Bitcoin prices.

Note: Not a full/complete project

General Information

Preprocessing is handled by utilizing NLP techniques provided by the Natural Language Toolkit³ (NLTK) to normalize textual data. Textual data is then converted into a Vector Space Model (VSM) with Term Frequency-Inverse Document Frequency⁴ (TF-IDF). A SVM is then used for binary classification of tweets to determine positive and non-positive sentiment of tweets. Tweets can be acquired using TweetStreamer.

In Progress

• Revised Stop Word Removal
• Negation Handling
• Emoji Support and Scoring

References

1. Natural Language Processing Wikipedia
2. Support Vector Machine
3. Natural Language Toolkit Official
4. TF-IDF
5. Sentiment 140 Training