Description
In an era of almost-unlimited textual data, accurate sentiment analysis can be the key for determining if our products, services and communities are delighting or aggravating others. We'll take a look at the sentiment analysis landscape in Python: touching on simple libraries and approaches to try as well as more complex systems based on machine learning.
Overview
This talk aims to introduce the audience to the wide array of tools available in Python focused on sentiment analysis. It will cover basic semantic mapping, emoticon mapping as well as some of the more recent developments in applying neural networks, machine learning and deep learning to natural language processing. Participants will also learn some of the pitfalls of the different approaches and see some hands-on code for sentiment analysis.
Outline
- NLP: then and now
- Why Emotions Are Hard
- Simple Analysis
- TextBlob (& other available libraries)
- Bag of Words
- Naive Bayes
- Complex Analysis
- Preprocessing with word2vec
- Metamind & RNLN
- Optimus & CNN
- TensorFlow
- Watson
- Live Demo
- Q&A