Getting Computers to Understand Us

Explore the future of human-computer interaction in this conference talk from WeAreDevelopers Conference 2017. Delve into the evolution of voice interfaces like Siri, Amazon Alexa, and Google Home, and understand why voice commands are becoming the preferred method of interaction. Learn about Natural Language Processing (NLP), sentiment analysis, and traditional approaches to computer understanding. Discover the Stanford Sentiment Analysis Classification method, feature extraction techniques, and dialogue classification. Examine the challenges of ambiguity in language processing and how they are addressed through engram tagging and contextual analysis. Gain insights into the limitations of current approaches and the potential of unsupervised training and neural networks in advancing computer understanding. Conclude with a look at the data and training requirements for these cutting-edge technologies.

Introduction
Evolution of humancomputer interaction
Applications of computers
NLP
Sentiment Analysis
Traditional Approach
Stanford Sentiment Analysis
Classification
Training
Feature Extractor
Dialogue Classification
Extract Relations
Example
Challenges
Ambiguity
Ambiguous
How we solve ambiguity
Engram tagging
Words around it
Summary
Limitations
Unsupervised Training
Neural Nets
Data and Training
Conclusion