SMS for Humans: Using NLP To Make Text Message Interfaces That Fat Fingers Can Use

05:15 PM - 06:00 PM on August 16, 2014, Room 704

Rob Spectre

Audience level:

intermediate

Watch:

http://youtu.be/QtD5uZpH3xo

Description

6.8 billion people in the world use SMS every day and yet text apps still are stuck with all caps KEYWORD interfaces for programmatic interaction. This live coding presentation explores the practical application of natural language processing to make SMS more human (and more forgiving for those of us with big thumbs) with a few Python tools.

Abstract

There are now more cell phones than there are people on planet Earth. But with so many people and so many languages using the medium every day, SMS apps are still stuck with poor user experiences dependent on all caps KEYWORD interfaces to get HELP for ANYTHING. With the wealth of natural language processing tools available to the Python developer, this paradigm for SMS interfaces no longer needs to exist. Using TextBlob, nltk and Google Translate, we'll investigate how to make a SMS app that works for humans with:

• Sentiment Analysis for Personalized Support Responses
• Lemmatization for Support Routing
• Spelling Correction for keyword recognition
• Language Detection for Internationalization

SMS is a 20 year old technology, but we can do better than a 20 year old interface. Practical Python NLP power to the rescue!