Articulate is a methodology that uses natural language to interact with visualizations. It is part of Project Lifelike - a research initiative to provide a more natural alternative for interacting with computers.

As computing becomes more accessible through the Internet, and more recently mobile phone applications, there is renewed interest in organic and seamless ways to harness the power of this technology. Human speech is a promising and attractive approach.

Articulate is a system that applies natural language processing and machine learning techniques in the area of data visualization and analysis. Its goal is to automatically translate natural language queries into meaningful representations of data. Articulate incorporates a natural language parser with machine learning algorithms to translate spoken language. The user’s input sentence is parsed into a syntactic tree tagged with part-of-speech labels, from which a feature vector can be extracted to represent the query. Based on this set of features, a supervised learning algorithm identifies the intended visual analytic task, which is used to guide a graph generation algorithm in determining the appropriate visualization to present to the user.

The ultimate goal of this research is to allow users to employ advanced visualization techniques without the burden of first having to learn a complex or cumbersome user interface. Articulate is a step towards a time when computers will understand the user's intentions as naturally as other humans do.