The latest issue of the New Scientist magazine writes about Per Ola Kristensson‘s work on using crowdsourcing and online web sources to create better statistical language models for AAC devices: Crowdsourcing improves predictive texting.

The research paper was published in the Association for Computational Linguistics’  2011 Conference on Empirical Methods in Natural Language Processing. It is published using the open access model and can be read  here. The language models are publicly released and can be found here.