The last time Hackerfall tried to access this page, it returned a not found error. A cached version of the page is below, or clickhereto continue anyway

Making Apps Understand Natural Language | Yahoo Labs

By Nick Cassimatis

Wouldn’t it be great if it took no knowledge of natural language processing or machine learning, and only a few dozen lines of code, to create apps that understand millions of natural language queries and commands? Now you can do just that thanks to scientific advances (and a lot of hard work) that have enabled us to create the SkyPhrase SDK.

Natural language understanding has recently been playing a greater role in many product categories from search to personal assistants to smartwatches. With intuitive commands and queries as brief as a text message, people are now able to accomplish tasks that were previously impossible or too complicated for traditional interfaces and device types. To date, these benefits have mostly been confined to products made by very large companies since it takes many highly-skilled people and huge amounts of training data to build these interfaces. The only cost-effective alternatives in many cases have been interfaces that understand a handful of very basic commands, but are otherwise very brittle.  

To address this issue, the SkyPhrase (Deep Natural Language Understanding) team at Yahoo Labs has been working on a new scientific framework for natural language understanding based on results from recent linguistics and cognitive science research. Until now, most work in computational linguistics has used grammatical and semantic theories that linguists moved beyond several decades ago. More recent research into human language syntax and semantics contains insights that are very difficult to express using these older theories. Also, much research in cognitive science has been converging on the conclusion that a modest set of concepts underlie language and thought about all topics. Little research in natural language processing has taken these developments into account until now. By rethinking the natural language understanding problem using these insights, we have been able to develop new algorithms that can understand language with great precision, while requiring several orders of magnitude less training data and human effort than ever before.

These scientific advances have enabled the creation of the SkyPhrase SDK. In short, by using the SDK, developers can make tasks that now require filling out complex forms and other user interactions much simpler. You can go here to sign up to learn more.

Continue reading on yahoolabs.tumblr.com