Microsoft Research has provided a public beta of LUIS, Language Understanding Intelligent Service, the fourth element of Project Oxford. Meanwhile beta versions of the other three SDKs are now available on GitHub.
Project Oxford from Microsoft Research provides pre-trained artificial intelligence services in four categories - Face, Speech, Vision and Language Understanding. Reporting the announcement of Project Oxford at this year's Build, Mike James explained by this is such a welcome idea:
The big problem with incorporating AI into your own project is that it is not only complicated but time consuming. Even if you understand the theory to build your own working AI, you need a lot of data and a lot of time spent training it to produce a good performance. Of course, if what you need is something that tackles a standard task, you could use something that had been trained by someone else.
The remaining API for creating language understanding models that can be used with the speech API and its intent recognition is the most flexible and technically demanding and up until now has been in invitation only beta.
Now, as announced in a blog post Ryan Galgon, Senior Program Manager in the Technology & Research group at Microsoft, Project Oxford’s Language Understanding Intelligent Service (LUIS) has public beta availability anyone can create an account to create models to understand intents and entities and to deploy them to an HTTP endpoint or to a device.
Sample code is available to help new users:
In addition to wider availability of LUIS, Microsoft introduced the following new features:
Chinese language support: Responding to a common request it is now possible to create LUIS applications in English or Chinese. In addition, LUIS correctly processes Chinese utterances that include fragments of English.
Application import and export: Now, all of the data you've entered into a LUIS application can be downloaded to a JSON object and new applications can be created by importing that JSON object. This allows developers to copy applications, share applications with others and check their applications into source control – for example, to be versioned alongside the code for the client app that calls LUIS.
Increased coverage of pre-built models: LUIS provides access to many of the same models that power Microsoft products. The number of intents and entities available has more than tripled going from 56 to 196.
Galgon writes in his blog post:
Since launch, we’ve worked closely with early adopters on our MSDN forum, and we’ve added bulk import of unlabeled utterances, regular expression machine-learning features, a super-charged date-time entity and a lot of UI polish.
It is to be hoped that widening access to LUIS will also the rapid rate of development will continue, making it ever easier to add language understanding to apps using readily available models.