Making high-precision natural language understanding ubiquitous
The release of Wolfram|Alpha brought a breakthrough in broad high-precision natural language understanding. Now fully integrated into the Wolfram technology stack, the Wolfram Natural Language Understanding (NLU) System is a key enabler in a wide range of Wolfram products and services.
Anyone can immediately use Wolfram|Alpha or intelligent assistants based on it without learning anything. NLU is what makes that possible by providing a zero-length path into a complex computational system.
NLU is also crucial in making real-world connections. Nobody wants to read a manual to know how to refer to something; one just wants to use natural language. But NLU can convert that into a precise symbolic form that's suitable for computation mixing the best of precise computer language and natural language.
The world's portal to computational knowledge
The API powering intelligent assistants and more
Probabilistic query triage in ten milliseconds or less
Programming with natural language
Putting structure into unstructured data
Tagging what's meaningful in documents
Deploy your own natural language grammar
Canonical WDF from natural language
Letting forms use natural language
Wolfram NLU works by using breakthrough knowledge-based techniques to transform free-form language into a precise symbolic representation suitable for computation.
Wolfram NLU has interpreted many billions of queries in Wolfram|Alpha and in well-developed domains, the success rate for understanding web queries is now in excess of 95%.
Part of what lets Wolfram NLU succeed is its access to unprecedented broad and deep computational knowledge in the Wolfram Cloud.
Wolfram NLU is set up to handle complex lexical and grammatical structures, and translate them to precise symbolic forms, without resorting to imprecise meaning-independent statistical methods.
The high performance of today's Wolfram NLU has been achieved partly through analysis of billions of user queries in Wolfram|Alpha.
Wolfram NLU routinely combines outside information like a user's geolocation, or conversational context with its built-in knowledgebase to achieve extremely high success rates in disambiguating queries.
Wolfram NLU has a huge built-in lexical and grammatical knowledgebase, derived from extensive human curation and corpus analysis, and sometimes informed by statistical studies of the content of the web.
Wolfram NLU is set up not only to take input from written and spoken sources, but also to handle the more "stream-of-consciousness" forms that people type into input fields.
Wolfram NLU and its large knowledgebase is all part of the Wolfram Cloud, so it's easy for any system or program to access.
Successful natural language understanding lets even the most complex functionality be used with zero learning and without documentation.
Wolfram NLU has let many millions of people instantly use Wolfram|Alpha without ever explicitly having been told how.
Once speech has been turned into text, Wolfram NLU is broad enough to take whatever has been said and determine what to do.
With Wolfram Smart Fields powered by Wolfram NLU in the Wolfram Cloud, fields in forms, mobile apps, etc. can be interpreted semantically, so users never have to worry about the details of allowed formats.
Being able to use natural language within the Wolfram Language creates a system of great power, in which real-world constructs mix seamlessly with abstract computation.
Wolfram NLU lets you specify simple programs purely in natural language then translates them into precise Wolfram Language code.
Wolfram NLU in the standard Wolfram Language interface lets you type natural language fragments to specify real-world constructs, and have these automatically converted to precise symbolic form.
In some cases (like specifying units of measure), natural language can be much more succinct than precise symbolic language and Wolfram NLU lets you just use the natural language form.
Wolfram NLU can take large volumes of unstructured data and turn it into meaningful canonical WDF.
Whether there are dates or places or names of species, Wolfram NLU can understand them, and turn them into precise WDF with a unique standardized meaning.
Wolfram NLU technology can automatically decode not just individual data elements but also how tabular or other data is arranged and delimited.
When there's lots of data in tabular form, Wolfram NLU looks at whole columns etc. together, and uses machine learning techniques to adapt and optimize the interpretations it gives.
The Wolfram Semantic Server runs in the cloud, providing interactive, API and embeddable access to semantic conversion services.
Through its Programmable Linguistic Interface (PLI), Wolfram NLU is extensible, so you can define your own grammar to ask questions or specify actions then deploy the result to be used from the cloud.
With the Wolfram PLI, you can give grammars that define what natural language forms should generate what underlying Wolfram Language functions, and perform what actions.
The Wolfram PLI lets you start from the complete Wolfram NLU system, then add specific lexical and grammar elements to cover the particular domain you want.
The Wolfram PLI is deployed in the Wolfram Cloud, and is accessible directly in programs, or through an API.
With natural language and the Wolfram PLI, it's possible for users to interact with vastly more complex interfaces than before, routinely taking advantage of system capabilities that were previous inaccessible.