Introduction
Language Understanding Service, known as LUIS, is a cloud service that uses the power of machine learning to solve meaningful problems.
We can interact with LUIS using text or voice requests, as it is able to understand and interpret human conversation.
LUIS uses natural language to analyze user conversation intentions for customer applications like social network applications (Skype, Twitter, Messenger...), to extract the most important information.
Many powerful tools are implemented:
The LUIS model is created, configured and published
here, the output of a Luis app is a web service endpoint that uses HTTP protocol.
After the LUIS application is published, a client application sends statements to the LUIS Natural Language Processing Endpoint API and receives the results as JSON responses.
Natural language processing
This is a subfield of linguistics, computer science, information engineering, and artificial intelligence concerned with the interactions between computers and human (natural) languages, in particular how to program computers to process and analyze large amounts of natural language data.
LUIS contains a natural language model specific to a domain. You can start the LUIS application with a predefined domain template, create your own template, or merge the elements of a predefined domain with your own custom information.
- The Predefined Model LUIS has many predefined domain models, including predefined intentions, statements, and entities. You can use the predefined entities without having to use the intentions or statements of the predefined model. The predefined domain templates integrate the entire design for you and are a great way to get started with LUIS quickly.
- The LUIS custom template gives you several ways to identify your own custom templates, including intentions and entities. Entities include automatically learned entities, specific or literal entities, and a combination of both.
LUIS keys
Before developing a LUIS application you should understand:
- Utterance
An utterance is text input from the user that your application needs to understand. It may be a sentence, like “Where my next destination ?”, or a fragment of a sentence, like “destination”.
With active learning, your LUIS app is coached by you, based on real-world utterances. You can access or correct the endpoint prediction, retrain, and republish. LUIS learns quickly with this iterative process, taking the minimum amount of your time and effort. Also, note that all of this is automatable, so theoretically, end-users could rate the accuracy of recognition.
- Intent
The intent of a LUIS application is similar to verbs in a sentence, in real use it represents a task or action the user wants to perform.
We can define a set of intents that corresponds to users actions, for example,
All applications come with the predefined intent, "None", which is the fallback intent.
- Entity
An entity represents detailed information that is relevant in the utterance. For instance, in the utterance “1st of January, book to USA”, “1st of January ” is the key date. Entities mentioned in the user’s utterance can help LUIS choose a specific action to book the flight to respond to the user’s request.
To define an entity, we have multiple formats:
- Simple: The generic entity
- List: Defines a fixed list of entities composed from related words in your system
- Hierarchical: Defines a category and member composition in the form of a parent-child relationship.
- Composite: Made up of other entities
- Regex : to define a regular expression
Improve model prediction
Once your LUIS application is published and receives real user statements, LUIS provides active learning of endpoint statements to improve the accuracy of prediction.
LUIS APIs
You can broadly classify LUIS APIs into two main categories. There’s one that your applications call to access the domain-specific model you build. This domain-specific model can be built using point-and-click on the website www.luis.ai.