Teneo Platform 7.3 Overview
Conversational Language Understanding
Teneo 7.3 brings Microsoft's Conversational Language Understanding (CLU) closer to the Studio developer with the possibility to create and maintain CLU intent models directly in Teneo Studio.
CLU is a service provided by Microsoft in the Cognitive Services layer in the Azure ecosystem and the service allows users to train and query natural language understanding models using state-of-the-art, transformer-based, pre-trained language models which even provide multilingual capabilities!
A new Account page in Teneo Studio (before opening a solution) allows the Teneo user to manage CLU credentials completely by themselves as long as the user has the correct permission.
The CLU Manager allows to train CLU intent models based on the classes and training data examples added in the Class Manager; it is also here the developer can assign a trained model to the solution or set a specific assignment as the stable version.
The Class Manager brings support for test data examples and linked test examples which can be used when testing the solution's intent model; the import functionality is extended in Studio Desktop to support not only import of classes and training data but also test data.
Test Data Evaluation
With Teneo 7.3 it is now possible to perform Test Data Evaluations of the solution's intent model by using the test data examples provided either directly in a class or as linked test examples.
Due to the implementation of the Deferred Classification described further below, two new buttons are available in the Tryout to allow developers to test inputs either with the new lazy classification or by re-creating the behavior known from previous Teneo versions.
As of Teneo 7.3, developers can restrict the scope of the displayed data in the Classifier by selecting a Revision Range and / or a Date Range.
Flow Variable Description
Developers of Teneo Studio can now add descriptions to Flow Variables both in Studio Desktop and Studio Web.
Visual Warning for Non-Input Consuming Transitions
With the Give Response label in the flow editor, the flow graph kind of indicates that all transitions leaving a Give Response label are also "Get input before continuing" or input consuming. But not all the transitions leaving the Give Response label have to be input consuming, and when they are not, the behavior can be unexpected which is why Teneo 7.3 introduces a visual warning in the flow graph allowing developers to easily spot this difference in transitions and thereby review the wanted behavior of the highlighted transition.
Teneo Generative QnA
The Teneo Generative QnA template is a starter kit which enables projects to quickly set up a Conversational AI application using OpenAI services on the Teneo Platform to use it to answer questions based on a custom controlled source of information.
The Czech NLU Ontology and Semantic Network has been improved with + 9,000 Language Objects and Entities to cater for the completely new Czech Conversational Modules.
Variable names are updated in the Teneo Dialogue Resources to use camelCase conventions.
The Predict Input Processor is modified to cater for the usage of CLU models in Teneo; a new annotation variable is added with information of which intent classifier is used.
Note that due to the deferred or explained below, class annotations are now only created when Predict is called by Engine during the matching process.
Teneo 7.3 brings an update in the behavior of the Teneo Engine with deferred input classification allowing Engine to - on demand - call the Predict Input Processor during the matching process to create class annotations only when references to these are encountered.
IP Messages in Engine Logs
The Input Processors are extended to enable them to send log messages to the engine log after which Engine logs these together with the session Id and the transaction Id (when available).
Technology and Deployment
With Teneo 7.3 comes an entirely new Teneo Manager User Interface (UI) based on React presenting a more modern look and feel, but also allowing to keep the technology up-to-date and secure.
7.3 introduces versioning to the REST API endpoints in preparation for friction-free upgrades in future releases.
Updates to APIs
The Manager, Studio and Inquire APIs are all extended with new endpoints and several model updates.
The 7.3 release includes the resolution of various known issues.
Various pieces of the Teneo reference documentation is updated based on the changes coming with the Teneo 7.3 release; main updates are performed in the following pages and/or sections:
- Class management section:
- New page introducing Teneo Generative QnA
- Class Performance now includes Test Data Evaluation
- New Account page in the Teneo Studio section
- New Flow Scripts is added
- The Variable Naming Conventions have been reviewed by the team and the recommendations updated
- The conceptual overview From Request to Response is updated with a section to explain the new deferred intent classification in Engine
- See the full list of available flows in the new Czech Teneo Dialogue Resource
- The Teneo Manager section is reviewed and updated based on the new Manager in React frontend.