Cognitive Speech, Language, Knowledge and Search Services

Put intelligence APIs to work

Together with MSFT Azure we created some cognitive services (Speech, Language, Knowledge, Search) which can be embedded in Applications, Services, Websites, Mobiles or as Skype Contact.

It can work as a chat-bot or with Text-2-Speech and Speech-2-Text module (coming end of Q3) as a speech-interface in almost any language. In addition, the service is able to translate the input as well as the output in real time – Language and Speech (e.g. English to German; German to English):

DEMO BOTs: B2B AppAgile FAQ BOT & B2C Health Insurance Renewal

Both Use-Cases (see description below) are realized at the following BOT.  Check out how it works. Just enter a question to the chat in native Englisch or German language. Just type in only “Hi” or click the mic and say “Hi” 🙂  mic

Note: You also can add the AppAgile BOTs as Skype Contact or open it in a separate window (to use speech service).

Disclaimer: (i) The DEMO BOT shows the possibilities of Azure Cognitive Services. The DEMO BOT doesn’t provide end-to-end functionality! For a productive environment, an end-to-end process has to be defined as well as the system has to be trained in terms of specific customer terms (e.g. include Custom Speech Service [external link]); (ii) The DEMO BOT, which is done in English is using an automated translation service to translate the conversation into the customer language. For productive environments, the system has to be trained at the relevant language.

Get more information out of the BOT

Integrating Application Insights into the BOT provides the possibility to get more information out of the conversation.

Customers admin can see the flow of the entire conversation. Just check out that Analytics-Interface.


Use-Case 1: AppAgile FAQ BOT

As a first example we created a small AppAgile FAQ Chat-Bot-DEMO.

The questions which we used to train the bots are very simple and con be more precise on a real use case:

FAQ_ID Question Alternative question Answer External Links
5 What is AppAle? Can you explain AppAgile?

AppAgile is modular Platform-as-a-Service-Offering (PaaS), which combines technical deployment platforms with fast available business applications from the Cloud. Preconfigured Middleware-Services like databases, application- or Web-Server can be deployed in seconds and can scale according to your needs.

To get more information on how your business profits from AppAgile please tap on the button below [17]

To get more information on how your IT profits from AppAgile please tap on the button below [18]

17 How does my business benefits from AppAgile? What are the business benefits from AppAgile?
Does my business benefit from AppAgile?
The benefits for your business are the following:

  • Platform particularly suited to start with new business concepts – without any expenditures for IT infrastructure and software
  • Test applications for business innovations in the market – before all other competitors
  • Opex instead of capex, pay-per-use price model and open source
  • High costs savings potential in investments in IT, IT operation and licenses, firmly defined SLA
23 Can I deploy my own containers? Is it possible to deploy containers, that are not in the AppAgile repository?
Can I deploy containers from a different repository?

If you like to build your own containers, please consult our Container guideline.

You can find the Container guideline here: http://appagile.io/paas/docker-repository/container-guideline/

http://appagile.io/paas/docker-repository/container-guideline/

The complete example-file for the FAQ Chat BOT you will find >HERE< (FAQ template including FAQs, relations, conversation tree, entities and gimmicks as an simple example).

Used Azure Services

Language

Allow your apps to process natural language, evaluate sentiment and topics, and learn how to recognize what users want.

luis Language Understanding Intelligent Service (external link)

translatorapi Translator Text API (external link)

Knowledge

Map complex information and data in order to solve tasks such as intelligent recommendations and semantic search

qna QnA Maker API (external link)
Speech

Processing spoken language in your applications

 bingspeechBing Speech API  (external link)

Use-Case 2: Product Order BOT

As a second example we created a DEMO Product Order BOT, which covers a fictitious Health Insurance Policy renewal for existing customers (Customers who already have a policy number registered with the insurance provider).

The sample data needed to run the script successfully is shown below, you need only to select one record (Policy Number) in order to have a happy scenario:

Policy Number Name Mobile Current Health Insurance Plan
PI-123-45-67 Taiseer Joudeh +962795888450 Gold
PI-987-65-43 Rom Smith +4915145982267 Silver
PI-654-32-10 Steven Anderson +4915178248519 Basic

The dialog Flow of the Use-Case you’ll find >HERE<.
The below table will illustrates the steps needed to have a complete happy scenario as an example. The complete List you will find >HERE<.

Step Action Notes
1 User: Will start by typing something similar to:

  • “I want to update my insurance plan”
  • “I need to renew my insurance policy”
  • “Can you help me renewing my insurance policy?”
The user can type utterances which contains the Insurance Type s/he interested in, for example s/he can type:

  • “I want to renew my health insurance plan” —-> Health Insurance Type
  • “Can you help me updating my medical insurance plan” —-> Health Insurance Type
  • “I need to upgrade my healthcare plan” —-> Health Insurance Type
  • “Can you help me renewing my townhouse insurance policy?” —-> Property Insurance Type
  • “I want an upgrade on my villa insurance” —-> Property Insurance Type

Typing such utterances which includes the type of the insurance will help the bot to understand that the user is interested in a specific insurance type and will skip step 2, and directly show step 3.

2

Bot: Will reply back by saying “Please select the insurance type you are interested in” and displays a list of available insurance types.

 

User: Will select “Health Insurance” from the cards

For the Demo purposes, only the “Health Insurance” Type is supported, selecting any other insurance type will end the conversation and the bot will respond by: “Right now I can only help with health insurance. Please contact our call center instead on 1-800-555-5555 for other insurance types”
Used Azure Services

This scenario covers different Microsoft Bot Framework capabilities and uses different Cognitive Services, for example it will cover the below Bot Framework features:

  • Different UI elements and rich user controls such as (Hero Cards, Carousel cards, Confirm dialogs, Receipt Cards).
  • Deep conversation flow with various dialogs to help the user complete the flow.
  • Check out experience where the user can pay externally and return the flow to the bot.

For cognitive services, we have used the below:

  • LUIS to understand the user intent when he is asking for insurance renewal or upgrade.
  • Text Analytics which is used to understand the sentiment of the text provided by the user when he is asked for a feedback regarding the bot experience.
Language

Allow your apps to process natural language, evaluate sentiment and topics, and learn how to recognize what users want.

luis Language Understanding Intelligent Service (external link)

translatorapi Translator Text API (external link)

textanalytics Text Analytics API

Knowledge

Map complex information and data in order to solve tasks such as intelligent recommendations and semantic search

qna QnA Maker API (external link)
Speech

Processing spoken language in your applications

 bingspeechBing Speech API  (external link)

Some useful links for bot-usage with Azure

More information about additional AppAgile Azure Hybrid Services you will find >HERE<.