What new in Microsoft AI?
•Download as PPTX, PDF•
2 likes•46 views
The document discusses several Microsoft AI tools: Bot Framework Composer for building conversational bots visually; a healthcare bot service for the healthcare industry; building a sentiment analysis model using ML.Net; and Cognitive Services' Content Moderator for moderating inappropriate content. It provides overviews and demos of each tool, highlighting Bot Framework Composer's adaptive dialogs, language understanding and generation capabilities, and unified development experience, as well as how to create projects for sentiment analysis and use Content Moderator.
Recommended
More Related Content
What's hot
What's hot (20)
Similar to What new in Microsoft AI?
Similar to What new in Microsoft AI? (20)
More from Cheah Eng Soon
More from Cheah Eng Soon (20)
Recently uploaded
Recently uploaded (20)
What new in Microsoft AI?
- 1. What new in Microsoft AI? By Eng Soon Cheah Microsoft MVP
- 2. Agenda • Bot Framework Composer • Healthcare Bot • Build a Sentiment Model using ML.Net • Cognitive Services : Content Moderator
- 4. Introduction to Bot Framework Composer Within this tool, you'll find everything you need to build a sophisticated conversational experience: • A visual dialog editor • Tools to train and manage Language Understanding (LU) • A powerful language generation and templating system • A ready-to-use bot runtime executable
- 5. Advantage of developing bots with Composer Below are some of the advantages of developing bots in Composer: • Adaptive dialogs allow for Language Generation (LG), which can simplify interruption handling and give bots character • The visual representation and editing in Composer eliminates the need for boilerplate code and makes bot development more accessible • No longer need to navigate between experiences to maintain LU model as it is editable within the app • Time saved with fewer steps to set up your environment and start bots
- 8. With Language Generation, previously complex tasks can be quickly achieved, like: • Including dynamic elements in messages • Generating grammatically correct lists, pronouns, articles • Providing context-sensitive variation in messages • Creating Adaptive Cards attachments, as seen above
- 9. Unified toolset When building bots in Composer, developers will have access to: • Adaptive dialogs and the Bot Framework SDK • Language Understanding service from LUIS • Built-in Language Generation and expression library • QnA Maker capabilities • Bot Framework Emulator
- 10. DEMO Bot Framework Composer [Preview]
- 15. Healthcare Bot
- 16. Prerequisites • Microsoft Azure Subscription
- 17. Create a new Health Bot Service
- 21. Health Care Bot Service Portal
- 29. Build a Sentiment Model using ML.Net
- 30. Prerequisites • Visual Studio 2017/ Visual Studio 2019 • ML.NET Model Builder • The Sentiment140 Dataset.
- 37. Train your Model
- 41. Run the project
- 43. Cognitive Services : Content Moderator
- 44. Prerequisites • Microsoft Azure Account • Visual Studio 2017/ Visual Studio 2019
- 46. References • YouTube https://www.youtube.com/c/engsooncheah • Bot Builder SDK v4 HOL https://github.com/xlegend1024/az-bot-lab • ML.NET Model Builder https://dotnet.microsoft.com/apps/machinelearning-ai/ml-dotnet/model- builder
- 47. References • Content Moderator Tool https://contentmoderator.cognitive.microsoft.com/ • Sample Source Code https://github.com/cheahengsoon/CognitiveServicesContentModerator • Bot Framework Composer https://github.com/microsoft/BotFramework-Composer
Editor's Notes
- Bot Framework Composer is an integrated development tool for developers and multi-disciplinary teams to build bots and other types of conversational software with the Microsoft Bot Framework. Composer enables teams to create bots that use the latest features from the Bot Framework SDK.
- Developers familiar with the Bot Framework SDK will notice differences between bots developed with it and the Bot Framwork Composer. Below are some of the advantages of developing bots in Composer: Adaptive dialogs allow for Language Generation (LG), which can simplify interruption handling and give bots character The visual representation and editing in Composer eliminates the need for boilerplate code and makes bot development more accessible No longer need to navigate between experiences to maintain LU model as it is editable within the app Time saved with fewer steps to set up your environment and start bots A major difference between the current version of the Bot Framework SDK and Composer is that the apps created using Composer uses the Adaptive dialog format, a JSON specification shared by many tools provided by the Bot Framework. You can find more information about Adaptive dialog here. You can manage Composer assests, such as Dialogs, LU training data, and message templates like normal developer assets - files that can be committed to source control and deployed alongside code updates.
- Language Understanding (LU) is a core component of Composer, allowing developers and conversation designers to train language understanding directly in the context of editing a dialog. As dialogs are edited in Composer developers can continuously add to their bot's natural language capabilities through a simple markdown-like format that makes it easy to define new intents and provide sample utterances. Composer detects changes and updates the bot's cloud-based NLU model automatically so it is always up to date.
- Creating grammatically correct, data-driven responses that also use a consistent tone and convey a clear brand voice has always been a challenge for bot developers. Composer's integrated Language Generation (LG) system allows developers, and the writers who work with them, to create replies with a great deal of flexibility.
- Under the hood, Composer harnesses the power of many of the components from the Bot Framework SDK. When building bots in Composer, developers will have access to: Adaptive dialogs and the Bot Framework SDK Language Understanding service from LUIS Built-in Language Generation and expression library QnA Maker capabilities Bot Framework Emulator After creating bots with Composer the resulting bot project will contain reusable assets in the form of JSON and Markdown files that can be bundled and packaged with a bot's source code. These files can be used with a wide variety of compatible tools from the Bot Framework.
- The Health bot Service is a SaaS (Software as a Service) solution made from Microsoft to create almost zero-code bots as health care agents.
- Open the Azure Portal and search for Healthcare Bot Service. Click Create
- Fill out the create form and click Subscribe
- Wait for the deployment to finish and open the resource. Click Configure Account to configure your bot
- This will take you to the Healthcare Bot Service management portal. Fill in the fields and click Create
- In the menu you can see all Healthcare options of the Service.
- If you have already create a scenario you can see it in the list, otherwise all options to create a bot scenario will appear. Click Select a template scenario.
- You can choose any one to experiment. For the purpose of this tutorials I will choose the first Provider Lookup. As the description says it creates a bot that provides the closest health care provider near you.
- Give it a name and select Import Template
- The scenario editor will load. Here we can see a diagram of the bot. Based on each option the bot operates and responds accordingly. On the top right there is the chat box. Try writing help to see the options.
- You can click on the scenario prompts to see how they are configured
- Machine Learning has come to .Net. ML.NET is a library to create custom ML models using C# or F#. Can also be extended to other libraries and frameworks like Infer.NET, TensorFlow, and ONNX for more ML scenarios.
- Create the Project Download and install the ML.NET Model Builder tool from the Visual Studio Marketplace. You can use the link above. Close all VS instances to complete the installation. As soon as the installation has completed open Visual Studio and create a new .NET Core console app.
- Right-click on the project in Solution Explorer and select Add > Machine Learning. This opens the ML.NET Model Builder in a new window in Visual Studio. Model Builder will help us through the process of building a machine learning model.
- To generate your model, you need to select a machine learning scenario. Model Builder offers several templates. Now we are going to select the Sentiment Analysis scenario.
- In Model Builder, you can add data from a local file or connect to a SQL Server database. In this case, you will add the dataset file you downloaded in the Prerequisites step. Select File as the input data source in the drop-down, and in Select a file find and select the dataset file. In my case is the mini dataset I created from the Sentiment140 dataset. Under Column to predict (Label), select “0”. Since the Dataset has no headers it takes the first row. You can add headers to avoid misunderstandings. The Label is what you are predicting, which in this case is the Sentiment found in the first column of the dataset. The rest of the columns are Features, which are attributes that help predict the Label. If you are using the mini dataset 0 is for negative 1 is for positive. If you are using the original dataset 0 is negative 2 is neutral and 4 is positive, so make a note of this.
- After adding your data, go to the Train step. Model Builder evaluates many models with varying settings to give you the best performing model. The default Time to train (the amount of time you would like Model Builder to explore various models) is 10 seconds. Note that for larger datasets, you should set a longer training time. If you add very short duration you will get a pop up with the recommended times. Using random or very short durations can cause training to fail.
- Select Start Training to start the training process. You can keep track of the progress of model training in the Progress section. Status – This shows you the status of the model training process; this will tell you how much time is left in the training process and will also tell you when the training process has completed. Best accuracy – This shows you the accuracy of the best model that Model Builder has found so far. Higher accuracy means the model predicted more correctly on test data. Best algorithm – This shows you which algorithm performed the best so far during Model Builder’s exploration. Last algorithm – This shows you the last algorithm that was explored by Model Builder.
- After model training finishes, go to the Evaluate step. The Evaluate step shows you various outputs, including how many models were explored and the ML task (in this case binary classification). Model Builder also displays the top 5 models explored and displays several evaluation metrics for each of those top 5 models, including AUC, AUPRC, and F1-score. You can find more informatio on what those are here.
- After evaluating your model, move on to the Code step. In the Code step in Model Builder, select Add Projects. Model Builder adds both the machine learning model and the projects for training and consuming the model to your solution. In the Solution Explorer, you should see the code files that were generated by Model Builder.
- The Cognitive services Content Moderator is a very useful tool to find explicit or offensive content in images, videos and text. It also supports custom image and text lists to block or allow matching content and can also be used as a feedback tool to assist human moderators. The service response includes the following information: Profanity: term-based matching with built-in list of profane terms in various languages Classification: machine-assisted classification into three categories Personally Identifiable Information (PII) Auto-corrected text Original text Language