Azure AI Fundamental - AI 900 - PART 30

Introduction

Welcome to Part 30 of our deep dive into questions concerning Microsoft certified AI fundamentals! If you’re following along, you should already be a dedicated subscriber and a member of either the Cloud Colonel or Cloud Ninja community.

In this segment, we’ll discuss a hypothetical scenario: you’re tasked with creating a language understanding application to support a music festival. The tool you will employ is Language Understanding (LUIS), which is a cloud-based conversational AI service from Azure. LUIS utilizes custom machine learning intelligence to interpret users' natural language text and predict their overall intent.

Understanding User Queries

For example, a user may ask, “Which act is playing on the main stage?” This illustrates the types of elements found in LUIS, which we’ll categorize as follows:

1. Intent: This serves as the purpose behind user queries. Common examples include “find a restaurant” or “book a flight.”

2. Entities: Within the user’s input, specific pieces of information can be extracted. In the query “I want to book a flight to New York on Tuesday,” “New York” and “Tuesday” are entities.

3. Utterance: Different expressions of the same intent highlight the importance of utterances. Users may ask the same question in various ways, but the AI must still provide the correct answer. Thus, utterance is the correct type of element for handling varied user input at our music festival application.

Building a Q&A Chatbot

Next, we explore the creation of a Q&A Maker bot. Suppose you want to build a conversational layer on top of pre-existing data – specifically using frequently asked questions.

To enhance user interaction, you may wish to incorporate professional greetings and friendly responses. Here are some strategies to consider:

1. Increasing Confidence Threshold: While boosting response confidence might narrow down the accuracy of the answers, it doesn't aid in making the bot more personable.

2. Enable Active Learning: This feature helps enhance the knowledge base by suggesting addition from real user queries, but it will not assist in improving user friendliness.

3. Creating Multi-turn Questions: If users engage in a series of questions sequentially, this helps facilitate ongoing conversations but does not directly cater to improving bot interactions.

4. Add Chit Chat: This feature refers to informal conversational interactions that enhance user friendliness and create a more relatable bot experience.

Processing Documents for Chatbot Responses

In our next scenario, let’s assume you’re developing a chatbot that needs to answer user questions based on specific documents, such as Microsoft Word files and FAQ lists. Your goal is to identify which service should be used to process these documents.

• Azure Bot Service: This is primarily for creating conversational AI but doesn’t directly process documents.

• Text Analytics: Although useful for identifying sentiments and key phrases, it does not facilitate document processing.

• Q&A Maker: This service allows the import of documents and converts the data into a markdown format, allowing the chatbot to retrieve and respond to user queries accurately.

In this case, utilizing Q&A Maker is the most suitable option as it effectively addresses the need to process documents in conjunction with a chatbot.

Conclusion

We hope this overview has provided clarity on the various components of developing AI applications using Azure services. If you haven't already, consider becoming a member of our community to gain access to additional resources and questions tailored for Cloud certifications, whether for Azure, AWS, or Google Cloud.

See you in the next part!