Exam AI-102 topic 5 question 4 discussion

As per Udemy it is: - Add a new application. - Add the prebuilt domain ToDo. - Train the application. - Publish the application

Given answer is correct. (Taken from Udemy practice exam question) The order should be : 1. Add a new app, 2. Add example utterances, 3. Train the app, 4. Publish the app.

Answer Area Add a new application. (You need to start by creating a new LUIS application.) Add the prebuilt domain ToDo. (This is crucial for minimizing development time as it provides pre-built intents and entities related to task management, which aligns with the "task tracking" requirement.) Add example utterances. (Even with a prebuilt domain, you'll want to add your own specific examples to tailor the model to your chatbot's needs and user language.) Train the application. (After adding data, you need to train the model so it can learn and make predictions.) (Publish the application would be the next step after training and testing, but only four actions were requested.)

To build a Language Understanding model for a task tracking chatbot while minimizing development time, the correct sequence of four actions is: 1.Add a new application. 2. Add the prebuilt domain ToDo. 3. Add example utterances. 4. Train the application. This sequence provides the most efficient path because: - First, you create the application in the Language Understanding service - Next, you leverage the prebuilt ToDo domain which comes with pre-configured intents and entities for task tracking (saving significant development time) - Then, you add custom example utterances to tailor the model to your specific needs - Finally, you train the model so it can recognize the patterns in your utterances Publishing would be the next step after these four, but since the question asks for exactly four actions in sequence, these represent the core model building process that minimizes development time

1) Add a new application - No argument here 2) Add the prebuilt domain ToDo - to "minimize development time" 3) Train the application - to make the pre-built model better 4) Publish the application - last obvious thing to do

In the case that this question actually requires 3 steps as opposed to 4 (per pyguy's comment at the bottom), I believe the correct sequence would be: 1. Add a new app 2. add the prebuilt domain ToDo 3. Publish the application Prebuilt domains don't require training (it's ready to go out of the box, why would it need trained?) It's also worth noting that LUIS is retired next year, and you actually haven't been able to make any new LUIS resources since April 2023. I really doubt this would show up on the current exams (at least I hope) since I see no mention of prebuilt domains in the CLU documentation. If I missed it somewhere, please correct me.

Add a new app, Add example utterances, Train the app, Publish the app.

1. Add a new app 2. Add example utterances 3. Train the app 4. Publish the app

Final Answer 1. Add a new application 2. Add the prebuilt domain ToDo 3. Publish the application

Corrrection to answer: 1.add a new application 2. add the prebuilt domain ToDo 3. Train the application 4. Publish the application Select a pre-built domain (e.g., "Todo"), which provides a predefined set of intents and entities designed specifically for scenarios like task tracking. Train the application, even when using a pre-built domain. Training the application is necessary because it optimizes the model to better understand user queries specific to your data and use case. Even if you choose a pre-built domain, training the application remains a necessary step. Pre-built domains provide a good starting point, including a set of related intents and entities, but through training, you can optimize the model's performance for your specific data and scenario.

The question requires minimal time to build a model. Should we use prebuilt model with a custom domain ("ToDo") here? Ref: https://docs.microsoft.com/en-us/azure/cognitive-services/luis/luis-reference-prebuilt-domains#prebuilt-domains-per-language