Now that the Kylie Bot is returning some content, we need to extend this capability to search a number of sources simultaneously. The easiest way to achieve this is to create an index that the bot can search. This index in turn will store huge amounts of content that we can index as necessary and return to the bot to provide meaningful and empowered knowledge.
So, how do we do this? We make use of the conveniently available Azure Search service. For those of you familiar with my previous article on Relevance Search. Azure Search is what powers that. See the full release about Dynamics 365 and Azure Search here. To get started, log into the Azure Portal and you should be presented with the following dashboard:
Next, click on the ‘+New’ option in the top left of the screen and you should be presented with this screen:
Next, type in ‘Azure Search’ and select the Azure Search pop-up item as follows:
You should then be presented with this screen:
Select the ‘Azure Search’ option at the top of the list and you should be presented with the following:
Clicking on the ‘Create’ button in the bottom right corner which will result in the following:
Fill in the required details and then click on ‘Create’ in the bottom left hand corner as you should observe two notifications indicating that the deployment has started and then that the deployment has succeeded.
Navigate to and click on the ‘Resource groups’ menu item on the left hand side of the screen. You should then be presented with the following:
Click on the Resource Group where you just created your Azure Search service:
You will then be able to confirm that your service has been deployed correctly by clicking on the service name. Mine is called ‘KylieBot’ and you will then see something similar to this:
Next, we will need to add an index to this Azure Search service by clicking on the ‘+Add index’ button in the top left of the tile. You should be presented with the following:
Enter a name for the index and then click the fields menu item to specify the specific properties of the index. Add the following properties to the index:
Id
Title
Keywords
Content
Rating
NumberOfRatings
TotalRatingScore
Tags
Product
Version
Category
Source
LoadDate
ArticleDate
*Worth noting the guidance on index attributes available here
The index should now look like this:
Click on ‘OK’ in the bottom left of the screen to confirm the creation of the Fields and then click on the ‘OK’ button also in the bottom left hand corner to confirm the creation of the index. You should now see the Index you just created, in the middle of the screen (mine is called ‘generalcontent’).
*Note: be careful of the filed names for the index. Once created, you cannot change the names and you will need to recreate the index if you make a mistake.
Great job. Now that we have an Azure Search index, we are going to load it with data. Look out for the next post which will detail how to create a Dynamics 365 solution which can store the outputs of the attempts to populate the index.