Defining a knowledge search provider for chatbot

As an administrator you can specify a knowledge provider for a chatbot. If you have multiple chatbots, you can define a different provider for each chatbot. 

To learn about the support search providers and their benefits, see Search-providers.

Before you begin

To define BMC Helix Knowledge Management search

1. Log in to BMC Helix Innovation Studio. 
2. On the Workspace tab, select the BMC Helix Virtual Agent application or custom application for which you set up the chatbot instances. 
3. Click the Chatbots tab and select the chatbot instance for which you want to configure the BMC Helix Knowledge Management search.
4. On the Chatbot Configuration page, click the Search Settings tab. 
5. From the list of locales, click the locale of the chatbot application in which you want to configure BMC Helix Knowledge Management search; for example, English.
6. From the Provider list, select ComAround Knowledge.
7. (Optional) From the Search filter list, select the relevant search filter. For more information, see Creating search filters for BMC Helix Virtual Agent.
8. Click Save.

The following image shows how to define BMC Helix Knowledge Management search: 

You have defined BMC Helix Knowledge Management by ComAround as the default search provider for the chatbot.

End users can view the knowledge articles as shown in the following image: 

To define cognitive search

1. Log in to BMC Helix Innovation Studio. 
2. On the Workspace tab, select the BMC Helix Virtual Agent application or custom application for which you set up the chatbot instances. 
3. Click the Chatbots tab and select the chatbot instance for which you want to configure the cognitive search.
4. On the Chatbot Configuration page, click the Search Settings tab. 
5. From the list of locales, click the locale of the chatbot application in which you want to configure cognitive search; for example, English.
6. On the Edit Bot Locale Search Settings page, set the following fields:
   1. In the Provider list, select Cognitive.
   2. In the Data Set field, select the external data sets and BMC Helix Innovation Suite data sets that you want to include in the cognitive search.

1. 1. In the Passages Fields field, specify the passage text that you want to include in the cognitive search. 
      You need to first configure the passage text in IBM Watson Discovery settings and add it to the chatbot. You can provide multiple values separated by comma. These fields are used to extract content or passages from IBM Watson Discovery. These passages are returned from IBM Watson Discovery and are displayed on the tiles of searched knowledge articles. To learn more about passages, see To define Passage Fields.
   2. Click Save.
      

1. From the list of locales, select the locale for which you want to apply the display templates.
   
   
   
   The Display Template by Channel section displays two entries for each data set—one for the web channel and the other for the non-web channels.

1. From the Display Template by Channel section, do the following:
   1. Select an entry in the Data Set column.
      In this example, Template for Internal DS_Calbro.
   2. Click Upload Template, and browse to the location where you saved the out-of-the-box templates. 
   3. Select the HTML template that you want to apply for displaying knowledge articles of the selected channel, and click Open.
   4. Repeat the preceding steps for the remaining channels in the Display Template by Channel section to complete the configuration for cognitive search.

You have defined cognitive search provider for the chatbot. The end users can view the knowledge articles as shown in the following image: 

To highlight keywords while searching for knowledge articles

As an administrator, you can configure BMC Helix Virtual Agent to highlight keywords when users are searching for knowledge articles. 

1. Log in to BMC Helix Innovation Studio and navigate to the Administration tab.
2. Select Cognitive and chatbot > Chatbots. 
3. Select the chatbot instance for which you want to configure the cognitive search.
4. On the Chatbot Configuration page, click the General Settings tab. 
5. Turn on the Highlight Articles toggle key to highlight keywords in knowledge articles.
   
   

1. Click Save.

To define the passage fields

1. Open the IBM Watson Discovery instance.
2. Open the project. 
3. Go to the Improve and Customize pane.
4. Search for the required documents. 
   A list of documents is displayed.
5. Click View document to open the document.
6. Click Open advanced view.
7. Select the json format to view the document.
   
   The json file is displayed.
8. Check for key value pairs in the csv files.
   By default, the text field value is displayed on the knowledge article tiles in the chatbot. You can add multiple comma-separated values in Passage Field.
   

If a document has no text value, you can select other key values; for example, Passage Field: column_1. 


The column_1 value is displayed on the tiles of the knowledge article in the chatbot.

Troubleshooting

After configuring cognitive search, if search results are not displayed as expected, use the following checklist to troubleshoot the issues:

Unable to view the search results or irrelevant search results are displayed

• Check if the API key and the IBM Watson Assistant Skill ID are correct. 
• Check if the IBM Watson Discovery service credentials are correct.
• Check if the IBM Watson Discovery Collection IDs are correct.
• Check if you have selected Cognitive as the Provider in the Bot Locale Search Settings.
• Check if you have selected data sets in the Bot Locale Search Settings.
• On the Chatbot Configuration page > Search Settings tab, check if you have uploaded the correct templates. 

• If irrelevant search results are displayed, in IBM Watson Discovery, improve the search result relevance. 

• Verify if the Passage Fields values are correctly mapped with the json structure in the IBM Watson Discovery document.
  Verify if the collection name, project name, data set, and document name are correct. 

For further queries contact support.