Leveraging cognitive search in your application
The BMC Helix Cognitive Automation uses IBM Watson Discovery to provide the cognitive search capabilities. Cognitive search utilizes artificial intelligence (AI) to merge the scattered knowledge and provide more relevant search results. You can leverage cognitive search in BMC Helix Virtual Agent and custom applications to enable cognitive search across BMC Helix Innovation Studio data and external data.
Data crawler to include knowledge articles in cognitive search
As an administrator, you must crawl the knowledge articles to include them in cognitive search. You can run one of the following crawlers on the knowledge articles:
- BMC Crawler—For knowledge articles in BMC Helix ITSM: Knowledge Management only.
- IBM Watson Discovery Data Crawler—For data in any database or file system that is not in BMC Helix ITSM: Knowledge Management.
BMC Helix Innovation Studio cognitive search supports BMC Helix Business Workflows knowledge articles and does not require a crawler to include knowledge articles in cognitive search.
Data flow when you use BMC crawler utility
The following diagram explains the data flow after cognitive search is triggered:
Scenario: Enabling cognitive search in multiple chatbots
An organization uses a chatbot in their HR and IT service request applications. The knowledge base of the organization is in BMC Helix ITSM: Knowledge Management. As an administrator, you install and run the BMC crawler to include knowledge articles in BMC Helix ITSM: Knowledge Management in cognitive search. After a successful crawl, you enable cognitive search for the HR and IT chatbots so that the search results are displayed based on users' searches.
For example, a user is facing problems when installing an application and asks BMC Helix Virtual Agent for help. BMC Helix Virtual Agent uses cognitive search to find the matching paragraphs in the relevant knowledge articles in BMC Helix ITSM: Knowledge Management and provides the information to the user.
Result: End-user experience after enabling cognitive search in a bot
An employee searches for knowledge articles on how to change an ink cartridge. In their own words, they ask the chatbot for information. The bot uses cognitive search, identifies the intent of the user, and displays snippets from the most relevant knowledge articles as search cards. The search results are displayed in a carousel in the chatbot window as shown in the following image:
The employee can click the display results and view the complete knowledge article from the chatbot window.
Based on the usefulness of the knowledge article, the employee can rate the knowledge article. The ratings are then sent to the cognitive service to train IBM Watson Discovery to display most relevant knowledge articles.
IBM Watson Discovery requires at least 50 unique results for training.
Benefits of cognitive search
- Crawl, convert, and enrich data.
- Customize how your documents are indexed to better target relevant information.
- Securely explore your proprietary content.
- Apply the following additional enrichments:
- Keyword extraction
- Entity extraction
- Concept tagging
- Sentiment analysis through Natural Language Understanding (NLU)
- Category classification