Our previous chatbot explainer post discussed chatbot templates which can accelerate the chatbot development process. Today we will explore another feature that can have the same effect–knowledge extraction, which enables the chatbot to parse structured and unstructured data sources into content to be used for answering customer questions. Knowledge extraction is classified in the Training category of chatbot product features.

Chatbots have two main functions–transactional purposes such as completing tasks requested by the customer, and informational purposes such as delivering information inquired by the customer. For its informational purposes the enterprise user would usually have to train the chatbot by manually entering potential questions the customer might ask and their respective answers one by one. The knowledge extraction feature simplifies this process by ingesting structured or unstructured data sources and automatically parsing their content into questions and answers that the chatbot can use to answer customer questions.

The knowledge extraction feature is usually powered by the chatbot vendor’s own proprietary LLM (large language model), or by integrating with a third-party LLM such as OpenAI or Anthropic. The data sources that it can parse range from structured documents such as CSV files or FAQ lists to unstructured ones such as web pages or PDF files. After parsing the data source, the chatbot can either immediately use the content to respond to customer questions, or show the entire list of parsed questions and answers so that the enterprise user can edit them or add to them as necessary. In conclusion, the knowledge extraction feature provides a convenient shortcut for the enterprise user to quickly build informational chatbots without needing to manually enter questions and answers one by one.

The screenshot above shows the knowledge extraction feature from an example chatbot vendor – Intercom.

Our chatbot explainer series shares our knowledge of chatbot product features. If you have any questions, please reach out to me directly and follow me to get notifications on future posts.

Dong Liu, Founding Analyst
dong@daybreakinsights.com

@dongliu1226