Skip to main content

How document structure and question type affect Knowbot answers

Learn how the structure of your learning material affects what Knowbot finds and what admins can do to improve the results.

Written by Camilla Orloff

Knowbot searches the learning material available to a user and selects the information that appears most relevant to their question. The structure and wording of the material can therefore affect the answer.

Why document structure matters

The same topic may appear in several places in a document, including:

  • the table of contents

  • headings

  • introductory sections

  • instructions

  • information further down the document

If the same term appears several times, Knowbot must determine which sections are most relevant to the user’s question.

Clear headings and specific product, system, process, or policy names make it easier for Knowbot to identify the correct information.

Specific and broad questions

Specific questions are normally easier for Knowbot to answer.


For example:

“What are the dimensions of Product Z?” This question tells Knowbot which product and which information to search for.


A broader question such as: “What products are available?” requires Knowbot to find a complete overview. Several of the most relevant sections may concern the same product, such as its description, dimensions, installation, and maintenance.


If the selected information mainly concerns Product Z, information about Products X and Y may not be included in the answer.

This is why broad requests such as “list all products” may occasionally return an incomplete overview, even when additional information exists elsewhere in the learning material.


How admins can improve the learning material

The structure and wording of the learning material can make information easier for Knowbot to find.

When preparing material:

  • use clear headings that describe the information in each section

  • clearly name the relevant product, system, process, or policy

  • keep information about different topics or products separate

  • avoid using the same general heading for several different topics

  • include important information in the document itself, not only in the table of contents

  • include a clear overview when users need Knowbot to provide a complete list

These steps help Knowbot distinguish between similar topics and identify the information most closely related to the question.


Test the material with realistic questions

After adding or updating learning material, test it using questions your users are likely to ask.

Include:

  • specific questions

  • questions that require a complete overview

  • short or informal questions

  • questions containing common spelling mistakes

  • questions in the languages your users are likely to use

If a question could refer to several products or systems, Knowbot may ask for clarification instead of guessing. This means more context is needed to identify the correct information.

If an answer is incomplete, check whether the relevant information is clearly named, grouped, and described in the source material.

Did this answer your question?