Skip to main content

View FactCheck Session Results

Once your session has completed the results will be available for you in the project view.

Navigate to AI Trust Suite > FactCheck > Your Test Project

Project Overview

The project's Overview section shows a summary of the statements made, indicating how many were correct or false. It also includes direct links to the Chatbot Under Test, and the Involved Test Set.



  1. Chatbot Under Test
  2. Involved Test Set
  3. Repeat Test Session
  4. Recent Test Results

Test Results

The session results provide an overview of the statements made, indicating how many were correct or false. It also includes direct links to the Test Project, the Chatbot Under Test, and the Involved Test Set. Additionally, a complete list of all the questions asked along with the bot's answers is provided.

Each result includes an explanation to help clarify why the specific status was assigned.



Result Explanation

Each test result includes a breakdown of why the decision was made. For a status of ACCURATE, all statements must be true.

In the example below, a status of INACCURATE was given because one of the five statements contained discrepancies.

  • Question: The question asked to the bot.

    What types of testing does Botium perform for chatbots?

  • Bot Response: The response provided by the bot.

    Botium performs the following types of testing for chatbots: NLP score testing, conversational flow testing, GDPR and security testing, performance testing, and monitoring.

  • Aligned Response: The fact checker uses the source of truth you proved to adjust the bot's answer, creating what we call the 'Aligned Response.' Botium compares this aligned response with the original bot response to determine any discrepancies.

    Botium performs the following types of testing for chatbots: NLP score testing, conversational flow testing, and testing across every channel and platform.

  • Reason: The decision is based on discrepancies between the bot's response and the aligned response.

    The article mentioned that Botium performs tests on a chatbot's ability to understand (which could be referred to as NLP score testing), testing real-life scenarios (could be considered as conversational flow testing), and testing across every channel and platform (could be part of performance testing). However, the article does not mention anything about GDPR and security testing, and monitoring that the statement has mentioned.

  • Decision: The decision is made by comparing the statements' claims to the aligned response.

    This disagrees with the statement claims.



Was this article helpful?

0 out of 0 found this helpful