Introduction
The codeit AI is a powerful tool that enables you to automate much of your verbatim coding.
It is designed to operate in a simple, seamless, intuitive way, so you don't need to get involved in technical details.
However, there are some elements of the AI system where a "one size fits all" approach doesn't make sense.
There are certain aspects of the functionality where, generally, different people want different things.
To cater for this, codeit exposes a number of settings that can be used to alter the behaviour of the AI system.
Where to access AI settings
Within the Project Details page, you can alter the AI settings at three different levels, as follows:
Level | Description |
Project | Applies at the overall project level and therefore to all tasks and waves within the project. |
Task | Applies to a specific Task. By default, settings will inherit from the project level, but can be overridden at an individual Task level, if required. |
Wave | Applies to a specific Wave. By default, settings will inherit from the project level, but can be overridden at an individual Wave level, if required. |
AI setting types
There are four types of AI settings as follows:
- Text Matching
Controls whether the codeit AI can use Text Matching to autocode data - Text Rules
Controls whether the codeit AI can use Text Rules to autocode data - Machine Learning
Controls whether the codeit AI can use Machine Learning to autocode data - Guardrails
Applies constraints and controls on the behaviour of the codeit AI when autocoding
The following guardrails are available:Name Description Text Matching Frequency The minimum number of identical verbatim examples required to teach the codeit AI Text Matching system Autocoding Mode Controls whether the codeit AI Autocoding should prioritize coding volume over completeness. - Minimise Gaps restricts the codeit AI so that it will only autocode a verbatim if it is sure it has captured ALL of themes within that verbatim
- Maximise Coding allows the codeit AI to autocode a verbatim regardless of whether it has captured ALL of the themes within that verbatim
Autocoding Threshold The minimum confidence that the codeit AI must have in a code before it can be auto applied. Coverage Threshold The minimum proportion of a verbatim that must be coded in order to be considered "complete" when using "Minimise Gaps" autocoding mode