The ServiceNow similarity framework allows you to find new records that are similar to a given input record. This could help a support agent to find past customer cases similar to the case they are working on so that they can reuse knowledge or resolution details. It can also help prevent entering duplicate data or correlating records coming in, such as incidents related to an active major incident.
The similarity framework differs from the prior two predictive models because it will find entire records and not just a single field value.
A useful trick when employing a similarity search model is to use data values from a highly similar past data record to populate your current record. This allows a good similarity model to act as a classification or regression model. This approach is not without its drawbacks however and ServiceNow is not set up to calculate metrics for that use. In general, when using similarity models, it can be hard to assess...