Review solution similarity examples
Review the similarity examples generated during solution training to determine whether the similarity score threshold meets your business requirements.
시작하기 전에
- Train a similarity solution
- Role required: ml_admin or admin
이 태스크 정보
Solution training generates a set of paired record examples. Each pair includes a percentile score that estimates the degree of similarity between the two records. The higher the score, the higher the similarity, as follows:
- A score of 100 indicates identical records.
- A score of 0 indicates dissimilar records.
A deployed solution returns records only when the score is higher than the Similarity Score Threshold value:
- The higher the threshold value, the higher the precision and the lower the coverage.
- The lower the threshold value, the lower the precision and the higher the coverage.
On domain-separated instances, the following procedure displays records from all domains to ML admins. However, when solutions are applied in your forms and flows, similarity predictions are domain-aware. Examples from other domains on the instance are not displayed to users.
주:
The similarity filters specified in the solution definition aren't applied for similarity examples and are applied only during prediction.