The quality of recommendations provided by Machine Learning Match Recommendations (MLMR) in the Clerical Review Task List are dependent on the 'merge' and 'reject' decisions made by the user charged with training the matching agent. To improve the quality of the recommendations, Stibo Systems provides customers with a dedicated team prepared to engage in a collaborative process with customers to help improve the customer's understanding of the recommendations, and to improve the quality of those recommendations.
If the matching agent recommendations you receive results in questions for you or your team, in the Stibo Systems Service Portal, create a ticket with the Issue Type 'Customer Request.' Find below a list of the fields required when creating a support issue for the MLMR and descriptions of how to provide the requested content.
Note: Before you create a support issue, verify the relevant data to match on is mapped to the matching agent.
Summary: Add a short description of the issue you are experiencing in this field. Preface your summary content with ‘MLMR' so it is clear to the support team that the issue relates to the matching agent recommendations. The format will look like this: 'MLMR - <description of the issue>’.
Description: In this field, copy the data points listed below and paste it into the 'Description' field in the issue. Then add the requested information for each data point:
Description of problem:
Total number of recommendations:
Number of wrong recommendations identified:
Description of wrong recommendations: Describe what is wrong with the recommendations from the matching agent.
System Name / URL:
Training Process BGP ID:
Training Process BGP Started timestamp:
Training Process Execution Report:
<paste text>
Recommendation Process BGP ID:
Recommendation Process BGP Started timestamp:
Recommendation Process Execution Report:
<paste text>
Issue Category: ML Matching Agent
Business Domain: CMDM