AI Matching

By using the AI Matching feature (based on the Artificial Intelligence) of ibi MDM, an administrator can apply a trained AI model of ibi™ Patterns - Searchto match incoming new records with the existing records and subsequently identify possible duplicate records. The AI model uses actions taken on records in the data quality work item and trains itself. When similar records arrive in ibi MDM, ibi Patterns - Search uses the trained AI model for matching records, and then returns more accurate results.

• If only one match is detected, automatically two records are merged
• If multiple duplicate records are found, you can review and reject unwanted records, and merge accurate matched records

Finding and labeling record pairs for supervised training of any machine learning model is a time-consuming task. The AI Matching feature aims to simplify this task by periodically extracting labeled record pairs from the previous record merge decisions that have accumulated in the ibi MDM database. Newly extracted record pairs can make the retrained model perform better. You can then redeploy such retrained models to improve the performance of the record matching in ibi MDM over time.

An administrator can access the AI Matching page by clicking the Administration on the global navigation bar.