Configure Match and Merge > Configuring match and merge
Configuring match and merge
Configure the match and merge process to identify and resolve duplicate records. The match and merge process configuration together constitutes the match model. Based on the outcome of the match process and the merge configuration, the records are either automatically merged, queued for manual merge by a data steward, or skipped as not a candidate for merge.
To resolve duplicates, you can choose to use a machine learning (ML) model, declarative match rules, or both. The declarative match rules are based on an exact or fuzzy match strategy. An exact match strategy identifies identical records, and a fuzzy match strategy identifies similar records. To reduce overmatching and provide a better matching performance, consider adding at least one exact match rule in a match model.
An ML model needs to undergo a training process based on your business needs, where a user labels record pairs to train the ML model. An ML model training reduces the manual effort required for defining declarative match rules.
To configure the match and merge process, use the predefined match model or create a custom model to suit your business needs. You can create a custom match model from scratch or from a copy of an existing match model. You cannot edit the predefined model.
Important: You must configure and publish the match model after you configure the data model, but before the ingress job is run for the initial data load. If the match model configuration is published after the initial data load, the data that is already in the Business 360 data store is not considered by the match model and does not undergo matching.
After you publish a model, you can define match and merge jobs, and run the jobs. For information about defining and running match and merge jobs, see Define Jobs.