Filtering by Definition

Filtering by Definition

When reviewing match results, you often need to focus on specific subsets of your data. MatchLogic provides several filtering mechanisms that work together: definition tabs, confidence band pills, and datasource dropdowns.

Definition Tabs

Above the results table, tabs let you switch between definitions:

• All — shows all match pairs regardless of which definition produced the match. The Max Score column reflects the highest score across all definitions.
• Individual definition tabs — each tab filters results to show only pairs that matched on that specific definition. The score column shows the score for that definition only.

Switching tabs is useful for evaluating how each definition performs independently. If Definition 2 produces mostly low-scoring pairs, you might need to adjust its criteria or weights.

Confidence Band Filter Pills

Below the definition tabs, colored pills correspond to the confidence bands described in https://help.matchlogic.io/article/271-score-distribution-confidence-bands. Click a pill to show only pairs within that score range. Click it again to remove the filter.

For example, clicking the green Excellent pill shows only pairs scoring 95-100. Clicking the orange Moderate pill narrows the view to pairs scoring 40-59, which are the ones most likely to need manual review.

Datasource Dropdowns

In the Pairs view, Data Source A and Data Source B dropdown selectors let you filter by which datasources are involved in the pair. This is especially useful when you have multiple datasources and want to see only cross-source matches between two specific files.

Combined Filtering

All three filter types work together. You can:

1. Select a specific definition tab to focus on one set of criteria
2. Click a confidence band pill to narrow by score range
3. Choose specific datasources from the dropdowns

The filters are cumulative — the results shown must satisfy all active filters simultaneously. For example, selecting "Definition 1" + "High confidence" + "Source A = Customers.csv" shows only high-scoring pairs from Definition 1 where one record comes from the Customers file.