Discover why records match, which rules connect them and where potentially over-merged groups deserve closer investigation.
With WinPure Clean & Match Enterprise v12, MatchGraph™ introduces a more transparent way to investigate data matching results. Instead of simply listing records in the same duplicate group, this interactive data matching visualisation shows which record pairs actually match, their similarity scores and, when several rules are configured, the rule responsible for each connection.
MatchGraph™ also highlights bridge connections: relationships whose removal would split a group into separate parts. This helps data quality teams investigate potentially over-merged groups and understand the structure of complex matching results.
MatchGraph™ turns duplicate groups into explainable networks, helping users see not just which records belong together, but why.
WinPure MatchGraph™ reveals how four matched records are connected, displaying similarity scores, highlighting the Master Record and identifying a critical bridge connection in red. This helps users understand matching relationships and investigate potential over-merging.
Go Beyond Traditional Duplicate Matching Results
Imagine three customer records: A matches B at 98%, and B matches C at 96%, but A and C do not directly satisfy a matching rule. All three can still belong to the same group because B connects them. This is an example of transitive matching.
A traditional results table shows the group membership but may not make these direct and indirect relationships obvious. MatchGraph™ displays records as nodes and qualifying matches as connecting lines. Each line shows its pairwise matching score and, when more than one rule is configured, the first rule in the configured order that passes.
This explainable data matching view makes it easier to review matching logic, investigate unexpected connections and decide whether thresholds or rules need refining. To learn more about how similarity works, see our guide to fuzzy matching algorithms.
Visualise Record Relationships, Matching Rules and Scores
MatchGraph™ gives each group a structured, interactive view:
WinPure MatchGraph™ provides a visual explanation of data matching results, highlighting direct record matches, similarity scores and critical bridge connections. In this example, four records are connected by four qualifying matches, including one red bridge that would split the group if removed, helping users investigate potential over-merging.
- Record nodes: Each displayed record is labelled with the value of the first condition of Rule 1 and its source table in brackets.
- Qualifying match lines: Lines represent record pairs that pass the configured rules, with their scores and applicable rule labels.
- Master Record highlighting: The record selected by the Master Record Selector is highlighted in amber and labelled MASTER RECORD. Master status does not affect the graph layout.
- Full duplicate folding: Records identical in the values used by every matching rule are folded into one displayed record with a +n full duplicates badge.
- Relationship-based layout: The most connected record sits centrally, while other records appear on rings according to their distance in matching connections.
- Interactive navigation: Use Zoom In, Zoom Out and Fit to View; drag individual nodes, use scroll bars or move the canvas with the right or middle mouse button.
The status line reports records, folded duplicates, compared pairs, matches, bridges, elapsed comparison time and the number of lines currently displayed.
In Match Against Trusted Source mode, MatchGraph™ shows only qualifying matches between the main table and the other tables, reflecting the pairs that mode compares.
Filter Matching Relationships and Simplify Complex Groups
When a group contains many connections, MatchGraph™ helps users focus on the relationships relevant to their investigation:
- Inspect one record: Select a record in the Record list or click its node, then tick Only matches of the selected record. Only that record’s match lines remain visible; other nodes stay in place for context. The record whose View button opened the graph is selected automatically.
- Filter by rule: Use the Rule filter to display connections attributed to one matching rule.
- Simplified (strongest links only): Reduce the network to a strongest-link spanning tree that preserves how the group is connected. For a connected group with 50 displayed nodes, this means 49 lines rather than every qualifying pair.
The simplified view is useful when the question is how does this group hold together? rather than which pairs match? Red bridge highlighting retains its meaning from the full graph, rather than being recalculated from the simplified view. The status line shows how many lines are currently visible.
Identify Potentially Over-Merged Records with Bridge Detection
One of MatchGraph™’s most useful diagnostic capabilities is bridge detection. A bridge is a matching connection whose removal would split a group into disconnected parts. MatchGraph™ highlights these lines in red.
For example, a short or generic company name could match records from otherwise distinct clusters and connect them into one large group. A chain of records held together by one or two red bridges can therefore be a useful starting point when investigating over-merging.
A red bridge does not automatically mean an incorrect match. It indicates a structurally important connection worth checking against its matching rule, similarity score and record values. Users can then assess whether to refine a rule, adjust a threshold, improve cleansing or keep the connection.
This helps teams move beyond simple duplicate detection towards more transparent, explainable record matching and group validation.
How MatchGraph™ Works in WinPure Clean & Match Enterprise v12
MatchGraph™ complements AdaptiveMatch™, WinPure’s re-engineered matching framework with configurable, field-specific matching algorithms and similarity thresholds. For example, different fields can benefit from techniques such as Jaro-Winkler similarity or Levenshtein distance. MatchGraph™ explains relationships in qualifying Match V2 results; it is not a separate matching engine or a visualisation of Senzing MatchAI™ entity resolution results.
On-demand comparison, without stored graph edges
When the graph window opens, WinPure rechecks the relevant record pairs within the selected group using the rules and parameters from the original run. It applies the same preparation, including Knowledge Base Library dictionaries, Fuzzy Level and each condition’s matching algorithm, together with the strict rule check. A displayed line therefore represents a pair that passes a configured rule.
These relationships are calculated on demand, not stored as additional pairwise graph data during matching. This avoids growing the result table or saved project file with stored graph connections.
How to open MatchGraph™
For an eligible result, the results view contains a MatchGraph™ column with a View button. Click View on any row to open the graph for that row’s match group.
Availability, saved projects and limits
MatchGraph™ is available for results produced by the Match V2 engine (configured under Application Settings › Matching, where available) when the rules include at least one fuzzy condition. It is not available for classic-engine results or results from rules without a fuzzy condition.
Availability belongs to the result, not the current setting. Switching Match V2 off after a qualifying run does not remove the graph from that result; switching it on later does not add a graph to a result previously produced with the classic engine.
MatchGraph™ supports groups of up to 300 records. The parameters of the last matching run are saved with the project, so qualifying graphs can be opened after reopening a project. All columns referenced by the matching rules must be included in the result. If matching fields or rules have changed since the run, run matching again before opening the graph.
Discover MatchGraph™ in WinPure v12
Go beyond identifying duplicate groups. Understand exactly which matching relationships hold them together.
With explainable scores, matching rule labels, bridge detection, full duplicate folding and focused filters, WinPure MatchGraph™ helps data quality teams review duplicate groups with greater confidence. Whether you’re working with CRM data, customer records, supplier databases or enterprise master data, MatchGraph™ provides a clearer way to investigate matching outcomes.
Ready to See MatchGraph™ in Action?
Explore WinPure Clean & Match Enterprise v12 and discover a clearer way to investigate your data matching results. Try WinPure free for 30 days. No credit card required.
Frequently Asked Questions About MatchGraph™
Data matching visualisation presents records and their matching relationships as a network, making it easier to see which pairs match directly and how larger duplicate groups are connected.
It displays qualifying record-pair connections with their similarity scores and, when multiple rules are configured, the first matching rule in the configured order that passes.
MatchGraph™ highlights bridge connections in red. Removing one of these connections would split the group, making it a useful relationship to investigate. A bridge does not necessarily indicate an incorrect match.
Yes. You can show only matches for a selected record, filter connections by rule or use Simplified (strongest links only) to focus on the links that hold a group together.
No. MatchGraph™ recalculates qualifying pairwise relationships when the graph opens instead of storing additional graph connections during matching.
MatchGraph™ requires a qualifying Match V2 result with at least one fuzzy condition and supports groups of up to 300 records. Matching-rule columns must be present in the result.
Share this article
