QUA-1024: Restructure Entity Resolution check docs#1139
Conversation
…xamples/api/faq Mirrors the page set introduced for the Unique check (QUA-1806) and updates content to match the current Multi-Field Entity Resolution shape on develop: - Shape Anomaly (not Record), target_fields array with per-field match types, composite_match_threshold, weighted composite scoring, transitive grouping, blocking via match_type: exact, and Source Records that show one example row per distinct distinction-field value within each non-compliant cluster. - Target field shapes documented with the upickle_type discriminator (StringTargetField, NumericTargetField, DateTimeTargetField) so payloads copied from the docs deserialize correctly. - mkdocs.yml: nav expanded into the five-page set; redirects from checks/entity-resolution.md and data-quality-checks/entity-resolution.md to the new introduction so existing bookmarks keep working. - overview-of-a-check.md and rule-types-overview.md: link targets updated and descriptions revised to reflect the current rule shape.
Greptile SummaryThis PR replaces the single-page Entity Resolution doc with a five-page set (Introduction, How It Works, Examples, API, FAQ) aligned with the current multi-field rule shape, and wires up redirects so existing bookmarks continue to work.
Confidence Score: 5/5Documentation-only restructuring; no executable code changed. All internal cross-links resolve to existing files, include-markdown markers are present in the component files, and both legacy redirects are correctly wired in mkdocs.yml. The five new pages are internally consistent, cross-links to Unique/Not Null/Satisfies Expression all resolve, the include-markdown paths follow the same pattern used by the working Unique introduction page, and the redirect entries cover both legacy URL paths. The only discrepancy is the PR checklist calling out 12 questions while the FAQ contains 13 — a counting error in the PR description rather than a content defect. No files require special attention. The mkdocs.yml redirect entries and the introduction.md include-markdown markers are the two places most likely to cause a silent build failure, and both are correct. Important Files Changed
Flowchart%%{init: {'theme': 'neutral'}}%%
flowchart TD
OLD["data-quality-checks/entity-resolution.md\n(deleted)"] -- "redirect" --> INTRO
LEGACY["checks/entity-resolution.md\n(redirect)"] -- "redirect" --> INTRO
INTRO["entity-resolution/introduction.md\n• Definition\n• Field Scope\n• General Properties (filter-only)\n• Anomaly Types (shape-only)\n• Next Steps grid"]
INTRO --> HIW
INTRO --> EX
INTRO --> API
INTRO --> FAQ
HIW["how-it-works.md\n• 5-step evaluation flow\n• match_type tables\n• Composite formula\n• Shape Anomaly template"]
EX["examples.md\n• Customer dedup\n• Business homophones\n• Tenant-scoped blocking"]
API["api.md\n• Endpoints\n• Full payload\n• Target-field tables"]
FAQ["faq.md\n• 13 Q&As"]
OVR["overview-of-a-check.md"] -- "link updated" --> INTRO
RTO["rule-types-overview.md"] -- "link updated" --> INTRO
Reviews (1): Last reviewed commit: "docs(entity-resolution): restructure int..." | Re-trigger Greptile |
…s tables Replaces the legacy text-negative spans (red text only) in the three Source Records tables across entity-resolution/examples.md with the new .anomalous-cell utility class, which renders each anomalous cell with an orange outline and warning-tinted background. This mirrors the source-records-container.vue treatment in the Qualytics frontend and keeps the Sample Data visual treatment consistent with the new pattern introduced for Expected Values. The .anomalous-cell class is also added to docs/stylesheets/extra.css on this branch (the same block is in the Expected Values restructure PR; whichever PR merges first absorbs the other's overlap as a trivial duplicate-class conflict).
Overview
Restructures the Entity Resolution check page into the five-page set introduced for Unique (introduction, how-it-works, examples, api, faq) and aligns the content with the current Multi-Field Entity Resolution shape on develop.
Key Changes
match_typetables, three optional knobs on fuzzy strings, weighted composite formula, composite match threshold, filter behavior, cluster identifier_qualytics_entity_id, Shape Anomaly message template, source-records behavior, performance considerations, relationship with Unique/Not Null/Satisfies Expression.upickle_typediscriminator, top-level field notes, per-target-field-type tables (String/Numeric/Datetime) each with aupickle_typerow and correctmatch_typevalues.checks/entity-resolution.mdanddata-quality-checks/entity-resolution.mdto the new introduction so existing bookmarks keep working.entity-resolution/introduction.mdwith descriptions matching the current rule shape.Pages to Test
match_typecolumn at min-width 100px, three optional knobs on fuzzy strings, composite-score formula block, Shape Anomaly message template (no[filter: ...]suffix)._qualytics_entity_idcolumn; each payload hasupickle_typeas the first key in everytarget_fieldsentry.upickle_typediscriminator, top-level table withoutmax_distinct_records, three target-field-type tables (String/Numeric/Datetime) each with aupickle_typerow and correctmatch_typevalues (Numeric: absolute/relative/exact; Datetime: offset/granularity/exact with Day/Week/Month/Year).[filter: ...].