Atlas Graph Edge Quality Report

Scope

Input files:

• 07_visualization/full_graph_candidate_edges.csv
• 07_visualization/edge_list_expanded.csv
• 07_visualization/extreme_ductility_extension_edges.csv
• 07_visualization/sustainable_ecc_extension_edges.csv
• 07_visualization/by_lee_lab_position_edges.csv

Summary

• Raw input edge rows: 1587
• Unified records before filtering: 1587
• Paper-to-node edges: 365
• Concept-to-concept edges: 491
• Lab-to-global edges: 103
• Excluded/audit rows: 55

Duplicate removal

• lab_to_global: 206 duplicate row(s) removed
• paper_to_node: 367 duplicate row(s) removed

Output validation

paper_to_node

• Rows: 365
• Missing target files after refinement: 0
• Duplicate rows after refinement: 0
• Relation counts:
• supports: 365
• Confidence counts:
• confirmed: 4
• verified_from_source_note: 361

concept_to_concept

• Rows: 491
• Missing target files after refinement: 0
• Duplicate rows after refinement: 0
• Relation counts:
• affects: 11
• applies_to: 1
• characterizes: 1
• controls: 3
• depends_on: 2
• enables: 5
• extends: 2
• frames: 1
• includes: 8
• measured_by: 2
• measures: 11
• mechanism_of: 5
• parameterizes: 1
• part_of: 3
• reduces: 4
• requires: 2
• supports: 417
• uses: 3
• validates: 1
• variant_of: 8
• Confidence counts:
• book_anchor_seed: 40
• candidate: 61
• confirmed: 17
• verified_from_pdf: 373

lab_to_global

• Rows: 103
• Missing target files after refinement: 0
• Duplicate rows after refinement: 0
• Relation counts:
• lab_contributes_to: 103
• Confidence counts:
• verified_from_source_note: 103

Excluded edge reasons

• unclassified_source_paper_target_unknown: 40
• unclassified_source_unknown_target_unknown: 12
• missing_target_file: 3

Output files

• 07_visualization/graph_paper_to_node_edges.csv
• 07_visualization/graph_concept_to_concept_edges.csv
• 07_visualization/graph_lab_to_global_edges.csv
• 07_visualization/graph_excluded_edges.csv
• 07_visualization/lab_to_global_lineage_map.md

Notes

• Output files contain only edges whose target file exists.
• graph_excluded_edges.csv preserves records with missing or unclassified targets for later node cleanup.
• lab_contributes_to edges are separated from general paper-to-node support edges.