diff --git a/schemas/ae_chairs.schema.json b/schemas/ae_chairs.schema.json index d43d8ca..7d2d551 100644 --- a/schemas/ae_chairs.schema.json +++ b/schemas/ae_chairs.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/ae_chairs.schema.json", "title": "AEChair Collection", "description": "Array of AEChair records. Each element: An AE Chair with chairing history, pipeline status, and cross-conference service.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/AEChair" diff --git a/schemas/ae_members.schema.json b/schemas/ae_members.schema.json index e9fde43..d036bbe 100644 --- a/schemas/ae_members.schema.json +++ b/schemas/ae_members.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/ae_members.schema.json", "title": "AEMember Collection", "description": "Array of AEMember records. Each element: An Artifact Evaluation committee member with service history and institutional affiliation.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/AEMember" diff --git a/schemas/artifact_availability.schema.json b/schemas/artifact_availability.schema.json index 0da2312..98e5535 100644 --- a/schemas/artifact_availability.schema.json +++ b/schemas/artifact_availability.schema.json @@ -1,7 +1,7 @@ { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "https://reprodb.github.io/data-schemas/schemas/artifact_availability.schema.json", - "version": "0.2.0", + "version": "0.3.0", "$defs": { "AvailabilityRecord": { "additionalProperties": false, diff --git a/schemas/artifact_citations.schema.json b/schemas/artifact_citations.schema.json index 4d902b3..9c9f81d 100644 --- a/schemas/artifact_citations.schema.json +++ b/schemas/artifact_citations.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/artifact_citations.schema.json", "title": "ArtifactCitation Collection", "description": "Array of ArtifactCitation records. Each element: Citation counts for a single artifact DOI, aggregated from OpenAlex and Semantic Scholar.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/ArtifactCitation" diff --git a/schemas/artifacts.schema.json b/schemas/artifacts.schema.json index c0e4ce1..6f8bc86 100644 --- a/schemas/artifacts.schema.json +++ b/schemas/artifacts.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/artifacts.schema.json", "title": "Artifact Collection", "description": "Array of Artifact records. Each element: A single research artifact associated with a conference paper.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/Artifact" @@ -77,6 +77,19 @@ "title": "Artifact Urls", "type": "array" }, + "artifinder_urls": { + "description": "Artifact URLs discovered automatically by ArtiFinder (https://github.com/DistriNet/ArtiFinder) and matched to this paper by title + author list. These links were NOT manually verified by an artifact evaluation committee, carry no badges, and do not contribute to any artifact/reproducibility score. They may, however, be used for repository statistics when they point to a GitHub repo.", + "examples": [ + [ + "https://github.com/org/discovered-repo" + ] + ], + "items": { + "type": "string" + }, + "title": "Artifinder Urls", + "type": "array" + }, "doi": { "default": "", "description": "Canonical artifact DOI extracted from artifact_urls, e.g. '10.5281/zenodo.12345'. Empty string if none found.", diff --git a/schemas/artifacts_by_conference.schema.json b/schemas/artifacts_by_conference.schema.json index 13303c9..1c9c0ed 100644 --- a/schemas/artifacts_by_conference.schema.json +++ b/schemas/artifacts_by_conference.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/artifacts_by_conference.schema.json", "title": "ConferenceEntry Collection", "description": "Array of ConferenceEntry records. Each element: Artifact counts and badge breakdowns for a single conference across all tracked years.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/ConferenceEntry" diff --git a/schemas/artifacts_by_year.schema.json b/schemas/artifacts_by_year.schema.json index 501f149..5d13897 100644 --- a/schemas/artifacts_by_year.schema.json +++ b/schemas/artifacts_by_year.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/artifacts_by_year.schema.json", "title": "ArtifactsByYear Collection", "description": "Array of ArtifactsByYear records. Each element: Total artifact count for a single year, split into systems and security conferences.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/ArtifactsByYear" diff --git a/schemas/author_index.schema.json b/schemas/author_index.schema.json index fb257cb..4241e72 100644 --- a/schemas/author_index.schema.json +++ b/schemas/author_index.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/author_index.schema.json", "title": "AuthorIndexEntry Collection", "description": "Array of AuthorIndexEntry records. Each element: Canonical author record: stable ID, name, affiliation, external identifiers, and enrichment history.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/AuthorIndexEntry" diff --git a/schemas/author_profiles.schema.json b/schemas/author_profiles.schema.json index d179bb9..ffe96a4 100644 --- a/schemas/author_profiles.schema.json +++ b/schemas/author_profiles.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/author_profiles.schema.json", "title": "AuthorProfile Collection", "description": "Array of AuthorProfile records. Each element: Unified author profile combining publication record, artifact metrics, and AE committee service.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/AuthorProfile" diff --git a/schemas/author_stats.schema.json b/schemas/author_stats.schema.json index 1753ed9..4aa7df0 100644 --- a/schemas/author_stats.schema.json +++ b/schemas/author_stats.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/author_stats.schema.json", "title": "AuthorStats Collection", "description": "Array of AuthorStats records. Each element: Per-author artifact statistics including badge counts, paper breakdowns, and conference history.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/AuthorStats" diff --git a/schemas/chair_stats.schema.json b/schemas/chair_stats.schema.json index 4f580ea..288d209 100644 --- a/schemas/chair_stats.schema.json +++ b/schemas/chair_stats.schema.json @@ -1,7 +1,7 @@ { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "https://reprodb.github.io/data-schemas/schemas/chair_stats.schema.json", - "version": "0.2.0", + "version": "0.3.0", "$defs": { "ChairStatsSummary": { "additionalProperties": false, diff --git a/schemas/combined_rankings.schema.json b/schemas/combined_rankings.schema.json index 29c4583..f40faa5 100644 --- a/schemas/combined_rankings.schema.json +++ b/schemas/combined_rankings.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/combined_rankings.schema.json", "title": "AuthorRanking Collection", "description": "Array of AuthorRanking records. Each element: Author ranking entry combining artifact contributions, citation impact, and AE committee service.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/AuthorRanking" diff --git a/schemas/committee_stats.schema.json b/schemas/committee_stats.schema.json index 3f70bec..b153256 100644 --- a/schemas/committee_stats.schema.json +++ b/schemas/committee_stats.schema.json @@ -1,7 +1,7 @@ { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "https://reprodb.github.io/data-schemas/schemas/committee_stats.schema.json", - "version": "0.2.0", + "version": "0.3.0", "$defs": { "CommitteeSize": { "additionalProperties": false, diff --git a/schemas/institution_ranking_history.schema.json b/schemas/institution_ranking_history.schema.json index 08bf0ce..2640402 100644 --- a/schemas/institution_ranking_history.schema.json +++ b/schemas/institution_ranking_history.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/institution_ranking_history.schema.json", "title": "InstitutionRankingHistoryEntry Collection", "description": "Array of InstitutionRankingHistoryEntry records. Each element: Dated snapshot of all institution rankings, enabling rank-over-time analysis.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/InstitutionRankingHistoryEntry" diff --git a/schemas/institution_rankings.schema.json b/schemas/institution_rankings.schema.json index 375c3a2..145e39f 100644 --- a/schemas/institution_rankings.schema.json +++ b/schemas/institution_rankings.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/institution_rankings.schema.json", "title": "InstitutionRanking Collection", "description": "Array of InstitutionRanking records. Each element: Institution ranking aggregating artifact production, citation impact, and AE committee service across affiliated authors.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/InstitutionRanking" @@ -374,7 +374,8 @@ "artifact_pct", "author_count", "conferences", - "years" + "years", + "top_authors" ], "title": "InstitutionRanking", "type": "object" diff --git a/schemas/paper_index.schema.json b/schemas/paper_index.schema.json index 7907cfc..d1dc58a 100644 --- a/schemas/paper_index.schema.json +++ b/schemas/paper_index.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/paper_index.schema.json", "title": "Paper Collection", "description": "Array of Paper records. Each element: A paper published at a tracked conference, with artifact badges and citation data.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/Paper" diff --git a/schemas/participation_stats.schema.json b/schemas/participation_stats.schema.json index 4c10474..d27e56e 100644 --- a/schemas/participation_stats.schema.json +++ b/schemas/participation_stats.schema.json @@ -1,7 +1,7 @@ { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "https://reprodb.github.io/data-schemas/schemas/participation_stats.schema.json", - "version": "0.2.0", + "version": "0.3.0", "$defs": { "AreaTrend": { "additionalProperties": false, diff --git a/schemas/ranking_history.schema.json b/schemas/ranking_history.schema.json index 9cc47dc..10589db 100644 --- a/schemas/ranking_history.schema.json +++ b/schemas/ranking_history.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/ranking_history.schema.json", "title": "RankingHistoryEntry Collection", "description": "Array of RankingHistoryEntry records. Each element: Dated snapshot of all author rankings, enabling rank-over-time analysis.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/RankingHistoryEntry" diff --git a/schemas/repo_stats.schema.json b/schemas/repo_stats.schema.json index c92e6d0..11e4320 100644 --- a/schemas/repo_stats.schema.json +++ b/schemas/repo_stats.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/repo_stats.schema.json", "title": "RepoStatsEntry Collection", "description": "Array of RepoStatsEntry records. Each element: GitHub/Zenodo repository metrics for a single artifact (stars, forks, views, downloads).", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/RepoStatsEntry" diff --git a/schemas/repo_stats_summary.schema.json b/schemas/repo_stats_summary.schema.json index ab3e241..cef4c67 100644 --- a/schemas/repo_stats_summary.schema.json +++ b/schemas/repo_stats_summary.schema.json @@ -1,7 +1,7 @@ { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "https://reprodb.github.io/data-schemas/schemas/repo_stats_summary.schema.json", - "version": "0.2.0", + "version": "0.3.0", "$defs": { "AreaRepoStats": { "additionalProperties": false, diff --git a/schemas/repo_stats_yearly.schema.json b/schemas/repo_stats_yearly.schema.json index 98e7a28..492223f 100644 --- a/schemas/repo_stats_yearly.schema.json +++ b/schemas/repo_stats_yearly.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/repo_stats_yearly.schema.json", "title": "RepoStatsYearly Collection", "description": "Array of RepoStatsYearly records. Each element: Yearly repository statistics split by area (all, systems, security) with aggregate GitHub metrics.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/RepoStatsYearly" diff --git a/schemas/search_data.schema.json b/schemas/search_data.schema.json index 6392d4e..9901500 100644 --- a/schemas/search_data.schema.json +++ b/schemas/search_data.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/search_data.schema.json", "title": "SearchEntry Collection", "description": "Array of SearchEntry records. Each element: Searchable artifact entry with paper metadata, badges, author names, and institutional affiliations.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/SearchEntry" @@ -75,6 +75,19 @@ "title": "Artifact Urls", "type": "array" }, + "artifinder_urls": { + "description": "Artifact URLs discovered by ArtiFinder and matched to this paper. Not manually verified, no badges, excluded from all scores. Shown in the UI with an 'Artifinder' marker.", + "examples": [ + [ + "https://github.com/org/discovered-repo" + ] + ], + "items": { + "type": "string" + }, + "title": "Artifinder Urls", + "type": "array" + }, "doi_url": { "description": "Paper DOI URL, e.g. 'https://doi.org/10.1145/...'. Empty string if not available.", "examples": [ @@ -175,6 +188,19 @@ "Distinguished Artifact" ], "title": "Award" + }, + "source": { + "default": "ae", + "description": "Provenance of the entry: 'ae' for papers that went through artifact evaluation (they may still carry additional artifinder_urls), or 'artifinder' for papers that were only discovered by ArtiFinder and never went through AE (no badges, no scores).", + "enum": [ + "ae", + "artifinder" + ], + "examples": [ + "ae" + ], + "title": "Source", + "type": "string" } }, "required": [ diff --git a/schemas/summary.schema.json b/schemas/summary.schema.json index a765da1..4c00157 100644 --- a/schemas/summary.schema.json +++ b/schemas/summary.schema.json @@ -1,7 +1,7 @@ { "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "https://reprodb.github.io/data-schemas/schemas/summary.schema.json", - "version": "0.2.0", + "version": "0.3.0", "additionalProperties": false, "description": "High-level summary: total artifact and conference counts, year range, and last-updated timestamp.", "properties": { diff --git a/schemas/top_repos.schema.json b/schemas/top_repos.schema.json index 3294382..1659da8 100644 --- a/schemas/top_repos.schema.json +++ b/schemas/top_repos.schema.json @@ -3,7 +3,7 @@ "$id": "https://reprodb.github.io/data-schemas/schemas/top_repos.schema.json", "title": "TopRepo Collection", "description": "Array of TopRepo records. Each element: Top-ranked artifact repository by GitHub stars, with paper metadata and activity info.", - "version": "0.2.0", + "version": "0.3.0", "type": "array", "items": { "$ref": "#/$defs/TopRepo"