Update Croissant RAI metadata

croissant.json

@@ -2,286 +2,128 @@
   "@context": {
     "@language": "en",
     "@vocab": "https://schema.org/",
+    "arrayShape": "cr:arrayShape",
+    "citeAs": "cr:citeAs",
+    "column": "cr:column",
+    "conformsTo": "dct:conformsTo",
+    "containedIn": "cr:containedIn",
     "cr": "http://mlcommons.org/croissant/",
-    "sc": "https://schema.org/",
-    "recordSet": "cr:recordSet",
+    "data": {
+      "@id": "cr:data",
+      "@type": "@json"
+    },
+    "dataBiases": "cr:dataBiases",
+    "dataCollection": "cr:dataCollection",
+    "dataType": {
+      "@id": "cr:dataType",
+      "@type": "@vocab"
+    },
+    "dct": "http://purl.org/dc/terms/",
+    "extract": "cr:extract",
     "field": "cr:field",
-    "dataType": "cr:dataType",
-    "source": "cr:source",
+    "fileProperty": "cr:fileProperty",
     "fileObject": "cr:fileObject",
-    "extract": "cr:extract",
-    "column": "cr:column"
+    "fileSet": "cr:fileSet",
+    "format": "cr:format",
+    "includes": "cr:includes",
+    "isArray": "cr:isArray",
+    "isLiveDataset": "cr:isLiveDataset",
+    "jsonPath": "cr:jsonPath",
+    "key": "cr:key",
+    "md5": "cr:md5",
+    "parentField": "cr:parentField",
+    "path": "cr:path",
+    "personalSensitiveInformation": "cr:personalSensitiveInformation",
+    "recordSet": "cr:recordSet",
+    "references": "cr:references",
+    "regex": "cr:regex",
+    "repeated": "cr:repeated",
+    "replace": "cr:replace",
+    "sc": "https://schema.org/",
+    "separator": "cr:separator",
+    "source": "cr:source",
+    "subField": "cr:subField",
+    "transform": "cr:transform",
+    "rai": "http://mlcommons.org/croissant/RAI/",
+    "prov": "http://www.w3.org/ns/prov#"
   },
   "@type": "sc:Dataset",
-  "name": "RSFaith-Bench",
-  "alternateName": "RSFaith-Bench",
-  "description": "RSFaith-Bench is a remote-sensing vision-language benchmark with multiple-choice question-answer records, compact scene graphs, supporting evidence, and executable reasoning programs.",
-  "version": "1.0.0",
-  "license": "https://creativecommons.org/licenses/by-nc/4.0/deed.en",
-  "inLanguage": "en",
-  "keywords": [
-    "remote sensing",
-    "visual question answering",
-    "vision-language benchmark",
-    "scene graph",
-    "multiple choice",
-    "geospatial"
-  ],
   "distribution": [
     {
       "@type": "cr:FileObject",
-      "@id": "metadata_jsonl",
-      "name": "metadata.jsonl",
-      "contentUrl": "metadata.jsonl",
-      "encodingFormat": "application/jsonl",
-      "contentSize": 42232446,
-      "description": "Flat index over all RSFaith-Bench records."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "dataset_manifest_json",
-      "name": "dataset_manifest.json",
-      "contentUrl": "dataset_manifest.json",
-      "encodingFormat": "application/json",
-      "contentSize": 4475,
-      "description": "Dataset counts and archive manifest."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__fine_grained_recognition__records_json",
-      "name": "fine_grained_recognition.json",
-      "contentUrl": "Perception/fine_grained_recognition/fine_grained_recognition.json",
-      "encodingFormat": "application/json",
-      "contentSize": 3110650,
-      "description": "Question-answer records for Perception/fine_grained_recognition."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__fine_grained_recognition__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Perception/fine_grained_recognition/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 1417644607,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Perception/fine_grained_recognition."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__object_counting__records_json",
-      "name": "object_counting.json",
-      "contentUrl": "Perception/object_counting/object_counting.json",
-      "encodingFormat": "application/json",
-      "contentSize": 2898226,
-      "description": "Question-answer records for Perception/object_counting."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__object_counting__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Perception/object_counting/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 2131892267,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Perception/object_counting."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__object_localization__records_json",
-      "name": "object_localization.json",
-      "contentUrl": "Perception/object_localization/object_localization.json",
-      "encodingFormat": "application/json",
-      "contentSize": 2803453,
-      "description": "Question-answer records for Perception/object_localization."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__object_localization__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Perception/object_localization/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 2495263000,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Perception/object_localization."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__object_presence__records_json",
-      "name": "object_presence.json",
-      "contentUrl": "Perception/object_presence/object_presence.json",
-      "encodingFormat": "application/json",
-      "contentSize": 1491906,
-      "description": "Question-answer records for Perception/object_presence."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Perception__object_presence__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Perception/object_presence/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 1554646842,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Perception/object_presence."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__aggregate_distribution__records_json",
-      "name": "aggregate_distribution.json",
-      "contentUrl": "Relational reasoning/aggregate_distribution/aggregate_distribution.json",
-      "encodingFormat": "application/json",
-      "contentSize": 7018837,
-      "description": "Question-answer records for Relational reasoning/aggregate_distribution."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__aggregate_distribution__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Relational reasoning/aggregate_distribution/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 5081368574,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Relational reasoning/aggregate_distribution."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__directional__records_json",
-      "name": "directional.json",
-      "contentUrl": "Relational reasoning/directional/directional.json",
-      "encodingFormat": "application/json",
-      "contentSize": 4697127,
-      "description": "Question-answer records for Relational reasoning/directional."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__directional__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Relational reasoning/directional/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 1833785421,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Relational reasoning/directional."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__projective_ordering__records_json",
-      "name": "projective_ordering.json",
-      "contentUrl": "Relational reasoning/projective_ordering/projective_ordering.json",
-      "encodingFormat": "application/json",
-      "contentSize": 5354614,
-      "description": "Question-answer records for Relational reasoning/projective_ordering."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__projective_ordering__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Relational reasoning/projective_ordering/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 1473705066,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Relational reasoning/projective_ordering."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__proximity__records_json",
-      "name": "proximity.json",
-      "contentUrl": "Relational reasoning/proximity/proximity.json",
-      "encodingFormat": "application/json",
-      "contentSize": 10594022,
-      "description": "Question-answer records for Relational reasoning/proximity."
+      "@id": "repo",
+      "name": "repo",
+      "description": "The Hugging Face git repository.",
+      "contentUrl": "https://huggingface.co/datasets/anon-neurips2026/RSFaith-Bench/tree/refs%2Fconvert%2Fparquet",
+      "encodingFormat": "git+https",
+      "sha256": "https://github.com/mlcommons/croissant/issues/80"
     },
     {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__proximity__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Relational reasoning/proximity/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 1490183220,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Relational reasoning/proximity."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__topological__records_json",
-      "name": "topological.json",
-      "contentUrl": "Relational reasoning/topological/topological.json",
-      "encodingFormat": "application/json",
-      "contentSize": 3784579,
-      "description": "Question-answer records for Relational reasoning/topological."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Relational_reasoning__topological__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Relational reasoning/topological/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 1238362287,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Relational reasoning/topological."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Temporal_reasoning__category_turnover__records_json",
-      "name": "category_turnover.json",
-      "contentUrl": "Temporal reasoning/category_turnover/category_turnover.json",
-      "encodingFormat": "application/json",
-      "contentSize": 4609363,
-      "description": "Question-answer records for Temporal reasoning/category_turnover."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Temporal_reasoning__category_turnover__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Temporal reasoning/category_turnover/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 1103133575,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Temporal reasoning/category_turnover."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Temporal_reasoning__net_change__records_json",
-      "name": "net_change.json",
-      "contentUrl": "Temporal reasoning/net_change/net_change.json",
-      "encodingFormat": "application/json",
-      "contentSize": 9985480,
-      "description": "Question-answer records for Temporal reasoning/net_change."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Temporal_reasoning__net_change__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Temporal reasoning/net_change/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 2299874948,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Temporal reasoning/net_change."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Temporal_reasoning__semantic_transition__records_json",
-      "name": "semantic_transition.json",
-      "contentUrl": "Temporal reasoning/semantic_transition/semantic_transition.json",
-      "encodingFormat": "application/json",
-      "contentSize": 9784610,
-      "description": "Question-answer records for Temporal reasoning/semantic_transition."
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "Temporal_reasoning__semantic_transition__assets_archive",
-      "name": "assets.tar.zst",
-      "contentUrl": "Temporal reasoning/semantic_transition/assets.tar.zst",
-      "encodingFormat": "application/zstd",
-      "contentSize": 689231000,
-      "description": "Zstandard-compressed tar archive containing images/ and scene_graphs/ for Temporal reasoning/semantic_transition."
+      "@type": "cr:FileSet",
+      "@id": "parquet-files-for-config-default",
+      "containedIn": {
+        "@id": "repo"
+      },
+      "encodingFormat": "application/x-parquet",
+      "includes": "default/*/*.parquet"
     }
   ],
   "recordSet": [
     {
       "@type": "cr:RecordSet",
-      "name": "qa_records",
-      "description": "Flat metadata index over all RSFaith-Bench question-answer records.",
-      "source": {
-        "fileObject": {
-          "@id": "metadata_jsonl"
-        }
+      "dataType": "cr:Split",
+      "key": {
+        "@id": "default_splits/split_name"
       },
+      "@id": "default_splits",
+      "name": "default_splits",
+      "description": "Splits for the default config.",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "@id": "default_splits/split_name",
+          "dataType": "sc:Text"
+        }
+      ],
+      "data": [
+        {
+          "default_splits/split_name": "benchmark"
+        }
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "default",
+      "description": "anon-neurips2026/RSFaith-Bench - 'default' subset",
       "field": [
         {
           "@type": "cr:Field",
-          "name": "question_id",
-          "description": "Anonymized question identifier.",
+          "@id": "default/split",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "fileProperty": "fullpath"
+            },
+            "transform": {
+              "regex": "default/(?:partial-)?(benchmark)/.+parquet$"
+            }
+          },
+          "references": {
+            "field": {
+              "@id": "default_splits/split_name"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/question_id",
+          "dataType": "sc:Text",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "question_id"
@@ -290,12 +132,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "scene_id",
-          "description": "Anonymized scene identifier.",
+          "@id": "default/scene_id",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "scene_id"
@@ -304,12 +145,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "level",
-          "description": "High-level reasoning category.",
+          "@id": "default/level",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "level"
@@ -318,12 +158,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "subcategory",
-          "description": "Fine-grained reasoning category.",
+          "@id": "default/subcategory",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "subcategory"
@@ -332,12 +171,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "question",
-          "description": "Natural-language question.",
+          "@id": "default/question",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "question"
@@ -346,12 +184,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "answer",
-          "description": "Correct answer.",
+          "@id": "default/answer",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "answer"
@@ -360,12 +197,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "answer_type",
-          "description": "Answer representation.",
+          "@id": "default/answer_type",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "answer_type"
@@ -374,26 +210,26 @@
         },
         {
           "@type": "cr:Field",
-          "name": "choices",
-          "description": "Multiple-choice options encoded as JSON.",
+          "@id": "default/choices",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "choices"
             }
-          }
+          },
+          "isArray": true,
+          "arrayShape": "-1"
         },
         {
           "@type": "cr:Field",
-          "name": "image_t1",
-          "description": "Relative path to the first image after archive extraction.",
+          "@id": "default/image_t1",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "image_t1"
@@ -402,12 +238,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "image_t2",
-          "description": "Relative path to the second image for temporal examples, if present.",
+          "@id": "default/image_t2",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "image_t2"
@@ -416,12 +251,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "scene_graph",
-          "description": "Relative path to the compact scene graph after archive extraction.",
+          "@id": "default/scene_graph",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "scene_graph"
@@ -430,12 +264,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "record_json",
-          "description": "Relative path to the subcategory JSON file.",
+          "@id": "default/record_json",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "record_json"
@@ -444,12 +277,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "support_json",
-          "description": "Grounded support evidence encoded as JSON.",
+          "@id": "default/support_json",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "support_json"
@@ -458,12 +290,11 @@
         },
         {
           "@type": "cr:Field",
-          "name": "program_json",
-          "description": "Executable reasoning program encoded as JSON.",
+          "@id": "default/program_json",
           "dataType": "sc:Text",
           "source": {
-            "fileObject": {
-              "@id": "metadata_jsonl"
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
             },
             "extract": {
               "column": "program_json"
@@ -472,5 +303,74 @@
         }
       ]
     }
-  ]
+  ],
+  "conformsTo": "http://mlcommons.org/croissant/1.1",
+  "name": "RSFaith-Bench",
+  "description": "\n\t\n\t\t\n\t\tRSFaith-Bench\n\t\n\nRSFaith-Bench is a remote-sensing vision-language benchmark designed to\nevaluate grounded visual reasoning beyond surface-level object recognition.\nThe benchmark covers perception, relational reasoning, and temporal reasoning\nover remote-sensing imagery. Each example is formulated as a multiple-choice\nquestion and is paired with a compact scene graph, supporting evidence, and an\nexecutable reasoning program.\nThe release contains 13,511 question-answer records, 16,288… See the full description on the dataset page: https://huggingface.co/datasets/anon-neurips2026/RSFaith-Bench.",
+  "alternateName": [
+    "anon-neurips2026/RSFaith-Bench",
+    "RSFaith-Bench"
+  ],
+  "creator": {
+    "@type": "Person",
+    "name": "anon-neurips2026",
+    "url": "https://huggingface.co/anon-neurips2026"
+  },
+  "keywords": [
+    "visual-question-answering",
+    "English",
+    "cc-by-nc-4.0",
+    "10K - 100K",
+    "json",
+    "Text",
+    "Geospatial",
+    "Datasets",
+    "pandas",
+    "Polars",
+    "Croissant",
+    "arxiv:2406.06028",
+    "🇺🇸 Region: US",
+    "remote-sensing",
+    "vision-language",
+    "benchmark",
+    "scene-graph",
+    "multiple-choice",
+    "geospatial"
+  ],
+  "license": "https://choosealicense.com/licenses/cc-by-nc-4.0/",
+  "url": "https://huggingface.co/datasets/anon-neurips2026/RSFaith-Bench",
+  "rai:dataLimitations": [
+    "RSFaith-Bench is intended for diagnostic evaluation of remote-sensing vision-language models, not for operational geospatial decision making. The dataset is derived from selected public remote-sensing research datasets and therefore does not represent all sensors, resolutions, geographies, seasons, land-use patterns, disaster types, or object taxonomies. Performance measured on this benchmark may not generalize to unseen regions, sensors, task formulations, or safety-critical deployments.",
+    "The benchmark uses compact scene graphs, templated/constructed questions, multiple-choice options, support evidence, and executable reasoning programs. These derived annotations may contain residual extraction, normalization, or construction errors even after validation. The dataset should not be used as the sole evidence source for surveillance, emergency response, legal, financial, insurance, military, or other high-stakes decisions."
+  ],
+  "rai:dataBiases": [
+    "The data distribution reflects the source datasets and the benchmark construction process. It may over-represent object categories and scenes common in remote-sensing benchmarks, including urban infrastructure, vehicles, aircraft, ships, buildings, land-cover classes, and disaster-damage imagery, while under-representing other regions, rural settings, seasons, sensors, and rare object classes.",
+    "Question templates, answer choices, scene-graph construction, and filtering rules may introduce benchmark-specific biases, including lexical regularities, answer-distribution artifacts, class imbalance, and task imbalance across perception, relational reasoning, and temporal reasoning. Models may exploit such artifacts without robust grounded reasoning."
+  ],
+  "rai:personalSensitiveInformation": [
+    "The dataset is not designed to contain person-level demographic, biometric, health, political, religious, or directly identifying personal information. The imagery is remote-sensing imagery and the annotations use anonymized question and scene identifiers.",
+    "Remote-sensing images may nevertheless depict human-built environments, infrastructure, vehicles, buildings, or disaster-affected areas. Users should not use the dataset for identifying individuals, inferring protected attributes, tracking people, or making decisions about specific persons or communities."
+  ],
+  "rai:dataUseCases": [
+    "Recommended uses include academic research and benchmark evaluation of remote-sensing multimodal large language models, especially grounded visual reasoning, faithfulness of supporting evidence, perception, relational reasoning, temporal reasoning, error analysis, and reproducibility studies.",
+    "The dataset is not recommended for training or validating systems for operational surveillance, autonomous geospatial decision making, disaster response triage, law enforcement, military targeting, insurance assessment, legal adjudication, or any other safety-critical or rights-affecting application without independent domain validation and additional governance."
+  ],
+  "rai:dataSocialImpact": "RSFaith-Bench may help the research community identify hallucination, unfaithful evidence use, and reasoning failures in remote-sensing vision-language models before such systems are deployed. Potential negative impacts include overclaiming model reliability from benchmark scores, reinforcing biases inherited from source datasets, or repurposing the benchmark toward surveillance or high-stakes geospatial decisions. The release mitigates these risks through non-commercial licensing, anonymized identifiers, explicit usage limitations, and documentation of known biases and limitations.",
+  "rai:hasSyntheticData": true,
+  "rai:dataCollection": "RSFaith-Bench is derived from public remote-sensing research datasets and their associated imagery or annotations. The benchmark packages selected images, compact scene graphs, question-answer records, choices, support evidence, and executable reasoning programs for remote-sensing visual reasoning evaluation.",
+  "rai:dataAnnotationProtocol": "Question-answer records, answer choices, support evidence, compact scene graphs, and executable reasoning programs were constructed from source annotations and benchmark-generation rules, then checked for record consistency, answer-choice validity, file references, image readability, and scene-graph consistency before release.",
+  "rai:dataPreprocessingProtocol": "The release uses anonymized sequential question and scene identifiers, removes original dataset names from record-level metadata, packages images and scene graphs into per-subcategory archives, filters or replaces images larger than 100 MiB before archiving, and provides a flat metadata index for the 13,511 records.",
+  "prov:wasDerivedFrom": [
+    "https://gcheng-nwpu.github.io/",
+    "https://captain-whu.github.io/DOTA/dataset.html",
+    "https://www.gaofen-challenge.com/benchmark",
+    "https://github.com/GeoZcx/A-deeply-supervised-image-fusion-network-for-change-detection-in-remote-sensing-images/tree/master/dataset",
+    "https://xview2.org/dataset",
+    "https://arxiv.org/abs/2406.06028"
+  ],
+  "prov:wasGeneratedBy": "RSFaith-Bench benchmark construction pipeline: source-dataset selection, image and annotation normalization, scene-graph construction, question-answer generation, support/program construction, replacement of oversized images with same-subcategory alternatives, asset archiving, and final consistency validation.",
+  "version": "1.0.0",
+  "datePublished": "2026-05-06",
+  "citation": "RSFaith-Bench: When Correct Answers Come with Unfaithful Evidence in Remote Sensing MLLMs. Anonymous, 2026.",
+  "citeAs": "RSFaith-Bench: When Correct Answers Come with Unfaithful Evidence in Remote Sensing MLLMs. Anonymous, 2026."
 }