feat: user and org data

Author: caviri

sample-data/parser-json/MMathisLab.json

@@ -1,7 +1,7 @@
 {
   "link": "https://github.com/MMathisLab",
   "type": "user",
-  "parsedTimestamp": "2025-11-05T11:55:43.928045",
+  "parsedTimestamp": "2025-11-06T10:40:22.532462",
   "output": {
     "name": "Mackenzie Mathis",
     "fullname": "Mackenzie Mathis",
@@ -47,161 +47,108 @@
       "readme_content": "## Hi there, this is Mackenzie! 👋\n\nI am a Prof. at EPFL leading a lab that studies adaptive intelligence. What is that? Check out  my prespective piece [here](https://www.arxiv.org/abs/2411.15234). My handle reflects some historical use of GitHub before Orgs 👀 where I used to host all lab repos, but now most are under our Org ➡️ [here](https://github.com/AdaptiveMotorControlLab).\n\nBelow is a collection of projects I contribute to, with a few highlights being:\n\n- 💜 [DeepLabCut](https://www.mackenziemathislab.org/deeplabcut)\n- 🦓 [CEBRA](https://github.com/AdaptiveMotorControlLab/CEBRA)\n- 🎻 [AmadeusGPT](https://www.mackenziemathislab.org/amadeusgpt)\n- 🟠 [CellSeg3D](https://github.com/AdaptiveMotorControlLab/CellSeg3D)\n\n\nThanks for visiting! \n\n\n\n"
     },
     "relatedToOrganization": [
-      "EPFL",
       "AdaptiveMotorControlLab",
-      "Swiss Data Science Center",
       "DeepLabCut",
       "DeepLabCutAIResidency",
       "SCENE-Collaboration",
+      "CEBRA",
       "kinematik-ai",
-      "Prof Alexander Mathis Group",
-      "Brain Mind Institute",
-      "École Polytechnique Fédérale de Lausanne"
-    ],
-    "relatedToOrganizationsROR": [
+      "Swiss Data Science Center",
+      "EPFL",
+      "Cochrane",
+      "AdaptiveMotorControlLab",
+      "DeepLabCut",
+      "DeepLabCutAIResidency",
       {
         "type": "Organization",
-        "legalName": "Prof Alexander Mathis Group",
-        "hasRorId": null,
+        "legalName": "Cochrane",
+        "hasRorId": "https://ror.org/00dbkrq51",
         "alternateNames": [
-          "Alexander Mathis Lab"
+          "Cochrane Collaboration"
         ],
-        "organizationType": "Research Group",
-        "parentOrganization": "Brain Mind Institute",
-        "country": "Switzerland",
-        "website": "https://mathislab.epfl.ch/",
-        "attributionConfidence": 0.8,
-        "academicCatalogRelations": [
-          {
-            "catalogType": "infoscience",
-            "entityType": "orgunit",
-            "entity": {
-              "uuid": null,
-              "name": "Prof Alexander Mathis Group",
-              "email": null,
-              "orcid": null,
-              "affiliation": null,
-              "profile_url": null,
-              "publication_count": null,
-              "research_interests": []
-            },
-            "confidence": 0.9,
-            "justification": "Identified via EPFL Infoscience lab search for 'Mathis Lab'; corresponds to the GitHub 'MMathisLab' repository owner.",
-            "externalId": null,
-            "matchedOn": [
-              "name"
-            ]
-          }
-        ]
+        "organizationType": "Research organization",
+        "parentOrganization": null,
+        "country": "United Kingdom",
+        "website": "http://uk.cochrane.org/",
+        "attributionConfidence": 0.6,
+        "academicCatalogRelations": []
       },
       {
         "type": "Organization",
-        "legalName": "Brain Mind Institute",
+        "legalName": "AdaptiveMotorControlLab",
         "hasRorId": null,
-        "alternateNames": [
-          "BMI"
-        ],
-        "organizationType": "Research Institute",
-        "parentOrganization": "École Polytechnique Fédérale de Lausanne",
-        "country": "Switzerland",
-        "website": "https://bmi.epfl.ch/",
-        "attributionConfidence": 0.7,
-        "academicCatalogRelations": [
-          {
-            "catalogType": "infoscience",
-            "entityType": "orgunit",
-            "entity": {
-              "uuid": null,
-              "name": "Bertarelli Foundation Chair of Integrative Neuroscience",
-              "email": null,
-              "orcid": null,
-              "affiliation": null,
-              "profile_url": null,
-              "publication_count": null,
-              "research_interests": []
-            },
-            "confidence": 0.6,
-            "justification": "Inferred relationship: the Prof Alexander Mathis Group is part of BMI, confirmed by Infoscience records for EPFL units.",
-            "externalId": null,
-            "matchedOn": [
-              "partOf"
-            ]
-          }
-        ]
+        "alternateNames": null,
+        "organizationType": "Research lab",
+        "parentOrganization": null,
+        "country": null,
+        "website": null,
+        "attributionConfidence": 0.1,
+        "academicCatalogRelations": []
       },
       {
         "type": "Organization",
-        "legalName": "École Polytechnique Fédérale de Lausanne",
-        "hasRorId": "https://ror.org/03yrm5c26",
-        "alternateNames": [
-          "EPFL",
-          "Ecole Polytechnique Federale de Lausanne"
-        ],
-        "organizationType": "University",
+        "legalName": "DeepLabCut",
+        "hasRorId": null,
+        "alternateNames": null,
+        "organizationType": "Software project",
         "parentOrganization": null,
-        "country": "Switzerland",
-        "website": "https://www.epfl.ch/",
-        "attributionConfidence": 0.9,
+        "country": null,
+        "website": null,
+        "attributionConfidence": 0.2,
+        "academicCatalogRelations": []
+      },
+      {
+        "type": "Organization",
+        "legalName": "DeepLabCutAIResidency",
+        "hasRorId": null,
+        "alternateNames": null,
+        "organizationType": "Program",
+        "parentOrganization": null,
+        "country": null,
+        "website": null,
+        "attributionConfidence": 0.1,
         "academicCatalogRelations": []
       }
     ],
     "relatedToOrganizationJustification": [
-      "README: 'I am a Prof. at EPFL leading a lab…' indicates faculty appointment at EPFL.",
-      "Bio & company field: 'PI @AdaptiveMotorControlLab' shows leadership of the EPFL lab.",
-      "README (explicit critical phrase): 'Currently, I am working as a Data Engineer at the Swiss Data Science Center at EPFL.'",
-      "Bio & README highlights: contributor to DeepLabCut, GitHub org 'DeepLabCut'.",
-      "Organizations list includes 'DeepLabCutAIResidency', indicating participation in that program.",
-      "Organizations list includes 'SCENE-Collaboration', showing membership in that collaborative network.",
-      "Bio: 'co-founder @kinematik-ai' indicates startup affiliation."
+      "Company field '@AdaptiveMotorControlLab' and bio 'PI @AdaptiveMotorControlLab' indicate affiliation with AdaptiveMotorControlLab.",
+      "Bio 'DeepLabCut' and membership in GitHub org 'DeepLabCut' show involvement with the DeepLabCut project.",
+      "Membership in GitHub org 'DeepLabCutAIResidency' lists DeepLabCutAIResidency as an affiliated organization.",
+      "Membership in GitHub org 'SCENE-Collaboration' lists SCENE-Collaboration as an affiliated organization.",
+      "Membership in GitHub org 'CEBRA' and bio listing CEBRA indicate involvement with the CEBRA project organization.",
+      "Bio 'co-founder @kinematik-ai' shows a co-founder role at kinematik-ai.",
+      "README states 'Currently, I am working as a Data Engineer at the Swiss Data Science Center at EPFL.'",
+      "README content 'I am a Prof. at EPFL leading a lab...' shows EPFL affiliation."
     ],
     "discipline": [
       "Information engineering",
-      "Biological engineering"
+      "Applied sciences"
     ],
     "disciplineJustification": [
-      "Development of DeepLabCut and CEBRA involves computational algorithms and data pipelines – core to information engineering.",
-      "Research on adaptive intelligence for biological systems (animal behavior tracking) aligns with biological engineering."
+      "Development of machine learning software and computational models like DeepLabCut and CEBRA indicates work in information engineering.",
+      "Research focus on adaptive intelligence and bioimaging tools such as CellSeg3D demonstrates an applied sciences orientation."
     ],
     "position": [
+      "Data Engineer",
       "Professor",
       "Principal Investigator",
-      "Data Engineer",
-      "Co-Founder"
+      "Co-founder"
     ],
     "positionJustification": [
-      "README: 'I am a Prof. at EPFL leading a lab…' explicitly states professorship.",
-      "Bio: 'PI @AdaptiveMotorControlLab' indicates role as Principal Investigator.",
-      "README (critical text): 'Currently, I am working as a Data Engineer at the Swiss Data Science Center at EPFL.'",
-      "Bio: 'co-founder @kinematik-ai' explicitly notes co-founder role."
+      "README states 'Currently, I am working as a Data Engineer at the Swiss Data Science Center at EPFL.'",
+      "README content says 'I am a Prof. at EPFL leading a lab that studies adaptive intelligence.'",
+      "Bio 'PI @AdaptiveMotorControlLab' indicates a Principal Investigator role.",
+      "Bio 'co-founder @kinematik-ai' indicates a co-founder position at kinematik-ai."
     ],
     "relatedToEPFL": true,
-    "relatedToEPFLJustification": "The assessment identifies four independent strong signals of affiliation with EPFL:\n\n1. README mention (0.25): The user’s GitHub README explicitly states “I am a Prof. at EPFL leading a lab…,” directly indicating a faculty appointment at EPFL.\n2. Related organization – EPFL (0.25): Enrichment data lists “EPFL” as a related organization for this user.\n3. Related organization – Swiss Data Science Center (0.25): Enrichment data also lists “Swiss Data Science Center,” a joint EPFL+ETH initiative, further linking the user to EPFL.\n4. Infoscience entity (0.40): The user has a person record in EPFL’s Infoscience repository, confirming an institutional affiliation.\n\nSumming these contributions yields 1.15, capped at 1.00. This exceeds the 0.50 threshold, yielding a high-confidence affiliation with EPFL.",
+    "relatedToEPFLJustification": "We found multiple independent and high-quality signals linking Mackenzie Mathis to EPFL:\n\n1. Infoscience Publications (0.4): Four publications authored by Mackenzie Mathis are indexed in EPFL's Infoscience repository, providing the strongest confirmation of an EPFL affiliation.\n2. README Mention of EPFL (0.25): The GitHub README explicitly states \"I am a Prof. at EPFL leading a lab that studies adaptive intelligence.\", clearly indicating an EPFL position.\n3. README Mention of Swiss Data Science Center at EPFL (0.25): The README also notes \"Currently, I am working as a Data Engineer at the Swiss Data Science Center at EPFL.\", reinforcing an institutional role at EPFL.\n4. Related Organization \"EPFL\" (0.25): The user is explicitly listed as related to the organization \"EPFL\" in provided metadata.\n5. Related Organization \"Swiss Data Science Center\" (0.25): The user is also listed as related to the Swiss Data Science Center, a joint EPFL/ETH Zürich initiative, further tying them to EPFL.\n\nSumming these contributions yields 1.40, capped at 1.00. Because the confidence is >= 0.5, Mackenzie Mathis is conclusively related to EPFL.",
     "relatedToEPFLConfidence": 1,
     "academicCatalogRelations": [
-      {
-        "catalogType": "infoscience",
-        "entityType": "person",
-        "entity": {
-          "uuid": "2e985179-c5f5-41b2-aa2d-367f2564acca",
-          "name": "Mathis, Mackenzie",
-          "email": null,
-          "orcid": null,
-          "affiliation": "UPMWMATHIS",
-          "profile_url": "https://infoscience.epfl.ch/entities/person/2e985179-c5f5-41b2-aa2d-367f2564acca",
-          "publication_count": null,
-          "research_interests": []
-        },
-        "confidence": 0.95,
-        "justification": "Exact match found via search_infoscience_authors_tool for \"Mackenzie Mathis\"",
-        "externalId": null,
-        "matchedOn": [
-          "name"
-        ]
-      },
       {
         "catalogType": "infoscience",
         "entityType": "publication",
         "entity": {
+          "type": "InfosciencePublication",
           "uuid": "3b0aba34-2850-400b-b711-b80e35b30478",
           "title": "The neocortical column as a universal template for perception and world-model learning",
           "authors": [
@@ -211,26 +158,24 @@
           "doi": "10.1038/s41583-022-00658-6",
           "publication_date": "2022-11-11",
           "publication_type": "text::journal::journal article::research article",
-          "url": "https://infoscience.epfl.ch/record/20.500.14299/192908",
-          "repository_url": null,
-          "lab": null,
+          "url": "https://infoscience.epfl.ch/entities/publication/3b0aba34-2850-400b-b711-b80e35b30478",
           "subjects": [
             "Neurosciences",
             "Neurosciences & Neurology"
           ]
         },
         "confidence": 0.9,
-        "justification": "Found via get_author_publications_tool search for publications by Mackenzie Mathis; exact author match and DOI provided",
-        "externalId": "10.1038/s41583-022-00658-6",
+        "justification": "Retrieved via get_author_publications_tool; Mackenzie Mathis is listed as author in the publication metadata.",
         "matchedOn": [
-          "doi",
-          "authors"
+          "author_name",
+          "uuid"
         ]
       },
       {
         "catalogType": "infoscience",
         "entityType": "publication",
         "entity": {
+          "type": "InfosciencePublication",
           "uuid": "c664f9d5-3560-40b4-abb2-a13357efdb37",
           "title": "A new spin on fidgets",
           "authors": [
@@ -240,23 +185,21 @@
           "doi": "10.1038/s41593-019-0504-2",
           "publication_date": "2019-10-01",
           "publication_type": "text::journal::journal article::research article",
-          "url": "https://infoscience.epfl.ch/record/20.500.14299/173029",
-          "repository_url": null,
-          "lab": null,
+          "url": "https://infoscience.epfl.ch/entities/publication/c664f9d5-3560-40b4-abb2-a13357efdb37",
           "subjects": []
         },
         "confidence": 0.9,
-        "justification": "Found via get_author_publications_tool; exact author match and DOI",
-        "externalId": "10.1038/s41593-019-0504-2",
+        "justification": "Retrieved via get_author_publications_tool; Mackenzie Mathis is author.",
         "matchedOn": [
-          "doi",
-          "authors"
+          "author_name",
+          "uuid"
         ]
       },
       {
         "catalogType": "infoscience",
         "entityType": "publication",
         "entity": {
+          "type": "InfosciencePublication",
           "uuid": "1e4351b3-671c-4cc7-91e9-7fed068f72e2",
           "title": "Brain dynamics uncovered using a machine-learning algorithm",
           "authors": [
@@ -266,9 +209,7 @@
           "doi": "10.1038/d41586-023-01339-9",
           "publication_date": "2023-05-03",
           "publication_type": "text::journal::journal article::research article",
-          "url": "https://infoscience.epfl.ch/record/20.500.14299/198631",
-          "repository_url": null,
-          "lab": null,
+          "url": "https://infoscience.epfl.ch/entities/publication/1e4351b3-671c-4cc7-91e9-7fed068f72e2",
           "subjects": [
             "Multidisciplinary Sciences",
             "Science & Technology - Other Topics",
@@ -277,17 +218,17 @@
           ]
         },
         "confidence": 0.9,
-        "justification": "Found via get_author_publications_tool; exact author match and DOI",
-        "externalId": "10.1038/d41586-023-01339-9",
+        "justification": "Retrieved via get_author_publications_tool; Mackenzie Mathis is author, relating to CEBRA work mentioned in bio.",
         "matchedOn": [
-          "doi",
-          "authors"
+          "author_name",
+          "uuid"
         ]
       },
       {
         "catalogType": "infoscience",
         "entityType": "publication",
         "entity": {
+          "type": "InfosciencePublication",
           "uuid": "955800f0-af9c-443b-b240-33dbcc24fe4c",
           "title": "Deep learning tools for the measurement of animal behavior in neuroscience",
           "authors": [
@@ -298,17 +239,14 @@
           "doi": "10.1016/j.conb.2019.10.008",
           "publication_date": "2020",
           "publication_type": "text::journal::journal article::research article",
-          "url": "https://infoscience.epfl.ch/record/20.500.14299/173050",
-          "repository_url": null,
-          "lab": null,
+          "url": "https://infoscience.epfl.ch/entities/publication/955800f0-af9c-443b-b240-33dbcc24fe4c",
           "subjects": []
         },
         "confidence": 0.9,
-        "justification": "Found via get_author_publications_tool; exact author match and DOI",
-        "externalId": "10.1016/j.conb.2019.10.008",
+        "justification": "Retrieved via get_author_publications_tool; co-authored with Alexander Mathis, relates to DeepLabCut theme in bio.",
         "matchedOn": [
-          "doi",
-          "authors"
+          "author_name",
+          "uuid"
         ]
       }
     ]
@@ -317,12 +255,15 @@
     "agent_input_tokens": 0,
     "agent_output_tokens": 0,
     "total_tokens": 0,
-    "estimated_input_tokens": 9443,
-    "estimated_output_tokens": 2626,
-    "estimated_total_tokens": 12069,
-    "duration": 209.54886,
-    "start_time": "2025-11-05T11:52:14.378945",
-    "end_time": "2025-11-05T11:55:43.927805",
-    "status_code": 200
+    "estimated_input_tokens": 9549,
+    "estimated_output_tokens": 2402,
+    "estimated_total_tokens": 11951,
+    "duration": 305.513017,
+    "start_time": "2025-11-06T10:35:17.019173",
+    "end_time": "2025-11-06T10:40:22.532190",
+    "status_code": 200,
+    "github_rate_limit": 5000,
+    "github_rate_remaining": 4990,
+    "github_rate_reset": "2025-11-06T11:24:45"
   }
-}
+}