Changes

On December 3, 2024 at 9:58:09 AM UTC, admin:
Changed title to arXiv dataset (previously arXiv Dataset)
Set author of arXiv dataset to Shrey Mishra (previously Hamed Rahimi)
Updated description of arXiv dataset from
The arXiv dataset is a collection of documents published between 2006 and 2022.
to
The dataset used in this paper is a collection of arXiv papers in English, filtered to include only those written in English, with LATEX source available, compilable on a modern LATEX distribution, and containing at least a theorem or a proof environment.
Removed the following tags from arXiv dataset
- Research Papers
- Academic Research
Added the following tags to arXiv dataset
Changed value of field extra_authors to [{'extra_author': 'Antoine Gauquier', 'orcid': ''}, {'extra_author': 'Pierre Senellart', 'orcid': ''}] in arXiv dataset
Changed value of field landing_page to https://github.com/mv96/mm_extraction in arXiv dataset
Deleted resource Original Metadata from arXiv dataset
              
    
          
          
        
        
            f 1 { f 1 {
            2   "access_rights": "", 2   "access_rights": "",
            n 3   "author": "Hamed Rahimi", n 3   "author": "Shrey Mishra",
            4   "author_email": "", 4   "author_email": "",
            5   "citation": [], 5   "citation": [],
            6   "creator_user_id": "17755db4-395a-4b3b-ac09-e8e3484ca700", 6   "creator_user_id": "17755db4-395a-4b3b-ac09-e8e3484ca700",
            7   "defined_in": "", 7   "defined_in": "",
            8   "doi": "10.57702/ep6960zn", 8   "doi": "10.57702/ep6960zn",
            9   "doi_date_published": "2024-12-03", 9   "doi_date_published": "2024-12-03",
            10   "doi_publisher": "TIB", 10   "doi_publisher": "TIB",
            11   "doi_status": true, 11   "doi_status": true,
            12   "domain": "https://service.tib.eu/ldmservice", 12   "domain": "https://service.tib.eu/ldmservice",
            13   "extra_authors": [ 13   "extra_authors": [
            14     { 14     {
            n 15       "extra_author": "Camelia Constantin", n 15       "extra_author": "Antoine Gauquier",
            16       "orcid": "" 16       "orcid": ""
            17     }, 17     },
            18     { 18     {
            n 19       "extra_author": "Hubert Naacke", n 
            20       "orcid": "" 
            21     }, 
            22     { 
            23       "extra_author": "Bernd Amann", 19       "extra_author": "Pierre Senellart",
            24       "orcid": "" 20       "orcid": ""
            25     } 21     }
            26   ], 22   ],
            27   "groups": [ 23   "groups": [
            28     { 24     {
            29       "description": "", 25       "description": "",
            n 30       "display_name": "Academic Research", n 26       "display_name": "Mathematical Text Analysis",
            31       "id": "95de0ad3-0ca3-4849-8aeb-08116ee55f4e", 27       "id": "fc53fb65-8b01-4f2b-b6c5-e4df54a4d3e8",
            32       "image_display_url": "", 28       "image_display_url": "",
            n 33       "name": "academic-research", n 29       "name": "mathematical-text-analysis",
            34       "title": "Academic Research" 30       "title": "Mathematical Text Analysis"
            35     }, 
            36     { 
            37       "description": "", 
            38       "display_name": "Research Papers", 
            39       "id": "bb4db54c-8561-44b6-bd27-53e3a92ed112", 
            40       "image_display_url": "", 
            41       "name": "research-papers", 
            42       "title": "Research Papers" 
            43     } 31     }
            44   ], 32   ],
            45   "id": "fc9b3adf-6ee5-4ef1-9ba6-5c6868035de0", 33   "id": "fc9b3adf-6ee5-4ef1-9ba6-5c6868035de0",
            46   "isopen": false, 34   "isopen": false,
            n 47   "landing_page": "https://arxiv.org/", n 35   "landing_page": "https://github.com/mv96/mm_extraction",
            48   "license_title": null, 36   "license_title": null,
            49   "link_orkg": "", 37   "link_orkg": "",
            50   "metadata_created": "2024-12-03T09:19:11.793172", 38   "metadata_created": "2024-12-03T09:19:11.793172",
            n 51   "metadata_modified": "2024-12-03T09:19:12.214684", n 39   "metadata_modified": "2024-12-03T09:58:08.329440",
            52   "name": "arxiv-dataset", 40   "name": "arxiv-dataset",
            n 53   "notes": "The arXiv dataset is a collection of documents published  n 41   "notes": "The dataset used in this paper is a collection of arXiv 
            54 between 2006 and 2022.", 42 papers in English, filtered to include only those written in English, 
            43 with LATEX source available, compilable on a modern LATEX 
            44 distribution, and containing at least a theorem or a proof 
            45 environment.",
            55   "num_resources": 1, 46   "num_resources": 0,
            56   "num_tags": 3, 47   "num_tags": 4,
            57   "organization": { 48   "organization": {
            58     "approval_status": "approved", 49     "approval_status": "approved",
            59     "created": "2024-11-25T12:11:38.292601", 50     "created": "2024-11-25T12:11:38.292601",
            60     "description": "", 51     "description": "",
            61     "id": "079d46db-32df-4b48-91f3-0a8bc8f69559", 52     "id": "079d46db-32df-4b48-91f3-0a8bc8f69559",
            62     "image_url": "", 53     "image_url": "",
            63     "is_organization": true, 54     "is_organization": true,
            64     "name": "no-organization", 55     "name": "no-organization",
            65     "state": "active", 56     "state": "active",
            66     "title": "No Organization", 57     "title": "No Organization",
            67     "type": "organization" 58     "type": "organization"
            68   }, 59   },
            69   "owner_org": "079d46db-32df-4b48-91f3-0a8bc8f69559", 60   "owner_org": "079d46db-32df-4b48-91f3-0a8bc8f69559",
            70   "private": false, 61   "private": false,
            71   "relationships_as_object": [], 62   "relationships_as_object": [],
            72   "relationships_as_subject": [], 63   "relationships_as_subject": [],
            n 73   "resources": [ n 64   "resources": [],
            74     { 
            75       "cache_last_updated": null, 
            76       "cache_url": null, 
            77       "created": "2024-12-03T10:17:29", 
            78       "data": [ 
            79         "dcterms:title", 
            80         "dcterms:accessRights", 
            81         "dcterms:creator", 
            82         "dcterms:description", 
            83         "dcterms:issued", 
            84         "dcterms:language", 
            85         "dcterms:identifier", 
            86         "dcat:theme", 
            87         "dcterms:type", 
            88         "dcat:keyword", 
            89         "dcat:landingPage", 
            90         "dcterms:hasVersion", 
            91         "dcterms:format", 
            92         "mls:task" 
            93       ], 
            94       "description": "The json representation of the dataset with its  
            95 distributions based on DCAT.", 
            96       "format": "JSON", 
            97       "hash": "", 
            98       "id": "e03c18ec-c28b-405d-bf03-266c8baf7e5d", 
            99       "last_modified": "2024-12-03T09:19:12.207644", 
            100       "metadata_modified": "2024-12-03T09:19:12.217554", 
            101       "mimetype": "application/json", 
            102       "mimetype_inner": null, 
            103       "name": "Original Metadata", 
            104       "package_id": "fc9b3adf-6ee5-4ef1-9ba6-5c6868035de0", 
            105       "position": 0, 
            106       "resource_type": null, 
            107       "size": 655, 
            108       "state": "active", 
            109       "url":  
            110 resource/e03c18ec-c28b-405d-bf03-266c8baf7e5d/download/metadata.json", 
            111       "url_type": "upload" 
            112     } 
            113   ], 
            114   "services_used_list": "", 65   "services_used_list": "",
            115   "state": "active", 66   "state": "active",
            116   "tags": [ 67   "tags": [
            117     { 68     {
            n 118       "display_name": "Academic Research", n 69       "display_name": "Mathematical Text",
            119       "id": "58c40ec8-33f5-49d1-8577-9c806bde6882", 70       "id": "c0891439-57e0-472c-9c44-223644b3fabc",
            120       "name": "Academic Research", 71       "name": "Mathematical Text",
            121       "state": "active", 72       "state": "active",
            122       "vocabulary_id": null 73       "vocabulary_id": null
            123     }, 74     },
            124     { 75     {
            n 125       "display_name": "Research Papers", n 76       "display_name": "Proof Extraction",
            126       "id": "1143b85e-8248-495e-9c25-8595e7908fc4", 77       "id": "095634fd-9818-401f-a509-3c7edfa709db",
            127       "name": "Research Papers", 78       "name": "Proof Extraction",
            79       "state": "active",
            80       "vocabulary_id": null
            81     },
            82     {
            83       "display_name": "Theorem Extraction",
            84       "id": "411264b5-632c-426d-96a3-d696229e0722",
            85       "name": "Theorem Extraction",
            128       "state": "active", 86       "state": "active",
            129       "vocabulary_id": null 87       "vocabulary_id": null
            130     }, 88     },
            131     { 89     {
            132       "display_name": "arXiv", 90       "display_name": "arXiv",
            133       "id": "c0082e38-3626-4894-835f-759512446c29", 91       "id": "c0082e38-3626-4894-835f-759512446c29",
            134       "name": "arXiv", 92       "name": "arXiv",
            135       "state": "active", 93       "state": "active",
            136       "vocabulary_id": null 94       "vocabulary_id": null
            137     } 95     }
            138   ], 96   ],
            t 139   "title": "arXiv Dataset", t 97   "title": "arXiv dataset",
            140   "type": "dataset", 98   "type": "dataset",
            141   "version": "" 99   "version": ""
            142 } 100 }