genevera ishumilin commited on
Commit
595c8f2
·
0 Parent(s):

Duplicate from ishumilin/epstein-files-ocr-complete

Browse files

Co-authored-by: ishumilin <ishumilin@users.noreply.huggingface.co>

Files changed (5) hide show
  1. .gitattributes +1 -0
  2. .gitignore +1 -0
  3. LICENSE +127 -0
  4. README.md +159 -0
  5. data/dataset.parquet +3 -0
.gitattributes ADDED
@@ -0,0 +1 @@
 
 
1
+ *.parquet filter=lfs diff=lfs merge=lfs -text
.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ .DS_Store
LICENSE ADDED
@@ -0,0 +1,127 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Creative Commons CC0 1.0 Universal
2
+
3
+ CREATIVE COMMONS CORPORATION IS NOT A LAW FIRM AND DOES NOT PROVIDE LEGAL
4
+ SERVICES. DISTRIBUTION OF THIS DOCUMENT DOES NOT CREATE AN
5
+ ATTORNEY-CLIENT RELATIONSHIP. CREATIVE COMMONS PROVIDES THIS
6
+ INFORMATION ON AN "AS-IS" BASIS. CREATIVE COMMONS MAKES NO WARRANTIES
7
+ REGARDING THE USE OF THIS DOCUMENT OR THE INFORMATION OR WORKS
8
+ PROVIDED HEREUNDER, AND DISCLAIMS LIABILITY FOR DAMAGES RESULTING FROM
9
+ THE USE OF THIS DOCUMENT OR THE INFORMATION OR WORKS PROVIDED
10
+ HEREUNDER.
11
+
12
+ Statement of Purpose
13
+
14
+ The laws of most jurisdictions throughout the world automatically confer
15
+ exclusive Copyright and Related Rights (defined below) upon the creator
16
+ and subsequent owner(s) (each and all, an "owner") of an original work of
17
+ authorship and/or a database (each, a "Work").
18
+
19
+ Certain owners wish to permanently relinquish those rights to a Work for
20
+ the purpose of contributing to a commons of creative, cultural and
21
+ scientific works ("Commons") that the public can reliably and without fear
22
+ of later claims of infringement build upon, modify, incorporate in other
23
+ works, reuse and redistribute as freely as possible in any form whatsoever
24
+ and for any purposes, including without limitation commercial purposes.
25
+ These owners may contribute to the Commons to promote the ideal of a free
26
+ culture and the further production of creative, cultural and scientific
27
+ works, or to gain reputation or greater distribution for their Work in
28
+ part through the use and efforts of others.
29
+
30
+ For these and/or other purposes and motivations, and without any
31
+ expectation of additional consideration or compensation, the person
32
+ associating CC0 with a Work (the "Affirmer"), to the extent that he or she
33
+ is an owner of Copyright and Related Rights in the Work, voluntarily
34
+ elects to apply CC0 to the Work and publicly distribute the Work under its
35
+ terms, with knowledge of his or her Copyright and Related Rights in the
36
+ Work and the meaning and intended legal effect of CC0 on those rights.
37
+
38
+ 1. Copyright and Related Rights.
39
+
40
+ A Work made available under CC0 may be protected by copyright and related
41
+ or neighboring rights ("Copyright and Related Rights"). Copyright and
42
+ Related Rights include, but are not limited to, the following:
43
+
44
+ i. the right to reproduce, adapt, distribute, perform, display,
45
+ communicate, and translate a Work;
46
+ ii. moral rights retained by the original author(s) and/or performer(s);
47
+ iii. publicity and privacy rights pertaining to a person's image or
48
+ likeness depicted in a Work;
49
+ iv. rights protecting against unfair competition in regards to a Work,
50
+ subject to the limitations in paragraph 4(a), below;
51
+ v. rights protecting the extraction, dissemination, use and reuse of
52
+ data in a Work;
53
+ vi. database rights (such as those arising under Directive 96/9/EC of the
54
+ European Parliament and of the Council of 11 March 1996 on the legal
55
+ protection of databases, and under any national implementation
56
+ thereof, including any amended or successor version of such
57
+ directive); and
58
+ vii. other similar, equivalent or corresponding rights throughout the
59
+ world based on applicable law or treaty, and any national
60
+ implementations thereof.
61
+
62
+ 2. Waiver.
63
+
64
+ To the greatest extent permitted by, but not in contravention of,
65
+ applicable law, Affirmer hereby overtly, fully, permanently, irrevocably
66
+ and unconditionally waives, abandons, and surrenders all of Affirmer's
67
+ Copyright and Related Rights and associated claims and causes of action,
68
+ whether now known or unknown (including existing as well as future
69
+ claims and causes of action), in the Work (i) in all territories
70
+ worldwide, (ii) for the maximum duration provided by applicable law or
71
+ treaty (including future time extensions), (iii) in any current or future
72
+ medium and for any number of copies, and (iv) for any purpose whatsoever,
73
+ including without limitation commercial, advertising or promotional
74
+ purposes (the "Waiver"). Affirmer makes the Waiver for the benefit of each
75
+ member of the public at large and to the detriment of Affirmer's heirs and
76
+ successors, fully intending that such Waiver shall not be subject to
77
+ revocation, rescission, cancellation, termination, or any other legal or
78
+ equitable action to disrupt the quiet enjoyment of the Work by the public
79
+ as contemplated by Affirmer's express Statement of Purpose.
80
+
81
+ 3. Public License Fallback.
82
+
83
+ Should any part of the Waiver for any reason be judged legally invalid or
84
+ ineffective under applicable law, then the Waiver shall be preserved to
85
+ the maximum extent permitted taking into account Affirmer's express
86
+ Statement of Purpose. In addition, to the extent the Waiver is so judged
87
+ Affirmer hereby grants to each affected person a royalty-free,
88
+ non transferable, non sublicensable, non exclusive, irrevocable and
89
+ unconditional license to exercise Affirmer's Copyright and Related Rights
90
+ in the Work (i) in all territories worldwide, (ii) for the maximum
91
+ duration provided by applicable law or treaty (including future time
92
+ extensions), (iii) in any current or future medium and for any number of
93
+ copies, and (iv) for any purpose whatsoever, including without limitation
94
+ commercial, advertising or promotional purposes (the "License"). The
95
+ License shall be deemed effective as of the date CC0 was applied by
96
+ Affirmer to the Work. Should any part of the License for any reason be
97
+ judged legally invalid or ineffective under applicable law, such partial
98
+ invalidity or ineffectiveness shall not invalidate the remainder of the
99
+ License, and in such case Affirmer hereby affirms that he or she will not
100
+ (i) exercise any of his or her remaining Copyright and Related Rights in
101
+ the Work or (ii) assert any associated claims and causes of action with
102
+ respect to the Work, in either case contrary to Affirmer's express
103
+ Statement of Purpose.
104
+
105
+ 4. Limitations and Disclaimers.
106
+
107
+ a. No trademark or patent rights held by Affirmer are waived, abandoned,
108
+ surrendered, licensed or otherwise affected by this document.
109
+ b. Affirmer offers the Work as-is and makes no representations or
110
+ warranties of any kind concerning the Work, express, implied,
111
+ statutory or otherwise, including without limitation warranties of
112
+ title, merchantability, fitness for a particular purpose, non
113
+ infringement, or the absence of latent or other defects, accuracy, or
114
+ the present or absence of errors, whether or not discoverable, all to
115
+ the greatest extent permissible under applicable law.
116
+ c. Affirmer disclaims responsibility for clearing rights of other persons
117
+ that may apply to the Work or any use thereof, including without
118
+ limitation any person's Copyright and Related Rights in the Work.
119
+ Further, Affirmer disclaims responsibility for obtaining any necessary
120
+ consents, permissions or other rights required for any use of the
121
+ Work.
122
+ d. Affirmer understands and acknowledges that Creative Commons is not a
123
+ party to this document and has no duty or obligation with respect to
124
+ this CC0 or use of the Work.
125
+
126
+ For more information, please see:
127
+ https://creativecommons.org/publicdomain/zero/1.0/
README.md ADDED
@@ -0,0 +1,159 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc0-1.0
3
+ task_categories:
4
+ - question-answering
5
+ - text-classification
6
+ - text-retrieval
7
+ language:
8
+ - en
9
+ tags:
10
+ - epstein
11
+ - jeffrey-epstein
12
+ - epstein-files
13
+ - epstein-case
14
+ - court-documents
15
+ - depositions
16
+ - unsealed-documents
17
+ - fbi-files
18
+ - legal
19
+ - flight-logs
20
+ - private-jet
21
+ - passenger-list
22
+ - island-visits
23
+ - us-law
24
+ - news
25
+ - politics
26
+ - corruption
27
+ - elite-networks
28
+ - power-networks
29
+ - social-graph
30
+ - network-analysis
31
+ - named-entities
32
+ - entity-linking
33
+ - relationship-extraction
34
+ - relation-extraction
35
+ - summarization
36
+ - investigative-journalism
37
+ - open-source-intelligence
38
+ - osint
39
+ - ocr
40
+ size_categories:
41
+ - 1M<n<10M
42
+ ---
43
+
44
+ # Epstein Files — Complete OCR Dataset
45
+
46
+ >
47
+ > This is a comprehensive, structured publication of the Epstein Files OCR dataset, significantly expanding upon the earlier [Datasets 1-8 release](https://huggingface.co/datasets/ishumilin/epstein-files-ocr-datasets-1-8-early-release).
48
+ >
49
+
50
+ ## Dataset Summary
51
+
52
+ This dataset contains **page-level OCR output** compiled from an extensive release of documents related to **Jeffrey Epstein / the Epstein case**.
53
+
54
+ Each row in this dataset represents **one scanned PDF document** from the original release using a proprietary automated OCR pipeline provided by [Wild Ma-Gässli](https://wildma.ch).
55
+
56
+ The dataset is designed for:
57
+ * Question answering
58
+ * Information retrieval
59
+ * Downstream NLP tasks such as named entity recognition (NER), entity linking, and relationship extraction.
60
+
61
+ ### Enhancements from Previous Versions
62
+
63
+ - **Scale:** This structured release covers **1,380,935 PDF documents**, comprising over **2,700,000 total pages**.
64
+ - **Format:** Restructured from individual `.md` files into a more efficient **Parquet** format.
65
+ - **Document Linking:** Each page retains its original `document_id` (e.g., `EFTA00000001`), resolving the limitation from earlier releases where pages could not be easily traced back to their source PDFs.
66
+
67
+ ## Supported Tasks
68
+
69
+ * Text retrieval / search (BM25, hybrid, dense retrieval)
70
+ * Question answering over retrieved context (RAG)
71
+ * Entity extraction (names, places, phone numbers, dates) from noisy OCR
72
+ * Social graph and network analysis
73
+
74
+ ## Languages
75
+
76
+ Primarily English (`en`).
77
+
78
+ ## Related Tools
79
+
80
+ This dataset is designed to be used with the **Epstein Chat** analysis tool, which provides a RAG (Retrieval-Augmented Generation) interface for querying these documents.
81
+
82
+ * **GitHub Repository**: [ishumilin/epstein-chat](https://github.com/ishumilin/epstein-chat)
83
+
84
+ ## Dataset Structure
85
+
86
+ The dataset is provided as a Parquet file, which works natively with Hugging Face's `datasets` library.
87
+
88
+ ### Data Fields
89
+
90
+ The schema contains the following fields:
91
+
92
+ - `document_id` (`string`): The identifier of the original document/page (e.g., `EFTA00146767`).
93
+ - `content` (`string`): The full OCR-extracted content for that specific document.
94
+
95
+ **Example Row:**
96
+ ```json
97
+ {
98
+ "document_id": "EFTA00146767",
99
+ "content": "Hey beautiful. Tried to call you back..."
100
+ }
101
+ ```
102
+
103
+ ### Splits
104
+
105
+ No predefined train/validation/test splits.
106
+
107
+ ## Dataset Creation
108
+
109
+ ### Source Data
110
+
111
+ * **Primary source**: The upstream Epstein Files release hosted at:
112
+ * Torrent: https://github.com/yung-megafone/Epstein-Files/blob/main/Torrent%20Files/epstein-files-structured-full-20250204.tar.zst.torrent
113
+
114
+ **Coverage in this dataset:** All PDF files from the upstream release.
115
+
116
+ ### OCR / Preprocessing
117
+
118
+ OCR was performed on this dataset using a **proprietary model** provided by [Wild Ma-Gässli](https://wildma.ch).
119
+
120
+ ## Considerations for Using the Data
121
+
122
+ ### Personal / Sensitive Information
123
+
124
+ These documents contain **personal data** (names, phone numbers, addresses, emails) and/or information about alleged criminal activity.
125
+
126
+ **Redaction Policy:**
127
+ * This dataset is published as **verbatim OCR output** derived from the public source files.
128
+ * **No additional redaction** (masking/removal) has been applied beyond what was already redacted by the DOJ or the original releasing entity.
129
+
130
+ **Use Responsibly:**
131
+ * Comply with applicable laws and platform policies.
132
+ * Avoid doxxing or harassment.
133
+ * Do not treat OCR text as ground truth; always verify against the original page images/PDFs for high-stakes use.
134
+
135
+ ### Known Limitations
136
+
137
+ * **OCR noise**: While improved, automated extraction can produce recognition errors, incorrect formatting artifacts, or miss obscure characters (especially on poor-quality scans or handwriting). Some pages contain explicit placeholders such as `[hidden text]` reflecting original redactions made by DOJ.
138
+ * **Content variance**: Documents range from dense narrative text to unformatted tables and metadata tags.
139
+ * **Corrupted Source Files**: Three files from the original release were severely corrupted and their contents remain unknown and unextracted:
140
+ * `EFTA00645624.pdf`
141
+ * `EFTA01175426.pdf`
142
+ * `EFTA01220934.pdf`
143
+
144
+ ### Biases
145
+
146
+ This dataset reflects:
147
+
148
+ * The selection, redaction, and presentation choices of the original releasing institution.
149
+ * OCR model performance characteristics (better on clean text, worse on handwriting / low-quality scans).
150
+
151
+ ## Licensing
152
+
153
+ See [`LICENSE`](./LICENSE) for the full CC0 1.0 legal text.
154
+
155
+ ## Citation
156
+
157
+ If you use this dataset, please cite:
158
+ 1. The original [public release](https://www.justice.gov/epstein/doj-disclosures).
159
+ 2. This [dataset](https://huggingface.co/datasets/ishumilin/epstein-files-ocr-complete).
data/dataset.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49022b8dfa77f6ee4d2df318e3190d2434562af444c69187c6bd71cb9f53ad02
3
+ size 1414456548