ZexiK commited on
Commit
e864c25
·
verified ·
1 Parent(s): a1eab76

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +109 -0
README.md ADDED
@@ -0,0 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Nature Communications Articles Dataset - natcomm_2512_001
2
+
3
+ ## Dataset Description
4
+
5
+ This dataset contains scraped Nature Communications articles with comprehensive extraction of:
6
+ - Main figures with captions and citations
7
+ - Tables with full HTML and data
8
+ - Supplementary figures with citation contexts
9
+ - Supplementary tables with citation contexts
10
+ - Supplementary files (PDFs, Excel files)
11
+ - Article sections with hierarchical structure (h2=section, h3=subsection)
12
+ - Figure/table citation contexts linking to sections
13
+
14
+ ## Dataset Structure
15
+
16
+ ### Files
17
+
18
+ - `images_tables_natcomm_2512_001.zip`: Contains all article images and tables
19
+ - `files/[article_id]/images/`: Main figure images
20
+ - `files/[article_id]/tables/`: Table HTML files
21
+
22
+ - `metadata_natcomm_2512_001.zip`: Contains JSON metadata for each article
23
+ - `metadata/[article_id].json`: Complete metadata including:
24
+ - `article_id`: Article identifier
25
+ - `url`: Original article URL
26
+ - `title`: Article title
27
+ - `image_set`: Dict of main figures with captions and citations
28
+ - `table_set`: Dict of tables with captions and data
29
+ - `supplementary_image_set`: Dict of supplementary figures with citations
30
+ - `supplementary_table_set`: Dict of supplementary tables with citations
31
+ - `supplementary_files`: List of supplementary file links
32
+ - `section_set`: Dict of article sections with hierarchical structure
33
+
34
+ ## Statistics
35
+
36
+ - **Total Articles**: 277
37
+ - **Scraping Date**: 2025-12-11
38
+
39
+ ### Content Summary
40
+
41
+ - **Main Figures**: 1182 (4.3 per article)
42
+ - **Tables**: 105 (0.4 per article)
43
+ - **Supplementary Figures**: 1605 (5.8 per article)
44
+ - **Supplementary Tables**: 322 (1.2 per article)
45
+ - **Supplementary Files**: 2211 (8.0 per article)
46
+ - **Sections**: 4876 (17.6 per article)
47
+
48
+ ## Features
49
+
50
+ ### Section Hierarchy
51
+ Each section includes:
52
+ - `level`: 2 (h2 main section) or 3 (h3 subsection)
53
+ - `type`: "section" or "subsection"
54
+ - `title`: Section heading
55
+ - `content`: Full section text
56
+
57
+ ### Excluded Metadata Sections
58
+ Administrative sections are filtered out:
59
+ - Acknowledgements, Author information, Contributions
60
+ - Peer review, Ethics declarations, Rights and permissions
61
+ - Similar content recommendations
62
+
63
+ ### Citation Tracking
64
+ Each figure/table has `contexts` array with:
65
+ - `section`: Section title where cited
66
+ - `text`: ±100 character context around citation
67
+ - `full_content`: Complete section content
68
+
69
+ ## Usage
70
+
71
+ ```python
72
+ import json
73
+ import zipfile
74
+
75
+ # Load article metadata
76
+ with open('metadata_natcomm_2512_001/metadata/s41467-025-65722-y.json', 'r') as f:
77
+ article = json.load(f)
78
+
79
+ # Access sections with hierarchy
80
+ for sec_id, sec_info in article['section_set'].items():
81
+ if sec_info['level'] == 2:
82
+ print(f"Section: {sec_info['title']}")
83
+ else:
84
+ print(f" Subsection: {sec_info['title']}")
85
+
86
+ # Access figures with citations
87
+ for fig_id, fig_info in article['image_set'].items():
88
+ print(f"{fig_id}: {fig_info['caption']}")
89
+ print(f" Cited in: {[c['section'] for c in fig_info['contexts']]}")
90
+ ```
91
+
92
+ ## Version Notes
93
+
94
+ **natcomm_2512_001** (December 2025):
95
+ - Fixed section extraction to exclude recommendation sections
96
+ - Added section hierarchy tracking (h2/h3 levels)
97
+ - Improved metadata section filtering
98
+ - Enhanced supplementary material citation tracking
99
+
100
+ ## License
101
+
102
+ This dataset is derived from Nature Communications articles. Please respect original article licenses and cite appropriately.
103
+
104
+ ## Created By
105
+
106
+ Batch Name: natcomm_2512_001
107
+ Scraping Tool: Nature Communications Comprehensive Scraper
108
+ Date: 2025-12-11
109
+ Repository: https://huggingface.co/datasets/ZexiK/natcomSS