Upload README.md with huggingface_hub
Browse files
README.md
ADDED
@@ -0,0 +1,145 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
task_categories:
|
4 |
+
- feature-extraction
|
5 |
+
- question-answering
|
6 |
+
language:
|
7 |
+
- en
|
8 |
+
tags:
|
9 |
+
- code
|
10 |
+
pretty_name: LOTUS Scraped Data (2025/06/07)
|
11 |
+
size_categories:
|
12 |
+
- 1K<n<10K
|
13 |
+
---
|
14 |
+
# Lotus Deep Research Dataset
|
15 |
+
|
16 |
+
[](https://huggingface.co/datasets/lotus-data/scraped_related_works_20250607_180022)
|
17 |
+
[](https://github.com/harshitgupta412/lotus-deep-research)
|
18 |
+
[](https://github.com/harshitgupta412/lotus-deep-research/blob/main/LICENSE)
|
19 |
+
|
20 |
+
---
|
21 |
+
|
22 |
+
A comprehensive dataset of academic papers with extracted related works sections and recovered citations, designed for training and evaluating research generation systems.
|
23 |
+
|
24 |
+
## π Dataset Overview
|
25 |
+
|
26 |
+
This dataset contains **67 academic papers** from ArXiv with their related works sections and **1663 recovered citations**, providing a rich resource for research generation and citation analysis tasks.
|
27 |
+
|
28 |
+
### π― Use Cases
|
29 |
+
|
30 |
+
- **Research Generation**: Train models to generate related works sections
|
31 |
+
- **Citation Analysis**: Study citation patterns and relationships
|
32 |
+
- **Academic NLP**: Develop tools for academic text processing
|
33 |
+
- **Evaluation**: Benchmark research generation systems
|
34 |
+
- **Knowledge Discovery**: Analyze research trends and connections
|
35 |
+
|
36 |
+
## π Dataset Structure
|
37 |
+
|
38 |
+
### 1. `papers_with_related_works.csv` (67 papers)
|
39 |
+
|
40 |
+
Contains academic papers with extracted related works sections in multiple formats:
|
41 |
+
|
42 |
+
| Column | Description |
|
43 |
+
|--------|-------------|
|
44 |
+
| `arxiv_id` | ArXiv identifier (e.g., "2506.02838v1") |
|
45 |
+
| `title` | Paper title |
|
46 |
+
| `authors` | Author names |
|
47 |
+
| `abstract` | Paper abstract |
|
48 |
+
| `categories` | ArXiv categories (e.g., "cs.AI, econ.GN") |
|
49 |
+
| `published_date` | Publication date |
|
50 |
+
| `updated_date` | Last update date |
|
51 |
+
| `abs_url` | ArXiv abstract URL |
|
52 |
+
| `arxiv_link` | Full ArXiv link |
|
53 |
+
| `raw_latex_related_works` | Raw LaTeX related works section |
|
54 |
+
| `clean_latex_related_works` | Cleaned LaTeX related works section |
|
55 |
+
| `pdf_related_works` | Related works extracted from PDF |
|
56 |
+
|
57 |
+
### 2. `citations_with_recovered_res.csv` (1663 citations)
|
58 |
+
|
59 |
+
Contains individual citations with recovered metadata:
|
60 |
+
|
61 |
+
| Column | Description |
|
62 |
+
|--------|-------------|
|
63 |
+
| `parent_paper_title` | Title of the paper containing the citation |
|
64 |
+
| `parent_paper_arxiv_id` | ArXiv ID of the parent paper |
|
65 |
+
| `citation_shorthand` | Citation key (e.g., "NBERw21340") |
|
66 |
+
| `raw_citation_text` | Raw citation text from LaTeX |
|
67 |
+
| `cited_paper_title` | Title of the cited paper |
|
68 |
+
| `cited_paper_arxiv_link` | ArXiv link if available |
|
69 |
+
| `cited_paper_abstract` | Abstract of the cited paper |
|
70 |
+
| `has_metadata` | Whether metadata was successfully recovered |
|
71 |
+
| `is_arxiv_paper` | Whether the cited paper is from ArXiv |
|
72 |
+
| `bib_paper_authors` | Authors of the cited paper |
|
73 |
+
| `bib_paper_year` | Publication year |
|
74 |
+
| `bib_paper_month` | Publication month |
|
75 |
+
| `bib_paper_url` | URL of the cited paper |
|
76 |
+
| `bib_paper_doi` | DOI of the cited paper |
|
77 |
+
| `bib_paper_journal` | Journal name |
|
78 |
+
| `search_res_title` | Title from search results |
|
79 |
+
| `search_res_url` | URL from search results |
|
80 |
+
| `search_res_content` | Content snippet from search results |
|
81 |
+
|
82 |
+
## π Quick Start
|
83 |
+
|
84 |
+
### Loading the Dataset
|
85 |
+
|
86 |
+
```python
|
87 |
+
import pandas as pd
|
88 |
+
|
89 |
+
# Load papers dataset
|
90 |
+
papers_df = pd.read_csv('papers_with_related_works.csv')
|
91 |
+
print(f"Loaded {len(papers_df)} papers")
|
92 |
+
|
93 |
+
# Load citations dataset
|
94 |
+
citations_df = pd.read_csv('citations_with_recovered_res.csv')
|
95 |
+
print(f"Loaded {len(citations_df)} citations")
|
96 |
+
```
|
97 |
+
|
98 |
+
|
99 |
+
|
100 |
+
### Example: Extract Related Works for a Paper
|
101 |
+
|
102 |
+
```python
|
103 |
+
# Get a specific paper
|
104 |
+
paper = papers_df[papers_df['arxiv_id'] == '2506.02838v1'].iloc[0]
|
105 |
+
print(f"Title: {paper['title']}")
|
106 |
+
print(f"Related Works:\n{paper['clean_latex_related_works']}")
|
107 |
+
|
108 |
+
# Get all citations for this paper
|
109 |
+
paper_citations = citations_df[citations_df['parent_paper_arxiv_id'] == '2506.02838v1']
|
110 |
+
print(f"Number of citations: {len(paper_citations)}")
|
111 |
+
```
|
112 |
+
|
113 |
+
## π Dataset Statistics
|
114 |
+
|
115 |
+
- **Total Papers**: 67
|
116 |
+
- **Total Citations**: 1663
|
117 |
+
- **Date Range**: 2024-2025 (recent papers)
|
118 |
+
|
119 |
+
## π§ Data Collection Process
|
120 |
+
|
121 |
+
This dataset was created using the [Lotus Deep Research](https://github.com/harshitgupta412/lotus-deep-research) pipeline:
|
122 |
+
|
123 |
+
1. **ArXiv Scraping**: Collected papers by category and date range
|
124 |
+
2. **Author Filtering**: Focused on high-impact researchers (h-index β₯ 25)
|
125 |
+
3. **LaTeX Extraction**: Extracted related works sections from LaTeX source
|
126 |
+
4. **Citation Recovery**: Resolved citations and recovered metadata
|
127 |
+
5. **Quality Filtering**: Ensured data quality and completeness
|
128 |
+
|
129 |
+
## π Related Resources
|
130 |
+
|
131 |
+
- **[GitHub Repository](https://github.com/harshitgupta412/lotus-deep-research)**: Full source code and documentation
|
132 |
+
- **[Data Pipeline](https://github.com/harshitgupta412/lotus-deep-research/tree/main/data_pipeline)**: Tools for collecting similar datasets
|
133 |
+
- **[Evaluation Framework](https://github.com/harshitgupta412/lotus-deep-research/tree/main/eval)**: Framework for evaluating research generation systems
|
134 |
+
|
135 |
+
## π€ Contributing
|
136 |
+
|
137 |
+
We welcome contributions to improve this dataset! Please see the [main repository](https://github.com/harshitgupta412/lotus-deep-research) for contribution guidelines.
|
138 |
+
|
139 |
+
## π License
|
140 |
+
|
141 |
+
This dataset is released under the MIT License. See the [LICENSE](https://github.com/harshitgupta412/lotus-deep-research/blob/main/LICENSE) file for details.
|
142 |
+
|
143 |
+
---
|
144 |
+
|
145 |
+
**Note**: This dataset is actively maintained and updated. Check the GitHub repository for the latest version and additional resources.
|