Update README.md

README.md

@@ -58,32 +58,6 @@ Paper appears 3 times:
 - Day 3: Rank 10 → 41 points
 - **Total Score: 137**
 
-## 📋 Dataset Structure
-
-### Columns
-
-| Column | Type | Description |
-|--------|------|-------------|
-| `rank` | integer | Overall rank (1 = highest score) |
-| `name` | string | Paper title |
-| `times_trended` | integer | Number of days appeared on trending |
-| `best_rank` | integer | Highest rank achieved (lowest number) |
-| `avg_rank` | float | Average rank across all appearances |
-| `median_rank` | integer | Median rank |
-| `publish_date` | string | Paper publication date |
-| `max_upvotes` | integer | Maximum upvotes received |
-| `max_github_stars` | integer | Maximum GitHub stars for associated repo |
-| `arxiv_link` | string | arXiv link to paper |
-
-### Sample Data
-
-```csv
-rank,name,times_trended,best_rank,avg_rank,median_rank,publish_date,max_upvotes,max_github_stars,arxiv_link
-1,LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models,432,2,11.28,11,"Mar 20, 2024",173,63300,https://arxiv.org/abs/2403.13372
-2,PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Model,124,1,3.35,3,"Oct 16, 2025",98,65400,https://arxiv.org/abs/2510.14528
-3,MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing,179,2,10.82,9,"Sep 26, 2025",134,49600,https://arxiv.org/abs/2509.22186
-```
-
 ## 🌟 Key Insights
 
 ### 1. Top 10 Most Trending Papers (2025)
@@ -159,141 +133,3 @@ rank,name,times_trended,best_rank,avg_rank,median_rank,publish_date,max_upvotes,
 - PaddleOCR 3.0: 57,600 stars
 - OmniFlatten: 50,300 stars
 - MinerU2.5: 49,600 stars
-
-### 5. Publication Timeline Insights
-
-**Recent Papers (2025) Dominating:**
-- 80%+ of top papers published in 2025
-- Fast adoption cycle: trending within days of publication
-- Community actively tracking latest research
-
-**Older Papers with Staying Power:**
-- LlamaFactory (Mar 2024) still trending heavily in 2025
-- DINOv3 maintains consistent visibility
-- High-impact tools have long trending tails
-
-## 🔍 Use Cases
-
-This dataset is valuable for:
-
-1. **Research Trend Analysis**: Understanding which ML/AI topics gained traction in 2025
-2. **Literature Review**: Identifying influential papers by community engagement
-3. **Topic Discovery**: Finding emerging research directions (agents, document AI)
-4. **Citation Strategy**: Papers with high trending scores often become highly cited
-5. **Tool Discovery**: Many top papers include open-source implementations
-6. **Academic Benchmarking**: Understanding what resonates with ML practitioners
-7. **Technology Forecasting**: Early indicators of which research will impact products
-8. **Educational Content**: Curating reading lists based on community interest
-
-## 📈 Data Quality
-
-- ✅ **Complete**: All 441 Wayback Machine snapshots processed
-- ✅ **Consistent**: Single scoring methodology throughout
-- ✅ **Rich Metadata**: Includes upvotes, GitHub stars, arXiv links
-- ✅ **Validated**: Cross-referenced with HuggingFace Daily Papers
-- ℹ️ **Coverage**: 2025 only (dataset is year-specific)
-- ℹ️ **GitHub Stars**: Captured at time of trending, may be higher now
-
-## 🚀 Quick Start
-
-### Load with Hugging Face Datasets
-
-```python
-from datasets import load_dataset
-
-# Load the dataset
-dataset = load_dataset("ronantakizawa/huggingface-top-papers")
-
-# Get top 10 papers
-df = dataset['train'].to_pandas()
-top_10 = df.head(10)
-print(top_10[['rank', 'name', 'times_trended', 'max_github_stars']])
-```
-
-### Load with Pandas
-
-```python
-import pandas as pd
-
-url = "https://huggingface.co/datasets/ronantakizawa/huggingface-top-papers/resolve/main/huggingface-top-papers.csv"
-df = pd.read_csv(url)
-
-# Find papers with most GitHub stars
-top_starred = df.nlargest(10, 'max_github_stars')
-print(top_starred[['name', 'max_github_stars', 'times_trended']])
-```
-
-### Example Analyses
-
-```python
-# Find agent-related papers
-agent_papers = df[df['name'].str.contains('Agent', case=False)]
-print(f"Agent papers: {len(agent_papers)}")
-
-# Papers that appeared most frequently
-most_consistent = df.nlargest(10, 'times_trended')
-
-# Papers with best average ranking
-highest_quality = df.nsmallest(10, 'avg_rank')
-
-# Community favorites (upvotes + stars)
-df['engagement'] = df['max_upvotes'] + (df['max_github_stars'] / 100)
-most_engaged = df.nlargest(10, 'engagement')
-```
-
-## 🔗 Related Datasets
-
-- **GitHub Top Developers** (2015-2025) - by same author
-- **TikTok Trending Hashtags** (2022-2025) - by same author
-- **Twitter Trending Hashtags** (2020-2025) - by same author
-- **HuggingFace Papers** (raw data - 441 snapshots)
-
-## 📝 Citation
-
-If you use this dataset in your research or project, please cite:
-
-```bibtex
-@dataset{huggingface_top_papers_2025,
-  title={HuggingFace Top Trending Papers (2025)},
-  author={Ronan Takizawa},
-  year={2025},
-  publisher={Hugging Face},
-  howpublished={\url{https://huggingface.co/datasets/ronantakizawa/huggingface-top-papers}},
-  note={Derived from Wayback Machine snapshots of HuggingFace Daily Papers, 441 snapshots}
-}
-```
-
-## 🤝 Contributing
-
-Found an issue? Suggestions for improvement?
-- Open an issue on the [GitHub repository](https://github.com/ronantakizawa)
-- Submit feedback through Hugging Face discussions
-
-## ⚠️ Disclaimer
-
-This dataset is derived from public Wayback Machine snapshots of HuggingFace Daily Papers. It represents:
-
-- **Community interest and visibility**, not absolute research quality
-- **Trending patterns**, which may favor practical tools over pure theory
-- **English-language papers** primarily (HuggingFace audience bias)
-- **2025 snapshot** of the ML/AI research landscape
-
-Trending algorithms and their criteria are not publicly documented. This dataset captures historical visibility on HuggingFace's curated daily papers feed.
-
-## 📄 License
-
-MIT License - Free to use with attribution
-
----
-
-**Dataset Details:**
-- **Created**: December 1, 2025
-- **Coverage**: January 2025 → November 2025
-- **Last Updated**: December 1, 2025
-- **Version**: 1.0
-- **Maintainer**: Ronan Takizawa
-- **Contact**: [Hugging Face Profile](https://huggingface.co/ronantakizawa)
-
----
-
-*Part of a series of trending data datasets capturing temporal patterns across different platforms (GitHub, TikTok, Twitter, HuggingFace Papers, Yahoo Finance).*