first commit

2026-06-11 03:33:14 +08:00
commit 5f555bf342
599 changed files with 142347 additions and 0 deletions
--- a/文档润色流和知识库构建流/claude-scholar/skills/ml-paper-writing/references/literature-research/arxiv-search-guide.md
+++ b/文档润色流和知识库构建流/claude-scholar/skills/ml-paper-writing/references/literature-research/arxiv-search-guide.md
@@ -0,0 +1,199 @@
+# arXiv Literature Search Guide
+
+## Overview
+
+This guide provides workflows for discovering and evaluating recent ML research papers on arXiv. Use this when conducting literature reviews, finding related work, or staying updated on recent publications.
+
+---
+
+## Search Strategies
+
+### 1. Keyword-Based Search
+
+**arXiv Search URL Pattern:**
+```
+https://arxiv.org/search/?searchtype=all&query=KEYWORDS&abstracts=show&order=-announced_date_first
+```
+
+**Common ML Search Keywords:**
+- **General ML**: `machine learning`, `deep learning`, `neural networks`
+- **Specific Areas**: `reinforcement learning`, `transformer`, `attention mechanism`, `graph neural networks`
+- **Applications**: `computer vision`, `natural language processing`, `reinforcement learning`
+- **Methods**: `self-supervised learning`, `contrastive learning`, `foundation models`
+
+**Tips:**
+- Combine keywords with `+` for AND operation
+- Use `|` for OR operation
+- Put phrases in quotes for exact matches: `"attention is all you need"`
+
+### 2. Category-Based Search
+
+**Recommended arXiv Categories for ML:**
+- `cs.LG` (Machine Learning)
+- `cs.AI` (Artificial Intelligence)
+- `cs.CV` (Computer Vision and Pattern Recognition)
+- `cs.CL` (Computation and Language)
+- `cs.NE` (Neural and Evolutionary Computing)
+- `stat.ML` (Machine Learning - Statistics)
+
+**Category Filter URL:**
+```
+https://arxiv.org/search/?cat:cs.LG+OR+cat:cs.AI+AND+all:transformer&abstracts=show&order=-announced_date_first
+```
+
+### 3. Time-Based Filtering
+
+**Recent Papers (Last 3 Months):**
+- Use `order=-announced_date_first` for newest first
+- Manually filter by submission date
+- Check paper metadata for submission date
+
+---
+
+## Using Chrome MCP for arXiv Search
+
+When available, use Chrome MCP tools for automated arXiv searching:
+
+1. **Navigate to arXiv search** using Chrome MCP navigation
+2. **Extract paper information** from search results:
+   - Paper title
+   - Authors
+   - arXiv ID
+   - Abstract preview
+   - Publication date
+
+3. **Navigate to individual papers** for detailed review
+
+---
+
+## Paper Quality Evaluation
+
+Evaluate papers using the 5-dimension criteria below:
+
+| Dimension | Weight | Key Points |
+|-----------|--------|------------|
+| **Innovation** | 30% | Novelty of contribution |
+| **Method Completeness** | 25% | Clarity and reproducibility |
+| **Experimental Thoroughness** | 25% | Validation depth |
+| **Writing Quality** | 10% | Clarity of expression |
+| **Relevance & Impact** | 10% | Domain importance |
+
+### Scoring Guidelines (1-5 scale)
+
+**Innovation (30%):**
+- 5: Breakthrough contribution, major impact
+- 4: Significant improvement, new insights
+- 3: Methodological innovation
+- 2: Incremental improvement
+- 1: Minor improvements
+
+**Method Completeness (25%):**
+- 5: Complete and rigorous, easily reproducible
+- 4: Very detailed, mostly reproducible
+- 3: Core method clear, basically reproducible
+- 2: Lacks key details
+- 1: Unclear description
+
+**Experimental Thoroughness (25%):**
+- 5: Comprehensive multi-dataset, ablation studies
+- 4: Multiple datasets, reasonable ablations
+- 3: Main experiments complete
+- 2: Limited experiments
+- 1: Minimal validation
+
+**Writing Quality (10%):**
+- 5: Excellent clarity and rigor
+- 4: Clear and well-structured
+- 3: Understandable
+- 2: Some ambiguity
+- 1: Confusing
+
+**Relevance & Impact (10%):**
+- 5: Solves important problem, wide impact
+- 4: Important domain problem
+- 3: Meaningful contribution
+- 2: Niche problem
+- 1: Limited impact
+
+### Selection Process
+
+1. **Screen by title/abstract** for relevance
+2. **Navigate to full paper** for detailed review
+3. **Score each dimension** (1-5)
+4. **Calculate weighted total**
+5. **Rank and select** top papers
+
+---
+
+## Extracting Paper Metadata
+
+**From arXiv Abstract Page (`https://arxiv.org/abs/ARXIV_ID`):**
+
+- Title (from `<h1>` tag)
+- Authors (from `.authors` element)
+- Abstract (from `blockquote.abstract`)
+- Submission date (from `.dateline`)
+- arXiv ID (from URL)
+- Categories (from `.subjects`)
+- Comments (if present)
+- Code repository (check abstract for GitHub links)
+
+---
+
+## Integration with Citation Workflow
+
+After finding relevant papers:
+
+1. **Verify citations** using Semantic Scholar API (see `../citation-workflow.md`)
+2. **Fetch BibTeX** programmatically via DOI
+3. **Store in bibliography** with verification status
+
+---
+
+## Common Use Cases
+
+### Finding Related Work
+
+When writing a paper, use arXiv search to:
+1. Find recent papers on your topic
+2. Identify state-of-the-art methods
+3. Discover competing approaches
+4. Find baseline comparisons
+
+### Staying Updated
+
+Set up regular searches for:
+- Your specific research area
+- Competing labs/researchers
+- New methods in your domain
+- Conference proceedings (preprints)
+
+### Literature Reviews
+
+For comprehensive reviews:
+1. Start with broad keyword searches
+2. Filter by recent publications (last 1-3 years)
+3. Use citation chaining (forward and backward)
+4. Evaluate and select high-quality papers
+5. Organize by theme/contribution
+
+---
+
+## Tips for Effective Searching
+
+1. **Use specific keywords** rather than broad terms
+2. **Combine techniques** (keywords + categories + time filters)
+3. **Check code availability** (many arXiv papers link to GitHub)
+4. **Look for citations** to understand impact
+5. **Read abstracts carefully** before full papers
+6. **Use paper metrics** (citation count, code stars) as indicators
+
+---
+
+## External Resources
+
+- **arXiv**: https://arxiv.org/
+- **Semantic Scholar**: https://www.semanticscholar.org/
+- **Papers With Code**: https://paperswithcode.com/
+- **Connected Papers**: https://www.connectedpapers.com/
+- **arXiv API**: http://export.arxiv.org/api_help/
--- a/文档润色流和知识库构建流/claude-scholar/skills/ml-paper-writing/references/literature-research/paper-quality-criteria.md
+++ b/文档润色流和知识库构建流/claude-scholar/skills/ml-paper-writing/references/literature-research/paper-quality-criteria.md
@@ -0,0 +1,303 @@
+# ML Paper Quality Evaluation Criteria
+
+## Overview
+
+Use these criteria to evaluate ML research papers found during literature search or when selecting papers for detailed review. The 5-dimension framework provides structured assessment for paper selection and comparison.
+
+---
+
+## Evaluation Dimensions
+
+| Dimension | Weight | Description |
+|-----------|--------|-------------|
+| **Innovation** | 30% | Novelty and originality of contribution |
+| **Method Completeness** | 25% | Clarity, rigor, and reproducibility |
+| **Experimental Thoroughness** | 25% | Validation depth and analysis quality |
+| **Writing Quality** | 10% | Clarity and presentation |
+| **Relevance & Impact** | 10% | Domain importance and potential impact |
+
+---
+
+## Detailed Scoring Rubrics
+
+### 1. Innovation (30%)
+
+**Score 5 - Breakthrough:**
+- Proposes entirely new paradigm or framework
+- Solves long-standing open problem
+- Major impact expected on the field
+- Challenges fundamental assumptions
+
+**Score 4 - Significant Innovation:**
+- Substantial improvement over existing methods
+- New insights or perspectives
+- Novel combination of techniques
+- Clear advancement over state-of-the-art
+
+**Score 3 - Methodological Innovation:**
+- New method or architecture proposed
+- Some novelty but incremental
+- Reasonable contribution
+- Standard type of innovation
+
+**Score 2 - Incremental Improvement:**
+- Minor improvements to existing methods
+- Limited novelty
+- Small advancement
+- Mostly derivative
+
+**Score 1 - Trivial:**
+- Minimal contribution
+- Obvious extension
+- No real innovation
+- Known results
+
+**Evaluation Questions:**
+- Does this paper propose something genuinely new?
+- Does it advance the state-of-the-art?
+- Will this influence future work?
+- Is the contribution significant or marginal?
+
+---
+
+### 2. Method Completeness (25%)
+
+**Score 5 - Complete and Rigorous:**
+- Full mathematical derivation
+- All hyperparameters specified
+- Complete algorithmic details
+- Easily reproducible
+- Code available
+
+**Score 4 - Very Complete:**
+- Detailed method description
+- Most important details included
+- Mostly reproducible
+- Minor gaps in documentation
+
+**Score 3 - Reproducible:**
+- Core method clearly described
+- Key details present
+- Can be reproduced with effort
+- Some ambiguity in details
+
+**Score 2 - Lacks Details:**
+- Key details missing
+- Difficult to reproduce
+- Incomplete description
+- Ambiguous in important areas
+
+**Score 1 - Unclear:**
+- Method description unclear
+- Missing critical information
+- Cannot determine validity
+- Poorly explained
+
+**Evaluation Questions:**
+- Can another researcher reproduce this work?
+- Are all important details specified?
+- Is mathematical derivation sound?
+- Is code available and documented?
+
+---
+
+### 3. Experimental Thoroughness (25%)
+
+**Score 5 - Comprehensive:**
+- Multiple diverse datasets
+- Extensive ablation studies
+- Statistical significance testing
+- Thorough analysis and discussion
+- Comparison with strong baselines
+
+**Score 4 - Very Thorough:**
+- Multiple datasets
+- Reasonable ablation studies
+- Proper baseline comparisons
+- Good analysis
+
+**Score 3 - Adequate:**
+- Main experiments complete
+- Standard datasets
+- Basic baselines
+- Results are credible
+
+**Score 2 - Limited:**
+- Limited experiments
+- Few datasets
+- Weak baselines
+- Minimal analysis
+
+**Score 1 - Insufficient:**
+- Minimal validation
+- Toy examples only
+- No meaningful comparisons
+- Results not convincing
+
+**Evaluation Questions:**
+- Are experiments comprehensive?
+- Are baselines strong and appropriate?
+- Are statistical tests used?
+- Is there ablation analysis?
+- Are results on standard datasets?
+
+---
+
+### 4. Writing Quality (10%)
+
+**Score 5 - Excellent:**
+- Clear, precise, well-structured
+- Logical flow throughout
+- Professional presentation
+- High-quality figures
+- No ambiguity
+
+**Score 4 - Very Good:**
+- Clear and well-written
+- Mostly logical structure
+- Good presentation
+- Minor issues
+
+**Score 3 - Understandable:**
+- Basically clear
+- Some organizational issues
+- Acceptable presentation
+- Understandable with effort
+
+**Score 2 - Fair:**
+- Some confusing sections
+- Organization problems
+- Presentation issues
+- Hard to follow at times
+
+**Score 1 - Poor:**
+- Unclear or confusing
+- Poor organization
+- Difficult to understand
+- Major presentation problems
+
+**Evaluation Questions:**
+- Is the paper easy to understand?
+- Is the structure logical?
+- Are figures/tables clear?
+- Is the writing professional?
+
+---
+
+### 5. Relevance & Impact (10%)
+
+**Score 5 - High Impact:**
+- Solves important problem
+- Broad applicability
+- Expected wide influence
+- Addresses fundamental challenge
+
+**Score 4 - Domain Important:**
+- Important problem in field
+- Significant potential impact
+- Relevant to many researchers
+
+**Score 3 - Meaningful:**
+- Meaningful contribution
+- Moderate impact expected
+- Relevant to subset of field
+
+**Score 2 - Niche:**
+- Specialized problem
+- Limited applicability
+- Narrow impact
+
+**Score 1 - Limited:**
+- Very narrow problem
+- Minimal impact expected
+- Limited relevance
+
+**Evaluation Questions:**
+- Is this an important problem?
+- Will this influence future work?
+- Is it relevant to current research needs?
+- Does it address a significant challenge?
+
+---
+
+## Scoring Calculation
+
+**Weighted Total:**
+```
+Total = (Innovation × 0.30) + (Method × 0.25) + (Experiments × 0.25) + (Writing × 0.10) + (Impact × 0.10)
+```
+
+**Example Calculation:**
+- Innovation: 4/5
+- Method: 3/5
+- Experiments: 4/5
+- Writing: 3/5
+- Impact: 4/5
+
+```
+Total = (4 × 0.30) + (3 × 0.25) + (4 × 0.25) + (3 × 0.10) + (4 × 0.10)
+      = 1.20 + 0.75 + 1.00 + 0.30 + 0.40
+      = 3.65 / 5.0
+```
+
+---
+
+## Selection Process
+
+### For Literature Reviews
+
+1. **Screen papers** by title/abstract for relevance
+2. **Full review** of potentially relevant papers
+3. **Score each paper** using all 5 dimensions
+4. **Rank by total score**
+5. **Select top papers** for detailed review
+
+### Quality Thresholds
+
+- **Excellent**: 4.0+ (include definitely)
+- **Good**: 3.5-3.9 (include if relevant)
+- **Fair**: 3.0-3.4 (include if highly relevant)
+- **Poor**: <3.0 (exclude unless essential)
+
+---
+
+## Quick Screening Indicators
+
+Before detailed review, check:
+
+**Positive Indicators:**
+- Published at top venue (NeurIPS, ICML, ICLR)
+- Citations in top papers
+- Code available with stars
+- Authors from top labs
+- Clear novelty in abstract
+
+**Negative Indicators:**
+- Vague abstract
+- Limited experiments mentioned
+- No baselines mentioned
+- Poor writing in abstract
+- incremental claims only
+
+---
+
+## Integration with Paper Discovery
+
+When using arXiv search (`arxiv-search-guide.md`):
+
+1. **Search** for relevant papers
+2. **Extract metadata** from arXiv pages
+3. **Quick screen** by abstract/relevance
+4. **Detailed review** of promising papers
+5. **Score using** these criteria
+6. **Rank and select** top candidates
+
+---
+
+## Notes
+
+- These criteria are designed for ML papers specifically
+- Adjust weights based on your specific needs
+- Use scores as relative comparisons, not absolute judgments
+- Consider venue reputation as additional signal
+- Code availability is increasingly important for reproducibility