中文 | English

PaperLens

Your AI Research Copilot

Stop searching papers. Start understanding research.

---

What is PaperLens?

PaperLens is an AI research copilot that turns the messy, fragmented process of academic literature work into a single, coherent workflow. It's like Cursor, but for research papers.

Most researchers spend their days bouncing between Google Scholar, arXiv, PDF readers, and half a dozen browser tabs. PaperLens replaces all of that with one tool:

Find → Filter → Read → Understand → Save → Cite

• 43 academic databases searched in parallel, including PubMed, Scopus, ScienceDirect, arXiv, CNKI, and more
• 300+ Chinese universities accessible via CARSI — no VPN needed for ScienceDirect, Scopus, JSTOR from anywhere
• 100% local-first — all data stays on your machine. Runs fully offline with Ollama. No account required.

---

Demo

| Search & Discover | Paper List | AI Deep Analysis | |:---:|:---:|:---:| | | | |

| Multi-Paper Compare | Citation Graph | |:---:|:---:| | | |

---

Why PaperLens?

The researcher's daily workflow (before PaperLens)

Google Scholar → arXiv → open PDF → not what I need → keep searching → open more tabs → 20 tabs later, still no clarity.

With PaperLens

Describe your research question in natural language. PaperLens searches 43 databases at once, ranks results by relevance, and lets you:

• Read a paper in 2 minutes instead of 30 — AI extracts what matters: method, innovation, strengths, weaknesses, future directions
• Find the right paper in seconds, not hours — AI understands your research intent, not just keyword matching
• Compare papers side-by-side — spot contradictions, identify research gaps, find complementary approaches
• Discover connections — interactive citation graph reveals how papers relate, who cites whom, and emerging research clusters

| Pain Point | PaperLens Solution | |---|---| | VPN needed for off-campus access to ScienceDirect / Scopus | CARSI authentication — 300+ universities, zero config | | Don't know how to write Boolean queries | Natural language — type your question in Chinese or English, AI builds the query | | Switching between PubMed, Scopus, Google Scholar, CNKI... | 43 sources, one search — unified, deduplicated results | | Reading 50 papers for a literature review | AI Deep Analysis — 7-dimension breakdown + multi-paper comparison | | Hard to find cross-disciplinary connections | Interactive citation graph (D3.js) — click to expand, drag to explore | | Data privacy concerns with cloud tools | 100% local — works fully offline with Ollama, no data leaves your machine |

---

The Research Workflow

PaperLens isn't just a search tool. It's a complete pipeline:

1. Find

Describe your research interest in natural language. PaperLens AI constructs the optimal query and searches 43 academic databases in parallel — PubMed, OpenAlex, CrossRef, arXiv, Semantic Scholar, ScienceDirect, Scopus, JSTOR, CNKI, Wanfang, VIP, Google Scholar, and more. Results are deduplicated, enriched, and ranked in seconds.

2. Filter

Narrow down results by journal, year, field tags, publication type, or metadata completeness. Every paper gets a metadata score (0-100%) so you know what's well-documented and what's missing data.

3. Read

Select any paper. AI extracts the essence:

Summary  →  Deep Analysis  →  Method  →  Innovation  →  Strength  →  Weakness  →  Future Work  →  Related Papers

This isn't a one-sentence summary. It's a 7-dimension structured breakdown — like having a senior colleague walk you through the paper.

4. Understand

Compare multiple papers side-by-side. AI identifies:

• Where approaches agree and diverge
• Research gaps no one has addressed
• Contradictions between findings
• Promising directions for future work

5. Save & Cite

Export in any format: RIS, BibTeX, CSV, EndNote XML. Generate citations in APA 7th, MLA 9th, GB/T 7714, Chicago 17th, Vancouver. Track reading status (unread / reading / read). Sync tags bidirectionally with Zotero. Download OA PDFs with one click.

---

Core Features

AI Search

• 43 data sources searched in parallel — see the full list in ROADMAP.md
• Natural language — Chinese or English, AI understands your intent
• Batch DOI import — paste a list of DOIs, enrich all at once
• Smart filtering — journal, year, field tags, publication type, metadata completeness
• Progressive rendering — results appear per-source as they arrive, no waiting

AI Deep Analysis

• Summary — one-paragraph overview: what, how, why it matters
• Deep Analysis — 7 dimensions: contribution, motivation, methods, results, limitations, context, future work
• Multi-Paper Compare — systematic comparison across approaches, innovations, and research gaps
• Novelty Detection — identifies under-explored areas, contradictions, and promising directions

CARSI Institutional Access

Connect to 300+ Chinese universities via CARSI. Access ScienceDirect, Scopus, JSTOR, and institutional databases without VPN, from anywhere in the world.

| Source | Status | Auth | |---|---|---| | ScienceDirect | ✅ | CARSI | | Scopus | ✅ | CARSI | | JSTOR | ✅ | CARSI | | CNKI | ✅ | CARSI + captcha |

Interactive Citation Graph

• D3.js force-directed graph — zoom, pan, drag, click to expand
• Keyword co-occurrence network — discover thematic clusters
• Author collaboration network — see co-authorship patterns
• Click any paper node to load its citing and referenced papers on-the-fly

Export & Integration

• Multi-format export — RIS, BibTeX, CSV, EndNote XML
• Citation formatting — APA 7th, MLA 9th, GB/T 7714, Chicago 17th, Vancouver
• Zotero integration — SQLite direct read + MCP plugin + bi-directional tag sync + PDF full-text extraction
• Tag system — custom tags, color picker, Zotero sync
• Reading status — unread / reading / read tracking
• Batch operations — multi-select, one-click export
• Smart export — auto-organized sub-folders: PDF / RIS / BibTeX / CSV / Citations

---

PaperLens vs The World

Why not just use Google Scholar? Or ChatGPT? Or Perplexity?

| | PaperLens | Google Scholar | ChatGPT / Perplexity | Zotero | Connected Papers | |---|:---:|:---:|:---:|:---:|:---:| | Complete research workflow | ✅ | ❌ | ❌ | ❌ | ❌ | | AI Deep Analysis (7-dim) | ✅ | ❌ | Surface-level | ❌ | ❌ | | Multi-paper comparison | ✅ | ❌ | Manual copy-paste | ❌ | ❌ | | 43 sources, one search | ✅ | 1 source | 0 sources (no real-time) | ❌ | 1 source | | Citation graph | ✅ | ❌ | ❌ | ❌ | ✅ | | Offline / local-first | ✅ | ❌ | ❌ | ✅ | ❌ | | CARSI 300+ universities | ✅ | ❌ | ❌ | ❌ | ❌ | | Citation formatting | ✅ | ❌ | ❌ | ✅ | ❌ | | Reference management | ✅ | ❌ | ❌ | ✅ | ❌ | | Chinese + English | ✅ | Limited | ✅ | ✅ | ❌ | | Free for personal use | ✅ | ❌ | ❌ | ✅ | ❌ |

---

Quick Start

30-Second Setup

git clone https://github.com/vanthree31/PaperLens.git
cd PaperLens
pip install -r requirements.txt
python main.py

The app opens in a native window (PyWebView) or your default browser.

Standalone Executable (No Python Required)

Download from GitHub Releases, or build:

build.bat          # Windows → dist/PaperLens.exe

Configure AI (Optional)

Open Settings in the app:

| Setting | Recommended | |---|---| | AI Search | qwen3.5:9b via Ollama (fast, local) | | AI Analysis | qwen3:14b via Ollama (quality, local) | | Or cloud API | DeepSeek / OpenAI / Anthropic |

All AI processing runs fully offline with Ollama — no API key, no data leaves your machine.

CARSI Setup (Chinese University Members)

• Open Settings → select your university from the CARSI list
• Log in with campus credentials
• ScienceDirect, Scopus, JSTOR are now accessible from anywhere

---

Platform Support

| Platform | Status | |---|---| | Windows 10/11 | Full support, standalone .exe | | macOS | python main.py | | Linux | python main.py |

Requirements: Python 3.9+, modern browser (Chrome/Edge recommended)

---

Tech Stack

| Layer | Technology | |---|---| | Backend | Python, Flask | | Frontend | ES Modules (7 JS + 9 CSS), D3.js | | AI | Ollama (offline) / DeepSeek / OpenAI / Anthropic | | Auth | CARSI / EZproxy | | Packaging | PyInstaller | | Extension | Chrome Manifest V3 |

---

Roadmap

See ROADMAP.md.

---

Contributing

Bug reports, feature ideas, and code contributions are all welcome.

Good first issues: good-first-issue

• Fork the repo → 2. Create a branch → 3. Commit → 4. Open a PR

See CONTRIBUTING.md for details.

---

Community & Support

Feedback & Discussion (QQ Group)

| QR Code | Group ID | |:---:|:---:| | | 632743351 |

If PaperLens helps your research, consider supporting the project. ☕

---

License

Free for personal, academic, and research use.

Commercial use (including SaaS, paid services, or integrating into a for-profit product) requires prior written authorization. See LICENSE for details.

📧 For commercial licensing or partnership: vanthree31@gmail.com

---

PaperLens — Your AI Research Copilot
_{Find. Read. Understand. Discover.}