[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80918":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":11,"openIssues":12,"contributorsCount":12,"subscribersCount":12,"size":12,"stars1d":12,"stars7d":12,"stars30d":12,"stars90d":12,"forks30d":12,"starsTrendScore":12,"compositeScore":12,"rankGlobal":9,"rankLanguage":9,"license":9,"archived":13,"fork":13,"defaultBranch":14,"hasWiki":15,"hasPages":13,"topics":16,"createdAt":9,"pushedAt":9,"updatedAt":17,"readmeContent":18,"aiSummary":19,"trendingCount":12,"starSnapshotCount":12,"syncStatus":20,"lastSyncTime":21,"discoverSource":22},80918,"GraphScholar","Vannico233\u002FGraphScholar","Vannico233","Graph-powered literature research assistant for graph learning and GraphRAG papers.",null,"Python",33,0,false,"main",true,[],"2026-06-12 02:04:08","\u003Ctable>\n  \u003Ctr>\n    \u003Ctd>\u003Ch1>GraphScholar\u003C\u002Fh1>\u003C\u002Ftd>\n    \u003Ctd align=\"right\">\u003Ca href=\".\u002FREADME_zh.md\">中文版\u003C\u002Fa> \u002F English\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n![GraphScholar](GraphScholar.png)\n\nGraphScholar is a research assistant for graph learning and GraphRAG papers. It turns local PDF papers into structured metadata, chunk-level evidence, and a graph-based literature store for retrieval and query answering.\n\n## What It Does\n\n- parses PDFs from `data\u002F`\n- extracts title, abstract, tasks, applications, datasets, method summary, and contribution summary\n- builds paper summaries and chunk-level evidence\n- constructs a paper-topic-method-task-application-dataset graph\n- answers research questions with retrieval and graph querying\n- saves each run into `outputs\u002Fanswer\u002F`\n\n## Main Outputs\n\n- `outputs\u002Fpaper_summaries.json`\n- `outputs\u002Fpaper_chunks.json`\n- `outputs\u002Fpaper_graph.json`\n- `outputs\u002Fpaper_metadata_cache.json`\n- `outputs\u002Fanswer\u002F*.md`\n\n## Workflow\n\n1. `paper_organization.py` reads PDFs and extracts structured metadata.\n2. `src\u002Fbuild_graph.py` builds the paper graph from the summaries.\n3. `src\u002Ftools.py` handles paper search, chunk search, and graph queries.\n4. `src\u002Fagent.py` routes the question, gathers evidence, and produces the answer.\n5. `run_agent.py` runs demo or single-question mode and writes a report.\n\n## Data Schema\n\nEach paper summary includes:\n\n- `title`\n- `abstract`\n- `tags`\n- `category`\n- `paper_type`\n- `tasks`\n- `applications`\n- `datasets`\n- `method_summary`\n- `contribution_summary`\n- `confidence`\n\n## How to Run\n\nRebuild the paper store:\n\n```powershell\npython paper_organization.py\n```\n\nRun in local deterministic mode:\n\n```powershell\npython run_agent.py --no-llm\n```\n\nRun one question:\n\n```powershell\npython run_agent.py --question \"If I am working on GraphRAG, help me organize the most representative papers in recent years by method, evaluation, and survey.\"\n```\n\nRun the preset demo set:\n\n```powershell\npython run_agent.py --demo\n```\n\n## LLM Configuration\n\n`src\u002Fllm_client.py` uses explicit in-code settings:\n\n- `DEFAULT_BASE_URL`\n- `DEFAULT_API_KEY`\n- `DEFAULT_MODEL_ID`\n\nFill those values before using LLM mode.\n\n## Why This Project Feels Different\n\nThis is not a plain keyword search tool. It combines:\n\n- structured paper extraction\n- graph-aware literature organization\n- task\u002Fapplication\u002Fdataset-aware retrieval\n- chunk evidence with page references\n- saved answer reports for review and reuse\n\n## Notes\n\nThe project is intentionally lightweight, but it already behaves like a compact research workflow for graph literature analysis.\n","GraphScholar 是一个基于图的文献研究助手，专为图学习和GraphRAG论文提供支持。该项目能够将本地PDF文件解析成结构化元数据、块级证据，并构建一个基于图的文献库以实现检索和查询回答。其核心技术特点包括从PDF中提取标题、摘要、任务、应用、数据集、方法概要及贡献总结，进而生成论文摘要与块级证据，并构建涵盖论文-主题-方法-任务-应用-数据集关系的知识图谱。此外，它还支持通过检索和图查询来回答研究问题，并保存每次运行的结果。GraphScholar适用于需要对特定领域内大量文献进行系统性整理与分析的研究场景，如学术研究或技术调研等。",2,"2026-06-11 04:02:50","CREATED_QUERY"]