প্রকল্প সম্পর্কে

ChatPaper is an open-source paper processing tool for researchers, aiming to accelerate literature reading and writing processes using large language models. The project is developed by a PhD student in the reinforcement learning direction at the University of Science and Technology of China, with README in Chinese and an English version provided. Main capabilities: - Paper summarization: Automatically search and download the latest papers on arXiv based on keywords, and use the ChatGPT API to summarize the paper in a fixed structure, including research background, existing solutions and problems, the solution and steps of the paper, and experimental tasks and results, for quick initial screening. - Local PDF processing: Can directly specify the path of the local PDF file or folder for batch summarization. - Full-text translation: Provides a solution for full-text translation of local PDFs, depending on scipdf_parser and grobid services, and requires a Java environment to be configured and run on Ubuntu or macOS. - Auxiliary tools: README lists related projects such as ChatReviewer (paper pros and cons analysis and improvement suggestions), ChatImprovement (polishing and translation), ChatResponse (review response), and ChatGenTitle (generating titles from abstracts), as well as community extensions such as one-click literature review and ChatPaper2Xmind. - Google Scholar organization: Includes google_scholar_spider.py, which can search and export CSV files based on keywords and sorted by references. Running modes: - Script mode: Fill in the OpenAI key in apikey.ini, install the dependencies in requirements.txt, and run chat_paper.py or chat_arxiv.py, supporting parameters such as query, filter_keys, max_results, sort, save_image, and file_format. - Flask service: Run app.py to provide a web interface on local port 5000, encapsulating multiple script interfaces, and save the results in export, pdf_files, and response_file directories. - Docker: Provides docker-compose.yaml, requiring the OpenAI key to be replaced and started, with a note in the README about potential issues with the path. - HuggingFace Space: Provides online deployment instructions, but the README notes that this part of the functionality is temporarily suspended. The README also notes that search words cannot recognize hyphens, summary articles are not supported, and network proxy is required for use, emphasizing academic ethics and proper usage. All project functions are free and open-source.