config.toml

# Configuration for Paper Aggregator

[general]
# Number of days to keep papers and cache summaries
days_back = 30

# Base URL of the GitHub Pages site (used for RSS feed links)
# Example: "https://username.github.io/paper-pulse"
site_url = "https://jamie-cui.github.io/paper-pulse"

# Data storage paths (relative to project root)
data_dir = "data"
papers_file = "papers.json"
failed_file = "failed.json"

[fetchers]
# arXiv fetcher settings
[fetchers.arxiv]
# Categories to fetch from arXiv
categories = ["cs.CR", "cs.AI", "cs.LG", "cs.CL"]
# Delay between requests (seconds) - arXiv recommends 3+ seconds
delay = 3.0
# Maximum number of results per category
max_results = 500
batch_size = 100

# IACR fetcher settings
[fetchers.iacr]
# Delay between requests (seconds)
delay = 2.0

[summarizer]
# AI model settings for DashScope API
model = "qwen-plus"  # Options: qwen-turbo, qwen-plus, qwen-max
max_tokens = 1500  # Increased for bilingual summaries (Chinese + English)
temperature = 0.7
timeout = 60

# Rate limiting
rate_limit_delay = 1.0  # Delay between summarization API calls (seconds)
max_retries = 3
retry_delay = 5.0

# Prompt template for paper summarization
# Note: The bilingual prompt is hard-coded in summarizer.py for better control
# This template is kept for backward compatibility and custom single-language summaries
# To modify the bilingual prompt, edit _create_bilingual_prompt() in scripts/summarizer.py
prompt_template = """你是一位精通各领域前沿研究的学术文献解读专家，面对一篇给定的论文，请你高效阅读并迅速提取出其核心内容。要求在解读过程中，先对文献的背景、研究目的和问题进行简明概述，再详细梳理研究方法、关键数据、主要发现及结论，同时对新颖概念进行通俗易懂的解释，帮助读者理解论文的逻辑与创新点；最后，请对文献的优缺点进行客观评价，并指出可能的后续研究方向。整体报告结构清晰、逻辑严谨。

Title: {title}

Abstract: {abstract}

Provide a concise summary:"""

[rss]
# Maximum number of papers to include in the RSS feed
max_items = 50

[frontend]
# Number of papers to display per page
papers_per_page = 10

[email]
# Which summary language to include in the daily email digest
# Options: "zh" (Chinese only), "en" (English only), "both" (Chinese + English)
summary_language = "zh"

[keywords]
# Path to keyword filter configuration file
file = "keywords.txt"

# Control which sources should apply keyword filtering
# Set to true to apply keyword filtering, false to fetch all papers from that source
apply_to_arxiv = true
apply_to_iacr = false