fix(robotdaily): 改进 fallback 模式的标签和简介生成逻辑

arxiv-digest/scripts/enrich_papers.py

@@ -43,20 +43,57 @@ def build_prompt(paper: Dict[str, Any]) -> str:
 
 
 def fallback_enrichment(paper: Dict[str, Any]) -> Dict[str, Any]:
-    tags = FALLBACK_TAGS.get(paper["domain"], ["AI 论文", "机器学习", "应用研究"])
     title = paper.get("title", "")
     summary = paper.get("summary", "")
+    summary_lower = summary.lower()
     
-    # 从标题提取核心方法
-    brief = title.split(':')[0].strip()
-    if '.' in brief:
-        brief = brief.split('.')[0].strip()
+    # 从标题提取核心方法名，优先使用冒号前的部分
+    if ':' in title:
+        core_method = title.split(':')[0].strip()
+    elif '.' in title:
+        core_method = title.split('.')[0].strip()
+    else:
+        core_method = title.strip()
     
-    # 判断方法类型
-    summary_lower = summary.lower()
-    if "diffusion" in summary_lower:
+    # 如果方法名太长（超过 20 字符），使用摘要中的关键词
+    if len(core_method) > 20:
+        # 从摘要中提取关键词
+        if "diffusion" in summary_lower:
+            core_method = "扩散模型框架"
+        elif "reinforcement learning" in summary_lower:
+            core_method = "强化学习框架"
+        elif "imitation learning" in summary_lower:
+            core_method = "模仿学习框架"
+        elif "contrastive" in summary_lower:
+            core_method = "对比学习框架"
+        elif "transformer" in summary_lower:
+            core_method = "Transformer 框架"
+        elif "self-supervised" in summary_lower:
+            core_method = "自监督学习框架"
+        elif "representation" in summary_lower:
+            core_method = "表征学习框架"
+        elif "adaptation" in summary_lower or "adaptive" in summary_lower:
+            core_method = "自适应框架"
+        elif "multi-agent" in summary_lower or "marl" in summary_lower:
+            core_method = "多智能体框架"
+        elif "world model" in summary_lower:
+            core_method = "世界模型框架"
+        elif "residual policy" in summary_lower:
+            core_method = "残差策略优化"
+        elif "preference optimization" in summary_lower:
+            core_method = "偏好优化"
+        else:
+            core_method = "创新框架"
+    
+    # 判断方法类型（优先级从高到低）
+    method = "多种技术"
+    if "residual policy" in summary_lower:
+        method = "残差策略优化"
+    elif "preference optimization" in summary_lower:
+        method = "偏好优化"
+    elif "diffusion" in summary_lower:
         method = "扩散模型"
-    elif "reinforcement learning" in summary_lower:
+    elif "reinforcement learning" in summary_lower or "rl" in summary_lower:
         method = "强化学习"
     elif "imitation learning" in summary_lower:
         method = "模仿学习"
@@ -64,24 +101,58 @@ def fallback_enrichment(paper: Dict[str, Any]) -> Dict[str, Any]:
         method = "对比学习"
     elif "transformer" in summary_lower:
         method = "Transformer"
-    elif "self-supervised" in summary_lower:
+    elif "self-supervised" in summary_lower or "self supervised" in summary_lower:
         method = "自监督学习"
-    elif "representation" in summary_lower:
+    elif "representation learning" in summary_lower:
         method = "表征学习"
-    else:
-        method = "多种技术"
+    elif "adaptation" in summary_lower or "adaptive" in summary_lower:
+        method = "自适应方法"
+    elif "multi-agent" in summary_lower or "marl" in summary_lower:
+        method = "多智能体强化学习"
+    elif "world model" in summary_lower:
+        method = "世界模型"
     
-    # 判断应用领域
-    if "robot" in summary_lower or "manipulation" in summary_lower:
-        field = "机器人操作"
-    elif "navigation" in summary_lower or "driving" in summary_lower:
+    # 判断应用领域（优先级从高到低）
+    field = "相关任务"
+    if "cloth" in summary_lower or "布料" in summary_lower:
+        field = "布料操作"
+    elif "piano" in summary_lower or "music" in summary_lower:
+        field = "音乐演奏"
+    elif "racing" in summary_lower or ("autonomous" in summary_lower and ("driving" in summary_lower or "racing" in summary_lower)):
+        field = "自动驾驶"
+    elif "medical" in summary_lower or "delivery" in summary_lower or "logistics" in summary_lower:
+        field = "医疗物流"
+    elif "motion" in summary_lower or "humanoid" in summary_lower:
+        field = "人类动作生成"
+    elif "navigation" in summary_lower and ("robot" in summary_lower or "policy" in summary_lower):
+        field = "机器人导航"
+    elif "navigation" in summary_lower:
         field = "导航控制"
-    elif "translation" in summary_lower or "generation" in summary_lower:
-        field = "生成任务"
+    elif "traffic" in summary_lower or "scene understanding" in summary_lower:
+        field = "交通场景理解"
+    elif "map" in summary_lower or "localization" in summary_lower or "pose estimation" in summary_lower:
+        field = "定位建图"
+    elif "physical systems" in summary_lower or "emulator" in summary_lower:
+        field = "物理系统模拟"
+    elif "robot" in summary_lower or "manipulation" in summary_lower or "dexterous" in summary_lower:
+        field = "机器人操作"
     else:
-        field = "相关任务"
+        # 从标题推断领域
+        title_lower = title.lower()
+        if "robot" in title_lower or "manipulation" in title_lower:
+            field = "机器人操作"
+        elif "navigation" in title_lower or "driving" in title_lower:
+            field = "导航控制"
+        elif "piano" in title_lower or "music" in title_lower:
+            field = "音乐演奏"
+        elif "cloth" in title_lower:
+            field = "布料操作"
+        elif "motion" in title_lower or "humanoid" in title_lower:
+            field = "人类动作生成"
+        elif "racing" in title_lower or "autonomous" in title_lower:
+            field = "自动驾驶"
     
-    # 判断结果
+    # 判断结果/创新点
     if "real-world" in summary_lower or "deployment" in summary_lower:
         result = "真实部署"
     elif "zero-shot" in summary_lower:
@@ -90,15 +161,112 @@ def fallback_enrichment(paper: Dict[str, Any]) -> Dict[str, Any]:
         result = "首次提出"
     elif "improve" in summary_lower or "better" in summary_lower:
         result = "性能提升"
+    elif "efficient" in summary_lower or "efficiently" in summary_lower:
+        result = "高效"
+    elif "robust" in summary_lower or "robustly" in summary_lower:
+        result = "鲁棒性强"
+    elif "generalize" in summary_lower or "generalization" in summary_lower:
+        result = "泛化能力强"
+    elif "few-shot" in summary_lower or "few shot" in summary_lower:
+        result = "少样本学习"
+    elif "sim-to-real" in summary_lower or "sim2real" in summary_lower:
+        result = "仿真到现实迁移"
     else:
         result = "性能优化"
     
-    brief = f"{brief}，采用{method}解决{field}，实现{result}"
+    # 格式：提出 XXX 框架，采用 XXX 技术，解决 XXX 问题，实现 XXX 效果
+    brief = f"提出{core_method}，采用{method}解决{field}，实现{result}"
+    
+    # 从摘要提取具体标签（4-6 个），优先提取论文具体技术标签
+    tags = []
+    
+    # 核心方法标签
+    if "diffusion" in summary_lower:
+        tags.append("扩散模型")
+    if "reinforcement learning" in summary_lower:
+        tags.append("强化学习")
+    if "imitation learning" in summary_lower:
+        tags.append("模仿学习")
+    if "contrastive" in summary_lower:
+        tags.append("对比学习")
+    if "transformer" in summary_lower:
+        tags.append("Transformer")
+    if "self-supervised" in summary_lower:
+        tags.append("自监督学习")
+    if "multi-agent" in summary_lower or "marl" in summary_lower:
+        tags.append("多智能体强化学习")
+    if "world model" in summary_lower:
+        tags.append("世界模型")
+    if "residual policy" in summary_lower:
+        tags.append("残差策略优化")
+    if "preference optimization" in summary_lower:
+        tags.append("偏好优化")
+    if "representation learning" in summary_lower:
+        tags.append("表征学习")
+    if "adaptation" in summary_lower or "adaptive" in summary_lower:
+        tags.append("自适应")
+    
+    # 具体任务标签
+    if "robot" in summary_lower and "manipulation" in summary_lower:
+        tags.append("机器人操作")
+    if "dexterous" in summary_lower:
+        tags.append("灵巧操作")
+    if "navigation" in summary_lower:
+        tags.append("导航")
+    if "driving" in summary_lower or "racing" in summary_lower:
+        tags.append("自动驾驶")
+    if "cloth" in summary_lower:
+        tags.append("布料操作")
+    if "piano" in summary_lower:
+        tags.append("音乐演奏")
+    if "humanoid" in summary_lower or "motion" in summary_lower:
+        tags.append("动作生成")
+    if "localization" in summary_lower or "pose estimation" in summary_lower:
+        tags.append("定位")
+    if "traffic" in summary_lower:
+        tags.append("交通场景")
+    if "map" in summary_lower:
+        tags.append("建图")
+    
+    # 结果标签
+    if "zero-shot" in summary_lower:
+        tags.append("零样本")
+    if "real-world" in summary_lower:
+        tags.append("真实部署")
+    if "deployment" in summary_lower:
+        tags.append("部署")
+    if "sim-to-real" in summary_lower or "sim2real" in summary_lower:
+        tags.append("仿真到现实")
+    if "generalization" in summary_lower:
+        tags.append("泛化能力")
+    if "few-shot" in summary_lower:
+        tags.append("少样本")
+    if "efficient" in summary_lower:
+        tags.append("高效")
+    if "robust" in summary_lower:
+        tags.append("鲁棒性")
+    
+    # 如果标签数量不足 4 个，添加领域特定标签
+    if len(tags) < 4:
+        domain_tags = {
+            "embodied": ["具身智能", "机器人", "真实部署", "操控", "灵巧操作"],
+            "representation": ["表征学习", "潜在空间", "世界模型", "预训练", "自监督"],
+            "reinforcement": ["强化学习", "策略优化", "奖励设计", "离线 RL", "模仿学习"],
+        }
+        fallback = domain_tags.get(paper["domain"], ["AI 论文", "机器学习", "应用研究", "深度学习"])
+        for tag in fallback:
+            if tag not in tags:
+                tags.append(tag)
+                if len(tags) >= 6:
+                    break
+    
+    # 去重并限制数量
+    tags = list(dict.fromkeys(tags))[:6]
     
     return {
         "translated_abstract_zh": f"【LLM 暂不可用，先保留英文摘要要点】{truncate(summary, 220)}",
         "brief_explanation_zh": truncate(brief, 86),
-        "tags": tags[:5],
+        "tags": tags,
     }
 
 

site/content/ai-daily/2026-03-16.md

@@ -1,6 +1,6 @@
 ---
 title: "2026-03-16 · AI 每日简报"
-date: 2026-03-16T16:31:30.888993+08:00
+date: 2026-03-16T17:40:48.752446+08:00
 draft: false
 summary: "RobotDaily 2026-03-16：共 7 篇，含 具身智能 2 篇，表征学习 3 篇，强化学习 2 篇。"
 tags: ["robotdaily", "ai-daily", "具身智能", "表征学习", "强化学习"]
@@ -15,54 +15,54 @@ tags: ["robotdaily", "ai-daily", "具身智能", "表征学习", "强化学习"]
 ## 具身智能（2 篇）
 
 ### 1. PhysMoDPO: Physically-Plausible Humanoid Motion with Preference Optimization
-> PhysMoDPO，采用扩散模型解决机器人操作，实现真实部署
+> 提出PhysMoDPO，采用偏好优化解决人类动作生成，实现真实部署
 - 作者：Yangsong Zhang, Anujith Muraleedharan, Rikhat Akizhanov, Abdul Ahad Butt 等另外4人
-- 标签：`具身智能` `机器人` `真实部署` `操控`
+- 标签：`扩散模型` `偏好优化` `动作生成` `零样本`
 - 中文摘要：【LLM 暂不可用，先保留英文摘要要点】Recent progress in text-conditioned human motion generation has been largely driven by diffusion models trained on large-scale human motion data. Building on this progress, recent methods attempt to transfer such models…
 - 链接：[DOI](https://doi.org/10.48550/arXiv.2603.13228) | [arXiv](https://arxiv.org/abs/2603.13228v1) | [PDF](https://arxiv.org/pdf/2603.13228v1)
 
 ### 2. HandelBot: Real-World Piano Playing via Fast Adaptation of Dexterous Robot Policies
-> HandelBot，采用强化学习解决机器人操作，实现真实部署
+> 提出HandelBot，采用强化学习解决音乐演奏，实现真实部署
 - 作者：Amber Xie, Haozhi Qi, Dorsa Sadigh
-- 标签：`具身智能` `机器人` `真实部署` `操控`
+- 标签：`强化学习` `自适应` `机器人操作` `灵巧操作`
 - 中文摘要：【LLM 暂不可用，先保留英文摘要要点】Mastering dexterous manipulation with multi-fingered hands has been a grand challenge in robotics for decades. Despite its potential, the difficulty of collecting high-quality data remains a primary bottleneck for high-…
 - 链接：[DOI](https://doi.org/10.48550/arXiv.2603.12243) | [arXiv](https://arxiv.org/abs/2603.12243v1) | [PDF](https://arxiv.org/pdf/2603.12243v1)
 
 ## 表征学习（3 篇）
 
 ### 1. Representation Learning for Spatiotemporal Physical Systems
-> Representation Learning for Spatiotemporal Physical Systems，采用自监督学习解决相关任务，实现性能优化
+> 提出自监督学习框架，采用自监督学习解决物理系统模拟，实现性能优化
 - 作者：Helen Qu, Rudy Morel, Michael McCabe, Alberto Bietti 等另外3人
-- 标签：`表征学习` `潜在空间` `世界模型` `预训练`
+- 标签：`自监督学习` `表征学习` `潜在空间` `世界模型`
 - 中文摘要：【LLM 暂不可用，先保留英文摘要要点】Machine learning approaches to spatiotemporal physical systems have primarily focused on next-frame prediction, with the goal of learning an accurate emulator for the system's evolution in time. However, these emulators…
 - 链接：[DOI](https://doi.org/10.48550/arXiv.2603.13227) | [arXiv](https://arxiv.org/abs/2603.13227v1) | [PDF](https://arxiv.org/pdf/2603.13227v1)
 
 ### 2. Separable neural architectures as a primitive for unified predictive and generative intelligence
-> Separable neural architectures as a primitive for unified predictive and generative i…
+> 提出强化学习框架，采用强化学习解决导航控制，实现性能优化
 - 作者：Reza T. Batley, Apurba Sarker, Rajib Mostakim, Andrew Klichine 等另外1人
-- 标签：`表征学习` `潜在空间` `世界模型` `预训练`
+- 标签：`强化学习` `导航` `建图` `表征学习`
 - 中文摘要：【LLM 暂不可用，先保留英文摘要要点】Intelligent systems across physics, language and perception often exhibit factorisable structure, yet are typically modelled by monolithic neural architectures that do not explicitly exploit this structure. The separabl…
 - 链接：[DOI](https://doi.org/10.48550/arXiv.2603.12244) | [arXiv](https://arxiv.org/abs/2603.12244v1) | [PDF](https://arxiv.org/pdf/2603.12244v1)
 
 ### 3. VIRD: View-Invariant Representation through Dual-Axis Transformation for Cross-View Pose Estimation
-> VIRD，采用表征学习解决机器人操作，实现首次提出
+> 提出VIRD，采用多种技术解决自动驾驶，实现首次提出
 - 作者：Juhye Park, Wooju Lee, Dasol Hong, Changki Sung 等另外3人
-- 标签：`表征学习` `潜在空间` `世界模型` `预训练`
+- 标签：`自动驾驶` `定位` `表征学习` `潜在空间`
 - 中文摘要：【LLM 暂不可用，先保留英文摘要要点】Accurate global localization is crucial for autonomous driving and robotics, but GNSS-based approaches often degrade due to occlusion and multipath effects. As an emerging alternative, cross-view pose estimation predict…
 - 链接：[DOI](https://doi.org/10.48550/arXiv.2603.12918) | [arXiv](https://arxiv.org/abs/2603.12918v1) | [PDF](https://arxiv.org/pdf/2603.12918v1)
 
 ## 强化学习（2 篇）
 
 ### 1. Efficient Real-World Autonomous Racing via Attenuated Residual Policy Optimization
-> Efficient Real-World Autonomous Racing via Attenuated Residual Policy Optimization，采用…
+> 提出强化学习框架，采用残差策略优化解决自动驾驶，实现真实部署
 - 作者：Raphael Trumpp, Denis Hoornaert, Mirco Theile, Marco Caccamo
-- 标签：`强化学习` `策略优化` `奖励设计` `离线 RL`
+- 标签：`强化学习` `残差策略优化` `自动驾驶` `零样本`
 - 中文摘要：【LLM 暂不可用，先保留英文摘要要点】Residual policy learning (RPL), in which a learned policy refines a static base policy using deep reinforcement learning (DRL), has shown strong performance across various robotic applications. Its effectiveness is part…
 - 链接：[DOI](https://doi.org/10.48550/arXiv.2603.12960) | [arXiv](https://arxiv.org/abs/2603.12960v1) | [PDF](https://arxiv.org/pdf/2603.12960v1)
 
 ### 2. Beyond Imitation: Reinforcement Learning Fine-Tuning for Adaptive Diffusion Navigation Policies
-> Beyond Imitation，采用扩散模型解决机器人操作，实现零样本泛化
+> 提出Beyond Imitation，采用扩散模型解决机器人导航，实现零样本泛化
 - 作者：Junhe Sheng, Ruofei Bai, Kuan Xu, Ruimeng Liu 等另外4人
-- 标签：`强化学习` `策略优化` `奖励设计` `离线 RL`
+- 标签：`扩散模型` `强化学习` `模仿学习` `自适应`
 - 中文摘要：【LLM 暂不可用，先保留英文摘要要点】Diffusion-based robot navigation policies trained on large-scale imitation learning datasets, can generate multi-modal trajectories directly from the robot's visual observations, bypassing the traditional localization-m…
 - 链接：[DOI](https://doi.org/10.48550/arXiv.2603.12868) | [arXiv](https://arxiv.org/abs/2603.12868v1) | [PDF](https://arxiv.org/pdf/2603.12868v1)