We introduce MCompassRAG, a metadata-guided retrieval framework that uses topic-level signals as a semantic compass for paragraph-level RAG. Instead of relying only on noisy dense chunk embeddings, MCompassRAG enriches chunks with topic metadata and trains a lightweight retriever via LLM-teacher distillation. At inference time, it performs topic-aware retrieval without extra LLM calls, improving evidence quality and efficiency across complex retrieval benchmarks.</p>\n","updatedAt":"2026-06-22T17:15:36.618Z","author":{"_id":"60e32baedc56466240084155","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/60e32baedc56466240084155/Q-cJh3Q3-vvMbe749Gt5B.jpeg","fullname":"Amirhossein Abaskohi","name":"AmirhosseinAbaskohi","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":4,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8072634339332581},"editors":["AmirhosseinAbaskohi"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/60e32baedc56466240084155/Q-cJh3Q3-vvMbe749Gt5B.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.18508","authors":[{"_id":"6a396d12fdcd3514343bb3df","name":"Amirhossein Abaskohi","hidden":false},{"_id":"6a396d12fdcd3514343bb3e0","name":"Raymond Li","hidden":false},{"_id":"6a396d12fdcd3514343bb3e1","name":"Gaetano Cimino","hidden":false},{"_id":"6a396d12fdcd3514343bb3e2","name":"Peter West","hidden":false},{"_id":"6a396d12fdcd3514343bb3e3","name":"Giuseppe Carenini","hidden":false},{"_id":"6a396d12fdcd3514343bb3e4","name":"Issam H. Laradji","hidden":false}],"publishedAt":"2026-06-16T00:00:00.000Z","submittedOnDailyAt":"2026-06-22T00:00:00.000Z","title":"MCompassRAG: Topic Metadata as a Semantic Compass for Paragraph-Level Retrieval","submittedOnDailyBy":{"_id":"60e32baedc56466240084155","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/60e32baedc56466240084155/Q-cJh3Q3-vvMbe749Gt5B.jpeg","isPro":false,"fullname":"Amirhossein Abaskohi","user":"AmirhosseinAbaskohi","type":"user","name":"AmirhosseinAbaskohi"},"summary":"Retrieval-augmented generation (RAG) systems depend critically on how documents are chunked and searched. Fine-grained chunks can improve retrieval precision but expand the search space, increasing latency and cost; larger chunks reduce the number of candidates but make dense similarity less reliable, as the representation for each chunk mixes multiple topics and introduces more semantic noise. This trade-off becomes especially limiting in deep research tasks, where retrieval must be both fast and precise across large, heterogeneous corpora. We introduce MCompassRAG, a metadata-guided retrieval framework that uses topic-level signals as a semantic compass for selecting relevant evidence. Instead of relying only on cosine similarity between queries and noisy chunk embeddings, MCompassRAG enriches chunk representations with topic metadata in the same embedding space and trains a lightweight retriever through LLM-teacher distillation. At inference time, MCompassRAG performs topic-aware retrieval without additional LLM calls, improving both efficiency and evidence quality. Across six complex retrieval benchmarks, MCompassRAG improves information efficiency (IE) by 8.24% on average with over 5 times lower latency than the strongest efficient RAG baselines. Code is available on https://github.com/AmirAbaskohi/MCompassRAG.","upvotes":10,"discussionId":"6a396d12fdcd3514343bb3e5","githubRepo":"https://github.com/AmirAbaskohi/MCompassRAG","githubRepoAddedBy":"user","ai_summary":"MCompassRAG enhances retrieval-augmented generation by using topic-level metadata to guide chunk selection, improving both efficiency and precision in complex research tasks.","ai_keywords":["retrieval-augmented generation","chunking","dense similarity","semantic noise","LLM-teacher distillation","topic-aware retrieval","information efficiency","embedding space"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":3,"organization":{"_id":"6723d7c5f0e98abd466cd7df","name":"UBC-VisionLanguage","fullname":"Vision and NLP Group at University of British Columbia","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6712caf88a321497c3a15673/0mNkKMu8h74rP2Hzo1O5U.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"60e32baedc56466240084155","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/60e32baedc56466240084155/Q-cJh3Q3-vvMbe749Gt5B.jpeg","isPro":false,"fullname":"Amirhossein Abaskohi","user":"AmirhosseinAbaskohi","type":"user"},{"_id":"6228ede94323cef93a956b24","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6228ede94323cef93a956b24/b0t3qorPnU-0M_u3dZWqF.jpeg","isPro":false,"fullname":"AmirHossein DabiriAghdam","user":"AmirHossein1378","type":"user"},{"_id":"6a2da6c8ca070ee12c6e396c","avatarUrl":"/avatars/0355287dcabaa67dbc7f0b10b87451f9.svg","isPro":false,"fullname":"Joe Mama","user":"JoeMama123123123","type":"user"},{"_id":"66105ffd174b378a724fb56d","avatarUrl":"/avatars/a32540ce3eb0cc06fe6aa39127dc6ad2.svg","isPro":false,"fullname":"Gaetano Cimino","user":"Gaetano98","type":"user"},{"_id":"6a2718c4ee8c9826408a267d","avatarUrl":"/avatars/75b672abbc4fba71f784503e86753088.svg","isPro":false,"fullname":"SARA VAKILI","user":"saravakili123","type":"user"},{"_id":"6a271c4928e4d750aa0e88cd","avatarUrl":"/avatars/1fd882587298d5a30080b54d1b6c96e5.svg","isPro":false,"fullname":"Asami tanaka","user":"Asami105","type":"user"},{"_id":"64062855692855e65ae31688","avatarUrl":"/avatars/e35d22a037b8b35422d3ee982f133076.svg","isPro":false,"fullname":"Issam Laradji","user":"issamlaradji","type":"user"},{"_id":"607f060442beb4da0f990182","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/607f060442beb4da0f990182/j5W2tLyU6JqkaTf3kv66s.jpeg","isPro":false,"fullname":"Patrice Bechard","user":"patricebechard","type":"user"},{"_id":"64229415bc8614c43a5bc2f5","avatarUrl":"/avatars/37ab3c70cdc39ef4b3a8892062788d08.svg","isPro":false,"fullname":"Aparna Mahalingam","user":"aparna-mahalingam","type":"user"},{"_id":"69e8d7e5feb97678322e500e","avatarUrl":"/avatars/111fb44198b2dc1eebc9eabd286846a5.svg","isPro":false,"fullname":"Alex","user":"bitzy21ro","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"6723d7c5f0e98abd466cd7df","name":"UBC-VisionLanguage","fullname":"Vision and NLP Group at University of British Columbia","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6712caf88a321497c3a15673/0mNkKMu8h74rP2Hzo1O5U.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.18508.md","query":{}}">
MCompassRAG: Topic Metadata as a Semantic Compass for Paragraph-Level Retrieval
Abstract
MCompassRAG enhances retrieval-augmented generation by using topic-level metadata to guide chunk selection, improving both efficiency and precision in complex research tasks.
Retrieval-augmented generation (RAG) systems depend critically on how documents are chunked and searched. Fine-grained chunks can improve retrieval precision but expand the search space, increasing latency and cost; larger chunks reduce the number of candidates but make dense similarity less reliable, as the representation for each chunk mixes multiple topics and introduces more semantic noise. This trade-off becomes especially limiting in deep research tasks, where retrieval must be both fast and precise across large, heterogeneous corpora. We introduce MCompassRAG, a metadata-guided retrieval framework that uses topic-level signals as a semantic compass for selecting relevant evidence. Instead of relying only on cosine similarity between queries and noisy chunk embeddings, MCompassRAG enriches chunk representations with topic metadata in the same embedding space and trains a lightweight retriever through LLM-teacher distillation. At inference time, MCompassRAG performs topic-aware retrieval without additional LLM calls, improving both efficiency and evidence quality. Across six complex retrieval benchmarks, MCompassRAG improves information efficiency (IE) by 8.24% on average with over 5 times lower latency than the strongest efficient RAG baselines. Code is available on https://github.com/AmirAbaskohi/MCompassRAG.
Community
We introduce MCompassRAG, a metadata-guided retrieval framework that uses topic-level signals as a semantic compass for paragraph-level RAG. Instead of relying only on noisy dense chunk embeddings, MCompassRAG enriches chunks with topic metadata and trains a lightweight retriever via LLM-teacher distillation. At inference time, it performs topic-aware retrieval without extra LLM calls, improving evidence quality and efficiency across complex retrieval benchmarks.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.18508 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.18508 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.18508 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.