MDER-DR: Multi‑Hop Question Answering with Entity‑Centric Summaries
Riccardo Campi, Nicolò Oreste Pinciroli Vago, Mathyas Giudici, Marco Brambilla, Piero Fraternali · 2026
Read MDER-DR: Multi‑Hop Question Answering with Entity‑Centric Summaries by Riccardo Campi, Nicolò Oreste Pinciroli Vago, Mathyas Giudici, Marco Brambilla, Piero Fraternali (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-11223.txt.
Description
Riccardo Campi and coauthors rebuild retrieval-augmented question answering over knowledge graphs so that answers keep the context that plain triples throw away. Standard graph indexing reduces text to triples and loses nuance, which hurts multi-hop questions that must compose facts about several entities. Their indexing writes context-derived descriptions for triples and merges them with entity-level summaries, avoiding explicit graph traversal at retrieval time. A decompose-resolve retriever breaks questions into resolvable triples and grounds them iteratively, gaining up to 66 percent over standard baselines.
Catalog record
| CREATOR | Riccardo Campi, Nicolò Oreste Pinciroli Vago, Mathyas Giudici, Marco Brambilla, Piero Fraternali |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50215 |
| LICENSE | http://creativecommons.org/licenses/by-sa/4.0/ |
| RIGHTS | CC BY SA |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.11223