We present HistoriQA-ThirdRepublic: a French-language dataset of multi-hop historical questions derived from parliamentary debates and newspapers of the French Third Republic. Designed in collaboration with a historian, the corpus captures complex reasoning patterns typical of historical inquiry, including cross-source synthesis, temporal reasoning, and the integration of sparse evidence. The dataset is made of 1782 questions and emphasizes multi-hop connections across heterogeneous historical documents, providing a resource for evaluating retrieval-augmented and large language model systems in domain-specific contexts. We describe the methodology for constructing the corpus, including the selection and alignment of sources, question validation, and metadata integration. While the dataset focuses on French historical documents, our methodology can be readily adapted to other languages and national corpora. Finally, we demonstrate how the corpus can support realistic evaluation scenarios for multi-hop question answering, bridging the gap between NLP benchmarks and the needs of historical scholarship.

Graph-based Retrieval-Augmented Generation (RAG) is increasingly used to explore long, heterogeneous, and weakly structured corpora, including historical archives. However, in such settings, naive full-corpus indexing is often computationally costly and sensitive to OCR noise, document redundancy, and topical dispersion. In this paper, we investigate corpus pre-targeting strategies as an intermediate layer to improve the efficiency and effectiveness of graph-based RAG for historical research.We evaluate a set of pre-targeting heuristics tailored to single-hop and multi-hop of historical questions on HistoriQA-ThirdRepublic, a French question-answering dataset derived from parliamentary debates and contemporary newspapers. Our results show that appropriate pre-targeting strategies can improve retrieval recall by 3–5% while reducing token consumption by 32–37% compared to full-corpus indexing, without degrading coverage of relevant documents.Beyond performance gains, this work highlights the importance of corpus-level optimization for applying RAG to large-scale historical collections, and provides practical insights for adapting graph-based RAG pipelines to the specific constraints of digitized archives.

Retrieval-Augmented Generation (RAG) pipelines typically rely on a fixed indexing and retrieval configuration determined at preprocessing time. This one-size-fits-all design is ill-suited to domain-expert settings, where heterogeneous queries require different chunking granularities, metadata constraints, and source-selection strategies. As a result, configurations that are effective for one family of queries often perform poorly for others. In this paper, we introduce ORDER (Optimal Routing for Dynamic Evidence Retrieval), a query-conditioned RAG framework that jointly adapts indexing and retrieval to the incoming query. Our approach first discovers semantic clusters over a given set of questions associated to a corpus and learns, for each cluster, a chunking strategy together with a suited metadata filtering and reranking configuration. At inference time, queries are routed to the appropriate pre-built index through nearest-centroid assignment. To further improve retrieval, we propose a supervised query router (QRe) that predicts which collections are most likely to contain relevant evidence, coupled with a Uniform Multi-source Sampler (UMS) that allocates the retrieval budget evenly across the selected sources. We evaluate our framework on large-scale, heterogeneous historical archives and show that conditioning both indexing and retrieval on the query consistently outperforms both naive baselines and strong state-of-the-art RAG systems in complex expert-domain environments.

Dans le contexte de l’utilisation croissante des LLM, le besoin d’un retour efficace et automatique aux sources devient essentiel, en particulier pour les documents historiques. La capacité des LLM à identifier les sources pertinentes ne constitue plus seulement un maillon dans une chaîne où l’objectif final est la génération de réponses ; elle représente un enjeu fondamental de l’analyse, justifiant une évaluation à part entière. Quelles stratégies, quels modèles et quels paramètres offrent aux historiens les meilleures capacités d’exploration d’un corpus vaste et bruité ? Cet article propose une première tentative d’évaluation du retriever dans un cadre de RAG appliqué aux débats parlementaires de la Troisième République.

We present the AGODA (Analyse sémantique et Graphes relationnels pour l’Ouverture des Débats à l’Assemblée nationale) project, which aims to create a platform for consulting and exploring digitised French parliamentary debates (1881-1940) available in the digital library of the National Library of France. This project brings together historians and NLP specialists: parliamentary debates are indeed an essential source for French history of the contemporary period, but also for linguistics. This project therefore aims to produce a corpus of texts that can be easily exploited with computational methods, and that respect the TEI standard. Ancient parliamentary debates are also an excellent case study for the development and application of tools for publishing and exploring large historical corpora. In this paper, we present the steps necessary to produce such a corpus. We detail the processing and publication chain of these documents, in particular by mentioning the problems linked to the extraction of texts from digitised images. We also introduce the first analyses that we have carried out on this corpus with “bag-of-words” techniques not too sensitive to OCR quality (namely topic modelling and word embedding).