Comparing Chunking and Embedding Strategies for Turkish RAG Systems

arXiv:2608.26192v1 Announce Type: cross Abstract: How documents are segmented into retrievable chunks and how those chunks are embedded strongly affect Retrieval-Augmented Generation (RAG) quality, yet neither has been systematically studied for morphologically rich languages such as Turkish. We compare Turkish document question answering across three chunking strategies (fixed-length, semantic, and layout-aware Docling), five embedding models, and two generator LLMs, over three documents with contrasting layouts. The fully crossed design yields 9,000 graded question-answer evaluations, each scored by an independent judge model, and component comparisons are tested by paired McNemar tests under Holm correction. Four findings follow. The chunking strategy determines how much the embedding choice matters: layout-aware chunking compresses the spread between the modern embedding models to about a point. The three leading embedding models are statistically indistinguishable, so language specialization yields no measurable retrieval advantage. The faster generator is not the more accurate one. And the preferred configuration depends on content type, since layout-aware chunking helps documents containing tables far more than prose. The best individual components therefore do not compose into the best complete configuration, which reaches 87.0%.

Abstract:How documents are segmented into retrievable chunks and how those chunks are embedded strongly affect Retrieval-Augmented Generation (RAG) quality, yet neither has been systematically studied for morphologically rich languages such as Turkish. We compare Turkish document question answering across three chunking strategies (fixed-length, semantic, and layout-aware Docling), five embedding models, and two generator LLMs, over three documents with contrasting layouts. The fully crossed design yields 9,000 graded question-answer evaluations, each scored by an independent judge model, and component comparisons are tested by paired McNemar tests under Holm correction. Four findings follow. The chunking strategy determines how much the embedding choice matters: layout-aware chunking compresses the spread between the modern embedding models to about a point. The three leading embedding models are statistically indistinguishable, so language specialization yields no measurable retrieval advantage. The faster generator is not the more accurate one. And the preferred configuration depends on content type, since layout-aware chunking helps documents containing tables far more than prose. The best individual components therefore do not compose into the best complete configuration, which reaches 87.0%.

Accepted to INTCEC 2026. This is the author's pre-print version. The final authenticated version will be available through the conference proceedings

Computation and Language (cs.CL); Artificial Intelligence (cs.AI)

arXiv:2608.26192 [cs.CL]

(or arXiv:2608.26192v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2608.26192 Focus to learn more arXiv-issued DOI via DataCite

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

aiqbench analysis

Comparing Chunking and Embedding Strategies for Turkish RAG Systems

Related coverage