A Storage-Retrieval Gap in Parametric Knowledge Graph Memory
MCML Authors
Volker Tresp
Prof. Dr.
Core PI
Abstract
Volker Tresp
Prof. Dr.
Core PI
Abstract
Graph retrieval-augmented generation places retrieved subgraphs into the model's context window at query time, paying a recurring token cost and exposing source data on every call. We study an alternative: compiling a knowledge graph offline into a bank of LoRA adapters, one per entity, that serve as a parametric knowledge layer queried by injecting weights rather than text, at zero query-time context cost. On the MetaQA dataset, we find that subgraph-trained adapters encode context-free factual knowledge that generalizes to unseen questions: on single-valued relations the adapter gains +0.243 exact-match score over a base model that is nearly blind closed-book (0.007), and only the correct adapter recovers this knowledge (an oracle gap of +0.283 over the base model). However, the stored knowledge is not recoverable by similarity: given a query with no subgraph, embedding-based and weight-space geometry retrieval both perform at chance, because a semantically neighbouring entity's adapter does not contain the answer - knowledge is stored locally and does not transfer. Weight geometry correlates with subgraph semantics (ρ=+0.329) but not with functional retrievability. We quantify the byte and context-token costs against graph retrieval-augmented generation and discuss deployment implications. Our results establish that parametric knowledge graph memory is feasible for storing knowledge, and identify selecting and composing the right adapters by a mechanism other than semantic similarity as the central open problem - motivating a learned, query-conditioned composition mechanism.
inproceedings PCK+26
SKGi @SEMANTICS 2026
Workshop on Scaling Knowledge Graphs for Industry at the 22th International Conference on Semantic Systems. Ghent, Belgium, Sep 15-17, 2026.Authors
M. M. L. Pulici • C. X. Chu • E. Kharlamov • V. TrespLinks
PDFResearch Area
BibTeXKey: PCK+26