Dense vector retrieval can miss evidence connected through intermediate entities, while direct graph retrieval requires access to the graph at query time. We compare five offline graph-based embedding transformations and strategies for using the resulting graph-modified embeddings alongside the originals, without query-time graph access. On a synthetic professional-network benchmark with complete supporting-evidence sets, direct replacement of the original embeddings reduces evidence recovery. Multi-stage strategies improve recovery by using graph-modified embeddings to guide the search while retaining the originals for semantic relevance. On questions requiring linked facts or comparisons among candidates, the strongest configuration improves recovery by up to 25.9% relative to plain vector retrieval. It also matches baseline recovery with 40% fewer returned nodes. Controls confirm that graph information contributes beyond multi-stage retrieval alone. Across the tested methods, retrieval performance depends more on how graph-modified embeddings are used than on how they are produced.
Figures & tables
Figure 1: Graph information used by the five transformations. Darker nodes are more similar to the source node; orange outlines mark embeddings included in an average, with thicker outlines indicating greater weight. Dashed regions and crosses show Louvain communities and their centroids.
Figure 2: Allocation of the 40-node retrieval budget. Each cell represents one returned node. Light blue and pink denote original- and modified-space searches; orange denotes fused scoring. The two darker blues denote successive searches with shifted queries in the original space. Vertical lines separate stages.
Figure 3: Gold-evidence recovery under direct retrieval across all 977 validation questions. The dashed line denotes plain vector retrieval.
Figure 4: Gold-evidence recovery across retrieval strategies on the 527 difficult questions. The dashed line denotes plain retrieval.
Figure 5: Alternating budget split on the 527 difficult questions.
Figure 6: Two-stage fused retrieval on the 527 difficult questions.
Figure 7: Gold-evidence recovery of the single requery and chained requery schedules across graph-modified embedding collections. The dashed line denotes plain vector retrieval.
Figure 8: Improvement of chained requery with the walk representation over plain retrieval, reported by question group. Annotations show absolute and relative gains.
Figure 9: Cumulative gold-evidence recovery over the first k nodes returned in stage order, without global reranking. The full chained requery method uses the walk representation; both controls retain its 6/6/14/14 schedule but remove graph conditioning. Vertical dotted lines mark stage boundaries. The horizontal dashed line shows plain retrieval’s recovery at 40 nodes.
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
Figure 10: The graph schema and one instance from the validation corpus. Shape marks the node type: a circle is a chunk, a rounded square a document, a square an entity. Counts on the schema edges are those of the validation graph. In the instance, a resume document is about a person, its chunks are part_of it, and each chunk mentions the entities named in its text. Because mentions always runs from a chunk to an entity, two hops lead from one chunk to another through an entity they share — here two employees of the same organisation.