Topology of Reasoning: Retrieved Cell Complex-Augmented Generation for Textual Graph Question Answering
Authors: Sen Zhao, Lincheng Zhou, Yue Chen, Ding Zou
Organizations: Academy of Advanced Interdisciplinary Studies, Chongqing University of Posts and Telecommunications, Chongqing, China · School of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing, China · Intelligent System Department, Zhongxing Telecom Equipment (ZTE), Changsha, Hunan, China
Retrieval-Augmented Generation (RAG) enhances the reasoning ability of Large Language Models (LLMs) by dynamically integrating external knowledge, thereby mitigating hallucinations and strengthening contextual grounding for structured data such as graphs. Nevertheless, most existing RAG variants for textual graphs concentrate on low-dimensional structures -- treating nodes as entities (0-dimensional) and edges or paths as pairwise or sequential relations (1-dimensional), but overlook cycles, which are crucial for reasoning over relational loops. Such cycles often arise in questions requiring closed-loop inference about similar objects or relative positions. This limitation often results in incomplete contextual grounding and restricted reasoning capability. In this work, we propose Topology-enhanced Retrieval-Augmented Generation (TopoRAG), a novel framework for textual graph question answering that effectively captures higher-dimensional topological and relational dependencies. Specifically, TopoRAG first lifts textual graphs into cellular complexes to model multi-dimensional topological structures. Leveraging these lifted representations, a topology-aware subcomplex retrieval mechanism is proposed to extract cellular complexes relevant to the input query, providing compact and informative topological context. Finally, a multi-dimensional topological reasoning mechanism operates over these complexes to propagate relational information and guide LLMs in performing structured, logic-aware inference. Empirical evaluations demonstrate that our method consistently surpasses existing baselines across diverse textual graph tasks.
Figures & tables
Figure 1: Illustration of question answering with varying dimensional topological characteristics.
Figure 2: The overview of Topology-enhanced Retrieval-Augmented Framework.
Setting
Method
ExplaGraphs
SceneGraphs
WebQSP
Inference-only
Zero-shot
0.5650
0.3974
41.06
Zero-CoT
0.5704
0.5260
51.30
CoT-BAG
0.5794
0.5680
39.60
KAPING
0.6227
0.4375
52.64
Frozen LLM w/ PT
Prompt tuning
0.5763
0.6341
48.34
GraphToken
0.8508
0.4903
57.05
Table 1: Performance comparison across ExplaGraphs , SceneGraphs , and WebQSP datasets under different configurations. The bold numbers indicate that the improvement of our model over the baselines is statistically significant with (p-value <0.01 ), and the best baseline results are underlined
Method
ExplaGraphs (Accuracy)
WebQSP (Hit)
w/o CRL
0.8576
84.96
w/o TSR
0.8524
84.23
w/o MTR
0.8611
85.46
TopoRAG
0.9151
90.66
Table 2: Ablation Study on ExplaGraphs and WebQSP Datasets.
Figure 3: (a) Effect of layers L∈{2,3,4,5} on TopoRAG performance: WebQSP (Hit) and ExplaGraphs (Accuracy). (b) Effect of top- kk∈{0,1,2,3} on TopoRAG performance: WebQSP (Hit) and SceneGraphs (Accuracy).
Appendix figures & tables5 assets
Supplementary material from the paper’s appendix.
Appendix
Notation
Definition
G
The input textual graph, defined as G=(V,E,{tn}n∈V,{te}e∈E) .
V , E
The sets of vertices (nodes) and edges in the graph G .
tn , te
The text attributes associated with node n∈V and edge e∈E .
X
The regular cell complex constructed from the graph G .
X(d)
d -skeleton.
xk
A k -dimensional cell in the complex (e.g., x0 : a 0-cell, x1 : a 1-cell).
Appendix
Table 3: Notation and definitions used in the TopoRAG framework.
Dataset
ExplaGraphs
SceneGraphs
WebQSP
Number of Graphs
2,766
100,000
4,737
Average Nodes
5.17
19.13
1,370.89
Average Edges
4.25
68.44
4,252.37
Node Features
Commonsense concepts
Object properties ( e.g., color, shape)
Freebase entities
Edge Features
Commonsense relations
Object interactions and spatial relations
Freebase relations
Task Type
Commonsense reasoning
Scene graph QA
Knowledge graph QA
Appendix
Table 4: Overview of datasets used in the GraphQA benchmark.
Method
ExplaGraphs (Accuracy)
SceneGraphs (Accuracy)
WebQSP (Hit)
DFS
0.9151
0.8768
90.66
BFS
0.9025
0.8725
90.18
Random
0.9187
0.8698
90.03
Appendix
Table 5: Performance comparison of different spanning tree algorithms
Dataset
k
Number of Cells per Dimension
0-cells
1-cells
2-cells
WebQSP
1
15
17
2
2
16
18
3
3
17
20
4
SceneGraphs
1
9
12
4
2
9
13
5
Appendix
Table 6: Impact of Top- K Selection on Subcomplex Retrieval.
Setting
Method
ExplaGraphs
SceneGraphs
Hit
Time
Accuracy
Time
Frozen LLM w/ PT
G-Retriever
0.8516
2.6 min/epoch
0.8131
267 min/epoch
TopoRAG
0.8899
3.0 min/epoch
0.8362
300 min/epoch
Tuned LLM
G-Retriever w/ LoRA
0.8705
3.0 min/epoch
0.8683
285 min/epoch
TopoRAG w/ LoRA
0.9151
3.3 min/epoch
0.8768
310 min/epoch
Appendix
Table 7: Performance and Efficiency Comparison of TopoRAG and G-Retriever on ExplaGraphs and SceneGraphs Datasets.
Retrieval-Augmented Generation (RAG) has become a core paradigm for enhancing factual grounding and multi-hop reasoning in Large Language Models (LLMs). Traditional text-based RAG often retrieves logically irrelevant pseudo-evidence, while graph-based RAG is frequently hindered by search-time pruning, which may discard potentially valid reasoning paths. Existing hybrid approaches primarily adopt simple evidence concatenation or unidirectional enhancement, which fails to address the fundamental "Information Island" problem caused by asymmetric reasoning flows between unstructured text and structured graphs. We propose TGS-RAG, a unified framework for Text-Graph Synergistic enhancement. TGS-RAG introduces a bidirectional mechanism: (i) a Graph-to-Text channel that employs a Global Voting strategy from visited graph nodes to re-rank and refine textual evidence, filtering out semantic noise; and (ii) a Text-to-Graph channel that utilizes the Memory-based Orphan Entity Bridging algorithm. This algorithm utilizes textual cues to proactively resurrect valid but previously pruned reasoning paths from the search history without additional database overhead. Experimental results on multiple multi-hop reasoning benchmarks demonstrate that TGS-RAG significantly outperforms state-of-the-art baselines, achieving a superior balance between retrieval precision and computational efficiency.
Jiarui Zhong, Hong Cai Chen
School of Automation, Southeast University, Nanjing 210096, China
Textual graph-based retrieval-augmented generation (GraphRAG) has emerged as a powerful paradigm for enhancing large language models (LLMs) in domain-specific question answering. While existing approaches primarily focus on zero-shot GraphRAG, selecting high-quality demonstrations is crucial for improving reasoning and answer accuracy. Furthermore, recent studies have shown that retrieved subgraphs often contain irrelevant information, which can degrade reasoning performance. In this paper, we propose MixDemo, a novel GraphRAG framework enhanced with a Mixture-of-Experts (MoE) mechanism for selecting the most informative demonstrations under diverse question contexts. To further reduce noise in the retrieved subgraphs, we introduce a query-specific graph encoder that selectively attends to information most relevant to the query. Extensive experiments across multiple textual graph benchmarks show that MixDemo significantly outperforms existing methods.
Retrieval-Augmented Generation (RAG) mitigates hallucinations in Large Language Models (LLMs) by grounding the generation process on external knowledge. However, standard RAG approaches struggle with multi-hop reasoning. While recent graph-based RAG methods improve the retrieval of interconnected chunks, they often rely on computationally expensive and error-prone LLM-based extraction pipelines. To address these issues, we propose TIGRAG (Token-Induced GraphRAG), an efficient graph-augmented RAG framework based on a token co-occurrence Knowledge Graph. TIGRAG directly models topological relationships between tokens using sliding-window co-occurrence statistics, thus enabling scalable graph construction. During inference, it combines graph-based semantic expansion and neural reranking to retrieve interconnected evidence for multi-hop reasoning. Specifically, it introduces an iterative entity-driven retrieval strategy that progressively expands the query using bridging entities extracted from previously retrieved contexts. We evaluated TIGRAG on three widely adopted multi-hop Question Answering (QA) benchmarks. Experimental results demonstrated that our framework consistently outperforms dense retrieval and graph-based RAG methods in both retrieval and downstream QA tasks, while substantially reducing indexing time, inference latency, and prompt footprint.
Gianluca Bonifazi, Christopher Buratti, Michele Marchetti +5
aUniversità Politecnica delle Marche, Ancona, Italy · bUniversità di Modena e Reggio Emilia, Modena, Italy