Aug 2026· Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2· 0 citations· 9 references
TL;DR
Results validate that KG-guided reasoning serves as a critical mechanism to compensate for parameter disparity, particularly in tasks requiring deep traversal of biological mechanisms.
Abstract
Biomolecular scientific questions often require multi-step mechanistic reasoning over structured knowledge such as protein–protein interactions, pathways, and disease associations. While large language models (LLMs) can generate chain-of-thought (CoT) rationales, biomolecular CoT is frequently unreliable due to biologically implausible steps and long-horizon inconsistencies, and naive knowledge augmentation remains brittle on large, noisy knowledge graphs. We propose Bio-KCoT, a knowledge-augmented long-CoT framework that emphasizes principled use of structured knowledge. Bio-KCoT transforms knowledge graphs (KGs) into high-fidelity mechanistic reasoning trajectories. Instead of relying on naive shortest paths or direct KG prompting, Bio-KCoT adopts a generative paradigm that synthesizes mechanistic explanations anchored on KG entities and distills them into structured reasoning topologies. These curated chains serve as high-quality target trajectories during Supervised Fine-Tuning (SFT), enabling the model to learn structured mechanistic decomposition. Subsequently, to further enforce logical rigor, we apply Group Relative Policy Optimization (GRPO) with a KG-aligned process reward. This mechanism explicitly scores intermediate steps against the curated evidence, penalizing unsupported hops and reducing hallucinations even when the final answer is correct. To rigorously evaluate these capabilities, we construct BioMolKGQA, a benchmark with curated multi-hop evidence paths spanning diverse biomolecular QA pairs and reasoning depths. Extensive experiments on this dataset confirm that Bio-KCoT delivers substantial performance gains, enabling parameter-constrained models to rival the reasoning fidelity of significantly larger baselines. These results validate that KG-guided reasoning serves as a critical mechanism to compensate for parameter disparity, particularly in tasks requiring deep traversal of biological mechanisms.
VitaGraph is presented, a comprehensive multi-purpose biological knowledge graph built by integrating and refining multiple public datasets and enabling benchmarking of graph-based models and offering the opportunity to tackle tasks such as drug repurposing, PPI prediction, and side-effect prediction, among others.
Francesco Madeddu, Lucia Testa, Gianluca De Carlo et al.· Scientific Data· 0 citations
THGAgents utilizes collaborative and dynamically updating agents to build a Traceable Causal Knowledge Graph, which serves as the foundation for the evidence-based knowledge structure and employs an LLM-driven heuristic search algorithm to traverse the complex network, balancing both novelty and rigor to deduce strict, evidence-based mechanistic causal chains.
Mingjia Yang, Kunhua Dong, K. Lim et al.· 0 citations
Case studies suggest that KLaR can recover biologically plausible disease–gene and drug–target associations missing from the original graphs, supporting hypothesis generation without domain-specific language model fine-tuning.
Adapting large language models (LLMs) to new domains using techniques like pre-training or fine-tuning can be computationally intensive. This challenge has motivated researchers and practitioners to explore alternative strategies, such as Retrieval-based augmentation mechanisms, to reduce training cost while preserving model adaptability. These mechanisms enable LLMs to generate responses grounded in external and previously unseen data. Among these approaches, RetrievalAugmented Generation (RAG) is the widely adopted due to its relatively low computational overhead. However, conventional RAG pipelines exhibit some limitations, including weak semantic coherence across retrieved contexts, limited structural reasoning capability, and the absence of explicit relational representations. To overcome these limitations, recent research has shifted toward the integration of Knowledge Graphs (KGs), which provide structured semantic representations and explicit relational modeling. KG-based approaches looks promising but the typically depend on manually designed ontologies, resulting in scalability constraints and domain adaptation challenges. To address these limitations, we propose OF-BIO-KG, a framework for automated, ontology-free knowledge graph construction. Our approach begins by projecting domain data into a vector database to identify semantically coherent clusters. An LLM then labels these clusters as conceptual nodes and infers both hierarchical and cross-cluster relationships. These nodes and links are persisted in a graph database. By eliminating the need for predefined schemas, our framework offers a scalable, computationally efficient solution for domain-specific LLM adaptation.
Ahlem Ben Younes, Baha Eddine Kalai, Laila Ben Ayed et al.· Annual International Compute...· 0 citations
This work introduces MolBioKG, a two-layer system that grounds unseen molecules in biomedical evidence via multi-resolution structural anchoring and outperforms strong baselines across in-graph link recovery, complex multi-hop reasoning, and out-of-graph generalization.
Yiming Zhang, Hikaru Shindo, Shuan Chen et al.· 0 citations
Recent advances in multimodal large reasoning models (MLRMs) have demonstrated impressive capabilities on complex multimodal tasks, yet their reliance on long Chain-of-Thoughts (CoTs) often leads to redundant reasoning and high computational cost. Existing chain-based distillation and refinement approaches alleviate redundancy to some extent but remain limited when handling ultra-long reasoning chains, suffering from weakened dependency modeling, accumulated redundancy, and the omission of critical reasoning steps. To this end, in this paper, we propose GRACE, a Graph-based Rationalized CoT Compression framework that transforms long CoTs into reasoning graphs to enable efficient CoT compression, where reasoning steps are represented as nodes and semantic dependencies as edges. Specifically, after constructing the reasoning graph, GRACE applies a graph rationale learning approach with a selector–predictor architecture. Among them, the selector estimates node-level rationale scores to sample a compact rationale subgraph, while the predictor leverages this subgraph for final prediction. The resulting subgraph serves as the distilled core rationale, retaining essential reasoning while removing irrelevant steps. Finally, the compressed CoTs are then used to build supervised fine-tuning datasets for further MLRMs tuning. Experiments on multiple multimodal benchmarks show that GRACE substantially improves inference efficiency and interpretability without sacrificing accuracy.
Yizhi Wang, Linan Yue, Deng-Bao Wang et al.· Proceedings of the 32nd ACM...· 0 citations