AdaFuse is an adaptive ensemble decoding framework for large language models that dynamically selects fusion units during generation. It uses an uncertainty-based criterion to decide when to ensemble, applying a diversity-aware scaling strategy in uncertain states while continuing direct generation when confident. Experiments on question answering, arithmetic reasoning, and machine translation show AdaFuse outperforms strong baselines with an average relative improvement of 6.88%.
By Chengming Cui, Tianxin Wei, Ziyi Chen, Ruizhong Qiu, Zhichen Zeng, Zhining Liu, Xuying Ning, Duo Zhou, Jingrui He
arXiv:2607. 25579v1 Announce Type: cross Abstract: Entity alignment (EA) identifies entities across knowledge graphs (KGs) that refer to the same real-world object.
By Xinran Liu, Shengtao Li, Shouqian Shi, Ge Wang, Xin-Wei Yao
arXiv:2608. 10149v1 Announce Type: new Abstract: Due to the diversity of real-world time series, no single forecasting model consistently dominates across all samples.
By Xu Zhang, Chang Xu, Hui Sun, Nan Ma, Zijian Zhang, Peng Wang, Wei Wang, Li Zhao
arXiv:2609.37574v1 Announce Type: cross
Abstract: Large Language Models (LLMs) are increasingly used to enrich user queries in information retrieval (IR) so that a standard retriever such as BM25 can...
By Tzu-I Ho, Yung-Yu Shih, Shang-Yu Su, Dongzhe Wang, Yun-Nung Chen
arXiv:2607. 01977v1 Announce Type: new Abstract: Ontology learning (OL) aims to automatically construct structured knowledge models from text, yet progress remains fragmented across methods, domains, and evaluation practices.
By Hamed Babaei Giglou, Jennifer D'Souza, Andrei Aioanei, Nandana Mihindukulasooriya, S\"oren Auer
The paper introduces LLM‑PeerReview, an unsupervised ensemble method that selects the best response from multiple LLM-generated candidates by scoring each answer with several LLMs, aggregating those scores via averaging or a graphical model, and choosing the highest-scoring response. The approach is peer‑review inspired, transparent, and interpretable, and it outperforms the Smoothie‑Global model by 6.9%–7.3% across factual recall QA, math reasoning, and instruction‑following tasks. The authors also provide a curated benchmark suite of 12 ensemble methods evaluated on four datasets and three task families to aid reproducibility.
By Zhijun Chen, Zeyu Ji, Qianren Mao, Hao Wu, Jinhuan Song, Junhang Cheng, Bangjie Qin, Zhuoran Li, Jingzheng Li, Kai Sun, Zizhe Wang, Yikun Ban, Zhu Sun, Xiangyang Ji, Hailong Sun, Xiao Huang
arXiv:2608.25200v2 Announce Type: replace-cross
Abstract: We study learning a mixture of $k$ Plackett-Luce models from multi-way ranking responses from annotators that may represent heterogeneous und...
By Dongyue Li, Ziniu Zhang, Lu Wang, Hongyang R. Zhang
arXiv:2607. 01972v1 Announce Type: cross Abstract: Large language models (LLMs) are often asked to produce JSON conforming to a fixed schema, powering information extraction, tool calling, agentic planning, and knowledge-graph construction.
By Jan Drchal
arXiv:2606. 06109v1 Announce Type: cross Abstract: Entity alignment (EA) aims to identify equivalent entities across heterogeneous knowledge graphs (KGs) and is a key component of knowledge fusion and cross-KG reasoning.
By Xingyu Chen, Yuanning Cui, Zequn Sun, Wei Hu
arXiv:2606. 09653v1 Announce Type: new Abstract: Learned representations across models and modalities often exhibit striking structural similarities, suggesting shared underlying concept decompositions.
By Gr\'egoire Dhimo\"ila, Victor Boutin, Agustin Martin Picard, Thomas Fel, Thomas Serre
The paper proposes a behavior-based fusion model that combines large language models (LLMs) with ontology rankers to improve rare-disease diagnosis. By examining ranked lists, agreement, and ontology support, the model learns how much to rely on each system per case, achieving significant recall gains on Phenopacket Store and RAMEDIS benchmarks. Importantly, the fused diagnoses retain candidate-level ontology evidence for inspection.
By Zhaoyang Jiang, Zhizhong Fu, Yunsoo Kim, Zicheng Li, Xuanqi Peng, Fei Teng, Jiacong Mi, Honghan Wu
The paper introduces a unified framework for aligning biomedical text with knowledge graphs using a lightweight projection learned via contrastive learning, keeping the text encoder and KG embedding model frozen. It evaluates six design choices—text encoder, KG embedding, projection head, triple composition, training direction, and hard‑negative sampling—on a newly created CTD‑Align corpus of 22K chemical‑gene interaction pairs linked to PubMed passages. The study finds that triple composition and training direction have the largest impact, while simpler linear projections over concatenated subject, predicate, and object embeddings yield the best performance.
By Artem Bisliouk, Elizaveta Nosova, Heiko Paulheim, Andreea Iana, Rita T. Sousa