Knowledge Base Question Answering (KBQA) aims to answer natural language questions using structured knowledge from KBs. While LLM-only approaches offer generalization, they suffer from outdated knowledge, hallucinations, and lack of transparency. Chain-based KG-RAG methods address these issues by incorporating external KBs, but are limited to simple chain-structured questions due to the absence of planning and logical structuring. Inspired by semantic parsing methods, we propose PDRR: a four-stage framework consisting of Predict, Decompose, Retrieve, and Reason. Our method first predicts the question type and decomposes the question into structured triples. Then retrieves relevant information from KBs and guides the LLM as an agent to reason over and complete the decomposed triples. Experimental results show that our proposed KBQA model, PDRR, consistently outperforms existing methods across different LLM backbones and achieves superior performance on various types of questions.

Yihua Zhu, Qianying Liu, Jiaxin Wang, Fei Cheng, Chaoran Liu, Akiko Aizawa, Sadao Kurohashi, Hidetoshi Shimodaira. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

We present BIS Reasoning 1.0, the first large-scale Japanese dataset of syllogistic reasoning problems explicitly designed to evaluate belief-inconsistent reasoning in large language models (LLMs). Unlike prior resources such as NeuBAROCO and JFLD, which emphasize general or belief-aligned logic, BIS Reasoning 1.0 systematically introduces logically valid yet belief-inconsistent syllogisms to expose belief bias, the tendency to accept believable conclusions irrespective of validity. We benchmark a representative suite of cutting-edge models, including OpenAI GPT-5 variants, GPT-4o, Qwen, and prominent Japanese LLMs, under a uniform, zero-shot protocol. Reasoning-centric models achieve near-perfect accuracy on BIS Reasoning 1.0 (e.g., Qwen3-32B $\approx$99% and GPT-5-mini up to $\approx$99.7%), while GPT-4o attains around 80%. Earlier Japanese-specialized models underperform, often well below 60%, whereas the latest llm-jp-3.1-13b-instruct4 markedly improves to the mid-80% range. These results indicate that robustness to belief-inconsistent inputs is driven more by explicit reasoning optimization than by language specialization or scale alone. Our analysis further shows that even top-tier systems falter when logical validity conflicts with intuitive or factual beliefs, and that performance is sensitive to prompt design and inference-time reasoning effort. We discuss implications for safety-critical domains, including law, healthcare, and scientific literature, where strict logical fidelity must override intuitive belief to ensure reliability.

Optical Coherence Tomography (OCT) plays a vital role in the early detection and monitoring of tractional retinal lesions (TRL), providing high-resolution visualization of retinal structures. However, automated TRL diagnosis remains challenging due to complex lesion morphology, large low-entropy background regions, and the scarcity of high-quality labeled data. Existing Self-Supervised Learning (SSL) approaches often treat all image patches equally, making them sensitive to background noise and limiting their ability to capture fine-grained lesion features. To address these issues, we propose Clustering Hetero-geneous Masked Image Modeling (CH-MIM), a novel SSL frame-work tailored for OCT-based TRL analysis. Our method lever-ages a large-scale clinical dataset containing 11,861 OCT scans collected over five years, including 3,950 expert-annotated images across six TRL severity levels (TO- T5). CH - MIM introduces a Weighted Feature Space Clustering (WFSC) module to selectively mask high-entropy regions, effectively filtering out irrelevant background information. A heterogeneous progressive masking strategy combines binary, Gaussian, and Poisson noise masks to provide diverse, informative reconstruction tasks. Furthermore, a Consistency Regularization Module (CRM) enforces stable predictions across masking branches, improving representation robustness and transferability to downstream classification. Ex-tensive experiments demonstrate that CH - MIM achieves a top-l accuracy of 97.7% and top-5 accuracy of 99.8%, surpassing state-of-the-art supervised and self-supervised baselines. These results highlight the potential of CH - MIM as an effective pretraining strategy for automated TRL screening and its applicability to broader OCT-based retinal disease diagnosis.

Automatic robotic facial expression generation is crucial for human–robot interaction (HRI), as handcrafted methods based on fixed joint configurations often yield rigid and unnatural behaviors. Although recent automated techniques reduce the need for manual tuning, they tend to fall short by not adequately bridging the gap between human preferences and model predictions—resulting in a deficiency of nuanced and realistic expressions due to limited degrees of freedom and insufficient perceptual integration. In this work, we propose a novel learning-to-rank framework that leverages human feedback to address this discrepancy and enhanced the expressiveness of robotic faces. Specifically, we conduct pairwise comparison annotations to collect human preference data and develop the Human Affective Pairwise Impressions (HAPI) model, a Siamese RankNet-based approach that refines expression evaluation. Results obtained via Bayesian Optimization and online expression survey on a 35-DOF android platform demonstrate that our approach produces significantly more realistic and socially resonant expressions of Anger, Happiness, and Surprise than those generated by baseline and expert-designed methods. This confirms that our framework effectively bridges the gap between human preferences and model predictions while robustly aligning robotic expression generation with human affective responses.

Semi-supervised medical image segmentation has shown promise in training models with limited labeled data and abundant unlabeled data. However, state-of-the-art methods ignore a potentially valuable source of unsupervised semantic information-spatial registration transforms between image volumes. To address this, we propose CCT-R, a contrastive cross-teaching framework incorporating registration information. To leverage the semantic information available in registrations between volume pairs, CCT-R incorporates two proposed modules: Registration Supervision Loss (RSL) and Registration-Enhanced Positive Sampling (REPS). The RSL leverages segmentation knowledge derived from transforms between labeled and unlabeled volume pairs, providing an additional source of pseudo-labels. REPS enhances contrastive learning by identifying anatomically-corresponding positives across volumes using registration transforms. Experimental results on two challenging medical segmentation benchmarks demonstrate the effectiveness and superiority of CCT-R across various semi-supervised settings, with as few as one labeled case. Our code is available at https://github.com/kathyliu579/ContrastiveCross-teachingWithRegistration.

Previous studies have introduced a weaklysupervised paradigm for solving math word problems requiring only the answer value annotation.While these methods search for correct value equation candidates as pseudo labels, they search among a narrow sub-space of the enormous equation space.To address this problem, we propose a novel search algorithm with combinatorial strategy ComSearch, which can compress the search space by excluding mathematically equivalent equations.The compression allows the searching algorithm to enumerate all possible equations and obtain high-quality data.We investigate the noise in the pseudo labels that hold wrong mathematical logic, which we refer to as the false-matching problem, and propose a ranking model to denoise the pseudo labels.Our approach holds a flexible framework to utilize two existing supervised math word problem solvers to train pseudo labels, and both achieve state-of-the-art performance in the weak supervision task. 1

In spite of the potential for ground-breaking achievements offered by large language models (LLMs) (e.g., GPT-3) via in-context learning (ICL), they still lag significantly behind fullysupervised baselines (e.g., fine-tuned BERT) in relation extraction (RE).This is due to the two major shortcomings of ICL for RE: (1) low relevance regarding entity and relation in existing sentence-level demonstration retrieval approaches for ICL; and (2) the lack of explaining input-label mappings of demonstrations leading to poor ICL effectiveness.In this paper, we propose GPT-RE to successfully address the aforementioned issues by (1) incorporating task-aware representations in demonstration retrieval; and (2) enriching the demonstrations with gold label-induced reasoning logic.We evaluate GPT-RE on four widely-used RE datasets and observe that GPT-RE achieves improvements over not only existing GPT-3 baselines, but also fully-supervised baselines as in Figure 1.Specifically, GPT-RE achieves SOTA performances on the Semeval and SciERC datasets, and competitive performances on the TACRED and ACE05 datasets.Additionally, a critical issue of LLMs revealed by previous work, the strong inclination to wrongly classify NULL examples into other predefined labels, is substantially alleviated by our method.We show an empirical analysis.1

Low-resource speech recognition has been long-suffering from insufficient training data. In this paper, we propose an approach that leverages neighboring languages to improve low-resource scenario performance, founded on the hypothesis that similar linguistic units in neighboring languages exhibit comparable term frequency distributions, which enables us to construct a Huffman tree for performing multilingual hierarchical Softmax decoding. This hierarchical structure enables cross-lingual knowledge sharing among similar tokens, thereby enhancing low-resource training outcomes. Empirical analyses demonstrate that our method is effective in improving the accuracy and efficiency of low-resource speech recognition.

For medical image semantic segmentation (MISS), Vision Transformers have emerged as strong alternatives to convolutional neural networks thanks to their inherent ability to capture long-range correlations. However, existing research uses off-the-shelf vision Transformer blocks based on linear projections and feature processing which lack spatial and local context to refine organ boundaries. Furthermore, Transformers do not generalize well on small medical imaging datasets and rely on large-scale pre-training due to limited inductive biases. To address these problems, we demonstrate the design of a compact and accurate Transformer network for MISS, CS-Unet, which introduces convolutions in a multi-stage design for hierarchically enhancing spatial and local modeling ability of Transformers. This is mainly achieved by our well-designed Convolutional Swin Transformer (CST) block which merges convolutions with Multi-Head Self-Attention and Feed-Forward Networks for providing inherent localized spatial context and inductive biases. Experiments demonstrate CS-Unet without pre-training out- performs other counterparts by large margins on multi-organ and cardiac datasets with fewer parameters and achieves state-of-the-art performance. Our code is available at Github1.

Relation extraction (RE) has achieved remarkable progress with the help of pre-trained language models.However, existing RE models are usually incapable of handling two situations: implicit expressions and long-tail relation types, caused by language complexity and data sparsity.In this paper, we introduce a simple enhancement of RE using k nearest neighbors (kNN-RE).kNN-RE allows the model to consult training relations at test time through a nearest-neighbor search and provides a simple yet effective means to tackle the two issues above.Additionally, we observe that kNN-RE serves as an effective way to leverage distant supervision (DS) data for RE.Experimental results show that the proposed kNN-RE achieves state-of-the-art performances on a variety of supervised RE datasets, i.e., ACE05, SciERC, and Wiki80, along with outperforming the best model to date on the i2b2 and Wiki80 datasets in the setting of allowing using DS.Our code and models are available at: https://github.com/YukinoWan/kNN-RE.

With the advent of the General Data Protection Regulation (GDPR) and increasing privacy concerns, the sharing of speech data is faced with significant challenges. Protecting the sensitive content of speech is the same important as the voiceprint. This paper proposes an effective speech content protection method by constructing a frame-by-frame adversarial speech generation system. We revisited the adversarial examples generating method in the recent machine learning field and selected the phonetic state sequence of sensitive speech for the adversarial examples generation. We build an adversarial speech collection. Moreover, based on the speech collection, we proposed a neural network-based frame-by-frame mapping method to recover the speech content by converting from the adversarial speech to the human speech. Experiment shows our proposed method can encode and recover any sensitive audio, and our method is easy to be conducted with publicly available resources of speech recognition technology.

This study proposes a general multiple parameter control framework by leveraging the ability of a reinforcement learning system to learn empirical knowledge for evolutionary computation. We design a feedback evaluation mechanism to define the rewards offered to agents, using which they can learn to choose appropriate parameters in formulated action sets. Moreover, a learning strategy is proposed to utilize the parameter selection-related knowledge that is gained during training episodes. Three famous evolutionary computation (EC) methods (i.e., particle swarm optimization, artificial bee colony, and differential evolution) are selected as the baseline algorithms and applied to the proposed framework. The aforementioned redesigned algorithms are tested on 15 common benchmark functions, as well as the CEC2017 benchmarks. In addition, the robustness of the algorithms is demonstrated through parameter sensitivity analysis. The results of the comparative analysis reveal that the three improved algorithms exhibit a faster overall convergence and higher accuracy than their state-of-the-art variants. It is also confirmed that our proposed framework has the capability to improve the performance of EC approach.

Automatically solving math word problems is a critical task in the field of natural language processing. Recent models have reached their performance bottleneck and require more high-quality data for training. We propose a novel data augmentation method that reverses the mathematical logic of math word problems to produce new high-quality math problems and introduce new knowledge points that can benefit learning the mathematical reasoning logic. We apply the augmented data on two SOTA math word problem solving models and compare our results with a strong data augmentation baseline. Experimental results show the effectiveness of our approach (we release our code and data athttps://github.com/yiyunya/RODA).

In the customer segmentation problem, a large number of features are manually designed and used to comprehensively describe the customer instances. However, some of these features are irrelevant, redundant, and noisy, which are not necessary and effective for customer segmentation. Feature selection is an important data preprocessing method by selecting important features from the original feature set. Particularly, feature selection in customer segmentation is a multiobjective problem that aims to minimize the feature number and maximize the classification performance. This paper proposes a multiobjective feature-selection method based on a meta-heuristic algorithm—hydrological cycling optimization (HCO)—to solve customer segmentation. The proposed method is able to automatically evolve a set of non-dominated solutions that select small numbers of features and achieve high classification accuracy. To this end, three strategies based on the global flow operator, possibility-based acceptance criteria, and density-based evaporation and precipitation are proposed to improve the global search ability and the solution diversity of the proposed approach. The performance of the proposed approach is examined on three customer-segmentation datasets and compared with original multiobjective HCO and six well-known evolutionary multiobjective algorithms. The results confirm the superiority of the proposed approach in solving multiobjective customer-segmentation problems by achieving higher calculation stability, search diversity, and solution quality compared with the other competing methods.

Joint extraction of entities and relations has received significant attention due to its potential of providing higher performance for both tasks. Among existing methods, CopyRE is effective and novel, which uses a sequence-to-sequence framework and copy mechanism to directly generate the relation triplets. However, it suffers from two fatal problems. The model is extremely weak at differing the head and tail entity, resulting in inaccurate entity extraction. It also cannot predict multi-token entities (e.g. Steven Jobs). To address these problems, we give a detailed analysis of the reasons behind the inaccurate entity extraction problem, and then propose a simple but extremely effective model structure to solve this problem. In addition, we propose a multi-task learning framework equipped with copy mechanism, called CopyMTL, to allow the model to predict multi-token entities. Experiments reveal the problems of CopyRE and show that our model achieves significant improvement over the current state-of-the-art method by 9% in NYT and 16% in WebNLG (F1 score). Our code is available at https://github.com/WindChimeRan/CopyMTL

Building the work timetables for staff in healthcare institutions is known to be a highly constrained and NP-hard problem. In this research, a mathematical programming model, maximizing nurses' preference for work shifts and rest days while minimizing hospital operating costs, is proposed to solve the nurse scheduling problem (NSP) optimally. Then, we apply a new optimization algorithm-HCOMA, HCO based memetic algorithm, combining entropy-based decision-making mechanism and local search, to heuristically solve the NSP. In the global search, the entropy is calculated to assess population diversity following by every specified iteration. By analyzing the change of diversity, the population can identify the stagnation of search and perform local search at the best time. In summary, the local search includes three core parts: Meta-Lamarckian learning strategy, cooling schedule and Metropolis Criterion. Three neighborhood structures are utilized to exchange or reset the nurse's shifts, expanding the feasible solution area of the search and generating high-quality solutions. The Meta-Lamarckian learning strategy is used to automatically choose the best search structure based on their performance. The performance of HCOMA was tested with sufficient experimentations. The test problems were generated based on the actual situation of a hospital, including an instance and 30 random problems. The results indicate that the proposed algorithm was superior to the standard HCO and three well-known evolutionary algorithms in solution quality and convergence rate.

Qianying Liu, Wenyv Guan, Sujian Li, Daisuke Kawahara. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.