Large language models (LLMs) have achieved remarkable performance across a wide range of natural language tasks. Understanding how LLMs internally represent knowledge remains a significant challenge. Despite Sparse Autoencoders (SAEs) have emerged as a promising technique for extracting interpretable features from LLMs, SAE features do not inherently align with human-understandable concepts, making their interpretation cumbersome and labor-intensive. To bridge the gap between SAE features and human concepts, we present ConceptViz, a visual analytics system designed for exploring concepts in LLMs. ConceptViz implements a novel Identification ⇒ Interpretation ⇒Validation pipeline, enabling users to query SAEs using concepts of interest, interactively explore concept-to-feature alignments, and validate the correspondences through model behavior verification. We demonstrate the effectiveness of ConceptViz through two usage scenarios and a user study. Our results show that ConceptViz enhances interpretability research by streamlining the discovery and validation of meaningful concept representations in LLMs, ultimately aiding researchers in building more accurate mental models of LLM features. Our code and user guide are publicly available at https://github.com/Happy-Hippo209/conceptViz.

Multi-modal entity alignment aims to identify equivalent entities between two different multi-modal knowledge graphs.Current methods have made significant progress by improving embedding and cross-modal fusion.However, most of them depend on using loss functions to capture the relationship between modalities or adopt a one-time strategy to directly compute modality weights using attention mechanisms, which overlooks the relative interactions between modalities at the entity level and the accuracy of modality weights, thereby hindering the generalization to diverse entities.To address this challenge, we propose RICEA, a relative interaction and calibration framework for multi-modal entity alignment, which dynamically computes weights based on the relative interaction and recalibrates the weights according to their uncertainties.Among these, we propose a novel method called ADC that utilizes attention mechanisms to perceive the uncertainty of the weight for each modality, rather than directly calculating the weight of each modality as in previous works.Across 5 datasets and 23 settings, our proposed framework significantly outperforms other baselines.Our code and data are available at https://github.com/ChenxiaoLi-Joe/RICEA.

Multi-modal entity alignment aims to identify equivalent entities between two different multi-modal knowledge graphs, which consist of structural triples and images associated with entities. Unfortunately, prior works fuse the multi-modal knowledge of all entities only via solely one single fusion strategy. Therefore, the impact of the fusion strategy on individual entities could be largely ignored. To solve this challenge, we propose AMF2SEA, an adaptive multi-modal feature fusion strategy for entity alignment, which dynamically selects the optimal entity-level feature fusion strategy. Additionally, we build a new dataset based on DBP15K, which includes a full set of entity images from multiple inconsistent web sources, making it more representative of the real world. Experimental results demonstrate that our model achieves state-of-the-art (SOTA) performance compared to models using the same modality on DBP15K and its variants with richer image sources and styles. Our code and data are available at https://github.com/ChenxiaoLiJoe/AMFFSEA.

Multi-modal entity alignment (MMEA) aims to identify equivalent entities between two multimodal knowledge graphs (MMKGs).Existing methods have made substantial advancements in enhancing multi-modal fusion.However, the intrinsic noise within modalities, such as the inconsistency in visual modality and redundant attributes, has not been thoroughly investigated.Excessive noise not only weakens semantic representation but also increases the risk of overfitting in attention-based fusion methods.To address this, we propose LGEA (LLM-Guided Entity Alignment), a novel LLM-guided MMEA framework that prioritizes noise reduction before fusion.Specifically, LGEA introduces two key strategies: (1) fine-grained visual filtering to remove irrelevant images at the semantic level, and (2) contextual summarization of attribute information to enhance entity semantics.To our knowledge, we are the first work to apply LLMs for both visual filtering and attribute-level semantic enhancement in MMEA.Experiments on multiple benchmarks, including the noisy FBYG dataset, show that LGEA sets a new state-of-the-art (SOTA) in robust multi-modal alignment, highlighting the potential of noiseaware strategies as a promising direction for future MMEA research 1 .

Music emotion recognition, an important branch of emotion recognition, has been a hotspot of multidisciplinary cross-research. Especially in the automated performing arts industry, the recognition of stage music emotion plays a crucial role in controlling lighting, camera speed, and other elements. However, there are few existing datasets and algorithms specifically designed and optimized for this segment. In this paper, we introduce three convolutional neural network (CNN) models based on the mel frequency cepstral coefficients (MFCC) audio features — Res2Net, ResNetSE, and pre-trained audio neural networks (PANNS) — and innovatively create the dataset SHARD (sadness, happiness, excitement, romance, and drama), which focuses on stage music emotion recognition and contains 2,500 stage music clips with different emotions. Based on this dataset, we compare the recognition effects of the three models on music emotions and select the best-performing ResNetSE model to improve the recognition accuracy from 91.12% to 94.71% by introducing noise enhancement and speech rate perturbation preprocessing techniques. This study provides a new research direction and technical support for stage music emotion recognition.