Medical question answering (QA) systems must deliver accurate, interpretable, and trustworthy responses to support clinical decision-making and patient education. However, existing knowledge-enhanced QA methods often overlook the structural richness of biomedical knowledge graphs (KGs) or delegate all retrieval and reasoning to large language models (LLMs), resulting in hallucinations and significant computational overhead. To address these limitations, we propose SAKGR (Structure-Augmented Knowledge Graph Representations), a novel framework that integrates structured knowledge into LLMbased reasoning. SAKGR constructs structure-aware representations for medical entities by combining semantic embeddings with graph structural features through contrastive learning. These enriched representations enable accurate entity linking and facilitate efficient evidence subgraph retrieval via a hybrid approach that combines path-based search with neighborhood pruning. The extracted subgraphs are then converted into natural language prompts to guide the LLM's reasoning process, effectively incorporating both contextual and structural cues. Experimental results demonstrate that SAKGR significantly enhances retrieval quality and medical QA accuracy while reducing reliance on computationally expensive LLM-driven graph operations, thus offering a scalable and interpretable solution for medical question answering. Our code and data are available on https://github.com/deeomnjobs/sakgr.

Adverse drug reactions (ADRs) can reduce therapeutic efficacy and pose serious risks. Pharmacovigilance monitors and prevents ADRs, but current evaluations rely on manual text assessments, causing inefficiency and potential errors. To address this, we introduce PIESCCA, a Chinese dataset for pharmacovigilance information extraction aligned with the Standardised Case Causality Assessment (SCCA) guidelines from the NMPA. PIESCCA frames causality assessment as four NLP tasks: named entity recognition (NER), relation extraction (RE), event extraction (EE), and question answering (QA). It is curated from over 50,000 real-world reports collected in the past five years, with$2,100+$representative cases selected and annotated under strict quality standards. We benchmark PIESCCA using four LLMs and seven PLM-based models, establishing a foundation for automated pharmacovigilance research. Our code and data are available at https://github.com/deeomnjobs/PIESCCA.

Time series anomaly detection is essential in various fields such as industrial monitoring, cybersecurity, and finance. Traditional supervised methods often face challenges due to the limited availability of labeled anomaly instances for training. Moreover, these methods struggle to deal with intricate systems that incorporate information about topological structures. In this paper, we propose a novel approach called the Partially-Supervised Graph Derivation Network with Meta Learning (PS-GDNML) for time series anomaly detection. PS-GDNML combines the power of graph-based representations, partially-supervised learning, and meta-learning to enhance the effectiveness and robustness of anomaly detection. The method represents time series data as a graph, where each data point is a node, and temporal relationships are captured through edges. By leveraging a graph attention neural network (GAT), the model effectively captures complex dependencies and relationships within the data. To address the scarcity of labeled anomaly instances, PS-GDNML adopts a partially-supervised learning framework. It utilizes both labeled and unlabeled data, enabling the model to learn from the available information and generalize to detect anomalies in unseen data. Additionally, to explore the underlying commonalities of data across different time periods and enhance the model’s adaptability, we adopted a new meta-learning method called Task Relation Meta-Learner (TRMLearner). The purpose of this project is to utilize task relationships to guide the meta-learning optimization process. We evaluated the performance of PS-GDNML on benchmark datasets and compared it with state-of-the-art anomaly detection methods. The experimental results demonstrate that, even with a restricted set of labeled instances, our method excels at accurately detecting anomalies. Furthermore, the meta-learning component enhances the model’s capacity to generalize to novel and evolving anomaly patterns.

Sentiment analysis in various languages has been a hot research topic with several applications. Most of the existing models have been reported to work well with widely used language. Were the lass directly applying these models to poor-quality corpora often leads to low results. Thus, to deal with these shortcoming we propose a cross-lingual sentiment topic model evolution over time (CLSTOT) which jointly models time with topic and sentiment. In CLSTOT, we consider the mapping between sentiment-aware topics under different cultures and analyze their evolution over time. The topic-specific sentiment is extracted using the entire data and not for each single document. As long as providing sentiment-topic, we can predict the timestamps for each test document by finding its most likely location over the timeline. This is achieved by using inference algorithm which is based on Gibbs Sampling. The experimental results on Chinese and English newsreader dataset; Chinese from SinaNews2, and English from Yahoo1, show that CLSTOT achieves significant improvement over the state-of-the-art.

This paper investigates a challenging problem,which is known as fine-grained image classification(FGIC). Different from conventional computer visionproblems, FGIC suffers from the large intraclassdiversities and subtle inter-class differences.Existing FGIC approaches are limited to exploreonly the visual information embedded in the images.In this paper, we present a novel approachwhich can use handy prior knowledge from eitherstructured knowledge bases or unstructured text tofacilitate FGIC. Specifically, we propose a visual-semanticembedding model which explores semanticembedding from knowledge bases and text, andfurther trains a novel end-to-end CNN frameworkto linearly map image features to a rich semanticembedding space. Experimental results on a challenginglarge-scale UCSD Bird-200-2011 datasetverify that our approach outperforms several state-of-the-art methods with significant advances.

Topic models have been widely applied in discovering topics that underly a collection of documents. Incorporating human knowledge can guide conventional topic models to produce topics which are easily interpreted and semantically coherent. Several knowledge-based topic models have been proposed, bu t these models just leverage lexical knowledge of words that are often not in accordance with topics. To solve the problem, we recognize entity mentions, besides words, in the documents and incorporate entity knowledge from external knowledge bases. In this paper, we study to utilize entity knowledge, concepts and categories in Wikipedia, as prior knowledge into topic models to discover more coherent topics. A novel knowledge-based topic model, WCM-LDA (Wikipedia-Category-concept-Mention Latent Dirichlet Allocation), is proposed, which not only models the relationship between words and topics, but also utilizes concept and category knowledge of entities to model the semantic relation of entities and topics. We compare WCM-LDA with the state-of-the-art knowledge-based topic models, on three datasets. Experimental results show that our approach outperforms the existing baseline methods on all three datasets. Moreover, our model can visualize topics with top words, concepts and categories such that topics are made easily to be interpreted and classified.

As more and more multilingual knowledge becomes available on the Web, knowledge sharing across languages has become an important task to benefit many applications. One of the most crucial kinds of knowledge on the Web is taxonomy, which is used to organize and classify the Web data. To facilitate knowledge sharing across languages, we need to deal with the problem of cross-lingual taxonomy alignment, which discovers the most relevant category in the target taxonomy of one language for each category in the source taxonomy of another language. Current approaches for aligning cross-lingual taxonomies strongly rely on domain-specific information and the features based on string similarities. In this paper, we present a new approach to deal with the problem of cross-lingual taxonomy alignment without using any domain-specific information. We first identify the candidate matched categories in the target taxonomy for each category in the source taxonomy using the cross-lingual string similarity. We then propose a novel bilingual topic model, called Bilingual Biterm Topic Model (BiBTM), to perform exact matching. BiBTM is trained by the textual contexts extracted from the Web. We conduct experiments on two kinds of real world datasets. The experimental results show that our approach significantly outperforms the designed state-of-the-art comparison methods.

Joint sentiment/topic models are widely applied in detecting sentiment-aware topics on the lengthy review data and they are achieved with Latent Dirichlet Allocation (LDA) based model. Nowadays plenty of user-generated posts, e.g., tweets and E-commerce short reviews, are published on the social media and the posts imply the public's sentiments (i.e., positive and negative) towards various topics. However, the existing sentiment/topic models are not applicable to detect sentiment-aware topics on the posts, i.e., short texts, because applying the models to the short texts directly will suffer from the context sparsity problem. In this paper, we propose a Time-User Sentiment/Topic Latent Dirichlet Allocation (TUS-LDA) which aggregates posts in the same timeslice or user as a pseudo-document to alleviate the context sparsity problem. Moreover, we design approaches for parameter inference and incorporating prior knowledge into TUS-LDA. Experiments on the Sentiment140 and tweets of electronic products from Twitter7 show that TUS-LDA outperforms previous models in the tasks of sentiment classification and sentiment-aware topic extraction. Finally, we visualize the sentiment-aware topics discovered by TUS-LDA.