Semantic Representation Research Articles

Def2Vec: you shall know a word by its definition

AbstractDef2Vec introduces a new perspective on building words embeddings by using dictionary definitions. By leveraging term-document matrices derived from dictionary definitions and employing Latent Semantic Analysis (LSA), our method, Def2Vec, yields embeddings characterized by robust performance and adaptability. Through comprehensive evaluations encompassing token classification, sequence classification and semantic similarity, we show empirically how Def2Vec consistently demonstrates competitiveness with established models like Word2Vec, GloVe, and FastText. Notably, our model’s utilization of all the matrices resulting from LSA factorisation facilitates efficient prediction of embeddings for out-of-vocabulary words, given their definition. By effectively integrating the benefits of dictionary definitions with LSA-based embeddings, Def2Vec builds informative semantic representations, all while minimizing data requirements. In this paper, we run different experiment to assess the quality of our embedding model at word level and at sequence level. Our findings contribute to the ongoing evolution of word embedding methodologies by incorporating structured lexical information and enabling efficient embedding prediction.

Exploring online public survey lifestyle datasets with statistical analysis, machine learning and semantic ontology

Lifestyle diseases significantly contribute to the global health burden, with lifestyle factors playing a crucial role in the development of depression. The COVID-19 pandemic has intensified many determinants of depression. This study aimed to identify lifestyle and demographic factors associated with depression symptoms among Indians during the pandemic, focusing on a sample from Kolkata, India. An online public survey was conducted, gathering data from 1,834 participants (with 1,767 retained post-cleaning) over three months via social media and email. The survey consisted of 44 questions and was distributed anonymously to ensure privacy. Data were analyzed using statistical methods and machine learning, with principal component analysis (PCA) and analysis of variance (ANOVA) employed for feature selection. K-means clustering divided the pre-processed dataset into five clusters, and a support vector machine (SVM) with a linear kernel achieved 96% accuracy in a multi-class classification problem. The Local Interpretable Model-agnostic Explanations (LIME) algorithm provided local explanations for the SVM model predictions. Additionally, an OWL (web ontology language) ontology facilitated the semantic representation and reasoning of the survey data. The study highlighted a pipeline for collecting, analyzing, and representing data from online public surveys during the pandemic. The identified factors were correlated with depressive symptoms, illustrating the significant influence of lifestyle and demographic variables on mental health. The online survey method proved advantageous for data collection, visualization, and cost-effectiveness while maintaining anonymity and reducing bias. Challenges included reaching the target population, addressing language barriers, ensuring digital literacy, and mitigating dishonest responses and sampling errors. In conclusion, lifestyle and demographic factors significantly impact depression during the COVID-19 pandemic. The study’s methodology offers valuable insights into addressing mental health challenges through scalable online surveys, aiding in the understanding and mitigation of depression risk factors.

Design method based on extensible semantic representation algorithm and its application in product packaging design

Design method based on extensible semantic representation algorithm and its application in product packaging design

A Unified Framework for Multi-Domain CTR Prediction via Large Language Models

Multi-Domain Click-Through Rate (MDCTR) prediction is crucial for online recommendation platforms, which involves providing personalized recommendation services to users in different domains. However, current MDCTR models are confronted with the following limitations. Firstly, due to varying data sparsity in different domains, models can easily be dominated by some specific domains, which leads to significant performance degradation in other domains (i.e., the “seesaw phenomenon”). Secondly, when new domain emerges, the scalability of existing methods is limited, making it difficult to adapt to the dynamic growth of the domain. Traditional MDCTR models usually use one-hot encoding for semantic information such as product titles, thus losing rich semantic information and leading to insufficient generalization of the model. In this paper, we propose a novel solution Uni-CTR to address these challenges. Uni-CTR leverages Large Language Model (LLM) to extract layer-wise semantic representations that capture domain commonalities, mitigating the seesaw phenomenon and enhancing generalization. Besides, it incorporates a pluggable domain-specific network to capture domain characteristics, ensuring scalability to dynamic domain growth. Experimental results on public datasets and industrial scenarios show that Uni-CTR significantly outperforms state-of-the-art (SOTA) models. In addition, Uni-CTR shows significant results in zero shot prediction. Code is available at https://anonymous.4open.science/r/multi-domain .

Effective Hard Negative Mining for Contrastive Learning-based Code Search

Background. Code search aims to find the most relevant code snippet in a large codebase based on a given natural language query. An accurate code search engine can increase code reuse and improve programming efficiency. The focus of code search is how to represent the semantic similarity of code and query. With the development of code pre-trained models, the pattern of using numeric feature vectors (embeddings) to represent code semantics and using vector distance to represent semantic similarity has replaced traditional string matching methods. The quality of semantic representations is critical to the effectiveness of downstream tasks such as code search. Currently, the state-of-the-art (SOTA) learning method uses the contrastive learning paradigm. The objective of contrastive learning is to maximize the similarity between matching code and query (positive samples) and minimize the similarity between mismatched pairs (negative samples). To increase the reusing of negative samples, prior contrastive learning approaches use a large queue (memory bank) to store embeddings. Problem. However, there is still a lot of room for improvement in using negative examples for code search: ① Due to the random selection of negative samples, semantic representations learned by existing models cannot distinguish similar codes well. ② Since semantic vectors in the memory bank are reused from previous inference results and then directly used for loss function calculation without gradient descent, the model cannot effectively learn the negative sample semantic information. Method. To solve the above problems, we propose a contrastive learning code search model with hard negative mining called CoCoHaNeRe: ❶ To enable the model to distinguish similar codes, we introduce hard negative examples into contrastive training, which are negative examples in the codebase that are most similar to positive examples. As a result, hard negative examples are most likely to make the model make mistakes. ❷ To improve the learning efficiency of negative samples during training, we add all hard negative examples to the model's gradient descent process. Result. To verify the effectiveness of CoCoHaNeRe, we conducted experiments on large code search datasets with six programming languages, as well as similar retrieval tasks code clone detection and code question answering. Experimental results show that our model achieves SOTA performance. In the code search task, the average MRR score of CoCoHaNeRe exceeds CodeBERT, GraphCodeBERT, and UniXcoder by 11.25%, 8.13%, and 7.38%, respectively. It has also made great progress in code clone detection and code question answering. In addition, our method performs well in different programming languages and code pre-training models. Furthermore, qualitative analysis shows that our model effectively distinguishes high-order semantic differences between similar codes.

Hierarchical Event-RGB Interaction Network for single-eye expression recognition

The Single-eye Expression Recognition task stands as a crucial vision task, aimed at decoding human emotional states through careful examination of the eye region. Nevertheless, traditional cameras face challenges in detecting and capturing relevant biological information, especially under demanding lighting conditions such as dim environments, high exposure scenarios, or when other radiation sources are present. In this regard, we use a new type of sensor data that can resist extreme lighting conditions, namely event camera data, to improve the performance of single-eye expression recognition. To this end, we propose a novel Hierarchical Event-RGB Interaction Network (HI-Net), to fully integrate RGB and event data to overcome the extreme lighting challenges faced by the single-eye expression recognition task. The HI-Net contains two novel designs: Event-RGB Semantic Interaction Mechanism (ER-SIM) and Hierarchical Semantics Modeling (HSM) Scheme. The former aims to achieve interaction between Event and RGB modality semantics, while the latter aims to obtain high-quality modality semantic representations. In the ER-SIM, we employ an effective cross-attention mechanism to facilitate information fusion, to adaptively integrate and complement multi-scale Event and RGB semantics to cope with extreme lighting conditions. In HSM Scheme, we first explore multi-scale contextual semantics for the event modality and the RGB modality respectively. Then, we perform a semantics interaction strategy for these multi-scale contextual semantics, to enhance each modality's semantic representation. Extensive experiments demonstrate that our HI-Net significantly outperforms many state-of-the-art methods on the single-eye expression recognition task, especially under degraded lighting conditions.

Language proficiency is associated with neural representational dimensionality of semantic concepts

Previous studies suggest that semantic concepts are characterized by high-dimensional neural representations and that language proficiency affects semantic processing. However, it is not clear whether language proficiency modulates the dimensional representations of semantic concepts at the neural level. To address this question, the present study adopted principal component analysis (PCA) and representational similarity analysis (RSA) to examine the differences in representational dimensionalities (RDs) and in semantic representations between words in highly proficient (Chinese) and less proficient (English) language. PCA results revealed that language proficiency increased the dimensions of lexical representations in the left inferior frontal gyrus, temporal pole, inferior temporal gyrus, supramarginal gyrus, angular gyrus, and fusiform gyrus. RSA results further showed that these regions represented semantic information and that higher semantic representations were observed in highly proficient language relative to less proficient language. These results suggest that language proficiency is associated with the neural representational dimensionality of semantic concepts.

LLM-enhanced Composed Image Retrieval: An Intent Uncertainty-aware Linguistic-Visual Dual Channel Matching Model

Composed image retrieval (CoIR) involves a multi-modal query of the reference image and modification text describing the desired changes, allowing users to express image retrieval intents flexibly and effectively. The key of CoIR lies in how to properly reason the search intent from the multi-modal query. Existing work either aligns the composite embedding of the multi-modal query and the target image embedding in the visual domain through late-fusion or converts all images into text descriptions and leverage large language models (LLM) for text semantic reasoning. However, this single-modality reasoning approach fails to comprehensively and interpretably capture the users’ ambiguous and uncertain intents in the multi-modal queries, incurring the inconsistency between retrieved results and ground truth. Besides, the expensive manually-annotated datasets limit the further performance improvement of CoIR. To this end, this paper proposes an LLM-enhanced Intent Uncertainty-aware Linguistic-Visual Dual Channel Matching Model (IUDC), which combines the strengths of multi-modal late-fusion and LLMs for composed image retrieval. We first construct an LLM-based triplet augmentation strategy to generate more synthetic training triplets. Based on this, the core of IUDC consists of two matching channels: the semantic matching channel is responsible for intent reasoning on the aspect-level attributes extracted by an LLM, and the visual matching channel accounts for the fine-grained visual matching between multi-modal fusion embedding and target images. Considering the intent uncertainty presented in the multi-modal queries, we introduce Probability Distribution Encoder (PDE) to project the intents as probabilistic distributions in the two matching channels. Consequently, a mutually enhanced module is designed to share knowledge between the visual and semantic representations for better representation learning. Finally, the matching scores of two channels are added to retrieve the target image. Extensive experiments conducted on two real datasets demonstrate the effectiveness and superiority of our model. Notably, with the help of the proposed LLM-based triplet augmentation strategy, our model achieves a new record of state-of-the-art performance among all datasets.

A behaviouristic semantic approach to blockchain-based e-commerce

Electronic commerce and finance are progressively supporting and including decentralized, shared and public ledgers such as the blockchain. This is reshaping traditional commercial activities by advancing them towards Decentralized Finance (DeFi) and Commerce 3.0, thereby supporting the latter’s potential to outpace the hurdles of central authority controllers and lawgivers. The quantity and entropy of the information that must be sought and managed to become active participants in such a relentlessly evolving scenario are increasing at a steady pace. For example, that information comprises asset or service description, general rules of the game, and specific technologies involved for decentralization. Moreover, the relevant information ought to be shared among innumerable and heterogeneous stakeholders, such as producers, buyers, digital identity providers, valuation services, and shipment services, to just name a few. A clear semantic representation of such a complex and multifaceted blockchain-based e-Commerce ecosystem would contribute dramatically to make it more usable, namely more automatically accessible to virtually anyone wanting to play the role of a stakeholder, thereby reducing programmers’ effort. However, we feel that reaching that goal still requires substantial effort in the tailoring of Semantic Web technologies, hence this article sets out on such a route and advances a stack of OWL 2 ontologies for the semantic description of decentralized e-commerce. The stack includes a number of relevant features, ranging from the applicable stakeholders through the supply chain of the offerings for an asset, up to the Ethereum blockchain, its tokens and smart contracts. Ontologies are defined by taking a behaviouristic approach to represent the various participants as agents in terms of their actions, inspired by the Theory of Agents and the related mentalistic notions. The stack is validated through appropriate metrics and SPARQL queries implementing suitable competency questions, then demonstrated through the representation of a real world use case, namely, the iExec marketplace.

Word Learning in Arabic Diglossia in Children With Typical Language Development and Developmental Language Disorder.

Word learning requires the creation of phonological and semantic representations and links in long-term memory. Phonological distance of a given word from the spoken language affects children's lexical-phonological representations and processing. The study investigates the role of the phonological distance of Modern Standard Arabic (StA) words from the child's Spoken Arabic (SpA) vernacular in word learning in Arabic diglossia. It also examines whether, given their vulnerable phonological skills, children with developmental language disorder (DLD) show a stronger impact of phonological distance on word learning than children with typical language development (TLD). One hundred children with TLD and DLD in kindergarten and in first grade (25 per group) were tested on comprehension and production word-learning probes manipulating phonological distance. Learning monosyllabic and disyllabic nonwords encoding only SpA phonemes was compared with the learning of parallel nonwords encoding one unique StA consonant each. Repeated-measures analyses of variance showed higher word learning scores in children with TLD on both probes and for both syllable lengths. Moreover, all children fared significantly lower, in both comprehension and production probes, when the target stimulus was phonologically distant from the spoken language. Finally, an interaction effect was observed on the production probes, revealing differences in the developmental dynamics of phonological distance effects between the groups: Phonological distance hindered word learning among children with TLD in kindergarten, but among children with DLD in the first grade. The results support the role of phonological distance as a phonological complexity factor in word learning in Arabic diglossia. Furthermore, they show that the effect of phonological distance is complex and it interacts with modality, language aptitude, and grade level. The theoretical and clinical implications are discussed.

Heterogeneous graph contrastive learning with adaptive data augmentation for semi‐supervised short text classification

AbstractShort text classification has been widely used in many fields. Due to the scarcity of labelled data, implementing short text classification under semi‐supervised learning setting has become increasingly popular. Semi‐supervised short text classification methods based on graph neural networks can achieve state‐of‐the‐art classification performance by utilizing the expressive power of graph neural networks. However, these methods usually fail to mine the hidden patterns of a large amount of short text node data in the graph to optimize the short text node embeddings, which limits the semantic representation power of the short texts, thus leading to suboptimal classification performance. To overcome the limitation, this paper proposes a novel semi‐supervised short text classification method called the Heterogeneous Graph Contrastive Learning with Adaptive Data Augmentation (HGCLADA). In the knowledge bases guided soft prompt‐based data augmentation component, the related words of the tag words are used to optimize the soft prompts for generating diverse augmented samples. In the heterogeneous graph contrastive learning framework component, a heterogeneous graph that is constructed using short texts and keywords and an effective edge augmentation scheme based on a short text clustering algorithm are proposed. The optimized short text embeddings can be obtained to achieve the effective semi‐supervised short text classification. Extensive experiments on six benchmark datasets show that our HGCLADA method outperforms four classes of state‐of‐the‐art methods in terms of classification accuracy, especially with significant performance improvements of 8.74% on the TagMyNews dataset when each class only contains 20 labelled data.

Multi-granularity self-attention mechanisms for few-shot Learning

Few-shot learning aims to classify novel data categories with limited labeled samples. Although metric-based meta-learning has shown better generalization ability as a few-shot classification method, it still faces challenges in handling data noise and maintaining inter-sample distance stability. To address these issues, our study proposes an innovative few-shot learning approach to enhance image features' global and local semantic representation. Initially, our method employs a multiscale residual module to facilitate extracting multi-granularity features within images. Subsequently, it optimizes the fusion of local and global features using the self- attention mechanism inherent in the Transformer module. Additionally, a weighted metric module is integrated to improve the model's resilience against noise interference. Empirical evaluations on CIFAR-FS and MiniImageNet few-shot datasets using 5-way 1-shot and 5-way 5-shot scenarios demonstrate the effectiveness of our approach in capturing multi-level and multi-granularity image representations. Compared to other methods, our method improves accuracy by 2.63% and 1.27% for 5-shot scenes on these two datasets. The experimental results validate the efficacy of our model in significantly enhancing few-shot image classification performance.

An embedded computational framework of memory: Accounting for the influence of semantic information in verbal short-term memory

We introduce the Embedded Computational Framework of Memory (eCFM), a model that integrates structured semantic word representations with an instance-based memory model to account for the influence of semantic information in verbal short-term memory. The eCFM combines principles from the episodic MINERVA 2 model and the semantic Latent Semantic Analysis model. After reviewing how semantic information impacts verbal short-term memory performance, we demonstrate eCFM’s ability to reconcile various phenomena within a common computational framework. Our model captures key findings, such as the influence of semantic information in serial recall, its reduction in serial reconstruction, and the impact of task difficulty on semantic information. In five experiments, we tested predictions derived from the eCFM. Experiments 1 and 2 manipulated list organization, with Experiment 1 using alternating lists of related or unrelated words and Experiment 2 using blocked lists. Experiment 3 varied presentation rates, Experiment 4 revisited the detrimental effect of semantic information on order information, and Experiment 5 explored false recall. We found that semantic information interacts with list composition, presentation rate affects the magnitude of its influence, and semantic information impacts order information contrary to the dominant view. Additionally, increasing the number of related study words to a non-studied semantic lure boosts false recall. The eCFM captured these findings as well as memory at the item level. Our demonstration provides insight into the cognitive mechanisms underlying verbal short-term memory and the interplay of semantic and episodic memory processes in recall.

The Impact of Lexical-semantic Impairment on Spoken Verb Production in Individuals With Mild Cognitive Impairment.

Although episodic memory is the primary concern in individuals with mild cognitive impairment (MCI), other cognitive functions may also be affected, including language. Language impairment in individuals with MCI has been attributed primarily to the breakdown of semantic representations, difficulties in accessing semantic information, and the weakening of executive functions. However, in most prior studies of word processing in individuals with MCI, researchers have used measures focused on noun production. To investigate how verb production tasks might aid in detecting cognitive impairment in individuals with MCI. We compared the performance of 45 individuals with MCI and 45 healthy controls on action naming and action fluency tasks. In the action naming task, the performance of participants with MCI was significantly impaired compared to healthy controls in terms of total score, the number of semantic errors produced, and the use of generic terms. In the action fluency task, participants with MCI produced significantly fewer verbs, fewer clusters, and fewer switches than healthy controls. The results of our study emphasize the utility of verb production tasks in the identification of cognitive impairment in individuals with MCI and provide evidence of the importance of including action naming and action fluency tasks in the assessment of individuals with MCI.

Probing the link between vision and language in material perception using psychophysics and unsupervised learning.

We can visually discriminate and recognize a wide range of materials. Meanwhile, we use language to describe what we see and communicate relevant information about the materials. Here, we investigate the relationship between visual judgment and language expression to understand how visual features relate to semantic representations in human cognition. We use deep generative models to generate images of realistic materials. Interpolating between the generative models enables us to systematically create material appearances in both well-defined and ambiguous categories. Using these stimuli, we compared the representations of materials from two behavioral tasks: visual material similarity judgments and free-form verbal descriptions. Our findings reveal a moderate but significant correlation between vision and language on a categorical level. However, analyzing the representations with an unsupervised alignment method, we discover structural differences that arise at the image-to-image level, especially among ambiguous materials morphed between known categories. Moreover, visual judgments exhibit more individual differences compared to verbal descriptions. Our results show that while verbal descriptions capture material qualities on the coarse level, they may not fully convey the visual nuances of material appearances. Analyzing the image representation of materials obtained from various pre-trained deep neural networks, we find that similarity structures in human visual judgments align more closely with those of the vision-language models than purely vision-based models. Our work illustrates the need to consider the vision-language relationship in building a comprehensive model for material perception. Moreover, we propose a novel framework for evaluating the alignment and misalignment between representations from different modalities, leveraging information from human behaviors and computational models.

Microstate Analyses to Study face Processing in Healthy Individuals and Psychiatric Disorders: A Review of ERP Findings.

Microstates represent brief periods of quasi-stable electroencephalography (EEG) scalp topography, offering insights into dynamic fluctuations in event-related potential (ERP) topographies. Despite this, there is a lack of a comprehensive systematic overview of microstate findings concerning cognitive face processing. This review aims to summarize ERP findings on face processing using microstate analyses and assess their effectiveness in characterizing face-related neural representations. A literature search was conducted for microstate ERP studies involving healthy individuals and psychiatric populations, utilizing PubMed, Google Scholar, Web of Science, PsychInfo, and Scopus databases. Twenty-two studies were identified, primarily focusing on healthy individuals (n = 16), with a smaller subset examining psychiatric populations (n = 6). The evidence reviewed in this study suggests that various microstates are consistently associated with distinct ERP stages involved in face processing, encompassing the processing of basic visual facial features to more complex functions such as analytical processing, facial recognition, and semantic representations. Furthermore, these studies shed light on atypical attentional neural mechanisms in Autism Spectrum Disorder (ASD), facial recognition deficits among emotional dysregulation disorders, and encoding and semantic dysfunctions in Post-Traumatic Stress Disorder (PTSD). In conclusion, this review underscores the practical utility of ERP microstate analyses in investigating face processing. Methodologies have evolved towards greater automation and data-driven approaches over time. Future research should aim to forecast clinical outcomes and conduct validation studies to directly demonstrate the efficacy of such analyses in inverse space.

HeGCL: Advance Self-Supervised Learning in Heterogeneous Graph-Level Representation.

Representation learning in heterogeneous graphs with massive unlabeled data has aroused great interest. The heterogeneity of graphs not only contains rich information, but also raises difficult barriers to designing unsupervised or self-supervised learning (SSL) strategies. Existing methods such as random walk-based approaches are mainly dependent on the proximity information of neighbors and lack the ability to integrate node features into a higher-level representation. Furthermore, previous self-supervised or unsupervised frameworks are usually designed for node-level tasks, which are commonly short of capturing global graph properties and may not perform well in graph-level tasks. Therefore, a label-free framework that can better capture the global properties of heterogeneous graphs is urgently required. In this article, we propose a self-supervised heterogeneous graph neural network (GNN) based on cross-view contrastive learning (HeGCL). The HeGCL presents two views for encoding heterogeneous graphs: the meta-path view and the outline view. Compared with the meta-path view that provides semantic information, the outline view encodes the complex edge relations and captures graph-level properties by using a nonlocal block. Thus, the HeGCL learns node embeddings through maximizing mutual information (MI) between global and semantic representations coming from the outline and meta-path view, respectively. Experiments on both node-level and graph-level tasks show the superiority of the proposed model over other methods, and further exploration studies also show that the introduction of nonlocal block brings a significant contribution to graph-level tasks.

Differential Mnemonic Contributions of Cortical Representations during Encoding and Retrieval.

Several recent fMRI studies of episodic and working memory representations converge on the finding that visual information is most strongly represented in occipito-temporal cortex during the encoding phase but in parietal regions during the retrieval phase. It has been suggested that this location shift reflects a change in the content of representations, from predominantly visual during encoding to primarily semantic during retrieval. Yet, direct evidence on the nature of encoding and retrieval representations is lacking. It is also unclear how the representations mediating the encoding-retrieval shift contribute to memory performance. To investigate these two issues, in the current fMRI study, participants encoded pictures (e.g., picture of a cardinal) and later performed a word recognition test (e.g., word "cardinal"). Representational similarity analyses examined how visual (e.g., red color) and semantic representations (e.g., what cardinals eat) support successful encoding and retrieval. These analyses revealed two novel findings. First, successful memory was associated with representational changes in cortical location (from occipito-temporal at encoding to parietal at retrieval) but not with changes in representational content (visual vs. semantic). Thus, the representational encoding-retrieval shift cannot be easily attributed to a change in the nature of representations. Second, in parietal regions, stronger representations predicted encoding failure but retrieval success. This encoding-retrieval "flip" in representations mimics the one previously reported in univariate activation studies. In summary, by answering important questions regarding the content and contributions to the performance of the representations mediating the encoding-retrieval shift, our findings clarify the neural mechanisms of this intriguing phenomenon.

BAMRE: Joint extraction model of Chinese medical entities and relations based on Biaffine transformation with relation attention

Electronic Health Records (EHRs) contain various valuable medical entities and their relationships. Although the extraction of biomedical relationships has achieved good results in the mining of electronic health records and the construction of biomedical knowledge bases, there are still some problems. There may be implied complex associations between entities and relationships in overlapping triplets, and ignoring these interactions may lead to a decrease in the accuracy of entity extraction. To address this issue, a joint extraction model for medical entity relations based on a relation attention mechanism is proposed. The relation extraction module identifies candidate relationships within a sentence. The attention mechanism based on these relationships assigns weights to contextual words in the sentence that are associated with different relationships. Additionally, it extracts the subject and object entities. Under a specific relationship, entity vector representations are utilized to construct a global entity matching matrix based on Biaffine transformations. This matrix is designed to enhance the semantic dependencies and relational representations between entities, enabling triplet extraction. This allows the two subtasks of named entity recognition and relation extraction to be interrelated, fully utilizing contextual information within the sentence, and effectively addresses the issue of overlapping triplets.Experimental observations from the CMeIE Chinese medical relation extraction dataset and the Baidu2019 Chinese dataset confirm that our approach yields the superior F1 score across all cutting-edge baselines. Moreover, it offers substantial performance improvements in intricate situations involving diverse overlapping patterns, multitudes of triplets, and cross-sentence triplets.

Rumor Detection With Hierarchical Representation on Bipartite Ad Hoc Event Trees.

The rapid growth of social media has caused tremendous effects on information propagation, raising extreme challenges in detecting rumors. Existing rumor detection methods typically exploit the reposting propagation of a rumor candidate for detection by regarding all reposts to a rumor candidate as a temporal sequence and learning semantics representations of the repost sequence. However, extracting informative support from the topological structure of propagation and the influence of reposting authors for debunking rumors is crucial, which generally has not been well addressed by existing methods. In this article, we organize a claim post in circulation as an ad hoc event tree, extract event elements, and convert it into bipartite ad hoc event trees in terms of both posts and authors, i.e., author tree and post tree. Accordingly, we propose a novel rumor detection model with hierarchical representation on the bipartite ad hoc event trees called BAET. Specifically, we introduce word embedding and feature encoder for the author and post tree, respectively, and design a root-aware attention module to perform node representation. Then we adopt the tree-like RNN model to capture the structural correlations and propose a tree-aware attention module to learn tree representation for the author tree and post tree, respectively. Extensive experimental results on two public Twitter datasets demonstrate the effectiveness of BAET in exploring and exploiting the rumor propagation structure and the superior detection performance of BAET over state-of-the-art baseline methods.