Semantic Patterns Research Articles

Accessible Stories within Mediascapes

This article presents the first steps in the investigation of the potential for digital storytelling and digital museums to be used as instruments for access, as enablers of epistemic and poietic agency. Digital storytelling and migration museums are used as a case study to explore in what ways digital storytelling impacts meaning-making processes performed by migrants, allowing them to become active creators and disseminators of their own experiences. Through the combination of corpus linguistics, systemic functional linguistics, and lexical semantic analysis, an ad-hoc comparable corpus of migrant narratives in English and in Italian was cross-examined in order to scrutinise the conceptual categories activated in these stories. The mixed-methods analysis led to the identification of the most common lexico-semantic features forming the knowledge frames of the experiential world of migrants. Results show that each national narrative makes use of its own grammar and lexico-semantic domains, a set of semantic and syntactic patterns associated with the production of stories of migration within the context of transmedia textual subtypes. These domains are strategic, as they give access to marginalised stories within digital museum settings. Lay summary Adopting the perspective of accessibility studies, digital storytelling is interpreted as a series of instruments for access, as an enabler of epistemic and poietic agency. That is, digital storytelling empowers people to express their voice and become active creators and disseminators of their own experiences. In order to investigate the potential of digital storytelling as a tool for access, we created a comparable corpus made of migrant narratives in English and in Italian spread across museum platforms. Then we cross-examined the corpus through a combination of quantitative and qualitative methodologies in order to examine the conceptual categories activated in these stories. More specifically, we analysed and identified which linguistic strategies have been employed by the storytellers to inform the audience, what knowledge frames have been adopted to tell their experiences and expectations, and what similarities or differences exist among stories told by migrants who have settled down in different parts of the world. Showing that each national narrative makes use of its own semantic and syntactic patterns, our results support the idea of digital storytelling as an access enabler and digital museums as novel environments which jointly empower people with the tools and conditions to take a proactive role.

Rumor detection with self-supervised learning on texts and social graph

Rumor detection has become an emerging and active research field in recent years. At the core is to model the rumor characteristics inherent in rich information, such as propagation patterns in social network and semantic patterns in post content, and differentiate them from the truth. However, existing works on rumor detection fall short in modeling heterogeneous information, either using one single information source only (e.g., social network, or post content) or ignoring the relations among multiple sources (e.g., fusing social and content features via simple concatenation). Therefore, they possibly have drawbacks in comprehensively understanding the rumors, and detecting them accurately. In this work, we explore contrastive self-supervised learning on heterogeneous information sources, so as to reveal their relations and characterize rumors better. Technically, we supplement the main supervised task of detection with an auxiliary self-supervised task, which enriches post representations via post self-discrimination. Specifically, given two heterogeneous views of a post (i.e., representations encoding social patterns and semantic patterns), the discrimination is done by maximizing the mutual information between different views of the same post compared to that of other posts. We devise cluster-wise and instance-wise approaches to generate the views and conduct the discrimination, considering different relations of information sources. We term this framework as self-supervised rumor detection (SRD). Extensive experiments on three real-world datasets validate the effectiveness of SRD for automatic rumor detection on social media.

Deficits in Different Cognitive Domains Predict the Progression of Amnestic Mild Cognitive Impairment: A Literature Review

Introduction: Mild cognitive impairment (MCI) is a heterogeneous syndrome in which older adults show cognitive deficits that do not interfere with daily living. Amnestic mild cognitive impairment (aMCI) is a subtype of MCI where episodic memory is significantly impaired and considered the transition stage between normal aging and Alzheimer’s disease (AD). The cognitive profiles of individuals with aMCI may predict various trajectories and inform the risk of AD conversion. Given that cognitive domains beyond memory including language and executive functioning (EF) may contribute to the progression of aMCI, this paper will examine how deficits in these three domains can be used to predict trajectories of the syndrome. Methods: PUBMED, EMBASE, and CINAHL databases were used to screen for studies to construct this systematic review. A total of 20 studies were reviewed. Results: Functional changes in memory were observed including the diminished performance in the encoding and recognition phases of episodic memory, associative recall, rapid forgetting, and pattern separation in individuals with aMCI relative to age-matched peers. Overall EF and all three core EF components studied were impaired to similar extents. Dynamic working memory and verbal memory performance was impaired to a greater extent in individuals with multiple- domain aMCI (a more severe subtype of aMCI) relative to single-domain aMCI. Language impairments were associated with AD pathology, including verbal fluency deficits and semantic verbal fluency patterns, all of which were further impaired in individuals with multiple-domain aMCI relative to single-domain aMCI. Discussion: Structural and functional changes in the medial temporal lobe (MTL) underlie various deficits in memory, EF, and language domains. Research suggests that individuals with single-domain aMCI perform more similarly with healthy controls, while those with multiple-domain aMCI perform more similarly with individuals with AD. This is likely attributed to cognitive domains beyond memory having a drastic impact on aMCI pathogenesis. Conclusion: Preliminary evidence offers support for a more nuanced use of cognitive profiles to determine future outcomes and take appropriate clinical action earlier for improved prognosis and therapeutic plan development. Earlier formulation of treatment methods could drastically increase the likelihood of slowing or even reversing aMCI pathogenesis.

Expressing Conflicts. Semantic Patterns Typifying Letters in the British Press

Expressing Conflicts. Semantic Patterns Typifying Letters in the British Press

NI-UDA: Graph Contrastive Domain Adaptation for Nonshared-and-Imbalanced Unsupervised Domain Adaptation

With the technology development, information networks continuously generate a large amount of integrated labeled Big Data. Some types of labeled data in real scenes are scarce and difficult to obtain, such as some aerospace data. It is important to address the problem of nonshared and imbalanced unsupervised domain adaptation (NI-UDA) from the labeled Big Data with nonshared and long-tail distribution to unlabeled specified small and imbalanced space applications, where nonshared classes mean the label space out of the target domain. Previous methods proposed to integrate the semantic knowledge of Big Data to help the unsupervised domain adaptation for sparse data. However, they have the challenges of limited effect of knowledge sharing for long-tail Big Data and the imbalanced domain adaptation. To solve them, our goal is to leverage priori hierarchy knowledge to enhance domain contrastive aligned feature representation with graph reasoning. Our method consists of hierarchy graph reasoning (HGR) layer and K-positive contrastive domain adaptation (K-CDA). Our HGR contributes to learn direct semantic patterns for sparse classes by hierarchy attention in self-attention, nonlinear mapping, and graph normalization. For alleviating imbalanced domain adaptation, we proposed K-CDA, which explores k-positive instances for each class to every mini-batch with contrastive learning to align imbalanced feature representations. Compared with the previous contrastive UDA, our K-CDA alleviates the problems of large memory consumption and high computational cost. Experiments on three benchmark datasets shows our methods consistently improve the state-of-the-art contrastive UDA algorithms.

О семантике славянских существительных с префиксом *na- (к проблеме влияния отглагольного именного словообразования на отыменное)

The prevailing semantics of Russian and Slavic nouns with the prefix na- is determined by the primary function of the prefix — it designates an object as located on top of the root noun: Slav. *narǫka ‘bracelet’ (lit. ‘thing on top of hand’). However, another type of correlation between the semantics of the prefixed noun and the root noun is possible: the root noun in this case denotes the object itself, which, thanks to the preceding prefix, is characterized as located on top, so that the derived noun corresponds semantically to a reverse arrangement of the root noun and prefix — ‘object (which) is on top’, see Slav. *naledъ ‘ice-crusted ground’ (lit. ‘ice that is on top’). The emergence of this semantic pattern of denominative nouns with the prefix *na- can be explained by the patterning influence of nouns derived from prefixed verbs: cf. Slav. *nasypati ‘pour on/into’ → *nasypь ‘embankment’.

Characterizing data patterns with core–periphery network modeling

Traditional classification techniques usually classify data samples according to the physical organization, such as similarity, distance, and distribution, of the data features, which lack a general and explicit mechanism to represent data classes with semantic data patterns. Therefore, the incorporation of data pattern formation in classification is still a challenge problem. Meanwhile, data classification techniques can only work well when data features present high level of similarity in the feature space within each class. Such a hypothesis is not always satisfied, since, in real-world applications, we frequently encounter the following situation: On one hand, the data samples of some classes (usually representing the normal cases) present well defined patterns; on the other hand, the data features of other classes (usually representing abnormal classes) present large variance, i.e., low similarity within each class. Such a situation makes data classification a difficult task. In this paper, we present a novel solution to deal with the above mentioned problems based on the mesostructure of a complex network, built from the original data set. Specifically, we construct a core–periphery network from the training data set in such way that the normal class is represented by the core sub-network and the abnormal class is characterized by the peripheral sub-network. The testing data sample is classified to the core class if it gets a high coreness value; otherwise, it is classified to the periphery class. The proposed method is tested on an artificial data set and then applied to classify x-ray images for COVID-19 diagnosis, which presents high classification precision. In this way, we introduce a novel method to describe data pattern of the data “without pattern” through a network approach, contributing to the general solution of classification.

Linguistic patterns in the lexical-semantic subsystem of new public administration: typology and features

The authors analyzed the confeptosphere of public administration from the position of representation in the Rus-sian language as an area of increased interest from the controlling structures in the field of state security. The au-thors relied on the postulates of the presence of three dimensions inherent in verbal signs: due to the indirect connec-tions of words and objects of semantics, determined by the relationship of words to each other syntagmatics and due to the relationship of words and communicants of pragmatics; about the discreteness and continuity of verbal signs, about the relative simplicity and minimal variability of intersystem relations between them, about the limited motivation of verbal signs and the inverse proportion of morphemic complexity and semantic structure of the word. The focus of attention is on studying the influence of language patterns on the functioning of the “dominant” lexi-cal-semantic subsystem of public administration in the Russian language of the recent period, associated with the ideas about the specifics of public administration focused on ensuring law and order and constitutionally estab-lished rights and freedoms of citizens. The paper concludes that lexical and semantic patterns are important in de-termining the main trends in the development of the lexical system of the Russian language in general, individual subsystems and language signs in particular.

Generative Memory-Guided Semantic Reasoning Model for Image Inpainting

The critical challenge of single image inpainting stems from accurate semantic inference via limited information while maintaining image quality. Typical methods for semantic image inpainting train an encoder-decoder network by learning a one-to-one mapping from the corrupted image to the inpainted version. While such methods perform well on images with small corrupted regions, it is challenging for these methods to deal with images with large corrupted area due to two potential limitations. 1) Such one-to-one mapping paradigm tends to overfit each single training pair of images; 2) The inter-image prior knowledge about the general distribution patterns of visual semantics, which can be transferred across images sharing similar semantics, is not explicitly exploited. In this paper, we propose the Generative Memory-guided Semantic Reasoning Model (GM-SRM), which infers the content of corrupted regions based on not only the known regions of the corrupted image, but also the learned inter-image reasoning priors characterizing the generalizable semantic distribution patterns between similar images. In particular, the proposed GM-SRM first pre-learns a generative memory from the whole training data to explicitly learn the distribution of different semantic patterns. Then the learned memory are leveraged to retrieve the matching semantics for the current corrupted image to perform semantic reasoning during image inpainting. While the encoder-decoder network is used for guaranteeing the pixel-level content consistency, our generative priors are favorable for performing high-level semantic reasoning, which is particularly effective for inferring semantic content for large corrupted area. Extensive experiments on Paris Street View, CelebA-HQ, and Places2 benchmarks demonstrate that our GM-SRM outperforms the state-of-the-art methods for image inpainting in terms of both visual quality and quantitative metrics.

A Corpus-Based Study on the Semantic Use of Reporting Verbs in English Majors’ Undergraduate Thesis Writing

This study aims to add further pedagogical knowledge on students’ academic writing by investigating the semantic patterns of reporting verbs (RVs) in L1 Chinese undergraduate English majors’ theses in a southern Chinese university based on the semantic categories by Hunston et al. (1996) and Charles (2006a). A comparative analysis was conducted across L2 and L1 students’ academic writing in the discipline of applied linguistics. The study yielded two major findings: 1) there was a significantly insufficient employment of RVs in general, particularly among three categories (Argue, Show, Find) by L2 students, who also presented a strong reliance on argumentation by intuition; 2) L2 students illustrated a restricted vocabulary repertoire of colloquial RVs and their usage of RVs was misrepresented in context, diverging from the intended rhetorical functions. These findings indicate that evidence-based argumentative writing practice and targeted lexical and rhetorical instructions on vocabulary knowledge require further promotion in L2 English learners’ academic writing training.

Distilling Knowledge From Object Classification to Aesthetics Assessment

In this work, we point out that the major dilemma of image aesthetics assessment (IAA) comes from the abstract nature of aesthetic labels. That is, a vast variety of distinct contents can correspond to the same aesthetic label. On the one hand, during inference, the IAA model is required to relate various distinct contents to the same aesthetic label. On the other hand, when training, it would be hard for the IAA model to learn to distinguish different contents merely with the supervision from aesthetic labels, since aesthetic labels are not directly related to any specific content. To deal with this dilemma, we propose to distill knowledge on semantic patterns for a vast variety of image contents from multiple pre-trained object classification (POC) models to an IAA model. Expecting the combination of multiple POC models can provide sufficient knowledge on various image contents, the IAA model can easier learn to relate various distinct contents to a limited number of aesthetic labels. By supervising an end-to-end single-backbone IAA model with the distilled knowledge, the performance of the IAA model is significantly improved by 4.8% in SRCC compared to the version trained only with ground-truth aesthetic labels. On specific categories of images, the SRCC improvement brought by the proposed method can achieve up to 7.2%. Peer comparison also shows that our method outperforms 10 previous IAA methods.

Systematically Detecting Patterns of Social, Historical and Linguistic Change: The Framing of Poverty in Times of Poverty

AbstractThe linguistic DNA project seeks to understand the evolution of philosophy, society and language during the Modern English period. Corpora like Early English Books Online (EEBO), Corpus of Late Modern English Texts (CLMET) and Corpus of Historical American English (COHA) allow us to apply statistical data‐driven models extracting patterns to confirm our expectations. As systems biology has revolutionised biology by systematically searching for all patterns, we detect patterns in our data systematically with contextual and distributional semantic approaches, an approach that could be called systems history. We uncover semantic patterns with methods from text mining, computational linguistics and digital humanities. We normalise the spelling automatically to present‐day variants and use bottom‐up analyses to step from words to concepts: collocations, topic modelling and distributional semantics. We illustrate the approaches with two case studies: associations of poverty changing across time, and Charles Dickens social criticism, his vision of helping to improve the situation of the poor. As no gold standard for our task exists, our approaches are exploratory, which entails considerable manual intervention, e.g. sifting candidate lists, reading excerpts and interpreting topic models. A fully automatic approach is currently neither feasible nor envisaged: semi‐automatic approaches give researchers the inspiring opportunity to interact with the texts in a constant move between distant and close reading. The different characteristics of the various statistical methods offer complementary perspectives.

Creativity in verbal associations is linked to semantic control.

Although memory is known to play a key role in creativity, previous studies have not isolated the critical component processes and networks. We asked participants to generate links between words that ranged from strongly related to completely unrelated in long-term memory, delineating the neurocognitive processes that underpin more unusual versus stereotypical patterns of retrieval. More creative responses to strongly associated word-pairs were associated with greater engagement of episodic memory: in highly familiar situations, semantic, and episodic stores converge on the same information enabling participants to form a personal link between items. This pattern of retrieval was associated with greater engagement of core default mode network (DMN). In contrast, more creative responses to weakly associated word-pairs were associated with the controlled retrieval of less dominant semantic information and greater recruitment of the semantic control network, which overlaps with the dorsomedial subsystem of DMN. Although both controlled semantic and episodic patterns of retrieval are associated with activation within DMN, these processes show little overlap in activation. These findings demonstrate that controlled aspects of semantic cognition play an important role in verbal creativity.

Semantic models for IoT sensing to infer environment–wellness relationships

Every time an Internet of Things (IoT) solution is deployed, every time a smartphone owner connects her/his wireless device to a wearable activity-tracker, every time groups of citizens use geo-mapping applications to move around the city, choosing the least crowded path, data are produced and information have to be exchanged appropriately via APIs. Even if novel added-value IoT-based applications appear on the market with increasing speed, true semantic interoperability is far from being achieved, thus limiting the large-scale exploitation, the scalability and the time-to-market of novel apps. Currently, connecting different data prosumers with multiple data sources is still hampered by the lack of standardized and sustainable solutions, especially due to the significant heterogeneity of IoT platforms. In such a landscape, ontologies come to the rescue, thanks to their formal semantics, knowledge representation formats, and shared vocabularies. In this paper we examine, from an ontological perspective, how to describe environmental sensing and wellness monitoring, two of the most popular application cases of Mobile Crowd Sensing (MCS) and IoT, respectively. To this purpose, an ontology of sensor-agnostic APIs is proposed, along with a set of MCS-dedicated ontology modules (and the supporting platform), leveraging on standard and reusable domain ontologies. Moreover, it will be shown how to properly combine the proposed ontologies in order to support complex functionalities based on inference rules addressing the environment–wellness relationships. Finally, specific semantic modeling patterns suitable for typical IoT and MCS scenarios will be discussed.

A Novel Hybrid Attention-Driven Multistream Hierarchical Graph Embedding Network for Remote Sensing Object Detection

Multiclass geospatial object detection in high-spatial-resolution remote-sensing images (HSRIs) has recently attracted considerable attention in many remote-sensing applications as a fundamental task. However, the complexity and uncertainty of spatial distribution among multiclass geospatial objects are still huge challenges for object detection in HSRIs. Most current remote-sensing object-detection approaches fall back on deep convolutional neural networks (CNNs). Nevertheless, most existing methods only focus on mining visual characteristics and lose sight of spatial or semantic relation discriminations, eventually degrading object-detection performance in HSRIs. To tackle these challenges, we propose a novel hybrid attention-driven multistream hierarchical graph embedding network (HA-MHGEN) to explore complementary spatial and semantic patterns for improving remote-sensing object-detection performance. Specifically, we first constructed hierarchical spatial graphs for multiscale spatial relation representation. Then, semantic graphs were also constructed by integrating them with the word embedding of object category labels on graph nodes. Afterwards, we developed a self-attention-aware multiscale graph convolutional network (GCN) to derive stronger for intra- and interobject hierarchical spatial relations and contextual semantic relations, respectively. These two relation networks were followed by a novel cross-attention-driven spatial- and semantic-feature fusion module that utilizes a multihead attention mechanism to learn associations between diverse spatial and semantic correlations, and guide them to endowing a more powerful discrimination ability. With the collaborative learning of the three relation networks, the proposed HA-MHGEN enables grasping explicit and implicit relations from spatial and semantic patterns, and boosts multiclass object-detection performance in HRSIs. Comprehensive and extensive experimental evaluation results on three benchmarks, namely, DOTA, DIOR, and NWPU VHR-10, demonstrate the effectiveness and superiority of our proposed method compared with that of other advanced remote-sensing object-detection methods.

Leveraging hierarchical semantic‐emotional memory in emotional conversation generation

AbstractHandling emotions in human‐computer dialogues has emerged as a challenging task which requires artificial intelligence systems to generate emotional responses by jointly perceiving the emotion involved in the input posts and incorporating it into the generation of semantically coherent and emotionally reasonable responses. However, most previous works generate emotional responses solely from input posts, which do not take full advantage of the training corpus and suffer from generating generic responses. In this study, we introduce a hierarchical semantic‐emotional memory module for emotional conversation generation (called HSEMEC), which can learn abstract semantic conversation patterns and emotional information from the large training corpus. The learnt semantic and emotional knowledge helps to enrich the post representation and assist the emotional conversation generation. Comprehensive experiments on a large real‐world conversation corpus show that HSEMEC can outperform the strong baselines on both automatic and manual evaluation. For reproducibility, we release the code and data publicly at: https://github.com/siat‐nlp/HSEMEC‐code‐data.

Self-supervised anomaly pattern detection for large scale industrial data

Detecting the anomalies in a large amounts of high-dimensional data has been a challenging task. In the Industry 4.0 environment, large-scale high-dimensional monitoring data features the complex pattern of high level semantics. In order to provide enterprise-wide monitoring solutions, it is necessary to identify the high-level semantic patterns of the anomalies in these data without splitting them. Existing end-to-end deep neural networks for time series are capable of recognizing the high-level semantics in natural language or speech signals, but they are barely applied in real-time anomaly detection of industrial data because of the large time costs. In this paper, we leverage the self-supervised contrastive learning methodology and propose a Composite Semantic Augmentation Encoder (CSAE) to provide an appropriate representation of industrial data and implement quick detection of anomalies in industrial application environments. CSAE is a non-sequential deep neural network with two augmentation layers and a mandatory layer. The two layers of data-augmentation are built to expand the size of samples of both low-level semantic anomalies and high-level semantic anomalies, which enables CSAE to discover diverse anomalies and improves its accuracy of high-level semantic pattern recognition. The mandatory layer is built to compress and reserve the temporal information in the industrial data to accelerate the anomaly detection. Therefore, as a non-sequential contrastive learning model, CSAE has faster training convergence than the usual sequence models. The experiment results have verified that CSAE can achieve higher prediction accuracy with less time consumption than existing machine learning models in the tasks of high dimensional anomaly pattern detection.

Targeting stopwords for quality assurance of SNOMED-CT

ObjectiveWe assess the potential of exploiting stopwords in biomedical concept names to complete the logical definitions of concepts that are not sufficiently defined. MethodsConcepts containing stopwords are selected from the Disorder hierarchy of Systematized NOmenclature of MEDicine (SNOMED-CT). SNOMED-CT consists of two types of concepts: Fully Defined (FD) concepts which are sufficiently defined and Partially Defined (PD) concepts which are not sufficiently defined. In this work, FD concepts containing stopwords are treated as a source of ground truth to complete the definitions of, lexically and semantically similar, PD concepts. FD and PD concepts are lexically and semantically analysed to create sample-sets. Mandatory attribute-relationships are calculated by using an intersection-set logic for each FD sample-set. PD sample-sets are audited against this mandatory attribute-relationship template to identify inconsistencies in modelling styles and potentially missing attribute-relationships. ResultsLexical and semantic patterns around 11 stopwords were analysed. 26 sample-sets were extracted for the 11 stopwords. Mandatory attribute-relationships were identified for 24 of the 26 sample-sets. The method identified 62.5% - 72.22% of the PD concepts, containing the stopwords in and due to, to be inconsistent in their modelling style and potentially missing at least one attribute-relationship according to the created template.

A genre-based approach in the secondary school English writing class: Voices from student-teachers in the teaching practicum.

While the genre-based approach (GBA) has assumed increasing prominence in discussions of writing pedagogy for diverse classrooms, little is known about how secondary school student-teachers understand and adopt genre pedagogies in the English as a foreign language (EFL) writing class. Based on the data from semi-structured interviews and teaching materials, this study examined Chinese EFL student-teachers’ knowledge and use of genre-based writing instruction (GBWI) during the teaching practicum and explored the challenges they encountered in enacting it. The findings demonstrated that teacher informants showed some familiarity with genre pedagogies, especially in terms of scaffolding the linguistic features and semantic patterns in the focused genres. However, they were generally confused over the connection between language, content, and context, and their GBWI practice scarcely involved the explicit teaching of the linguistic and semantic choices for a specific audience and context, which gave rise to some perceived tensions in the teaching reality. Further probing has revealed the complex interplay between Chinese EFL student-teachers’ professional knowledge, perceived difficulties, and genre instructional practice in the secondary school writing class. The study concludes with practical implications for the student-teachers’ professional development of effective GBA.

Veggie-Renate und Merci-Jens

Abstract This paper investigates personal name compounds (PN compounds) with a personal name in the second position and a lexeme or a proper name as a modifier (cf. Veggie-Renate). Based on 1194 types from the web corpus of the Digitales Wörterbuch der Deutschen Sprache (DWDS) and Twitter, the paper aims at testing the following hypotheses from the previous studies (Wildgen 1981, Kürschner 2020): PN compounds are infrequent, irregular and bear mainly an evaluative function. By drawing on the semantic relations holding between the components of nominal compounds in German (cf. Ortner et al. 1991), we provide evidence for 15 regular semantic patterns underlying PN compounds. However, these patterns cannot fully account for the properties of PN compounds as the patterns are too abstract and neglect the contextual factors and extra-linguistic knowledge about the name bearer. Therefore, to arrive at the properties of PN compounds, semantic patterns must be pragmatically enriched in the context by taking into account the linguistic knowledge about the modifier, the linguistic context and knowledge about the name bearer that is not included in the semantic pattern. Furthermore, apart from their evaluative function, PN compounds also evoke knowledge about the name bearer’s appearance, function, occupation, origin or participation in a social event. Finally, the analysis indicates a large frequency of PN compounds in DWDS and Twitter compared to the German Reference Corpus (DeReKo).