Evaluation of a Video Generation Method Linking Dance and Scenes

Owing to the impact of COVID-19, the venues for dancers to perform have shifted from the stage to the media. In this study, we focus on the creation of dance videos that allow audiences to feel a sense of excitement without disturbing their awareness of the dance subject and propose a video generation method that links the dance and the scene by utilizing a sound detection method and an object detection algorithm. The generated video was evaluated using the Semantic Differential method, and it was confirmed that the proposed method could transform the original video into an uplifting video without any sense of discomfort. © 2023 SPIE.

Social Distance and Face Mask Detection using vision-based sensors and IP cameras

Healthcare systems all over the world are strained as the COVID-19 pandemic's spread becomes more widespread. The only realistic strategy to avoid asymptomatic transmission is to monitor social distance, as there are no viable medical therapies or vaccinations for it. A unique computer vision-based framework that uses deep learning is to analyze the images that are needed to measure social distance. This technique uses the key point regressor to identify the important feature points utilizing the Visual Geometry Group (VGG19) which is a standard Convolutional Neural Network (CNN) architecture having multiple layers, MobileNetV2 which is a computer vision network that advances the-state-of-art for mobile visual identification, including semantic segmentation, classification and object identification. VGG19 and MobileNetV2 were trained on the Kaggle dataset. The border boxes for the item may be seen as well as the crowd is sizeable, and red identified faces are then analyzed by MobileNetV2 to detect whether the person is wearing a mask or not. The distance between the observed people has been calculated using the Euclidian distance. Pretrained models like (You only look once) YOLOV3 which is a real-time object detection system, RCNN, and Resnet50 are used in our embedded vision system environment to identify social distance on images. The framework YOLOV3 performs an overall accuracy of 95% using transfer learning technique runs in 22ms which is four times fast than other predefined models. In the proposed model we achieved an accuracy of 96.67% using VGG19 and 98.38% using MobileNetV2, this beats all other models in its ability to estimate social distance and face mask. © 2023 IEEE.

Characterizing the Entities in Harmful Memes: Who is the Hero, the Villain, the Victim?

Memes can sway people's opinions over social media as they combine visual and textual information in an easy-to-consume manner. Since memes instantly turn viral, it becomes crucial to infer their intent and potentially associated harmfulness to take timely measures as needed. A common problem associated with meme comprehension lies in detecting the entities referenced and characterizing the role of each of these entities. Here, we aim to understand whether the meme glorifies, vilifies, or victimizes each entity it refers to. To this end, we address the task of role identification of entities in harmful memes, i.e., detecting who is the 'hero', the 'villain', and the 'victim' in the meme, if any. We utilize HVVMemes - a memes dataset on US Politics and Covid-19 memes, released recently as part of the CONSTRAINT@ACL-2022 shared-task. It contains memes, entities referenced, and their associated roles: hero, villain, victim, and other. We further design VECTOR (Visual-semantic role dEteCToR), a robust multi-modal framework for the task, which integrates entity-based contextual information in the multi-modal representation and compare it to several standard unimodal (text-only or image-only) or multi-modal (image+text) models. Our experimental results show that our proposed model achieves an improvement of 4% over the best baseline and 1% over the best competing stand-alone submission from the shared-task. Besides divulging an extensive experimental setup with comparative analyses, we finally highlight the challenges encountered in addressing the complex task of semantic role labeling within memes. © 2023 Association for Computational Linguistics.

Detection of Semantic Duplicates in Temporal Domains Using Directed Acyclic Graphs

We demonstrate the feasibility of a generalized technique for semantic deduplication in temporal data domains using graph-based representations of data records. Structured data records with multiple timestamp attributes per record may be represented as a directed graph where the nodes represent the events and the edges represent event sequences. Edge weights are based on elapsed time between connecting nodes. In comparing two records, we may merge these directed graphs and determine a representative directed acyclic graph (DAG) inclusive of a subset of nodes and edges that maintain the transitive weights of the original graphs. This DAG may then be evaluated by weighting elapsed time equivalences between records at each node and measuring the fraction of nodes represented in the DAG versus the union of nodes between the records being compared. With this information, we establish a duplication score and use a specified threshold requirement to assert duplication. This method is referred to as Temporal Deduplication using Directed Acyclic Graphs (TD:DAG). TD:DAG significantly outperformed established ASNM and ASNM+LCS methods for datasets rep-resenting two disparate domains, COVID-19 government policy data and PlayStation Network (PSN) trophy data. TD:DAG produced highly effective and comparable F1 scores of 0.960 and 0.972 for the two datasets, respectively, versus 0.864/0.938 for ASNM+LCS and 0.817/0.708 for ASNM. © 2023 IEEE.

U-Net based Semantic Segmentation for Touchless Fingerprint Technology: A Reflective Review

Touch-based fingerprints are widely used in today's world;even with all the success, the touch-based nature of these is a threat, especially in this COVID-19 period. A solution to the same is the introduction of Touchless Fingerprint Technology. The workflow of a touchless system varies vastly from its touch-based counterpart in terms of acquisition, pre-processing, image enhancement, and fingerprint verification. One significant difference is the methods used to segment desired fingerprint regions. This literature focuses on pixel-level classification or semantic segmentation using U-Net, a key yet challenging task. A plethora of semantic segmentation methods have been applied in this field. In this literature, a spectrum of efforts in the field of semantic segmentation using U-Net is investigated along with the components that are integral while training and testing a model, like optimizers, loss functions, and metrics used for evaluation and enumeration of results obtained. © 2022 IEEE.

Towards Timeline Generation with Abstract Meaning Representation

Timeline summarization (TLS) is a challenging research task that requires researchers to distill extensive and intricate temporal data into a concise and easily comprehensible representation. This paper proposes a novel approach to timeline summarization using Meaning Representations (AMRs), a graphical representation of the text where the nodes are semantic concepts and the edges denote relationships between concepts. With AMR, sentences with different wordings, but similar semantics, have similar representations. To make use of this feature for timeline summarization, a two-step sentence selection method that leverages features extracted from both AMRs and the text is proposed. First, AMRs are generated for each sentence. Sentences are then filtered out by removing those with no named-entities and keeping the ones with the highest number of named-entities. In the next step, sentences to appear in the timeline are selected based on two scores: Inverse Document Frequency (IDF) of AMR nodes combined with the score obtained by applying a keyword extraction method to the text. Our experimental results on the TLS-Covid19 test collection demonstrate the potential of the proposed approach. © 2023 ACM.

COVID-19 Instance Segmentation Network Based on Semantic Information Generating Proposals

The CT images of lungs of COVID-19 patients have distinct pathological features, segmenting the lesion area accurately by the method of deep learning, which is of great significance for the diagnosis and treatment of COVID-19 patients. Instance segmentation has higher sensitivity and can output the Bounding Boxes of the lesion region, however, the traditional instance segmentation method is weak in the segmentation of small lesions, and there is still room for improvement in the segmentation accuracy. We propose a instance segmentation network which is called as Semantic R-CNN. Firstly, a semantic segmentation branch is added on the basis of Mask-RCNN, and utilizing the image processing tool Skimage in Python to label the connected domain for the result of semantic segmentation, extracting the rectangular boundaries of connected domain and using them as Proposals, which will replace the Regional Proposal Network in the instance segmentation. Secondly, the Atrous Spatial Pyramid Pooling is introduced into the Feature Pyramid Network, then improving the feature fusion method in FPN. Finally, the cascade method is introduced into the detection branch of the network to optimize the Proposals. Segmentation experiments were carried out on the pathological lesion segmentation data set of CC-CCII, the average accuracy of the semantic segmentation is 40.56mAP, and compared with the Mask-RCNN, it has improved by 9.98mAP. After fusing the results of semantic segmentation and instance segmentation, the Dice coefficient is 80.7%, the sensitivity is 85.8%, and compared with the Inf-Net, it has increased by 1.6% and 8.06% respectively. The proposed network has improved the segmentation accuracy and reduced the false-negatives. © 2022 ACM.

The semiotics of the anti-COVID-19 mask

The pandemic spreading of the COVID-19 virus has led to the global need to introduce, often by law, the medical face mask, which can undoubtedly be considered as "the object of 2020.” In a few months, most human faces around the world in the public space, but also often in the private space, have been covered with various kinds of protective masks. Very soon, these objects have become the centre of several discursive productions, going from medical reports to media coverage, from artistic representations to ironic memes. The medical face mask was not totally new in the west, where it was already present in special circumstances, like dentists' studios or emergency rooms, and was quite familiar in the east, especially in Japan, China, and Korea. Yet such massive introduction changed the meaning of the medical face mask in every context. Old habits were reconfigured or clashed with the new ones, giving rise to a novel syntax, semantics, and pragmatics of the human face in conjunction with this device and in the context of the global pandemic. The present paper offers an introduction to a semiotic mapping of such radical cultural change and its likely consequences.

TwIdw—A Novel Method for Feature Extraction from Unstructured Texts

Featured ApplicationThe research has a potential application in the field of fake news detection. By using the feature extraction technique, TwIdw, proposed in this paper, more relevant and informative features can be extracted from the text data, which can lead to an enhancement in the accuracy of the classification models employed in these tasks.This research proposes a novel technique for fake news classification using natural language processing (NLP) methods. The proposed technique, TwIdw (Term weight–inverse document weight), is used for feature extraction and is based on TfIdf, with the term frequencies replaced by the depth of the words in documents. The effectiveness of the TwIdw technique is compared to another feature extraction method—basic TfIdf. Classification models were created using the random forest and feedforward neural networks, and within those, three different datasets were used. The feedforward neural network method with the KaiDMML dataset showed an increase in accuracy of up to 3.9%. The random forest method with TwIdw was not as successful as the neural network method and only showed an increase in accuracy with the KaiDMML dataset (1%). The feedforward neural network, on the other hand, showed an increase in accuracy with the TwIdw technique for all datasets. Precision and recall measures also confirmed good results, particularly for the neural network method. The TwIdw technique has the potential to be used in various NLP applications, including fake news classification and other NLP classification problems.

An Interpretable Fake News Detection Method Based on Commonsense Knowledge Graph

Existing deep learning-based methods for detecting fake news are uninterpretable, and they do not use external knowledge related to the news. As a result, the authors of the paper propose a graph matching-based approach combined with external knowledge to detect fake news. The approach focuses on extracting commonsense knowledge from news texts through knowledge extraction, extracting background knowledge related to news content from a commonsense knowledge graph through entity extraction and entity disambiguation, using external knowledge as evidence for news identification, and interpreting the final identification results through such evidence. To achieve the identification of fake news containing commonsense errors, the algorithm uses random walks graph matching and compares the commonsense knowledge embedded in the news content with the relevant external knowledge in the commonsense knowledge graph. The news is then discriminated as true or false based on the results of the comparative analysis. From the experimental results, the method can achieve 91.07%, 85.00%, and 89.47% accuracy, precision, and recall rates, respectively, in the task of identifying fake news containing commonsense errors.

CovRelex-SE: Adding Semantic Information for Relation Search via Sequence Embedding

In recent years, COVID-19 has impacted all aspects of human life. As a result, numerous publications relating to this disease have been issued. Due to the massive volume of publications, some retrieval systems have been developed to provide researchers with useful information. In these systems, lexical searching methods are widely used, which raises many issues related to acronyms, synonyms, and rare keywrds. In this paper, we present a hybrid relation retrieval system, CovRelex-SE, based on embeddings to provide high-quality search results. Our system can be accessed through the following URL: https://www.jaist.ac.jp/is/labs/nguyen-lab/systems/covrelex-se/. © 2023 Association for Computational Linguistics.

Stacked Bi-directional Long Short Term Memory model for Multi-Class Arabic Sentiment Analysis on Covid-19

The world has been affected by the Covid-19 epidemic during the last three years. During that period, most people tended to use social networks, where by searching for topics related to Covid-19, information could be provided to manage decisions by organizations or governments about public health. With the importance of the Arabic language, despite the lack of research targeting it, using Arabic language as a source of data and analyzing it due to the large number of users on social networks gives an impetus to understand people's feelings about the Covid-19 pandemic. One of the challenges facing sentiment analysis in Arabic is the use of dialects. The most common and existing methods used have been quite ineffective as they are oblivious to contextual information and cannot handle long-distance word dependencies. The Iraqi Arabic dialect is one of the Arabic dialects that still suffers from a lack of research in sentiment analysis. In this study, the official page of the Iraqi Ministry of Health on Facebook was used to collect and analysis comments. Word2vec model is incorporated to extract words semantic characteristics. To capture contextual features, Stacked Bi-directional Long Short Term Memory model (Stacked Bi-LSTM) utilizes sequential word vectors derived from the Continuous Bag of Words model. When compared to most common and existing approaches, the proposed method performed well. © 2022 IEEE.

Addressing Issues of Cross-Linguality in Open-Retrieval Question Answering Systems For Emergent Domains

Open-retrieval question answering systems are generally trained and tested on large datasets in well-established domains. However, low-resource settings such as new and emerging domains would especially benefit from reliable question answering systems. Furthermore, multilingual and cross-lingual resources in emergent domains are scarce, leading to few or no such systems. In this paper, we demonstrate a cross-lingual open-retrieval question answering system for the emergent domain of COVID-19. Our system adopts a corpus of scientific articles to ensure that retrieved documents are reliable. To address the scarcity of cross-lingual training data in emergent domains, we present a method utilizing automatic translation, alignment, and filtering to produce English-to-all datasets. We show that a deep semantic retriever greatly benefits from training on our English-to-all data and significantly outperforms a BM25 baseline in the cross-lingual setting. We illustrate the capabilities of our system with examples and release all code necessary to train and deploy such a system1 © 2023 Association for Computational Linguistics.

Managing multi-granular probabilistic linguistic information in large-scale group decision making: A personalized individual semantics-based consensus model

The multi-granular probabilistic linguistic modeling allows decision makers to express cognitive information using multiple linguistic term sets based on their preferences. However, personalized individual semantics (PIS) can lead to different meanings of the same word within the linguistic context. To address this issue and manage consensus in large-scale group decision making, this study proposes a decision framework that employs multi-granular probabilistic linguistic preference relations (MGPLPRs). First, a transformation method is presented to unify different granularity levels of MGPLPRs, thus ensuring the consistency of granularity. Moreover, a consistency-driven optimization model is constructed to generate the numerical scales with PIS for different experts. Thereafter, a two-stage consensus reaching process (CRP) is developed, including both within-cluster and across-cluster CRP, to achieve group consensus. The experts' original weights are derived from a social network, taking into account the trust relationships among them. A dynamic weighting mechanism is used to update the experts' weights based on their contributions to group consensus, which better reflects the actual situation than fixed weights. The proposed method is exemplified through a case study of assessing and selecting campus surveillance measures for COVID-19. Finally, the effectiveness and robustness of the proposed framework are verified through comparative analysis and sensitivity analysis.

A Critical Discourse Analysis of the Microstructure of Covid-19 Health-Protocol Advertisements in Tourist Facilities

To alleviate the impact of the Covid-19 pandemic on tourism, tourist facilities in Bali are informing visitors of the relevant health protocols, using posters to describe the appropriate behaviours. Using critical discourse analysis, this study examines the microstructure of the texts in these posters to identify their semantic, syntactic, lexical, and rhetorical elements. The study findings show that the semantic aspects consist of background, intention, and detail. The syntactic elements involve coherence and the use of the pronouns 'you' and 'we', and of the imperative, and the declarative. The lexical aspects include abbreviations and vocabulary, related to the health protocol. The textual messages are delivered in official language, supported by pictures and photographs.

Lexico-Semantic Change in the Kazakh Language of the COVID Era

This article discusses lexical and semantic changes during the COVID-19 pandemic. In this article, we describe semantic shifts, new concepts, and neologisms associated with the COVID-19 pandemic based on the results of an associative survey. A total of 142 respondents voluntarily participated in our online survey. The term 'coronavirus' was taken as a stimulus word. Respondents had to answer what colour and number the word 'coronavirus' is associated with. The results of the study show that the stimulus 'coronavirus' in the minds of people activates the colours red, green, black, blue, yellow and very weakly causes associations with brown, white, gold, purple, colourless, as well as the frequency of the number 19. Additionally, according to the results of the study, it can be said that during the COVID-19 pandemic, negative meanings of colourative vocabulary were actualized (except green, because this colour began to symbolize safety), and numbers and some new concepts that have a nonpositive colouring appeared.

Unsupervised Numerical Information Extraction via Exploiting Syntactic Structures

Numerical information plays an important role in various fields such as scientific, financial, social, statistics, and news. Most prior studies adopt unsupervised methods by designing complex handcrafted pattern-matching rules to extract numerical information, which can be difficult to scale to the open domain. Other supervised methods require extra time, cost, and knowledge to design, understand, and annotate the training data. To address these limitations, we propose QuantityIE, a novel approach to extracting numerical information as structured representations by exploiting syntactic features of both constituency parsing (CP) and dependency parsing (DP). The extraction results may also serve as distant supervision for zero-shot model training. Our approach outperforms existing methods from two perspectives: (1) the rules are simple yet effective, and (2) the results are more self-contained. We further propose a numerical information retrieval approach based on QuantityIE to answer analytical queries. Experimental results on information extraction and retrieval demonstrate the effectiveness of QuantityIE in extracting numerical information with high fidelity.

Digital topics on cultural heritage investigated: how can data-driven and data-guided methods support to identify current topics and trends in digital heritage?

In research and policies, the identification of trends as well as emerging topics and topics in decline is an important source of information for both academic and innovation management. Since at present policy analysis mostly employs qualitative research methods, the following article presents and assesses different approaches – trend analysis based on questionnaires, quantitative bibliometric surveys, the use of computer-linguistic approaches and machine learning and qualitative investigations. Against this backdrop, this article examines digital applications in cultural heritage and, in particular, built heritage via various investigative frameworks to identify topics of relevance and trendlines, mainly for European Union (EU)-based research and policies. Furthermore, this article exemplifies and assesses the specific opportunities and limitations of the different methodical approaches against the backdrop of data-driven vs. data-guided analytical frameworks. As its major findings, our study shows that both research and policies related to digital applications for cultural heritage are mainly driven by the availability of new technologies. Since policies focus on meta-topics such as digitisation, openness or automation, the research descriptors are more granular. In general, data-driven approaches are promising for identifying topics and trendlines and even predicting the development of near future trends. Conversely, qualitative approaches are able to answer "why” questions with regard to whether topics are emerging due to disruptive innovations or due to new terminologies or whether topics are becoming obsolete because they are common knowledge, as is the case for the term "internet”.

Clarifying Organizational Generativity: A Future Forming Perspective for OD Practitioners and Researchers

As the catastrophic consequences of the COVID-19 pandemic have made clear, both the practice and research of organizational development (OD) urgently need alternative pathways to the future. Organizational generativity (OG) offers one such promising alternative. While much of OD practice and research are focused on enabling organizations to better prepare for an unknown future, OG accommodates new ways for organizations to proactively create their own future. As a nascent field of inquiry, however, research on OG is underdeveloped and characterized by a lack of clarity. Therefore, the purpose of this paper is to clarify the construct of organizational generativity to be more actionable by OD practitioners, researchers, and managers alike. Using grounded theory, we review and critique the literature on generativity, from the "ancestral" writers in psychology to current OD authors. Through successive rounds of inquiry, we reveal the syntax, the semantics, and the inherent processual nature of organizational generativity. We then derive a conceptual framework describing seven manifestations of generative organizational processes: relational, transformational, disruptive, future-focused, idea-giving, actionable, and procreative. Finally, we discuss implications for OD practice and opportunities for future research.

VMCTE: Visualization-Based Malware Classification Using Transfer and Ensemble Learning

The Corona Virus Disease 2019 (COVID-19) effect has made telecommuting and remote learning the norm. The growing number of Internet-connected devices provides cyber attackers with more attack vectors. The development of malware by criminals also incorporates a number of sophisticated obfuscation techniques, making it difficult to classify and detect malware using conventional approaches. Therefore, this paper proposes a novel visualization-based malware classification system using transfer and ensemble learning (VMCTE). VMCTE has a strong anti-interference ability. Even if malware uses obfuscation, fuzzing, encryption, and other techniques to evade detection, it can be accurately classified into its corresponding malware family. Unlike traditional dynamic and static analysis techniques, VMCTE does not require either reverse engineering or the aid of domain expert knowledge. The proposed classification system combines three strong deep convolutional neural networks (ResNet50, MobilenetV1, and MobilenetV2) as feature extractors, lessens the dimension of the extracted features using principal component analysis, and employs a support vector machine to establish the classification model. The semantic representations of malware images can be extracted using various convolutional neural network (CNN) architectures, obtaining higher-quality features than traditional methods. Integrating fine-tuned and non-fine-tuned classification models based on transfer learning can greatly enhance the capacity to classify various families of malware. The experimental findings on the Malimg dataset demonstrate that VMCTE can attain 99.64%, 99.64%, 99.66%, and 99.64% accuracy, F1-score, precision, and recall, respectively. © 2023 Tech Science Press. All rights reserved.