Search

There is an increased demand for detecting online hate speech, especially with the recent changing policies of hate content and free-of-speech right of online social media platforms. Detecting hate speech will reduce its negative impact on social media users. A lot of effort in the Natural Language Processing (NLP) field aimed to detect hate speech in general or detect specific hate speech such as religion, race, gender, or sexual orientation. Hate communities tend to use abbreviations, intentional spelling mistakes, and coded words in their communication to evade detection, which adds more challenges to hate speech detection tasks. Word representation from its domain will play an increasingly pivotal role in detecting hate speech. This paper investigates the feasibility of leveraging domain-specific word embedding as features and a bidirectional LSTM-based deep model as a classifier to automatically detect hate speech. This approach guarantees that the word is assigned its negative meaning, which is a very helpful technique to detect coded words. Furthermore, we investigate the use of the transfer learning language model (BERT) on the hate speech problem as a binary classification task as it provides high-performance results for many NLP tasks. The experiments showed that domain-specific word embedding with the bidirectional LSTM-based deep model achieved a 93% f1-score, while BERT achieved 96% f1-score on a combined balanced dataset from available hate speech datasets. The results proved that the performance of pre-trained models is influenced by the size of the trained data. Although there is a huge variation in the corpus size, the first approach achieved a very close result compared to BERT, which is trained on a huge data corpus, this is because it is trained on data related to the same domain. The first approach was very helpful to detect coded words while the second approach achieved better performance because it is trained on much larger data. To conclude, it is very helpful to build large pre-trained models from rich domains specific content in current social media platforms. [ABSTRACT FROM AUTHOR]

Individual abnormal behaviors vary depending on crowd sizes, contexts, and scenes. Challenges such as partial occlusions, blurring, a large number of abnormal behaviors, and camera viewing occur in large-scale crowds when detecting, tracking, and recognizing individuals with abnormalities. In this paper, our contribution is two-fold. First, we introduce an annotated and labeled large-scale crowd abnormal behavior Hajj dataset, HAJJv2. Second, we propose two methods of hybrid convolutional neural networks (CNNs) and random forests (RFs) to detect and recognize spatio-temporal abnormal behaviors in small and large-scale crowd videos. In small-scale crowd videos, a ResNet-50 pre-trained CNN model is fine-tuned to verify whether every frame is normal or abnormal in the spatial domain. If anomalous behaviors are observed, a motion-based individual detection method based on the magnitudes and orientations of Horn–Schunck optical flow is proposed to locate and track individuals with abnormal behaviors. A Kalman filter is employed in large-scale crowd videos to predict and track the detected individuals in the subsequent frames. Then, means and variances as statistical features are computed and fed to the RF classifier to classify individuals with abnormal behaviors in the temporal domain. In large-scale crowds, we fine-tune the ResNet-50 model using a YOLOv2 object detection technique to detect individuals with abnormal behaviors in the spatial domain. The proposed method achieves 99.76% and 93.71% of average area under the curves (AUCs) on two public benchmark small-scale crowd datasets, UMN and UCSD, respectively, while the large-scale crowd method achieves 76.08% average AUC using the HAJJv2 dataset. Our method outperforms state-of-the-art methods using the small-scale crowd datasets with a margin of 1.66%, 6.06%, and 2.85% on UMN, UCSD Ped1, and UCSD Ped2, respectively. It also produces an acceptable result in large-scale crowds. [ABSTRACT FROM AUTHOR]

Today, fatalities, physical injuries, and significant economic losses occur due to car accidents. Among the leading causes of car accidents is drowsiness behind the wheel, which can affect any driver. Drowsiness and sleepiness often have associated indicators that researchers can use to identify and promptly warn drowsy drivers to avoid potential accidents. This paper proposes a spatiotemporal model for monitoring drowsiness visual indicators from videos. This model depends on integrating a 3D convolutional neural network (3D-CNN) and long short-term memory (LSTM). The 3DCNN-LSTM can analyze long sequences by applying the 3D-CNN to extract spatiotemporal features within adjacent frames. The learned features are then used as the input of the LSTM component for modeling high-level temporal features. In addition, we investigate how the training of the proposed model can be affected by changing the position of the batch normalization (BN) layers in the 3D-CNN units. The BN layer is examined in two different placement settings: before the non-linear activation function and after the non-linear activation function. The study was conducted on two publicly available drowsy drivers datasets named 3MDAD and YawDD. 3MDAD is mainly composed of two synchronized datasets recorded from the frontal and side views of the drivers. We show that the position of the BN layers increases the convergence speed and reduces overfitting on one dataset but not the other. As a result, the model achieves a test detection accuracy of 96%, 93%, and 90% on YawDD, Side-3MDAD, and Front-3MDAD, respectively. [ABSTRACT FROM AUTHOR]

The presence of polycyclic aromatic hydrocarbons (PAHs) in wastewater has raised concerns about human health due to their potential carcinogenic and mutagenic properties. The widespread use of products containing acenaphthene (ACN, one of the 16 priority PAHs) in many industries and large-scale ACN release into the wastewater has resulted in dangerous concentrations of ACN in the environment. As a result, before discharge, it is required to eliminate or reduce its concentration to an acceptable level. Adsorption is an effective method of removing PAHs from wastewater. In this study, the ACN adsorption reaction in sample wastewater was evaluated using activated carbon produced by oil palm leaves. HPLC was used as an analytical method for quantifying ACN in wastewater samples. The initial concentration of ACN in water samples was 9.58 ± 0.5 mg/L. The experiments were conducted using the CCD combined with the RSM and using three independent variables, i.e., pH, activated carbon dosage (g/L), and contact time (min), and one dependent variable, i.e., ACN removal efficiency (%). The ANOVA was used to identify the significance of the developed model in the RSM. Lastly, the RSM was used to optimize the adsorption results. The experimental results determined that the removal of 98.73 ± 1% of ACN (the highest amount) was achieved at pH 7, while the removal of 88.44 ± 1% of ACN (the lowest amount) was achieved at pH 4.5. The adsorption efficiency of ACN was slightly increased by an increase in activated carbon dosage from 0.1 to 3 g/L (<4%). The contact time was the most significant factor in controlling the adsorption efficiency of ACN in wastewater, and not pH value or dosage. The adsorption reaction was quick, and 88–90% of ACN was removed within 5 min of the adsorption reaction, followed by slower adsorption for up to 90 min. The RSM model was developed on the basis of experimental results. An ANOVA determined that the developed model was significant enough to represent the adsorption data as the p-value was <0.05 for the model. The factors pH, adsorbent dosage, and contact time were also significant factors (p-value < 0.05). The optimization results showed that pH of 6.96, adsorbent dosage of 2.62 g/L, and contact time of 71.67 min were the optimal conditions for eliminating 98.88% of the ACN. The optimization results were verified in the lab, and a close agreement was found between the predicted results of the RSM and experimental results. The study found that the RSM is an effective tool for optimizing operating variables, as well as for significantly reducing time and experimentation costs. [ABSTRACT FROM AUTHOR]

We are currently experiencing a revolution in data production and artificial intelligence (AI) applications. Data are produced much faster than they can be consumed. Thus, there is an urgent need to develop AI algorithms for all aspects of modern life. Furthermore, the medical field is a fertile field in which to apply AI techniques. Breast cancer is one of the most common cancers and a leading cause of death around the world. Early detection is critical to treating the disease effectively. Breast density plays a significant role in determining the likelihood and risk of breast cancer. Breast density describes the amount of fibrous and glandular tissue compared with the amount of fatty tissue in the breast. Breast density is categorized using a system called the ACR BI-RADS. The ACR assigns breast density to one of four classes. In class A, breasts are almost entirely fatty. In class B, scattered areas of fibroglandular density appear in the breasts. In class C, the breasts are heterogeneously dense. In class D, the breasts are extremely dense. This paper applies pre-trained Convolutional Neural Network (CNN) on a local mammogram dataset to classify breast density. Several transfer learning models were tested on a dataset consisting of more than 800 mammogram screenings from King Abdulaziz Medical City (KAMC). Inception V3, EfficientNet 2B0, and Xception gave the highest accuracy for both four- and two-class classification. To enhance the accuracy of density classification, we applied weighted average ensembles, and performance was visibly improved. The overall accuracy of ACR classification with weighted average ensembles was 78.11%. [ABSTRACT FROM AUTHOR]

Recent years have witnessed an increased interest in online education, both massive open online courses (MOOCs) and small private online courses (SPOCs). This significant interest in online education has raised many challenges related to student engagement, performance, and retention assessments. With the increased demands and challenges in online education, several researchers have investigated ways to predict student outcomes, such as performance and dropout in online courses. This paper presents a comprehensive review of state-of-the-art studies that examine online learners' data to predict their outcomes using machine and deep learning techniques. The contribution of this study is to identify and categorize the features of online courses used for learners' outcome prediction, determine the prediction outputs, determine the strategies and feature extraction methodologies used to predict the outcomes, describe the metrics used for evaluation, provide a taxonomy to analyze related studies, and provide a summary of the challenges and limitations in the field. [ABSTRACT FROM AUTHOR]

Unmanned aerial vehicles (UAVs) have recently attracted widespread attention in civil and commercial applications. For example, UAVs (or drone) technology is increasingly used in crowd monitoring solutions due to its wider air footprint and the ability to capture data in real time. However, due to the open atmosphere, drones can easily be lost or captured by attackers when reporting information to the crowd management center. In addition, the attackers may initiate malicious detection to disrupt the crowd-sensing communication network. Therefore, security and privacy are one of the most significant challenges faced by drones or the Internet of Drones (IoD) that supports the Internet of Things (IoT). In the literature, we can find some authenticated key agreement (AKA) schemes to protect access control between entities involved in the IoD environment. However, the AKA scheme involves many vulnerabilities in terms of security and privacy. In this paper, we propose an enhanced AKA solution for crowd monitoring applications that require secure communication between drones and controlling entities. Our scheme supports key security features, including anti-forgery attacks, and confirms user privacy. The security characteristics of our scheme are analyzed by NS2 simulation and verified by a random oracle model. Our simulation results and proofs show that the proposed scheme sufficiently guarantees the security of crowd-aware communication. [ABSTRACT FROM AUTHOR]

THIS ARTICLE USES WORDS OR LANGUAGE THAT IS CONSIDERED PROFANE, VULGAR, OR OFFENSIVE BY SOME READERS. Hate speech detection in online social networks is a multidimensional problem, dependent on language and cultural factors. Most supervised learning resources for this task, such as labeled datasets and Natural Language Processing (NLP) tools, have been specifically tailored for English. However, a large portion of web users around the world speak different languages, creating an important need for efficient multilingual hate speech detection approaches. In particular, such approaches should be able to leverage the limited cross-lingual resources currently existing in their learning process. The cross-lingual transfer in this task has been difficult to achieve successfully. Therefore, we propose a simple yet effective method to approach this problem. To our knowledge, ours is the first attempt to create a multilingual embedding model specific to this problem. We validate the effectiveness of our approach by performing an extensive comparative evaluation against several well-known general-purpose language models that, unlike ours, have been trained on massive amounts of data. We focus on a zero-shot cross-lingual evaluation scenario in which we classify hate speech in one language without having access to any labeled data. Despite its simplicity, our embeddings outperform more complex models for most experimental settings we tested. In addition, we provide further evidence of the effectiveness of our approach through an ad hoc qualitative exploratory analysis, which captures how hate speech is displayed in different languages. This analysis allows us to find new cross-lingual relations between words in the hate-speech domain. Overall, our findings indicate common patterns in how hate speech is expressed across languages and that our proposed model can capture such relationships significantly. [ABSTRACT FROM AUTHOR]

Current theories of occupational status conceptualize it as either a function of cultural esteem or the symbolic aspect of the class structure. Based on Weber’s definition of status as rooted in either cultural or class conditions, we argue that a consistent operationalization of occupational status must account for both of these dimensions. Using quantitative measures of cultural sentiments for occupational identities, we use affect control theory to model the network deference relations across occupations. We calculate a measure of the extent to which one occupational actor deferring to another is incongruent with cultural expectations for all possible combinations of 304 occupational titles. Because high-status actors are less likely to defer to low-status actors, the degree to which these events violate cultural expectations provides an indicator of the relative status position of different occupations. We assess the construct validity of our new deference score measure using Harris Poll data. Deference scores are more predictive of status rankings from poll data than are occupational prestige scores. We establish criterion validity using five theoretically relevant workplace outcomes: subjective work attachment, job satisfaction, general happiness, the importance of meaningful work, and perceived respect at work. [ABSTRACT FROM AUTHOR]

We present a methodological innovation for analyzing archival data that involves the framing strategies from the failed 1980 Iowa Equal Rights Amendment (ERA). First, we conducted an archival analysis that suggested that pro-ERA groups used “frame resonance,” a strategy prominent in the social movement literature where activists align issues with ideologies. Meanwhile, anti-ERA groups used what we coin here as “frame dissonance” by depicting how passing the ERA clashed with ideologies. Next, we used affect control theory (ACT)’s Interact computer program to simulate how constituents likely responded to frames, given the distinct ideologies that existed during the time period. The simulations triangulated the archival analysis by (a) confirming our categorization of framing strategies as either resonance or dissonance and (b) identifying frame dissonance as potentially better for mobilizing than frame resonance. Our study demonstrates the value of Interact for triangulating archival analyses and adds a new framing strategy to the social movement literature. [ABSTRACT FROM AUTHOR]

The negative outcomes associated with cultural stereotypes based on race, class, and gender and related schema-consistency biases are well documented. How these biases become culturally entrenched is less well understood. In particular, previous research has neglected the role of information transmission processes in perpetuating cultural biases. In this study, I combine insights from the cultural cognition, affect control theory, and cultural transmission frameworks to examine how one form of internalized culture--fundamental cultural sentiments--affects the content of information shared in communication. I argue that individuals communicate narratives in ways that minimize deflection of internalized cultural sentiments, resulting in cultural-consistency bias. I test this proposition using a serial transmission study in which participants read and retell short stories. Results show that culturally inconsistent, high-deflection information experiences an initial boost in memorability, but consistency biases ultimately win out as information is altered to increase cultural consistency, demonstrating that deflection provides a promising measure of cultural schema-consistency. This measure is predictive of the information that individuals share in communication and changes to this information in the transmission process. [ABSTRACT FROM AUTHOR]

Drawing on Bayesian probability theory, we propose a generalization of affect control theory (BayesACT) that better accounts for the dynamic fluctuation of identity meanings for self and other during interactions, elucidates how people infer and adjust meanings through social experience, and shows how stable patterns of interaction can emerge from individuals’ uncertain perceptions of identities. Using simulations, we illustrate how this generalization offers a resolution to several issues of theoretical significance within sociology and social psychology by balancing cultural consensus with individual deviations from shared meanings, balancing meaning verification with the learning processes reflective of change, and accounting for noise in communicating identity. We also show how the model speaks to debates about core features of the self, which can be understood as stable and yet malleable, coherent and yet composed of multiple identities that may carry competing meanings. We discuss applications of the model in different areas of sociology, implications for understanding identity and social interaction, as well as the theoretical grounding of computational models of social behavior. [ABSTRACT FROM AUTHOR]