A Multi-Task Learning Approach to Hate Speech Detection Leveraging Sentiment Analysis

Plaza-del-Arco, Flor Miriam; Molina-González, M. Dolores; Ureña‐López, L. Alfonso; Martín–Valdivia, María Teresa

doi:10.1109/access.2021.3103697

Cited by 52 publications

(29 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Plaza-Del-Arco et al proposed an interactive method to obtain the important and distinguishing attributes of manual continuous annotation. Local attributes with distinguishing and semantic meaning are discovered from image data sets using only fine-grained category labels and object boundary box annotations [5]. A potential conditional random field model is used to discover candidate attributes that are detectable and differentiated, and then a recommendation system is used to select attributes that may have semantic significance.…”

Section: Literature Reviewmentioning

confidence: 99%

University Media Content Detection and Classification Based on Information Fusion Algorithm

Zhang

Yang

et al. 2022

Advances in Multimedia

View full text Add to dashboard Cite

In order to further solve the problems in promoting the classification of media content in colleges and universities, the effective analysis and understanding of multimedia data content can be better realized based on the characteristics of multimedia data in colleges and universities, combining with the characteristics of rich information, large differences in performance, and large amount of large-scale data. This essay mainly introduces the technology of university media content detection and classification based on information fusion algorithm and focuses on the application of university multimedia content detection, analysis, and understanding, to explore the image discrimination auxiliary attribute feature learning and content association prediction and classification. A benchmark model for media content detection and classification is constructed. Through the model test, it is found that the F 1 value of the model is more than 70%, the check rate is more than 80%, and the recall rate is more than 50%. On this basis, a content detection system based on campus network is constructed.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

University Media Content Detection and Classification Based on Information Fusion Algorithm

Zhang

Yang

et al. 2022

Advances in Multimedia

View full text Add to dashboard Cite

show abstract

“…Moreover, abusive words are ever-evolving and are usually not spoken completely and clearly, making the understanding of the overall context very important for this task. Incorporating emotion attributes (which are strongly linked to offensive behaviour in real-life) boosts the performance of abuse detection as opposed to using only the video or text based features as shown in [9,10,19]. Furthermore, research suggests that emotion can be detected with greater accuracy in speech than from its corresponding text [20,21], making emotion one of our main choices among modalities.…”

Section: Related Workmentioning

confidence: 99%

“…A two-stage process of transcribing the spoken audio into text using automatic speech recognition (ASR) systems followed by using natural language processing based abuse detection methods designed for text is a plausible approach. While the transcribed text captures the semantic information, it is not able to represent the audio cues like pitch, volume, tone, emotions and so on, which can often play an important role in abuse detection as humans are generally angry, agitated or loud while displaying abusive behaviour [9,10].…”

Section: Introductionmentioning

confidence: 99%

Multilingual and Multimodal Abuse Detection

Sharon¹,

Shah²,

Mukherjee³

et al. 2022

Preprint

View full text Add to dashboard Cite

The presence of abusive content on social media platforms is undesirable as it severely impedes healthy and safe social media interactions. While automatic abuse detection has been widely explored in textual domain, audio abuse detection still remains unexplored. In this paper, we attempt abuse detection in conversational audio from a multimodal perspective in a multilingual social media setting. Our key hypothesis is that along with the modelling of audio, incorporating discriminative information from other modalities can be highly beneficial for this task. Our proposed method, MADA, explicitly focuses on two modalities other than the audio itself, namely, the underlying emotions expressed in the abusive audio and the semantic information encapsulated in the corresponding textual form. Observations prove that MADA demonstrates gains over audio-only approaches on the ADIMA dataset. We test the proposed approach on 10 different languages and observe consistent gains in the range 0.6%-5.2% by leveraging multiple modalities. We also perform extensive ablation experiments for studying the contributions of every modality and observe the best results while leveraging all the modalities together. Additionally, we perform experiments to empirically confirm that there is a strong correlation between underlying emotions and abusive behaviour. 1

show abstract

“…is has a negative impact on the students' emotional tendencies as well as their mental health. Only by mining and analyzing the sentiment tendencies of college students in college network public opinion can we understand college students' mental health status in a timely manner and provide targeted treatment and prevention [3].…”

Section: Introductionmentioning

confidence: 99%

A Novel Affective Analysis System Modeling Method Integrating Affective Cognitive Model and Bi-LSTM Neural Network

Yang

Zhou

Jing

2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

The severity of mental health issues among college students has increased over the past few years, having a significant negative impact on not only their academic performance but also on their families and even society as a whole. Therefore, one of the pressing issues facing college administrators right now is finding a method that is both scientific and useful for determining the mental health of college students. In pace with the advancement of Internet technology, the Internet has become an important communication channel for contemporary college students. As one of the main forces in the huge Internet population, college students are at the stage of growing knowledge and being most enthusiastic about new things, and they like to express their opinions and views on study life and social issues and are brave to express their emotions. These subjective text data often contain some affective tendencies and psychological characteristics of college students, and it is beneficial to dig out their affective tendencies to further understand what they think and expect and to grasp their mental health as early as possible. In order to address the issue of assessing the mental health of college students, this study makes an effort to use public opinion data from the university network and suggests a college student sentiment analysis model based on the OCC affective cognitive model and Bi-LSTM neural network. In order to label three different types of positive, negative, and neutral sentiment on the microblog text of college network public opinion, we first design a sentiment rule system based on the OCC affective cognition elicitation mechanism. In order to effectively and automatically identify the sentiment state of college students in the network public opinion, this study uses a Bi-LSTM neural network to classify the preprocessed college network public opinion data. Finally, this study performs comparison experiments to confirm the validity of the Bi-LSTM neural network sentiment recognition algorithm and the accuracy of the OCC sentiment rule labeling system. The findings show that the college student sentiment recognition effect of the model is significantly enhanced when the OCC sentiment rule system is used to label the college network public opinion data set as opposed to the naturally labeled data set. In contrast to SVM and other classification models like CNN and LSTM, the Bi-LSTM neural network-based classification model achieves more satisfactory classification results in the recognition of college opinion sentiment.

show abstract

A Multi-Task Learning Approach to Hate Speech Detection Leveraging Sentiment Analysis

Cited by 52 publications

References 42 publications

University Media Content Detection and Classification Based on Information Fusion Algorithm

University Media Content Detection and Classification Based on Information Fusion Algorithm

Multilingual and Multimodal Abuse Detection

A Novel Affective Analysis System Modeling Method Integrating Affective Cognitive Model and Bi-LSTM Neural Network

Contact Info

Product

Resources

About