ArSa-Tweets: A novel Arabic sarcasm detection system based on deep learning model.

Sarcasm in Sentiment Analysis (SA) is important due to the sense of sarcasm in sentences that differs from their literal meaning. Analysis of Arabic sarcasm still has many challenges like implicit indirect idioms to express the opinion, and lack of Arabic sarcasm corpus. In this paper, we proposed a new detecting model for sarcasm in Arabic tweets called the ArSa-Tweet model. It is based on implementing and developing Deep Learning (DL) models to classify tweets as sarcastic or not. The development of our proposed model consists of adding main improvements by applying robust preprocessing steps before feeding the data to the adapted DL models. The adapted DL models are LSTM, Multi-headed CNN-LSTM-GRU, BERT, AraBert-V01, and AraBert-V02. In addition, we proposed ArSa-data as a golden corpus that consists of Arabic tweets. A comparative process shows that our proposed ArSa-Tweet method has the most impact accuracy rate based on deploying the AraBert-V02 model, which obtains the best performance results in all accuracy metrics when compared with other methods.

Evaluating Large Language Models' Ability Using a Psychiatric Screening Tool Based on Metaphor and Sarcasm Scenarios.

Metaphors and sarcasm are precious fruits of our highly evolved social communication skills. However, children with the condition then known as Asperger syndrome are known to have difficulties in comprehending sarcasm, even if they possess adequate verbal IQs for understanding metaphors. Accordingly, researchers had employed a screening test that assesses metaphor and sarcasm comprehension to distinguish Asperger syndrome from other conditions with similar external behaviors (e.g., attention-deficit/hyperactivity disorder). This study employs a standardized test to evaluate recent large language models' (LLMs) understanding of nuanced human communication. The results indicate improved metaphor comprehension with increased model parameters; however, no similar improvement was observed for sarcasm comprehension. Considering that a human's ability to grasp sarcasm has been associated with the amygdala, a pivotal cerebral region for emotional learning, a distinctive strategy for training LLMs would be imperative to imbue them with the ability in a cognitively grounded manner.

Investigating the effect of emoji position on eye movements and subjective evaluations on Chinese sarcasm comprehension.

Evidence indicated that emojis could influence sarcasm comprehension and sentence processing in English. However, the effect of emojis on Chinese sarcasm comprehension remains unclear. Therefore, this study investigated the impact of the smiley emoji position and semantics on eye movements and subjective assessments during Chinese online communication. Our results showed that the presence of a smiley emoji improved participants' interpretation and perception of sarcasm. We also found shorter dwell times on sarcastic words compared to literal words under the comment-final emoji condition. Additionally, we clarified the time course of emojified sentence processing during Chinese reading: the presence of emoji initially decreased first fixation durations compared to the absence of emoji and then the comment-final emoji shortened dwell times on sarcastic words compared to literal words in the critical area of interest. Our findings suggested that the comment-final emoji was the preferable choice for avoiding semantic comprehension bias in China.

We studied how emoji position influenced Chinese semantic processing by combining the indices of eye movements and subjective assessments. Our results revealed that the comment-final smiley emoji was preferable for avoiding sarcasm comprehension bias. The corresponding time course and recommendations for improving Chinese online interpersonal interactions were discussed.

Cross-cultural nuances in sarcasm comprehension: a comparative study of Chinese and American perspectives.

It is evident that sarcasm can be interpreted differently due to various factors, However, rare research was conducted to investigate the influence of national culture on sarcasm comprehension despite its valuable theoretical implication. This study used an online rating task to explore how national culture impacts the comprehension of sarcasm, focusing on the differences between Chinese and American cultural values (i.e., power distance, uncertainty avoidance, collectivism, long-term orientation, and masculinity) and their influence on comprehending sarcastic praise and criticism. The study showed that Chinese participants tend to understand sarcasm less than Americans. It also found that Power Distance is linked to better sarcasm comprehension in both cultures, while Uncertainty Avoidance has a negative effect on it, especially in Chinese participants. Collectivism is also associated with improved sarcasm comprehension, especially in Chinese participants. However, Masculinity and Long-Term Orientation do not seem to have a significant impact on sarcasm comprehension, regardless of nationality or the type of comment (praise or criticism). Overall, the study reveals nuanced differences in how cultural values shape the comprehension of sarcasm in Chinese and American contexts, underscoring the complex interplay between culture and communication.

A transformer-based generative adversarial learning to detect sarcasm from Bengali text with correct classification of confusing text.

Sarcasm detection research in Bengali is still limited due to a lack of relevant resources. In this context, getting high-quality annotated data is costly and time-consuming. Therefore, in this paper, we present a transformer-based generative adversarial learning for sarcasm detection from Bengali text based on available limited labeled data. Here, we use the Bengali sarcasm dataset 'Ben-Sarc'. Besides, we construct another dataset containing Bengali sarcastic and non-sarcastic comments from YouTube and newspapers to observe the model's performance on the new dataset. On top of that, we utilize another Bengali sarcasm dataset 'BanglaSarc' to further prove our models' robustness. Among all models, the Bangla BERT-based Generative Adversarial Model has achieved the highest accuracy with 77.1% for the 'Ben-Sarc' dataset. Besides, this model has achieved the highest accuracy of 68.2% for the dataset constructed from YouTube and newspaper, and 97.2% for the 'BanglaSarc' dataset.

An Experimental Study on Sarcasm Comprehension in School Children: The Possible Role of Contextual, Linguistics and Meta-Representative Factors.

Understanding sarcasm is a complex ability, which includes several processes. Previous studies demonstrated the possible roles of linguistic and meta-representative factors in understanding sarcasm in school children, while the influence of specific contextual variables still needs to be investigated. Here, we present two studies investigating the possible role of contextual, linguistics, and meta-representative factors in understanding sarcasm in school children. In Study 1, we investigated sarcasm comprehension in 8-9-year-old school children in three different contexts, in which both familiarity and authority were manipulated. We found that understanding sarcasm was facilitated when the conversational partner was characterized by a high level of authority and familiarity (the mother) rather than when the conversational partner was an adult with a lower level of both authority and familiarity (the cashier of a food store). In Study 2, we replicated and extended Study 1 by investigating the possible influence of the same contextual factors but in a more sizeable sample and at different ages: first, third, and fifth grades of primary school. We found that understanding sarcasm improved significantly with age. The results of both studies indicated that understanding sarcasm is influenced by contextual factors. Children at any age better understood sarcasm produced by a speaker with a high level of both familiarity and authority. This ability improved with age. These results expand our understanding of how children infer a speaker's intentions in sarcasm. This might be particularly of interest to develop possible interventions for children on the Autism Spectrum, who are known to misunderstand sarcasm at different levels of complexity.

Enhanced Semantic Representation Learning for Sarcasm Detection by Integrating Context-Aware Attention and Fusion Network.

Sarcasm is a sophisticated figurative language that is prevalent on social media platforms. Automatic sarcasm detection is significant for understanding the real sentiment tendencies of users. Traditional approaches mostly focus on content features by using lexicon, n-gram, and pragmatic feature-based models. However, these methods ignore the diverse contextual clues that could provide more evidence of the sarcastic nature of sentences. In this work, we propose a Contextual Sarcasm Detection Model (CSDM) by modeling enhanced semantic representations with user profiling and forum topic information, where context-aware attention and a user-forum fusion network are used to obtain diverse representations from distinct aspects. In particular, we employ a Bi-LSTM encoder with context-aware attention to obtain a refined comment representation by capturing sentence composition information and the corresponding context situations. Then, we employ a user-forum fusion network to obtain the comprehensive context representation by capturing the corresponding sarcastic tendencies of the user and the background knowledge about the comments. Our proposed method achieves values of 0.69, 0.70, and 0.83 in terms of accuracy on the Main balanced, Pol balanced and Pol imbalanced datasets, respectively. The experimental results on a large Reddit corpus, SARC, demonstrate that our proposed method achieves a significant performance improvement over state-of-art textual sarcasm detection methods.

Sentiment Analysis and Sarcasm Detection using Deep Multi-Task Learning.

Social media platforms such as Twitter and Facebook have become popular channels for people to record and express their feelings, opinions, and feedback in the last decades. With proper extraction techniques such as sentiment analysis, this information is useful in many aspects, including product marketing, behavior analysis, and pandemic management. Sentiment analysis is a technique to analyze people's thoughts, feelings and emotions, and to categorize them into positive, negative, or neutral. There are many ways for someone to express their feelings and emotions. These sentiments are sometimes accompanied by sarcasm, especially when conveying intense emotion. Sarcasm is defined as a positive sentence with underlying negative intention. Most of the current research work treats them as two distinct tasks. To date, most sentiment and sarcasm classification approaches have been treated primarily and standalone as a text categorization problem. In recent years, research work using deep learning algorithms have significantly improved performance for these standalone classifiers. One of the major issues faced by these approaches is that they could not correctly classify sarcastic sentences as negative. With this in mind, we claim that knowing how to spot sarcasm will help sentiment classification and vice versa. Our work has shown that these two tasks are correlated. This paper proposes a multi-task learning-based framework utilizing a deep neural network to model this correlation to improve sentiment analysis's overall performance. The proposed method outperforms the existing methods by a margin of 3%, with an F1-score of 94%.

Enhancing the Generalization for Text Classification through Fusion of Backward Features.

Generalization has always been a keyword in deep learning. Pretrained models and domain adaptation technology have received widespread attention in solving the problem of generalization. They are all focused on finding features in data to improve the generalization ability and to prevent overfitting. Although they have achieved good results in various tasks, those models are unstable when classifying a sentence whose label is positive but still contains negative phrases. In this article, we analyzed the attention heat map of the benchmarks and found that previous models pay more attention to the phrase rather than to the semantic information of the whole sentence. Moreover, we proposed a method to scatter the attention away from opposite sentiment words to avoid a one-sided judgment. We designed a two-stream network and stacked the gradient reversal layer and feature projection layer within the auxiliary network. The gradient reversal layer can reverse the gradient of features in the training stage so that the parameters are optimized following the reversed gradient in the backpropagation stage. We utilized an auxiliary network to extract the backward features and then fed them into the main network to merge them with normal features extracted by the main network. We applied this method to the three baselines of TextCNN, BERT, and RoBERTa using sentiment analysis and sarcasm detection datasets. The results show that our method can improve the sentiment analysis datasets by 0.5% and the sarcasm detection datasets by 2.1%.

The Adjustment of Complexity on Sarcasm Processing in Chinese: Evidence from Reading Time Indicators.

It is controversial whether sarcasm processing should go through literal meaning processing. There is also a lack of eye movement evidence for Chinese sarcasm processing. In this study, we used eye movement experiments to explore the processing differences between sarcastic and literal meaning in Chinese text and whether this was regulated by sentence complexity. We manipulated the variables of complexity and literality. We recorded 33 participants' eye movements when they were reading Chinese text and the results were analyzed by a linear mixed model. We found that, in the early stage of processing, there was no difference between the processing time of the sarcastic meaning and the literal meaning of simple remarks, whereas for complex remarks, the time needed to process the sarcastic meaning was longer than that needed to process the literal meaning. In the later stage of processing, regardless of complexity, the processing time of the sarcastic meaning was longer than that of the literal meaning. These results suggest that sarcastic speech processing in Chinese is influenced by literal meaning, and the effect of literal meaning on sarcastic remarks is regulated by complexity. Sarcastic meaning was expressed differently in different stages of processing. These results support the hierarchical salience hypothesis of the serial modular model.

Judging emoji by occupation: A case of emoji-based sarcasm interpretation.

The smiling emoji has been claimed to be a marker of sarcastic intention among young Chinese users in computer-mediated communication. However, it is not well understood whether people interpret the emoji differently based on the characteristics or traits of the sender, as conveyed by occupation stereotypes. We investigated the effect of sender occupation on emoji-based sarcasm interpretation in both unambiguous (Experiment 1) and ambiguous (Experiment 2) contexts. The results showed that contextual incongruity was privileged over sender occupation in cueing sarcastic intention. In unambiguous contexts, sender occupation exerted no significant influence on the interpretation of emoji-based sarcastic statements. In contrast, sender occupation played an important role in the interpretation of emoji-based statements in ambiguous contexts. Specifically, emoji-based ambiguous statements delivered by senders in high­irony occupations were more likely to be perceived as sarcastic than by those in low-irony occupations. However, sender occupation did not affect the interpretation of the emoji; instead, it biased the judgment of emoji in sarcasm interpretation. In a follow-up experiment (Experiment 3), we investigated the perceived characteristics of both high- and low-irony occupations. The results demonstrated that individuals in high-irony occupations were stereotyped with characteristics, including being humorous, insincere, easy to setting up close relationships, and of a lower social status. Taken together, our study suggests that stereotypical information about the sender could drive the interpretation of potentially sarcastic statements, and the contextual information modulates the effect of sender occupation on sarcasm interpretation.

Irony detection in patients with borderline personality disorder: an experimental study examining schizotypal traits, response biases and empathy.

BACKGROUND: In verbal irony we often convey meanings that oppose the literal words. To look behind these words, we need to integrate perspectives of ourselves, others, and their beliefs about us. Although patients with borderline personality disorder (BPD) experience problems in social cognition and schizotypal symptoms, research on irony comprehension mainly focused on the schizophrenic spectrum. Accounting for possible negative biases in BPD, the current study examined the detection of praising and critical irony in a text messaging interface. METHODS: The cross-sectional study included 30 patients and 30 matched controls, who completed measures of cognitive and affective empathy (Interpersonal Reactivity Index, IRI), schizotypal (Schizotypal Personality Questionnaire; SPQ), and borderline symptoms (Borderline Symptom List; BSL-23) and the irony detection task. The irony task contained critical and praising remarks embedded in text messages. Asking for literality (ironic vs. literal) and intention ratings (critical to praising) of the stimuli, it allowed to analyze the sensitivity of literality detection as well as implicit and explicit response biases in a signal detection framework. RESULTS: Borderline symptoms explained lower sensitivity for the detection of literal and ironic statements across groups. Whereas HC showed a negativity bias when implicitly asked about the literalness of the statement, patients with BPD perceived praising utterances as less praising when explicitly asked about their perceived intention. Neither empathy nor schizotypy explained outcomes beyond borderline symptoms. CONCLUSIONS: This was the first study to show lower detection of verbal irony in patients with BPD. While patients were less biased when asked about the literality of a statement, they perceived praising remarks as less positive on explicit measurements. The results highlight the importance of congruent, transparent communication in promoting epistemic trust in individuals with BPD.

Respecting the Old and Loving the Young: Emoji-Based Sarcasm Interpretation Between Younger and Older Adults.

The interpretation of sarcasm relies on many cues and constraints. In computer-mediated communication (CMC), paralinguistic cues, such as emoticons and emoji, play an important role in signaling sarcastic intention. Smiling emoji have been claimed to be a marker of sarcasm among Chinese senders. Shared knowledge between the sender and the recipient, such as age and relationship, has a substantial effect on irony or sarcasm interpretation. However, hardly any research has been done to integrate the two factors to explore their joint effect on sarcasm interpretation. The present study investigated the interaction effect of these factors on the interpretation of ambiguous statements accompanied by a smiling emoji. Two experiments were conducted to investigate the differences between younger and older adults in making judgments about ambiguous statements accompanied by a smiling emoji. The results showed that sender age and sender-receiver relationship have disparate influences on younger and older adults' interpretation of emoji-based ambiguous statements. For younger adults, sender age and sender-receiver relationship were significantly associated with the perceived sarcasm of emoji-based ambiguous statements, whereas for older adults, sender age had a null effect on the sarcastic interpretation of emoji-based ambiguous statements, but relationship was an important cue that might impact their interpretation.

Acoustic cues associated with Korean sarcastic utterances produced by right- and left-hemisphere damaged individuals.

PURPOSE: Sarcasm, prevalent in everyday conversation, refers to the use of words that express negative attitudes toward persons or events. Acoustic cues associated with sarcasm have been reported to vary across studies and the relative importance of particular acoustic parameters for signaling sarcasm has not been fully determined. The hemispheric specialization for the production of acoustic cues has been a matter of controversy. This study investigated the possible prosodic cues associated with Korean sarcastic utterances and the differential effect of left hemisphere damage (LHD) or right hemisphere damage (RHD) on the production of acoustic features of Korean sarcastic utterances. METHOD: Twenty one native speakers of Korean (7 individuals with LHD, 7 individuals with RHD, and 7 healthy controls (HC)) produced six Korean utterances in two different modes: sarcastic and literal. Utterances validated by sarcasm ratings by native listeners were analyzed acoustically utilizing durational and fundamental frequency (F0) measures. RESULTS: Listeners' ratings and acoustic analyses indicated that sarcastic utterances in Korean were produced with a combination of multiple acoustic cues. Discriminant function analyses and multiple linear regression showed that LHD and RHD differentially affected the production of acoustic cues associated with sarcasm. CONCLUSION: LHD negatively affects the production of durational cues, while RHD negatively affects the production of F0 cues.

No, No One Had Fun. Individual Differences in Nonliteral Language Perception.

Nonliteral language represents a complex form of communication that can be interpreted in numerous different ways. Our study explored how individual differences in personality and communication styles affect the evaluation of literal and nonliteral language in the context of assumptions made by the Tinge Hypothesis (Dews & Winner, 1995). Participants watched videos of social interactions focusing on positive, negative, sarcastic, and jocular statements. They evaluated speaker intentions and social impressions and completed several personality and communication style questionnaires. Individual differences in empathy, defense style, and sarcasm use correlated with the accuracy of identifying speaker intent. Additionally, positive statements were rated as friendlier when compared to jocular statements, thereby supporting the Tinge Hypothesis. However, literal negative statements were rated as more friendly than sarcastic statements, which is inconsistent with the Tinge Hypothesis. The current results provide novel evidence for the Tinge Hypothesis using multimodal, dynamic stimuli and highlight the role of the individual personality of the recipient in evaluating sarcasm and jocularity.

Innate immunity impacts social-cognitive functioning in people with multiple sclerosis and healthy individuals: Implications for IL-1ra and urinary immune markers.

Social-cognitive difficulties can negatively impact interpersonal communication, shared social experience, and meaningful relationships. This pilot investigation examined the relationship between social-cognitive functioning and inflammatory markers in people with multiple sclerosis (MS) and demographically-matched healthy individuals. Additionally, we compared the immune marker profile in serum and urine-matched samples. Social cognitive functioning was objectively assessed using The Awareness of Social Inference Test - Short (TASIT-S) and subjectively assessed using self-reports of abilities in emotion recognition, emotional empathy, and cognitive theory of mind. In people with MS and healthy individuals, there were moderate-to-large negative relationships between pro-inflammatory biomarkers (serum IL-1ß, IL-17, TNF-α, IP-10, MIP-1α, and urine IP-10, MIP-1ß) of the innate immune system and social-cognitive functioning. In MS, a higher serum concentration of the anti-inflammatory marker IL-1ra was associated with better social-cognitive functioning (i.e., self-reported emotional empathy and TASIT-S sarcasm detection performance). However, there were mixed findings for anti-inflammatory serum markers IL-4 and IL-10. Overall, our findings indicate a relationship between pro-inflammatory cytokines and social-cognitive abilities. Future studies may provide greater insight into biologically-derived inflammatory processes, sickness behaviour, and their connection with social cognition.

Detecting sarcasm in multi-domain datasets using convolutional neural networks and long short term memory network model.

Sarcasm emerges as a common phenomenon across social networking sites because people express their negative thoughts, hatred and opinions using positive vocabulary which makes it a challenging task to detect sarcasm. Although various studies have investigated the sarcasm detection on baseline datasets, this work is the first to detect sarcasm from a multi-domain dataset that is constructed by combining Twitter and News Headlines datasets. This study proposes a hybrid approach where the convolutional neural networks (CNN) are used for feature extraction while the long short-term memory (LSTM) is trained and tested on those features. For performance analysis, several machine learning algorithms such as random forest, support vector classifier, extra tree classifier and decision tree are used. The performance of both the proposed model and machine learning algorithms is analyzed using the term frequency-inverse document frequency, bag of words approach, and global vectors for word representations. Experimental results indicate that the proposed model surpasses the performance of the traditional machine learning algorithms with an accuracy of 91.60%. Several state-of-the-art approaches for sarcasm detection are compared with the proposed model and results suggest that the proposed model outperforms these approaches concerning the precision, recall and F1 scores. The proposed model is accurate, robust, and performs sarcasm detection on a multi-domain dataset.

Over a decade of social opinion mining: a systematic review.

Social media popularity and importance is on the increase due to people using it for various types of social interaction across multiple channels. This systematic review focuses on the evolving research area of Social Opinion Mining, tasked with the identification of multiple opinion dimensions, such as subjectivity, sentiment polarity, emotion, affect, sarcasm and irony, from user-generated content represented across multiple social media platforms and in various media formats, like text, image, video and audio. Through Social Opinion Mining, natural language can be understood in terms of the different opinion dimensions, as expressed by humans. This contributes towards the evolution of Artificial Intelligence which in turn helps the advancement of several real-world use cases, such as customer service and decision making. A thorough systematic review was carried out on Social Opinion Mining research which totals 485 published studies and spans a period of twelve years between 2007 and 2018. The in-depth analysis focuses on the social media platforms, techniques, social datasets, language, modality, tools and technologies, and other aspects derived. Social Opinion Mining can be utilised in many application areas, ranging from marketing, advertising and sales for product/service management, and in multiple domains and industries, such as politics, technology, finance, healthcare, sports and government. The latest developments in Social Opinion Mining beyond 2018 are also presented together with future research directions, with the aim of leaving a wider academic and societal impact in several real-world applications.

Change in humor and sarcasm use based on anxiety and depression symptom severity during the COVID-19 pandemic.

BACKGROUND: Humor and sarcasm may be used as potential coping strategies during challenging times, and to improve wellbeing. We investigated changes in humor and sarcasm use during the COVID-19 pandemic in participants with varying anxiety and depression symptom severity, and in those with versus without psychiatric disorders. METHODS: Online data was collected from N = 661 Canadian adults during the height of COVID-19-related restrictions. Depression and anxiety symptom severity were assessed using the Patient Health Questionnaire (PHQ-9) and General Anxiety Disorder (GAD-7) scale. Participants were queried on current presence/absence of a psychiatric disorder, and on changes in humor and sarcasm use during compared to before the pandemic. RESULTS: Participants with any vs. no depression symptoms reported more sarcasm and humor use during the pandemic. Respondents with mild and severe vs. no anxiety symptoms reported using more sarcasm. However, those with any vs. no anxiety symptoms used less humor during the pandemic. Finally, less humor use was noted in those without vs. with a psychiatric disorder. LIMITATIONS: The use of self-report measures and a sample that may limit generalizability. CONCLUSIONS: Adults with depression symptoms used more humor and sarcasm as a potential coping strategy during COVID-19. While individuals with anxiety symptoms used more sarcasm, this was not true of humor, suggesting their decreased reliance on levity during a crisis. Humor use was greater in those with psychiatric disorders, perhaps due to self-preservation mechanisms during times of distress. Our findings have implications for using humor in therapy in individuals with mental health struggles and mood disorders.