Affective Dependency Graph for Sarcasm Detection

Lou, Chenwei; Liang, Bin; Gui, Lin; He, Yulan; Dang, Yixue; Xu, Ruifeng

doi:10.1145/3404835.3463061

Cited by 34 publications

(26 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As attention mechanism has led to improvements in various NLP tasks, Tay et al (2018); Xiong et al (2019) use attention to capture the relationship of word pairs along with an LSTM to model the entire sentence. Lou et al (2021) design a GCN-based model combining SenticNet (Cambria et al, 2020), dependency tree and LSTM with GCN (Kipf and Welling, 2017) together, which achieves promising performance. Similar to previous studies, to better understand sarcasm, many approaches are able to utilize external information such as emoji expressions (Felbo et al, 2017), affective knowledge (Babanejad et al, 2020) and commonsense (Li et al, 2021).…”

Section: Neural Modelsmentioning

confidence: 99%

“…ADGCN (Lou et al, 2021) is a GCN-based method with sentic graph and dependency graph 7 . The initial input of GCN is the hidden state of Bi-LSTM.…”

Section: Cnn-lstm-dnnmentioning

confidence: 99%

“…Many existing studies adopt generic classification models for sarcasm recognition (Lou et al, 2021;Ghosh and Veale, 2016). However, these methods directly model the entire sentence without considering the contradictory meanings behind sarcastic texts.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Dual-Channel Framework for Sarcasm Recognition by Detecting Sentiment Conflict

Liu¹,

Wang²,

Sun³

et al. 2022

Findings of the Association for Computational Linguistics: NAACL 2022

View full text Add to dashboard Cite

Sarcasm employs ambivalence, where one says something positive but actually means negative, and vice versa. The essence of sarcasm, which is also a sufficient and necessary condition, is the conflict between literal and implied sentiments expressed in one sentence. However, it is difficult to recognize such sentiment conflict because the sentiments are mixed or even implicit. As a result, the recognition of sophisticated and obscure sentiment brings in a great challenge to sarcasm detection. In this paper, we propose a Dual-Channel Framework by modeling both literal and implied sentiments separately. Based on this dual-channel framework, we design the Dual-Channel Network (DC-Net) to recognize sentiment conflict. Experiments on political debates (i.e., IAC-V1 and IAC-V2) and Twitter datasets show that our proposed DC-Net achieves state-of-the-art performance on sarcasm recognition. Our code is released to support research 1 .

show abstract

Section: Neural Modelsmentioning

confidence: 99%

“…ADGCN (Lou et al, 2021) is a GCN-based method with sentic graph and dependency graph 7 . The initial input of GCN is the hidden state of Bi-LSTM.…”

Section: Cnn-lstm-dnnmentioning

confidence: 99%

See 1 more Smart Citation

A Dual-Channel Framework for Sarcasm Recognition by Detecting Sentiment Conflict

Liu¹,

Wang²,

Sun³

et al. 2022

Findings of the Association for Computational Linguistics: NAACL 2022

View full text Add to dashboard Cite

show abstract

“…Sarcasm detection had been implemented by taking the contextual information of a sentence into account in a sequential manner using the concept of pseudo-labeling (Kumar Jena et al, 2020), (Kalaivani and Thenmozhi, 2020). Long-range literal sentiment inconsistencies had been taken into account in sarcasm detection by constructing an affective graph and a dependency graph for each sentence and had then used an Affective Dependency Graph Convolutional Network (ADGCN) framework for the classification process (Lou et al, 2021). Statistical approach had been proposed for sarcasm detection by combining TF-IDF features with the important features related to sentiments and punctuations that are identified using chi-square test (Gupta et al, 2020 such case had been implemented using a multimodal framework using Coupled-Attention Networks (CANs) which captures and integrates information from both text and image for the classification task (Zhao et al, 2021).…”

Section: Related Workmentioning

confidence: 99%

GetSmartMSEC at SemEval-2022 Task 6: Sarcasm Detection using Contextual Word Embedding with Gaussian model for Irony Type Identification

Krishnan¹,

C²,

Thenmozhi³

2022

Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022)

View full text Add to dashboard Cite

Sarcasm refers to the use of words that have different literal and intended meanings. It represents the usage of words that are opposite of what is literally said, especially in order to insult, mock, criticise or irritate someone. These types of statements may be funny or amusing to others but may hurt or annoy the person towards whom it is intended. Identification of sarcastic phrases from social media posts finds its application in different domains like sentiment analysis, opinion mining, author profiling and harassment detection. We have proposed a model for the shared task iSarcasmEval -Intended Sarcasm Detection in English and Arabic by SemEval-2022 considering the language English. The Subtask A and Subtask C were implemented using a Convolutional Neural Network based classifier which makes use of ELMo embeddings. The Subtask B was implemented using Gaussian Naive Bayes classifier by extracting TF-IDF vectors. The proposed models resulted in macro-F1 scores of 0.2012, 0.0387 and 0.2794 for sarcastic texts in Subtasks A, B and C respectively.

show abstract

“…Then, we explore a novel solution to assign weights to the edges of the cross-modal graph by means of computing the word similarities between the object descriptors of the attribute-object pairs and textual words based on the WordNet (Miller, 1992). Further, to introduce the multi-modal sentiment relations into the cross-modal graphs, inspired by (Lou et al, 2021), we devise a modulat-ing factor of sentiment relation for each edge by retrieving the affective weights of attribute descriptors (usually adjectives with affective information) and textual words from external affective knowledge (SenticNet (Cambria et al, 2020)). As such, the modulating factors can be adopted to refine the edge weights of word similarities, allowing the capture of sentiment incongruities of the cross-modal nodes in the graph.…”

Section: Introductionmentioning

confidence: 99%

Multi-Modal Sarcasm Detection via Cross-Modal Graph Convolutional Network

Liang¹,

Lou²,

Li³

et al. 2022

Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

Self Cite

View full text Add to dashboard Cite

With the increasing popularity of posting multimodal messages online, many recent studies have been carried out utilizing both textual and visual information for multi-modal sarcasm detection. In this paper, we investigate multimodal sarcasm detection from a novel perspective by constructing a cross-modal graph for each instance to explicitly draw the ironic relations between textual and visual modalities. Specifically, we first detect the objects paired with descriptions of the image modality, enabling the learning of important visual information. Then, the descriptions of the objects are served as a bridge to determine the importance of the association between the objects of image modality and the contextual words of text modality, so as to build a cross-modal graph for each multi-modal instance. Furthermore, we devise a cross-modal graph convolutional network to make sense of the incongruity relations between modalities for multi-modal sarcasm detection. Extensive experimental results and in-depth analysis show that our model achieves state-of-the-art performance in multi-modal sarcasm detection 1 .

show abstract

Affective Dependency Graph for Sarcasm Detection

Cited by 34 publications

References 34 publications

A Dual-Channel Framework for Sarcasm Recognition by Detecting Sentiment Conflict

A Dual-Channel Framework for Sarcasm Recognition by Detecting Sentiment Conflict

GetSmartMSEC at SemEval-2022 Task 6: Sarcasm Detection using Contextual Word Embedding with Gaussian model for Irony Type Identification

Multi-Modal Sarcasm Detection via Cross-Modal Graph Convolutional Network

Contact Info

Product

Resources

About