Abstract:Sarcasm is common in social media, and people use it to express their opinions with stronger emotions indirectly. Although it belongs to a branch of sentiment analysis, traditional sentiment analysis methods cannot identify the rhetoric of irony as it requires a significant amount of background knowledge. Existing sarcasm detection approaches mainly focus on analyzing the text content of sarcasm using various natural language processing techniques. It is argued herein that the essential issue for detecting sarcasm is examining its context, including sentiments of texts that reply to the target text and user's expression habit. A dual-channel convolutional neural network is proposed that analyzes not only the semantics of the target text, but also its sentimental context. In addition, SenticNet is used to add common sense to the long short-term memory (LSTM) model. The attention mechanism is then applied to take the user's expression habits into account. A series of experiments were carried out on several public datasets, the results of which show that the proposed approach can significantly improve the performance of sarcasm detection tasks.

What problem does this paper attempt to address?

This paper aims to solve the problem of sarcasm detection in social media. Sarcasm is a common rhetorical device through which people indirectly express their views with stronger emotions. Although sarcasm is a branch of sentiment analysis, traditional sentiment analysis methods cannot identify sarcasm because it requires a large amount of background knowledge, which is difficult to obtain in practical applications. ### Specific problems the paper attempts to solve 1. **Identification of emotional context**: - Existing sarcasm detection methods mainly focus on the analysis of text content and use various natural language processing techniques to detect sarcasm. However, these methods often overlook the importance of emotional context. For example, Bob's sarcastic reply has a different emotion from Alice's post. Analyzing only Bob's reply may lead to incorrect results. Therefore, the paper proposes that the emotional context, that is, the emotion of the reply text and the user's emotional habits, needs to be considered. 2. **Consideration of user expression habits**: - Different users may use different phrases when expressing sarcasm. The same phrase may have different sarcastic meanings in different users' mouths. In addition, users usually use similar rhetorical devices to express opinions on social media platforms, showing consistent expression habits. Therefore, the paper emphasizes the need to consider user expression habits. ### Solutions To effectively solve the above problems, the paper proposes a method based on a two - channel convolutional neural network (CNN), which not only analyzes the semantics of the target text but also its emotional context. Specifically: 1. **Emotional context - inconsistent feature embedding**: - Use CNN to extract the emotional features of the target text and the context, and generate emotional - inconsistent features by comparing the emotional polarity labels of the two. These features are embedded into word vectors to form an emotional context - inconsistent feature embedding (SCIFE) model. 2. **User expression habit feature engineering**: - Use bidirectional LSTM (Bi - LSTM) combined with an attention mechanism to encode user expression habits. SenticNet is used to add common - sense information and enhance the model's ability to encode emotional associations. 3. **Multi - dimensional information fusion**: - Integrate multi - dimensional information such as semantics, emotional context, and user habits, and train a neural network classification model for sarcasm detection. ### Experimental results The paper conducted experiments on multiple public datasets, and the results showed that the proposed model significantly improved the performance in the sarcasm detection task, with the average F1 scores reaching 74.5% and 71% respectively, an average improvement of 7% over the baseline methods. ### Summary This paper proposes an effective sarcasm detection method by combining emotional context and user expression habits. This method not only improves the accuracy of sarcasm detection but also provides new directions for future research, such as the application of semi - supervised learning and multi - task learning.

An Effective Sarcasm Detection Approach Based on Sentimental Context and Individual Expression Habits

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

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

Multi-Modal Sarcasm Detection with Sentiment Word Embedding

A contextual-based approach for sarcasm detection

Sarcasm Detection over Social Media Platforms Using Hybrid Ensemble Model with Fuzzy Logic

Multi-modal sarcasm detection based on emotion perception and cross-modality attention fusion

Sarcasm detection using optimized bi-directional long short-term memory

BNS-Net: A Dual-channel Sarcasm Detection Method Considering Behavior-level and Sentence-level Conflicts

Sememe knowledge and auxiliary information enhanced approach for sarcasm detection

Sarcasm Detection: A Comparative Study

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

Sarcasm Analysis Using Conversation Context

A hybrid convolutional neural network for sarcasm detection from multilingual social media posts

Interpretable Multi-Head Self-Attention Architecture for Sarcasm Detection in Social Media

N-Gram Based Sarcasm Detection for News and Social Media Text Using Hybrid Deep Learning Models

A survey of automatic sarcasm detection: Fundamental theories, formulation, datasets, detection methods, and opportunities

Sarcasm driven by sentiment: A sentiment-aware hierarchical fusion network for multimodal sarcasm detection

Dual-level adaptive incongruity-enhanced model for multimodal sarcasm detection

A Multi-Level Embedding Framework for Decoding Sarcasm Using Context, Emotion, and Sentiment Feature

KnowleNet: Knowledge fusion network for multimodal sarcasm detection