Sarcasm_Detection/ADGCN

• Title: Affective Dependency Graph for Sarcasm Detection
• • Introduce
  • Model
  • Experimental Results
  • Installation and Environment Configuration
  • Instructions for Use
  • Reference

Title: Affective Dependency Graph for Sarcasm Detection

Introduce

In this paper, we revisit sarcasm detection from a novel perspective, so as to account for the long-range literal sentiment inconsistencies. More concretely, we explore a novel scenario of constructing an affective graph and a dependency graph for each sentence based on the affective information retrieved from external affective commonsense knowledge and the syntactical information of the sentence. Based on it, an Affective Dependency Graph Convolutional Network (ADGCN) framework is proposed to draw long-range incongruity patterns and inconsistent expressions over the context for sarcasm detection by means with interactively modeling the affective and dependency information.

Model

As demonstrated in Figure 2, the architecture of the proposed ADGCN framework contains three main components:

• Deriving affective and dependency graphs, which constructs an affective graph and a syntax-aware dependency graph for each sentence based on affective commonsense knowledge and dependency tree;
• Learning context representation, which learns the vector representations of the context with bidirectional LSTMs (Bi-LSTM);
• Learning graph representation, which leverages the affective dependencies of the context with multi-layer GCNs for sarcasm detection.

Experimental Results

Installation and Environment Configuration

• Please refer to requirements.txt
• Download glove.42B.300d.txt from glove
• pip3 install https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-2.2.0/en_core_web_sm-2.2.0.tar.gz

Instructions for Use

python3 dependency_graph.py
python3 train.py

Reference

@inproceedings{
        author = {Lou, Chenwei and Liang, Bin and Gui, Lin and He, Yulan and Dang, Yixue and Xu, Ruifeng},
        title = {Affective Dependency Graph for Sarcasm Detection},
        year = {2021},
        publisher = {Association for Computing Machinery},
        address = {New York, NY, USA},
        url = {https://doi.org/10.1145/3404835.3463061},
        booktitle = {Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval},
        pages = {1844–1849},
        series = {SIGIR '21}
}