S+PAGE: A Speaker and Position-Aware Graph Neural Network Model for Emotion Recognition in Conversation

Chen, Liang; Chong, Yap Seng; Xu, Jing; Huang, Juyang; Wang, Yongliang; Yang, Dong

doi:10.48550/arxiv.2112.12389

Cited by 4 publications

(4 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The study conducted by Shen et al [24] introduces DialogXL, an improved version of XLNET [25], which incorporates sophisticated memory processes and dialog-aware self-attention. The Speaker and Position-aware Graph Neural Network (GNN) model, introduced by Liang et al [26] under the name S+PAGE, serves as a pioneering approach in the field. This model is specifically tailored to incorporate interspeaker and intra-speaker contextual dynamics into conversational graphs.…”

Section: A Emotion Recognition In Conversationmentioning

confidence: 99%

Residual Relation-Aware Attention Deep Graph-Recurrent Model for Emotion Recognition in Conversation

Duong,

Ho,

Pant

et al. 2024

IEEE Access

View full text Add to dashboard Cite

This work addresses Emotion Recognition in Conversation (ERC), a task with substantial implications for the classification of the underlying emotions in spoken encounters. Our focus is on utilizing a fully connected directed acyclic graph to represent conversations, presenting inter-locutor and intra-locutor ties to capture intricate relationships. Therefore, we propose a novel methodology, Residual Relation-Aware Attention (RRAA) with Positional Encoding, enhancing speaker relations' contexts for improved emotion recognition in conversation. The purpose of this mechanism is to facilitate a thorough comprehension of the connections between speakers, hence enhancing the sophistication and contextual awareness of an emotion recognition framework. We utilized the Gated recurrent units (GRU) to regulate context transmission, ensuring adaptability to changing emotional dynamics. It regulates the transmission of conversation context across all layers of the graph, guaranteeing a flexible and responsive representation of the changing emotional dynamics within the discourse. Evaluations on IEMOCAP, MELD, and EmoryNLP datasets disclose our model's superior performance (F1 scores: 69.1%, 63.82%, 39.85%, respectively), outperforming state-ofthe-art approaches. In general, this work enhances speaker interactions by utilizing a fully connected graph, and providing a more concise and efficient ERC framework. INDEX TERMSEmotion recognition in conversation, residual relation-aware attention, deep graphrecurrent model, fully connected directed acyclic graph, positional encoding, .

show abstract

Section: A Emotion Recognition In Conversationmentioning

confidence: 99%

Residual Relation-Aware Attention Deep Graph-Recurrent Model for Emotion Recognition in Conversation

Duong,

Ho,

Pant

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Another branch of work leverages the strong context modelling ability of Transformer-based networks to model the dialogue as a whole [10], [30], [31]. To introduce more interpretable structures, there are also many works [32], [33], [34] that construct a graph on the dialogue, and devise graph neural networks to model ERC as a node-classification task.…”

Section: Emotion Recognition In Conversationsmentioning

confidence: 99%

Cluster-Level Contrastive Learning for Emotion Recognition in Conversations

Yang

Zhang

Alhuzali

et al. 2023

IEEE Trans. Affective Comput.

View full text Add to dashboard Cite

A key challenge for Emotion Recognition in Conversations (ERC) is to distinguish semantically similar emotions. Some works utilise Supervised Contrastive Learning (SCL) which uses categorical emotion labels as supervision signals and contrasts in high-dimensional semantic space. However, categorical labels fail to provide quantitative information between emotions. ERC is also not equally dependent on all embedded features in the semantic space, which makes the high-dimensional SCL inefficient. To address these issues, we propose a novel low-dimensional Supervised Cluster-level Contrastive Learning (SCCL) method, which first reduces the high-dimensional SCL space to a three-dimensional affect representation space Valence-Arousal-Dominance (VAD), then performs cluster-level contrastive learning to incorporate measurable emotion prototypes. To help modelling the dialogue and enriching the context, we leverage the pre-trained knowledge adapters to infuse linguistic and factual knowledge. Experiments show that our method achieves new state-of-the-art results with 69.81% on IEMOCAP, 65.7% on MELD, and 62.51% on DailyDialog datasets. The analysis also proves that the VAD space is not only suitable for ERC but also interpretable, with VAD prototypes enhancing its performance and stabilising the training of SCCL. In addition, the pre-trained knowledge adapters benefit the performance of the utterance encoder and SCCL. Our code will be available online upon acceptance.

show abstract

“…DialogXL [21] used XLNet, which is an extension of Transformer-XL, to improve understanding of the context. S+PAGE [4] sought to understand the context using both Transformer and the R-GCN [22], and it is currently the state-of-the-art model on one of the benchmark datasets. However, graph-based models still suffer from capturing dependencies between distant utterances or sequential information.…”

Section: B Gnn Based Modelmentioning

confidence: 99%

Directed Acyclic Graphs With Prototypical Networks for Few-Shot Emotion Recognition in Conversation

Kang,

Cho

2023

IEEE Access

View full text Add to dashboard Cite

Emotion recognition in conversation is the task of recognizing the emotion in each utterance of a conversation, and it is an active field of study with various applications. Many studies have been conducted in well-designed settings in which all of the emotion labels in the training set are available. However, a rich dataset with emotion labels for all utterances is rare. In this study, we address this problem by resorting to few-shot learning. Specifically, we propose Directed Acyclic Graph (DAG)-based prototypical networks, namely ProtoDAG, to better capture contextual information in conversations, which in turn leads to accurate emotion predictions. In our model, the DAG layers are tailored into prototypical networks, which is learned end-to-end. Our experiments with popular benchmark datasets demonstrate that our model achieves stateof-the-art results outperforming the existing few-shot learning model by a significant margin and is even competitive with fully supervised baseline models for emotion recognition in conversation.

show abstract

S+PAGE: A Speaker and Position-Aware Graph Neural Network Model for Emotion Recognition in Conversation

Cited by 4 publications

References 28 publications

Residual Relation-Aware Attention Deep Graph-Recurrent Model for Emotion Recognition in Conversation

Residual Relation-Aware Attention Deep Graph-Recurrent Model for Emotion Recognition in Conversation

Cluster-Level Contrastive Learning for Emotion Recognition in Conversations

Directed Acyclic Graphs With Prototypical Networks for Few-Shot Emotion Recognition in Conversation

Contact Info

Product

Resources

About