Emotion recognition in conversation (ERC) is a sub-field ofemotion recognition, that focuses on mining humanemotions from conversations ordialogues having two or moreinterlocutors.[1] The datasets in this field are usually derived fromsocial platforms that allow free and plenty of samples, often containingmultimodal data (i.e., some combination of textual, visual, and acoustic data).[2] Self- and inter-personal influences play critical role[3] in identifying some basic emotions, such as,fear,anger, joy, surprise, etc. The more fine grained the emotion labels are the harder it is to detect the correct emotion. ERC poses a number of challenges,[1] such as, conversational-context modeling, speaker-state modeling, presence of sarcasm in conversation, emotion shift across consecutive utterances of the sameinterlocutor.
The task of ERC deals with detecting emotions expressed by the speakers in each utterance of the conversation. ERC depends on three primaryfactors – the conversational context,interlocutors' mental state, and intent.[1]
IEMOCAP,[4] SEMAINE,[5] DailyDialogue,[6] and MELD[7] are the four widely used datasets in ERC. Among these four datasets, MELD contains multiparty dialogues.
Approaches to ERC consist ofunsupervised, semi-unsupervised, andsupervised[8] methods. Popular supervised methods include using or combining pre-defined features,recurrent neural networks[9] (DialogueRNN[10]), graph convolutional networks[11] (DialogueGCN[12]), and attention gated hierarchical memory network.[13] Most of the contemporary methods for ERC are deep learning based and rely on the idea of latent speaker-state modeling.
Recently a new subtask of ERC has emerged that focuses on recognising emotion cause in conversation.[14] Methods to solve this task rely on language models-based question answering mechanism. RECCON[14] is one of the key datasets for this task.
^Hazarika, Devamanyu; Poria, Soujanya; Zimmermann, Roger; Mihalcea, Rada (Oct 2019). "Emotion Recognition in Conversations with Transfer Learning from Generative Conversation Modeling".arXiv:1910.04980 [cs.CL].
^Li, Yanran, Hui Su, Xiaoyu Shen, Wenjie Li, Ziqiang Cao, and Shuzi Niu. "DailyDialog: A Manually Labelled Multi-turn Dialogue Dataset." InProceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pp. 986-995. 2017.
^Poria, Soujanya; Hazarika, Devamanyu; Majumder, Navonil; Naik, Gautam; Cambria, Erik; Mihalcea, Rada (2019). "MELD: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversations".Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics:527–536.arXiv:1810.02508.doi:10.18653/v1/p19-1050.S2CID52932143.
^Abdelwahab, Mohammed; Busso, Carlos (March 2005). "Supervised domain adaptation for emotion recognition from speech".2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). pp. 5058–5062.doi:10.1109/ICASSP.2015.7178934.ISBN978-1-4673-6997-8.S2CID8207841.{{cite book}}:|journal= ignored (help)
^Chernykh, Vladimir; Prikhodko, Pavel; King, Irwin (Jul 2019). "Emotion Recognition From Speech With Recurrent Neural Networks".arXiv:1701.08071 [cs.CL].
^Ghosal, Deepanway; Majumder, Navonil; Soujanya, Poria (Aug 2019).DialogueGCN: A Graph Convolutional Neural Network for Emotion Recognition in Conversation. Conference on Empirical Methods in Natural Language Processing (EMNLP).
^Jiao, Wenxiang; R. Lyu, Michael; King, Irwin (November 2019). "Real-Time Emotion Recognition via Attention Gated Hierarchical Memory Network".arXiv:1911.09075 [cs.CL].
