Sentence encoding for Dialogue Act classification

Duran, Nathan; Battle, Steve; Smith, Jim Q.

doi:10.1017/s1351324921000310

Cited by 7 publications

(7 citation statements)

References 61 publications

(230 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…utilize attention mechanisms and time seriesLSTM for act recognition. Duran et al [14].during the encoding phase, integrate act and sentence encoding using BERT, systematically comparing and validating various encoding mechanisms' performance differences in act classification. Malhotra et al [15].…”

Section: Based On Sequence Labelingmentioning

confidence: 99%

A Personalized Model Integrating Personality for Joint Dialogue Emotion Classification and Act Recognition

Ma,

Zhou,

et al. 2024

Preprint

View full text Add to dashboard Cite

The joint dialogue emotion classification and act recognition tasks essentially involves the simultaneous classification of emotional and act labels for each utterance in a dialogue. Existing dual-tasks models can recognize emotion and act based on utterance context but overlook the interaction between discourse under the influence of personality and the correlation between dual-tasks. Utter- ances within the same personality and across different personalities exhibit explicit temporal dependencies in the past and future. However, existing models fail to explicitly address the temporal relationships in the interaction between personality and dual-tasks, resulting in a reasoning gap in the temporal corre- lation of interactions under long-term personality. In this paper, A Personalized Model Integrating Personality for Joint Dialogue Emotion Classification and Act RecognitionPEDR-GATis proposed to solve these problems.Firstly, we uti- lize personality information and label distributions to provide initial estimates in semantics, strengthening the association between local semantics guided by personality and task labels. Secondly, the model employs graph attention mech- anisms to extract global semantic temporal information and causal temporal relationships between the two tasks. Finally, PEDR-GAT, by acquiring initial local label information guided by personality and integrating personality-guided global semantic label information, introduces a common attention mechanism to maintain long-term dependencies and supplement short-term information.The experiments were conducted on three benchmark datasets: Mastodon social network (Mastodon), A Manually Labelled Multi-turn Dialogue Dataset (Dai- lyDialog), and A Large-Scale Chinese Personalized and Emotional Dialogue Dataset for Conversational AI (CPED). Compared to the baseline models, there were improvements in F1 scores for emotion classification and act recognition by 1.9% and 0.7% (Mastodon),3.3% and 1.3% (DailyDialog), and 5.3% and 2.3%(CPED), respectively. The results indicate that personality-guided semantic label acquisition enhances the association between semantics and labels, and the graph attention mechanism with personality fusion reasoning can stimulate the cognitive system’s understanding of potential causal information in abstraction.

show abstract

Section: Based On Sequence Labelingmentioning

confidence: 99%

A Personalized Model Integrating Personality for Joint Dialogue Emotion Classification and Act Recognition

Ma,

Zhou,

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Context is either modeled through a condensed vector of contextual cues (e.g., Chen et al, 2018;Colombo et al, 2020;Kumar et al, 2018;Raheja & Tetreault, 2019) or directly in the neural network structure with a node for each utterance (e.g., Cerisara et al, 2018;Kalchbrenner & Blunsom, 2013;Ortega & Vu, 2017;Ribeiro et al, 2019b). Some studies that used deep learning for dialog act classification did not include contextual cues (e.g., Duran & Battle, 2018;Duran et al, 2023;Khanpour et al, 2016;Ribeiro et al, 2019a), but most encoded context through using a condensed surface encoding of the previous utterances (e.g., Chen et al, 2018;Kumar et al, 2018;Yano et al, 2021;Zhao & Kawahara, 2019). Most deep learning models have regarded dialog act classification as classifying a sequence of dialog acts, without paying attention to the speaker or the structure of utterances into turns.…”

Section: Approaches To Dialog Act Classificationmentioning

confidence: 99%

“…Similarly, many different contextual aspects, ranging from predicted speaker intentions of previous utterances to cues on who is or was the speaker of the current or previous utterances, have been successfully used. But while all studies make use of one or more surface cues, not all studies make use of contextual cues (e.g., Ang et al., 2005; Duran & Battle, 2018; Duran et al., 2023; Novielli & Strapparava, 2009).…”

Section: Identifying Cues In Existing Dialog Act Classification Studiesmentioning

confidence: 99%

“…Noncontextualized distributional semantic approaches to encoding surface information might be suboptimal, because dialog acts seem to rely more on the statistical regulations of words in an utterance than on the semantic interpretation of the utterance (Cerisara et al., 2018; Duran & Battle 2018; Milajevs & Purver, 2014). More recently, also contextualized word embeddings, such as BERT (Duran et al., 2023; Ribeiro et al., 2019b) have been used in deep learning models. The main difference between noncontextualized and contextualized word embeddings is that contextualized embeddings encode full sentences at once, thus preserving word order information, and subsequently, the meaning of the whole sentence, while noncontextualized embeddings encode individual words, thereby fully relying on the semantic interpretation of individual words.…”

Section: Identifying Cues In Existing Dialog Act Classification Studiesmentioning

confidence: 99%

See 1 more Smart Citation

Surface and Contextual Linguistic Cues in Dialog Act Classification: A Cognitive Science View

Linders,

Louwerse

2023

Cognitive Science

View full text Add to dashboard Cite

What role do linguistic cues on a surface and contextual level have in identifying the intention behind an utterance? Drawing on the wealth of studies and corpora from the computational task of dialog act classification, we studied this question from a cognitive science perspective. We first reviewed the role of linguistic cues in dialog act classification studies that evaluated model performance on three of the most commonly used English dialog act corpora. Findings show that frequency‐based, machine learning, and deep learning methods all yield similar performance. Classification accuracies, moreover, generally do not explain which specific cues yield high performance. Using a cognitive science approach, in two analyses, we systematically investigated the role of cues in the surface structure of the utterance and cues of the surrounding context individually and combined. By comparing the explained variance, rather than the prediction accuracy of these cues in a logistic regression model, we found that (1) while surface and contextual linguistic cues can complement each other, surface linguistic cues form the backbone in human dialog act identification, (2) with word frequency statistics being particularly important for the dialog act, and (3) the similar trends across corpora, despite differences in the type of dialog, corpus setup, and dialog act tagset. The importance of surface linguistic cues in dialog act classification sheds light on how both computers and humans take advantage of these cues in speech act recognition.

show abstract

“…Most existing studies on dialogue classification focus on sentence-level or utterancelevel intent recognition of user statements [10]. These studies commonly employ hierarchical neural networks to model the sequential and structural information within words, characters, and utterances [11].…”

Section: Dialogue Classificationmentioning

confidence: 99%

DialGNN: Heterogeneous Graph Neural Networks for Dialogue Classification

Yan

Zhang

Min

et al. 2023

Preprint

View full text Add to dashboard Cite

Dialogue systems have attracted growing research interests due to its widespread applications in various domains. Existing studies on classification tasks in dialogue systems majorly focus on the sentence-level intent recognition of users’ utterances. While in real-world applications, classification of the entire dialogue, also benefits many downstream tasks such as customer satisfaction analysis, service quality assurance, dialogue topic categorization, etc. In this paper, we propose DialGNN, a heterogeneous graph neural network framework tailored for the problem of dialogue classification which takes the entire dialogue as input. Specifically, a heterogeneous graph is constructed with nodes in different levels of semantic granularity. The graph framework allows flexible integration of various pre-trained language representation models, such as BERT and its variants, which endows DialGNN with powerful text representational capabilities. The experimental results on two real-world datasets demonstrate the robustness and the effectiveness of the proposed DialGNN framework. The implementation of DialGNN and related data are shared through https://github.com/anonymous-auth/ DialGNN.

show abstract

Sentence encoding for Dialogue Act classification

Cited by 7 publications

References 61 publications

A Personalized Model Integrating Personality for Joint Dialogue Emotion Classification and Act Recognition

A Personalized Model Integrating Personality for Joint Dialogue Emotion Classification and Act Recognition

Surface and Contextual Linguistic Cues in Dialog Act Classification: A Cognitive Science View

DialGNN: Heterogeneous Graph Neural Networks for Dialogue Classification

Contact Info

Product

Resources

About