Graph-Based Knowledge Integration for Question Answering over Dialogue

Liu, Jian; Sui, Dianbo; Liu, Kang; Zhao, Jun

doi:10.18653/v1/2020.coling-main.219

Cited by 11 publications

(7 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is essential that the ESC system chooses an appropriate strategy based on the seeker's mental states and generates a strategy-constrained response. Inspired by DialogEIN ( 45 ), we propose the strategy tendency encoder to capture the tendency of each utterance and the latent strategy information. As shown in Figure 3 , the embedding of each category is depicted by the circles with different colors.…”

Section: Methodsmentioning

confidence: 99%

Enhancing the conversational agent with an emotional support system for mental health digital therapeutics

et al. 2023

View full text Add to dashboard Cite

As psychological diseases become more prevalent and are identified as the leading cause of acquired disability, it is essential to assist people in improving their mental health. Digital therapeutics (DTx) has been widely studied to treat psychological diseases with the advantage of cost savings. Among the techniques of DTx, a conversational agent can interact with patients through natural language dialog and has become the most promising one. However, conversational agents' ability to accurately show emotional support (ES) limits their role in DTx solutions, especially in mental health support. One of the main reasons is that the prediction of emotional support systems does not extract effective information from historical dialog data and only depends on the data derived from one single-turn interaction with users. To address this issue, we propose a novel emotional support conversation agent called the STEF agent that generates more supportive responses based on a thorough view of past emotions. The proposed STEF agent consists of the emotional fusion mechanism and strategy tendency encoder. The emotional fusion mechanism focuses on capturing the subtle emotional changes throughout a conversation. The strategy tendency encoder aims at foreseeing strategy evolution through multi-source interactions and extracting latent strategy semantic embedding. Experimental results on the benchmark dataset ESConv demonstrate the effectiveness of the STEF agent compared with competitive baselines.

show abstract

Section: Methodsmentioning

confidence: 99%

Enhancing the conversational agent with an emotional support system for mental health digital therapeutics

et al. 2023

View full text Add to dashboard Cite

show abstract

“…A graph node represents an entity, an edge represents the relationship between nodes, and the basic composition unit is the triple (SPO) composed of entities and relationships. [10]. The authors implemented several preprocessing steps on the unstructured text, including Chinese word segmentation, stop word removal, and semantic annotation.…”

Section: Knowledge Graph Methodsmentioning

confidence: 99%

The Advance of Multi-Round Dialogue System with Deep Learning

Yang

Zhang³

et al. 2023

ACE

View full text Add to dashboard Cite

Multi-turn dialog systems have seen significant advances in recent years, driven by various approaches. The Multi-layer Semantic Method, Reinforcement Learning Method, Knowledge Graph Method, and Medicine Knowledge Graph Method have all shown promising results in advancing the state-of-the-art in this field. However, challenges remain in developing models that can handle complex and diverse user inputs and generate responses that are not only informative but also engaging and natural. Further research is needed to address these challenges and advance state of art in multi-turn dialogue systems. This paper reviews four critical methods for improving the quality of multi-turn dialogue systems: Multi-layer Semantic Method, Reinforcement Learning Method, Knowledge Graph Method, and Medicine Knowledge Graph Method. The Multi-layer Semantic Method utilizes multi-layer neural networks to model dialogue context and generate responses with improved coherence and relevance. The reinforcement Learning Method employs a reward-based approach to optimize response generation by training models to maximize long-term dialogue success. The knowledge Graph Method incorporates external knowledge sources, such as knowledge graphs, to enrich the dialogue context and improve response quality. The Medicine Knowledge Graph Method focuses on integrating medical knowledge into dialogue systems for healthcare applications. Each of these methods has demonstrated promising results in enhancing the quality of multi-turn dialogue systems.

show abstract

“…TF-IDF [43] 0.279 0.536 0.542 Dual LSTM [43] 0.260 0.491 0.743 SMN [12] 0.299 0.585 0.595 DAM [16] 0.241 0.465 0.518 GPT-2 [73] 0.332 0.602 0.584 GPT-2-FT [73] 0.392 0.670 0.629 BERT [24] 0.648 0.847 0.795 RoBERTa [25] 0.713 0.892 0.836 RoBERTa + OCN [69] 0.867 0.958 0.926 ALBERT [74] 0.847 0.962 0.916 GRN-v2 [74] 0 revealing the strength of BiGRU in modeling the sentencelevel transition between the turns of utterances.…”

Section: Analysis a Ablation Studymentioning

confidence: 99%

Channel-Aware Decoupling Network for Multiturn Dialog Comprehension

Zhang

Zhao

Liu

2024

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

Training machines to understand natural language and interact with humans is one of the major goals of artificial intelligence. Recent years have witnessed an evolution from matching networks to pre-trained language models (PrLMs). In contrast to the plain-text modeling as the focus of the PrLMs, dialogue texts involve multiple speakers and reflect special characteristics such as topic transitions and structure dependencies between distant utterances. However, the related PrLM models commonly represent dialogues sequentially by processing the pairwise dialogue history as a whole. Thus the hierarchical information on either utterance interrelation or speaker roles coupled in such representations is not well addressed. In this work, we propose compositional learning for holistic interaction across the utterances beyond the sequential contextualization from PrLMs, in order to capture the utteranceaware and speaker-aware representations entailed in a dialogue history. We decouple the contextualized word representations by masking mechanisms in Transformer-based PrLM, making each word only focus on the words in current utterance, other utterances, and two speaker roles (i.e., utterances of sender and utterances of the receiver), respectively. In addition, we employ domain-adaptive training strategies to help the model adapt to the dialogue domains. Experimental results show that our method substantially boosts the strong PrLM baselines in four public benchmark datasets, achieving new state-of-the-art performance over previous methods.

show abstract

Graph-Based Knowledge Integration for Question Answering over Dialogue

Cited by 11 publications

References 21 publications

Enhancing the conversational agent with an emotional support system for mental health digital therapeutics

Enhancing the conversational agent with an emotional support system for mental health digital therapeutics

The Advance of Multi-Round Dialogue System with Deep Learning

Channel-Aware Decoupling Network for Multiturn Dialog Comprehension

Contact Info

Product

Resources

About