Controllable Dialogue Generation with Disentangled Multi-grained Style Specification and Attribute Consistency Reward

• URL: http://arxiv.org/abs/2109.06717v1
• Date: Tue, 14 Sep 2021 14:29:38 GMT
• Title: Controllable Dialogue Generation with Disentangled Multi-grained Style Specification and Attribute Consistency Reward
• Authors: Zhe Hu, Zhiwei Cao, Hou Pong Chan, Jiachen Liu, Xinyan Xiao, Jinsong Su, Hua Wu
• Abstract summary: We propose a controllable dialogue generation model to steer response generation under multi-attribute constraints. We categorize the commonly used control attributes into global and local ones, which possess different granularities of effects on response generation. Our model can significantly outperform competitive baselines in terms of response quality, content diversity and controllability.
• Score: 47.96949534259019
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Controllable text generation is an appealing but challenging task, which allows users to specify particular attributes of the generated outputs. In this paper, we propose a controllable dialogue generation model to steer response generation under multi-attribute constraints. Specifically, we define and categorize the commonly used control attributes into global and local ones, which possess different granularities of effects on response generation. Then, we significantly extend the conventional seq2seq framework by introducing a novel two-stage decoder, which first uses a multi-grained style specification layer to impose the stylistic constraints and determine word-level control states of responses based on the attributes, and then employs a response generation layer to generate final responses maintaining both semantic relevancy to the contexts and fidelity to the attributes. Furthermore, we train our model with an attribute consistency reward to promote response control with explicit supervision signals. Extensive experiments and in-depth analyses on two datasets indicate that our model can significantly outperform competitive baselines in terms of response quality, content diversity and controllability.

Related papers

• Evaluating the Smooth Control of Attribute Intensity in Text Generation with LLMs [36.89780636600556]
  Large language models (LLMs) have revolutionized text generation. We propose metrics to assess the range, calibration, and consistency of the generated text's attribute intensity.
  arXiv  Detail & Related papers  (2024-06-06T19:35:51Z)
• Continuous, Subject-Specific Attribute Control in T2I Models by Identifying Semantic Directions [21.371773126590874]
  We show that there exist directions in the commonly used token-level CLIP text embeddings that enable fine-grained subject-specific control of high-level attributes in text-to-image models. We introduce one efficient optimization-free and one robust optimization-based method to identify these directions for specific attributes from contrastive text prompts.
  arXiv  Detail & Related papers  (2024-03-25T18:00:42Z)
• Fine-grained Controllable Video Generation via Object Appearance and Context [74.23066823064575]
  We propose fine-grained controllable video generation (FACTOR) to achieve detailed control. FACTOR aims to control objects' appearances and context, including their location and category. Our method achieves controllability of object appearances without finetuning, which reduces the per-subject optimization efforts for the users.
  arXiv  Detail & Related papers  (2023-12-05T17:47:33Z)
• Seen to Unseen: Exploring Compositional Generalization of Multi-Attribute Controllable Dialogue Generation [23.79168163871952]
  Existing controllable dialogue generation work focuses on the single-attribute control. We propose a prompt-based disentangled controllable dialogue generation model, DCG.
  arXiv  Detail & Related papers  (2023-06-17T10:50:19Z)
• MacLaSa: Multi-Aspect Controllable Text Generation via Efficient Sampling from Compact Latent Space [110.85888003111653]
  Multi-aspect controllable text generation aims to generate fluent sentences that possess multiple desired attributes simultaneously. We introduce a novel approach for multi-aspect control, namely MacLaSa, that estimates compact latent space for multiple aspects. We show that MacLaSa outperforms several strong baselines on attribute relevance and textual quality while maintaining a high inference speed.
  arXiv  Detail & Related papers  (2023-05-22T07:30:35Z)
• Dual Semantic Knowledge Composed Multimodal Dialog Systems [114.52730430047589]
  We propose a novel multimodal task-oriented dialog system named MDS-S2. It acquires the context related attribute and relation knowledge from the knowledge base. We also devise a set of latent query variables to distill the semantic information from the composed response representation.
  arXiv  Detail & Related papers  (2023-05-17T06:33:26Z)
• Semantic Space Grounded Weighted Decoding for Multi-Attribute Controllable Dialogue Generation [41.23970507903113]
  We propose a novel framework called DASC that possesses strong controllability with a weighted decoding paradigm. Generation with multiple attributes is then intuitively implemented with an utterance of multiple attribute embeddings. Experiments show that DASC can achieve high control accuracy in generation task with the simultaneous control of 3 aspects.
  arXiv  Detail & Related papers  (2023-05-04T13:35:27Z)
• The Whole Truth and Nothing But the Truth: Faithful and Controllable Dialogue Response Generation with Dataflow Transduction and Constrained Decoding [65.34601470417967]
  We describe a hybrid architecture for dialogue response generation that combines the strengths of neural language modeling and rule-based generation. Our experiments show that this system outperforms both rule-based and learned approaches in human evaluations of fluency, relevance, and truthfulness.
  arXiv  Detail & Related papers  (2022-09-16T09:00:49Z)
• Progressive Open-Domain Response Generation with Multiple Controllable Attributes [13.599621571488033]
  We propose a Progressively trained Hierarchical Vari-Decoder (PHED) to tackle this task. PHED deploys Conditional AutoEncoder (CVAE) on Transformer to include one aspect of attributes at one stage. PHED significantly outperforms the state-of-the-art neural generation models and produces more diverse responses as expected.
  arXiv  Detail & Related papers  (2021-06-07T08:48:39Z)
• A Controllable Model of Grounded Response Generation [122.7121624884747]
  Current end-to-end neural conversation models inherently lack the flexibility to impose semantic control in the response generation process. We propose a framework that we call controllable grounded response generation (CGRG) We show that using this framework, a transformer based model with a novel inductive attention mechanism, trained on a conversation-like Reddit dataset, outperforms strong generation baselines.
  arXiv  Detail & Related papers  (2020-05-01T21:22:08Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.