Task-oriented Dialogue Research Articles

PluDG: enhancing task-oriented dialogue system with knowledge graph plug-in module

Task-oriented dialogue systems continue to face significant challenges as they require not only an understanding of dialogue history but also domain-specific knowledge. However, knowledge is often dynamic, making it difficult to effectively integrate into the learning process. Existing large language model approaches primarily treat knowledge bases as textual resources, neglecting to capture the underlying relationships between facts within the knowledge base. To address this limitation, we propose a novel dialogue system called PluDG. We regard the knowledge as a knowledge graph and propose a knowledge extraction plug-in, Kg-Plug, to capture the features of the graph and generate prompt entities to assist the system’s dialogue generation. Besides, we propose Unified Memory Integration, a module that enhances the comprehension of the sentence’s internal structure and optimizes the knowledge base’s encoding location. We conduct experiments on three public datasets and compare PluDG with several state-of-the-art dialogue models. The experimental results indicate that PluDG achieves significant improvements in both accuracy and diversity, outperforming the current state-of-the-art dialogue system models and achieving state-of-the-art performance.

Multi 3 WOZ: A Multilingual, Multi-Domain, Multi-Parallel Dataset for Training and Evaluating Culturally Adapted Task-Oriented Dialog Systems

Abstract Creating high-quality annotated data for task-oriented dialog (ToD) is known to be notoriously difficult, and the challenges are amplified when the goal is to create equitable, culturally adapted, and large-scale ToD datasets for multiple languages. Therefore, the current datasets are still very scarce and suffer from limitations such as translation-based non-native dialogs with translation artefacts, small scale, or lack of cultural adaptation, among others. In this work, we first take stock of the current landscape of multilingual ToD datasets, offering a systematic overview of their properties and limitations. Aiming to reduce all the detected limitations, we then introduce Multi3WOZ, a novel multilingual, multi-domain, multi-parallel ToD dataset. It is large-scale and offers culturally adapted dialogs in 4 languages to enable training and evaluation of multilingual and cross-lingual ToD systems. We describe a complex bottom–up data collection process that yielded the final dataset, and offer the first sets of baseline scores across different ToD-related tasks for future reference, also highlighting its challenging nature.

Cross-domain coreference modeling in dialogue state tracking with prompt learning

Dialogue state tracking (DST) is to identify user goals from dialogue context and it is an essential component in task-oriented dialogue systems. In multi-domain task-oriented dialogues, a user often refers to the information mentioned in the previous context when the topic is transferred to another domain. Accurate modeling of the cross-domain coreference plays an important role in building qualified DST systems. As far as we know, there is only one work attempting to model the cross-domain coreference phenomenon in DST. However, it still has low efficiency and suffers from the difficulty of handling the complex reasoning over the dialogue context. To tackle these problems, in this paper, we propose a simple but effective DST model, called Coref-DST, to track the cross-domain coreference slots. Instead of predicting the actual values via complex reasoning, Coref-DST directly identifies the coreferred domains and slots from the dialogue context. Moreover, a domain-specific prompt method is proposed to predict all the slot values in a domain simultaneously, so as to better capture the relationship among the slots. The extensive experimental results on the MultiWOZ 2.3 dataset demonstrate that Coref-DST not only outperforms the state-of-the-art DST baselines but also has higher efficiency of training and inference.

Multimodal Dialog Systems with Dual Knowledge-enhanced Generative Pretrained Language Model

Text response generation for multimodal task-oriented dialog systems, which aims to generate the proper text response given the multimodal context, is an essential yet challenging task. Although existing efforts have achieved compelling success, they still suffer from two pivotal limitations: (1) overlook the benefit of generative pretraining and (2) ignore the textual context-related knowledge . To address these limitations, we propose a novel dual knowledge-enhanced generative pretrained language mode for multimodal task-oriented dialog systems (DKMD), consisting of three key components: dual knowledge selection , dual knowledge-enhanced context learning , and knowledge-enhanced response generation . To be specific, the dual knowledge selection component aims to select the related knowledge according to both textual and visual modalities of the given context. Thereafter, the dual knowledge-enhanced context learning component targets seamlessly, integrating the selected knowledge into the multimodal context learning from both global and local perspectives, where the cross-modal semantic relation is also explored. Moreover, the knowledge-enhanced response generation component comprises a revised BART decoder, where an additional dot-product knowledge-decoder attention sub-layer is introduced for explicitly utilizing the knowledge to advance the text response generation. Extensive experiments on a public dataset verify the superiority of the proposed DKMD over state-of-the-art competitors.

Multiple Information-Aware Recurrent Reasoning Network for Joint Dialogue Act Recognition and Sentiment Classification

The task of joint dialogue act recognition (DAR) and sentiment classification (DSC) aims to predict both the act and sentiment labels of each utterance in a dialogue. Existing methods mainly focus on local or global semantic features of the dialogue from a single perspective, disregarding the impact of the other part. Therefore, we propose a multiple information-aware recurrent reasoning network (MIRER). Firstly, the sequence information is smoothly sent to multiple local information layers for fine-grained feature extraction through a BiLSTM-connected hybrid CNN group method. Secondly, to obtain global semantic features that are speaker-, context-, and temporal-sensitive, we design a speaker-aware temporal reasoning heterogeneous graph to characterize interactions between utterances spoken by different speakers, incorporating different types of nodes and meta-relations with node-edge-type-dependent parameters. We also design a dual-task temporal reasoning heterogeneous graph to realize the semantic-level and prediction-level self-interaction and interaction, and we constantly revise and improve the label in the process of dual-task recurrent reasoning. MIRER fully integrates context-level features, fine-grained features, and global semantic features, including speaker, context, and temporal sensitivity, to better simulate conversation scenarios. We validated the method on two public dialogue datasets, Mastodon and DailyDialog, and the experimental results show that MIRER outperforms various existing baseline models.

Design of a Modified Transformer Architecture Based on Relative Position Coding

Natural language processing (NLP) based on deep learning provides a positive performance for generative dialogue system, and the transformer model is a new boost in NLP after the advent of word vectors. In this paper, a Chinese generative dialogue system based on transformer is designed, which only uses a multi-layer transformer decoder to build the system and uses the design of an incomplete mask to realize one-way language generation. That is, questions can perceive context information in both directions, while reply sentences can only output one-way autoregressive. The above system improvements make the one-way generation of dialogue tasks more logical and reasonable, and the performance is better than the traditional dialogue system scheme. In consideration of the long-distance information weakness of absolute position coding, we put forward the improvement of relative position coding in theory, and verify it in subsequent experiments. In the transformer module, the calculation formula of self-attention is modified, and the relative position information is added to replace the absolute position coding of the position embedding layer. The performance of the modified model in BLEU, embedding average, grammatical and semantic coherence is ideal, to enhance long-distance attention.

Infer unseen from seen: Relation regularized zero-shot visual dialog

The Visual Dialog task requires retrieving the correct answer based on detected objects, a current question, and history dialogs. However, in real-world scenarios, most existing models face the hard-positive problem and are unable to reason about unseen features, which limits their generalization ability. To address this issue, we propose two Relation Regularized Modules (RRM) in this article. The first is the Visual Relation Regularized Module (VRRM), which seeks known visual features that have semantic relations with unknown visual features and leverages these known features to assist in understanding the unknown features. The second is the Text Relation Regularized Module (TRRM), which enhances the keywords in the answers to strengthen the understanding of unknown text features. To evaluate the effectiveness of these modules, we propose two zero-shot Visual Dialog splits for verification: Visual Zero-shot VisDial with unseen visual features and Text Zero-shot VisDial with unseen answers. Experimental results demonstrate that our proposed modules achieve state-of-the-art performance in zero-shot Visual Dialog with unseen visual features and unseen answers, while also producing comparable results on the benchmark VisDial v1.0 test dataset.

MidGAN: Mutual information in GAN-based dialogue models

Despite a large amount of research into task-oriented dialogue systems, open-ended dialogue agents have received little attention. Recently, researchers have explored using Generative Adversarial Network (GAN) to build such models, however, these require extensive computation and data. This paper propose MidGAN to address prior limitations of GAN-based dialogue models. It does this by trying to maximize the mutual information of the responses generated by the model. To this end, we propose a new metric, MMI-Like. This is based on Maximizing Mutual Information (MMI), yet unlike MMI, does not rely on an auxiliary generative model. We evaluate MidGAN based on the diversity, informativeness by measuring similarity and relevance of the responses it generates by BLEU metric. Our evaluation results, based on the three benchmark datasets, show that MidGAN outperforms the existing state-of-the-art framework, ADV.

Mutually improved response generation and dialogue summarization for multi-domain task-oriented dialogue systems

With the development of pre-trained language models (PrLM), the research of PrLM-based multi-domain task-oriented dialogue systems (TOD) has attracted growing attention and has achieved great progress. However, most current studies suffer from two problems. First, they model dialogue state tracking as an independent subtask supervised by slot-value pairs, resulting in poor adaptability when transferring to new task domains. Second, these studies ignore the fact that not all dialogue histories are valuable for the ensuing turns as they increase in length. To tackle these two issues, we propose a simple and novel framework to explore multi-domain TOD by jointly training response generation and dialogue summarization with PrLM as the backbone. Specifically, first, we use fluent text generated by the dialogue summarization model to replace formatted dialogue states, treating the dialogue state identification task as a natural language generation task, which allows dialogue state tracking to be easily extended to new task domains. Second, dialogue summarization removes redundant and useless information from the current dialogue process and is fed into the response decoder enabling the system to focus on crucial details in long dialogues. Furthermore, we employ a dialogue chunk detector to assist the dialogue summarization model and design a fusion mechanism to dynamically integrate helpful dialogue summarization into the response generation process. Experimental results show that the proposed model achieves state-of-the-art performance in both automatic and human evaluations on two public datasets and demonstrate that the joint framework can benefit both tasks from each other.

Metaphorical User Simulators for Evaluating Task-oriented Dialogue Systems

Task-oriented dialogue systems (TDSs) are assessed mainly in an offline setting or through human evaluation. The evaluation is often limited to single-turn or is very time-intensive. As an alternative, user simulators that mimic user behavior allow us to consider a broad set of user goals to generate human-like conversations for simulated evaluation. Employing existing user simulators to evaluate TDSs is challenging as user simulators are primarily designed to optimize dialogue policies for TDSs and have limited evaluation capabilities. Moreover, the evaluation of user simulators is an open challenge. In this work, we propose a metaphorical user simulator for end-to-end TDS evaluation, where we define a simulator to be metaphorical if it simulates a user’s analogical thinking in interactions with systems. We also propose a tester-based evaluation framework to generate variants, i.e., dialogue systems with different capabilities. Our user simulator constructs a metaphorical user model that assists the simulator in reasoning by referring to prior knowledge when encountering new items. We estimate the quality of simulators by checking the simulated interactions between simulators and variants. Our experiments are conducted using three TDS datasets. The proposed user simulator demonstrates better consistency with manual evaluation than an agenda-based simulator and a seq2seq model on three datasets; our tester framework demonstrates efficiency and has been tested on multiple tasks, such as conversational recommendation and e-commerce dialogues.

Building a hospitable and reliable dialogue system for android robots: a scenario-based approach with large language models

Dialogue systems implemented with android robots are expected to provide not only advanced conversational ability but also reliability and hospitality due to their human-like appearance. In this study, we aim to develop a hospitable dialogue system by encouraging open-ended utterances and responding adaptively to give users the feeling of being heard. To achieve this, utilizing large language models (LLMs) is a promising option, but task-oriented dialogue systems implemented with only LLMs often generate irrelevant, inconsistent, or contradictory utterances. Therefore, we propose a scenario-based dialogue system that subdivides the task into smaller sub-tasks, such as summarization, information extraction, and response generation, and uses LLMs in a fine-grained manner to overcome such shortcomings of the LLM-based dialogue system. Our system was evaluated in the tourist-spot recommendation task of the Dialogue Robot Competition 2022 and achieved second place in the preliminary round and first place in the final round, outperforming other rule-based dialogue systems. However, we also identified several challenges when using LLMs for android dialogue systems, including response delays due to computational complexity, hallucinations, and coordination issues between generated utterances and robot control.

KRP-DS: A Knowledge Graph-Based Dialogue System with Inference-Aided Prediction.

With the popularity of ChatGPT, there has been increasing attention towards dialogue systems. Researchers are dedicated to designing a knowledgeable model that can engage in conversations like humans. Traditional seq2seq dialogue models often suffer from limited performance and the issue of generating safe responses. In recent years, large-scale pretrained language models have demonstrated their powerful capabilities across various domains. Many studies have leveraged these pretrained models for dialogue tasks to address concerns such as safe response generation. Pretrained models can enhance responses by carrying certain knowledge information after being pre-trained on large-scale data. However, when specific knowledge is required in a particular domain, the model may still generate bland or inappropriate responses, and the interpretability of such models is poor. Therefore, in this paper, we propose the KRP-DS model. We design a knowledge module that incorporates a knowledge graph as external knowledge in the dialogue system. The module utilizes contextual information for path reasoning and guides knowledge prediction. Finally, the predicted knowledge is used to enhance response generation. Experimental results show that our proposed model can effectively improve the quality and diversity of responses while having better interpretability, and outperforms baseline models in both automatic and human evaluations.

A task-oriented neural dialogue system capable of knowledge accessing

As one of the most important AI applications, task-oriented dialogue systems have been a hot topic in this field. These systems are primarily designed to fulfill tasks and provide answers to queries, such as checking the weather or booking flights. However, there are two major challenges in implementing task-oriented dialogue systems: selecting the right knowledge from the relevant knowledge base and combining the knowledge to produce grammatically correct and fluent replies. In this paper, we present a task-oriented dialogue system using an end-to-end trainable neural model. This dialogue system consists of four components: an encoder for encoding the dialogue history, a three-hop memory network for storing the relevant knowledge entries and dialogue history words, a knowledge screener for selecting the knowledge that is really related to the dialogue history, and a decoder for generating the replies. Finally, we validate the proposed dialogue system on a benchmark dataset (KVRET). The experimental results show that the proposed model generates more accurate replies than baselines in automatic and human evaluations. Also, we verify the superiorities of the two improvements by ablation experiments.

Parameter-Efficient Fine-Tuning Method for Task-Oriented Dialogue Systems

The use of Transformer-based pre-trained language models has become prevalent in enhancing the performance of task-oriented dialogue systems. These models, which are pre-trained on large text data to grasp the language syntax and semantics, fine-tune the entire parameter set according to a specific task. However, as the scale of the pre-trained language model increases, several challenges arise during the fine-tuning process. For example, the training time escalates as the model scale grows, since the complete parameter set needs to be trained. Furthermore, additional storage space is required to accommodate the larger model size. To address these challenges, we propose a new new task-oriented dialogue system called PEFTTOD. Our proposal leverages a method called the Parameter-Efficient Fine-Tuning method (PEFT), which incorporates an Adapter Layer and prefix tuning into the pre-trained language model. It significantly reduces the overall parameter count used during training and efficiently transfers the dialogue knowledge. We evaluated the performance of PEFTTOD on the Multi-WOZ 2.0 dataset, a benchmark dataset commonly used in task-oriented dialogue systems. Compared to the traditional method, PEFTTOD utilizes only about 4% of the parameters for training, resulting in a 4% improvement in the combined score compared to the existing T5-based baseline. Moreover, PEFTTOD achieved an efficiency gain by reducing the training time by 20% and saving up to 95% of the required storage space.

Unsupervised Dialogue State Tracking for End-to-End Task-Oriented Dialogue with a Multi-Span Prediction Network

Unsupervised Dialogue State Tracking for End-to-End Task-Oriented Dialogue with a Multi-Span Prediction Network

Taming Continuous Posteriors for Latent Variational Dialogue Policies

Utilizing amortized variational inference for latent-action reinforcement learning (RL) has been shown to be an effective approach in Task-oriented Dialogue (ToD) systems for optimizing dialogue success.Until now, categorical posteriors have been argued to be one of the main drivers of performance. In this work we revisit Gaussian variational posteriors for latent-action RL and show that they can yield even better performance than categoricals. We achieve this by introducing an improved variational inference objective for learning continuous representations without auxiliary learning objectives, which streamlines the training procedure. Moreover, we propose ways to regularize the latent dialogue policy, which helps to retain good response coherence. Using continuous latent representations our model achieves state of the art dialogue success rate on the MultiWOZ benchmark, and also compares well to categorical latent methods in response coherence.

Dialogue Rewriting via Skeleton-Guided Generation

Dialogue rewriting aims to transform multi-turn, context-dependent dialogues into well-formed, context-independent text for most NLP systems. Previous dialogue rewriting benchmarks and systems assume a fluent and informative utterance to rewrite. Unfortunately, dialogue utterances from real-world systems are frequently noisy and with various kinds of errors that can make them almost uninformative. In this paper, we first present Real-world Dialogue Rewriting Corpus (RealDia), a new benchmark to evaluate how well current dialogue rewriting systems can deal with real-world noisy and uninformative dialogue utterances. RealDia contains annotated multi-turn dialogues from real scenes with ASR errors, spelling errors, redundancies and other noises that are ignored by previous dialogue rewriting benchmarks. We show that previous dialogue rewriting approaches are neither effective nor data-efficient to resolve RealDia. Then this paper presents Skeleton-Guided Rewriter (SGR), which can resolve the task of dialogue rewriting via a skeleton-guided generation paradigm. Experiments show that RealDia is a much more challenging benchmark for real-world dialogue rewriting, and SGR can effectively resolve the task and outperform previous approaches by a large margin.

Dialogue State Distillation Network with Inter-slot Contrastive Learning for Dialogue State Tracking

In task-oriented dialogue systems, Dialogue State Tracking (DST) aims to extract users' intentions from the dialogue history. Currently, most existing approaches suffer from error propagation and are unable to dynamically select relevant information when utilizing previous dialogue states. Moreover, the relations between the updates of different slots provide vital clues for DST. However, the existing approaches rely only on predefined graphs to indirectly capture the relations. In this paper, we propose a Dialogue State Distillation Network (DSDN) to utilize relevant information of previous dialogue states and migrate the gap of utilization between training and testing. Thus, it can dynamically exploit previous dialogue states and avoid introducing error propagation simultaneously. Further, we propose an inter-slot contrastive learning loss to effectively capture the slot co-update relations from dialogue context. Experiments are conducted on the widely used MultiWOZ 2.0 and MultiWOZ 2.1 datasets. The experimental results show that our proposed model achieves the state-of-the-art performance for DST.

Balanced Meta Learning and Diverse Sampling for Lifelong Task-Oriented Dialogue Systems

In real-world scenarios, it is crucial to build a lifelong taskoriented dialogue system (TDS) that continually adapts to new knowledge without forgetting previously acquired experiences. Existing approaches mainly focus on mitigating the catastrophic forgetting in lifelong TDS. However, the transfer ability to generalize the accumulated old knowledge to new tasks is underexplored. In this paper, we propose a two-stage lifelong task-oriented dialogue generation method to mitigate catastrophic forgetting and encourage knowledge transfer simultaneously, inspired by the learning process. In the first stage, we learn task-specific masks which adaptively preserve the knowledge of each visited task so as to mitigate catastrophic forgetting. In this stage, we are expected to learn the task-specific knowledge which is tailored for each task. In the second stage, we bring the knowledge from the encountered tasks together and understand thoroughly. To this end, we devise a balanced meta learning strategy for both forward and backward knowledge transfer in the lifelong learning process. In particular, we perform meta-update with a meta-test set sampled from the current training data for forward knowledge transfer. In addition, we employ an uncertainty-based sampling strategy to select and store representative dialogue samples into episodic memory and perform meta-update with a meta-test set sampled from the memory for backward knowledge transfer. With extensive experiments on 29 tasks, we show that MetaLTDS outperforms the strong baselines in terms of both effectiveness and efficiency. For reproducibility, we submit our code at: https: //github.com/travis-xu/MetaLTDS.

Multi-Action Dialog Policy Learning from Logged User Feedback

Multi-action dialog policy (MADP), which generates multiple atomic dialog actions per turn, has been widely applied in task-oriented dialog systems to provide expressive and efficient system responses. Existing MADP models usually imitate action combinations from the labeled multi-action dialog samples. Due to data limitations, they generalize poorly toward unseen dialog flows. While reinforcement learning-based methods are proposed to incorporate the service ratings from real users and user simulators as external supervision signals, they suffer from sparse and less credible dialog-level rewards. To cope with this problem, we explore to improve MADPL with explicit and implicit turn-level user feedback received for historical predictions (i.e., logged user feedback) that are cost-efficient to collect and faithful to real-world scenarios. The task is challenging since the logged user feedback provides only partial label feedback limited to the particular historical dialog actions predicted by the agent. To fully exploit such feedback information, we propose BanditMatch, which addresses the task from a feedback-enhanced semi-supervised learning perspective with a hybrid learning objective of SSL and bandit learning. BanditMatch integrates pseudo-labeling methods to better explore the action space through constructing full label feedback. Extensive experiments show that our BanditMatch improves MADPL over the state-of-the-art methods by generating more concise and informative responses. The source code and the appendix of this paper can be obtained from https://github.com/ShuoZhangXJTU/BanditMatch.