[2017 ACL] 对话系统

Long Papers

[Domain adaptation ]

1. Adversarial Adaptation of Synthetic or Stale Data ( Cited by 14 )

Young-Bum Kim, Karl Stratos and Dongchan Kim

Two types of data shift common in practice are 1. transferring from synthetic data to live user data (a deployment shift), and 2. transferring from stale data to current data (a temporal shift). Both cause a distribution mismatch between training and evaluation, leading to a model that overfits the flawed(有缺陷的) training data and performs poorly on the test data. We propose a solution to this mismatch problem by framing it as domain adaptation, treating the flawed training dataset as a source domain and the evaluation dataset as a target domain. To this end, we use and build on several recent advances in neural domain adaptation such as adversarial training (Ganin et al., 2016) and domain separation network (Bousmalis et al., 2016), proposing a new effective adversarial training scheme. In both supervised and unsupervised adaptation scenarios, our approach yields clear improvement over strong baselines.

[NLG - language model + affect info]

2. Affect-LM: A Neural Language Model for Customizable Affective Text Generation ( Cited by 27)
Sayan Ghosh, Mathieu Chollet, Eugene Laksana, Stefan Scherer and Louis-Philippe Morency

Human verbal communication includes affective messages which are conveyed through use of emotionally colored words. There has been a lot of research in this direction but the problem of integrating state-of-the-art neural language models with affective information remains an area ripe for exploration. In this paper, we propose an extension to an LSTM (Long Short-Term Memory) language model for generating conversational text, conditioned on affect(感情，情感，心情) categories. Our proposed model, Affect-LM enables us to customize the degree of emotional content in generated sentences through an additional design parameter. Perception studies conducted using Amazon Mechanical Turk show that Affect-LM generates naturally looking emotional sentences without sacrificing grammatical correctness. Affect-LM also learns affect-discriminative word representations, and perplexity experiments show that additional affective information in conversational text can improve language model prediction.

[task + open - whether to have a chat - dataset]

3. Chat Detection in Intelligent Assistant: Combining Task-oriented and Non-task-oriented Spoken Dialogue Systems (‎Cited by 7)

Authors: Satoshi Akasaki and Nobuhiro Kaji

Recently emerged intelligent assistants on smartphones and home electronics (e.g., Siri and Alexa) can be seen as novel hybrids of domain-specific task-oriented spoken dialogue systems and open-domain non-task-oriented ones. To realize such hybrid dialogue systems, this paper investigates determining whether or not a user is going to have a chat with the system. To address the lack of benchmark datasets for this task, we construct a new dataset consisting of 15; 160 utterances collected from the real log data of a commercial intelligent assistant (and will release the dataset to facilitate future research activity). In addition, we investigate using tweets and Web search queries for handling open-domain user utterances, which characterize the task of chat detection. Experiments demonstrated that, while simple supervised methods are effective, the use of the tweets and search queries further improves the F1-score from 86.21 to 87.53.

[domain adaptation]

4. Domain Attention with an Ensemble of Experts (‎Cited by 17)
Authors: Young-Bum Kim, Karl Stratos and Dongchan Kim

An important problem in domain adaptation is to quickly generalize to a new domain with limited supervision given K existing domains. One approach is to retrain a global model across all K + 1 domains using standard techniques, for instance Daume III ´ (2009). However, it is desirable to adapt without having to reestimate a global model from scratch each time a new domain with potentially new intents and slots is added. We describe a solution based on attending an ensemble of domain experts. We assume K domainspecific intent and slot models trained on respective domains. When given domain K + 1, our model uses a weighted combination of the K domain experts’ feedback along with its own opinion to make predictions on the new domain. In experiments, the model significantly outperforms baselines that do not use domain adaptation and also performs better than the full retraining approach.

[NLG - refering expression]

5. Generating Contrastive Referring Expressions (Cited by 0)

Authors: Martin Villalba, Christoph Teichmann and Alexander Koller

The referring expressions (REs) produced by a natural language generation (NLG) system can be misunderstood by the hearer, even when they are semantically correct. In an interactive setting, the NLG system can try to recognize such misunderstandings and correct them. We present an algorithm for generating corrective REs that use contrastive focus (“no, the BLUE button”) to emphasize the information the hearer most likely misunderstood. We show empirically that these contrastive REs are preferred over REs without contrast marking.

[E2E - Framework - RNN + knowledge]

6. Hybrid Code Networks: practical and efficient end-to-end dialog control with supervised and reinforcement learning (‎Cited by 87)

Authors: Jason D Williams, Kavosh Asadi and Geoffrey Zweig

End-to-end learning of recurrent neural networks (RNNs) is an attractive solution for dialog systems; however, current techniques are data-intensive and require thousands of dialogs to learn simple behaviors. We introduce Hybrid Code Networks (HCNs), which combine an RNN with domain-specific knowledge encoded as software and system action templates. Compared to existing end-to-end approaches, HCNs considerably reduce the amount of training data required, while retaining the key benefit of inferring a latent representation of dialog state. In addition, HCNs can be optimized with supervised learning, reinforcement learning, or a mixture of both. HCNs attain state-of-the-art performance on the bAbI dialog dataset, and outperform two commercially deployed customer-facing dialog systems.

[NLU - identity discussion points and discourse relations ? ]

7. Joint Modeling of Content and Discourse Relations in Dialogues ( ‎Cited by 7 )
Authors: Kechen Qin, Lu Wang, Joseph Kim and Julie Shah

We present a joint modeling approach to identify salient(显著的) discussion points in spoken meetings as well as to label the discourse(交谈) relations between speaker turns. A variation of our model is also discussed when discourse relations are treated as latent variables. Experimental results on two popular meeting corpora show that our joint model can outperform state-of-the-art approaches for both phrase-based content selection and discourse relation prediction tasks. We also evaluate our model on predicting the consistency among team members' understanding of their group decisions. Classifiers trained with features constructed from our model achieve significant better predictive performance than the state-of-the-art.

[NLG - Framework  - conditional variaional autoencoders]

8. Learning Discourse-level Diversity for Neural Dialog Models using Conditional Variational Autoencoders (‎Cited by 69)
Authors: Tiancheng Zhao, Ran Zhao and Maxine Eskenazi

While recent neural encoder-decoder models have shown great promise in modeling open-domain conversations, they often generate dull and generic responses. Unlike past work that has focused on diversifying the output of the decoder at word-level to alleviate this problem, we present a novel framework based on conditional variational autoencoders that captures the discourse-level diversity in the encoder. Our model uses latent variables to learn a distribution over potential conversational intents and generates diverse responses using only greedy decoders. We have further developed a novel variant that is integrated with linguistic prior knowledge for better performance. Finally, the training procedure is improved by introducing a bag-of-word loss. Our proposed models have been validated to generate significantly more diverse responses than baseline approaches and exhibit competence(能力) in discourse-level decision-making.

[Symmetric Collaborative Dialogue]

9. Learning Symmetric Collaborative Dialogue Agents with Dynamic Knowledge Graph Embeddings ( ‎Cited by 21)
Authors: He He, Anusha Balakrishnan, Mihail Eric and Percy Liang

We study a symmetric collaborative(对称协作) dialogue setting in which two agents, each with private knowledge, must strategically communicate to achieve a common goal. The open-ended dialogue state in this setting poses new challenges for existing dialogue systems. We collected a dataset of 11K human-human dialogues, which exhibits interesting lexical, semantic, and strategic elements. To model both structured knowledge and unstructured language, we propose a neural model with dynamic knowledge graph embeddings that evolve as the dialogue progresses. Automatic and human evaluations show that our model is both more effective at achieving the goal and more human-like than baseline neural and rule-based models.

[Dialogue state tracking - Framework - representation learning]

10. Neural Belief Tracker: Data-Driven Dialogue State Tracking ( ‎Cited by 63 )
Authors: Nikola Mrkšić, Diarmuid Ó Séaghdha, Tsung-Hsien Wen, Blaise Thomson and Steve Young

One of the core components of modern spoken dialogue systems is the belief tracker, which estimates the user's goal at every step of the dialogue. However, most current approaches have difficulty scaling to larger, more complex dialogue domains. This is due to their dependency on either: a) Spoken Language Understanding models that require large amounts of annotated training data; or b) hand-crafted lexicons for capturing some of the linguistic variation in users' language. We propose a novel Neural Belief Tracking (NBT) framework which overcomes these problems by building on recent advances in representation learning. NBT models reason over pre-trained word vectors, learning to compose them into distributed representations of user utterances and dialogue context. Our evaluation on two datasets shows that this approach surpasses past limitations, matching the performance of state-of-the-art models which rely on hand-crafted semantic lexicons and outperforming them when such lexicons are not provided.

[NLG - muli-turn response selection]

11. Sequential Matching Network: A New Architecture for Multi-turn Response Selection in Retrieval-based Chatbots （ ‎Cited by 48 ）

Authors: Yu Wu, Wei Wu, Chen Xing, Ming Zhou and Zhoujun Li

We study response selection for multi-turn conversation in retrieval-based chatbots. Existing work either concatenates utterances in context or matches a response with a highly abstract context vector finally, which may lose relationships among utterances or important contextual information. We propose a sequential matching network (SMN) to address both problems. SMN first matches a response with each utterance in the context on multiple levels of granularity, and distills important matching information from each pair as a vector with convolution and pooling operations. The vectors are then accumulated in a chronological(按时间的前后顺序排列的) order through a recurrent neural network (RNN) which models relationships among utterances. The final matching score is calculated with the hidden states of the RNN. An empirical study on two public data sets shows that SMN can significantly outperform state-of-the-art methods for response selection in multi-turn conversation.

[NLG - auto eval Metric]

12. Towards an Automatic Turing Test: Learning to Evaluate Dialogue Responses ( ‎Cited by 47)
Authors: Ryan Lowe, Michael Noseworthy, Iulian Vlad Serban, Nicolas Angelard-Gontier, Yoshua Bengio and Joelle Pineau

Automatically evaluating the quality of dialogue responses for unstructured domains is a challenging problem. Unfortunately, existing automatic evaluation metrics are biased and correlate very poorly with human judgements of response quality. Yet having an accurate automatic evaluation procedure is crucial for dialogue research, as it allows rapid prototyping(原型机制造) and testing of new models with fewer expensive human evaluations. In response to this challenge, we formulate automatic dialogue evaluation as a learning problem. We present an evaluation model (ADEM) that learns to predict human-like scores to input responses, using a new dataset of human response scores. We show that the ADEM model's predictions correlate significantly, and at a level much higher than word-overlap metrics such as BLEU, with human judgements at both the utterance and system-level. We also show that ADEM can generalize to evaluating dialogue models unseen during training, an important step for automatic dialogue evaluation.

[E2E - task - multi-turn - ]

13. Towards End-to-End Reinforcement Learning of Dialogue Agents for Information Access ( ‎Cited by 82)
Authors: Bhuwan Dhingra, Lihong Li, Xiujun Li, Jianfeng Gao, Yun-Nung Chen, Faisal Ahmed and Li Deng

This paper proposes KB-InfoBot -- a multi-turn dialogue agent which helps users search Knowledge Bases (KBs) without composing complicated queries. Such goal-oriented dialogue agents typically need to interact with an external database to access real-world knowledge. Previous systems achieved this by issuing a symbolic query to the KB to retrieve entries based on their attributes. However, such symbolic operations break the differentiability(可辨性) of the system and prevent end-to-end training of neural dialogue agents. In this paper, we address this limitation by replacing symbolic queries with an induced "soft" posterior distribution over the KB that indicates which entities the user is interested in. Integrating the soft retrieval process with a reinforcement learner leads to higher task success rate and reward in both simulations and against real users. We also present a fully neural end-to-end agent, trained entirely from user feedback, and discuss its application towards personalized dialogue agents. The source code is available at this https URL.

Short Papers

[NLG - Framework - response based on attributes ]

14. A Conditional Variational Framework for Dialog Generation ( ‎Cited by 20 )

Xiaoyu Shen, Hui Su, Yanran Li, Wenjie Li, Shuzi Niu, Yang Zhao, Akiko Aizawa, Guoping Long

Deep latent variable models have been shown to facilitate the response generation for open-domain dialog systems. However, these latent variables are highly randomized, leading to uncontrollable generated responses. In this paper, we propose a framework allowing conditional response generation based on specific attributes. These attributes can be either manually assigned or automatically detected. Moreover, the dialog states for both speakers are modeled separately in order to reflect personal features. We validate this framework on two different scenarios, where the attribute refers to genericness and sentiment states respectively. The experiment result testified the potential of our model, where meaningful responses can be generated in accordance with the specified attributes.

[NLU - context utiliztion eval]

15. How to Make Contexts More Useful? An Empirical Study to Context-Aware Neural Conversation Models ( ‎Cited by 18)
Authors: Zhiliang Tian, Rui Yan, Lili Mou, Yiping Song, Yansong Feng and Dongyan Zhao

Generative conversational systems are attracting increasing attention in natural language processing (NLP). Recently, researchers have noticed the importance of context information in dialog processing, and built various models to utilize context. However, there is no systematic comparison to analyze how to use context effectively. In this paper, we conduct an empirical study to compare various models and investigate the effect of context information in dialog systems. We also propose a variant that explicitly weights context vectors by context-query relevance, outperforming the other baselines.

[NLG - Open-domain - Engine - generation (info retrieval + Seq2Seq) ]

16. AliMe Chat: A Sequence to Sequence and Rerank based Chatbot Engine ( ‎Cited by 27)
Authors: Minghui Qiu and Feng-Lin Li

We propose AliMe Chat, an open-domain chatbot engine that integrates the joint results of Information Retrieval (IR) and Sequence to Sequence (Seq2Seq) based generation models. AliMe Chat uses an attentive Seq2Seq based rerank model to optimize the joint results. Extensive experiments show our engine outperforms both IR and generation based models. We launch AliMe Chat for a real-world industrial application and observe better results than another public chatbot.