Two are Better than One: An Ensemble of Retrieval- and Generation-Based Dialog Systems

By Yiping Song, Rui Yan, Xiang Li, Dongyan Zhao, Ming Zhang. 2016

114 citations recorded by the source corpus.

Open-domain human-computer conversation has attracted much attention in the field of NLP. Contrary to rule- or template-based domain-specific dialog systems, open-domain conversation usually requires data-driven approaches, which can be roughly divided into two categories: retrieval-based and generation-based systems. Retrieval systems search a user-issued utterance (called a query) in a large database, and return a reply that best matches the query. Generative approaches, typically based on recurrent neural networks (RNNs), can synthesize new replies, but they suffer from the problem of generating short, meaningless utterances. In this paper, we propose a novel ensemble of retrieval-based and generation-based dialog systems in the open domain. In our approach, the retrieved candidate, in addition to the original query, is fed to an RNN-based reply generator, so that the neural model is aware of more information. The generated reply is then fed back as a new candidate for post-reranking. Experimental results show that such ensemble outperforms each single part of it by a large margin.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

A Retrieve-and-Edit Framework for Predicting Structured Outputs (2018, methodology): “However, the desired output might be a variation of another, previously-observed example [14, 13, 30, 18, 24].”
Keep CALM and Explore: Language Models for Action Generation in Text-based Games (2020, background): “More broadly, the concept of generating candidates and re-ranking has been studied in other interactive lanugage tasks such as dialogue (Zhao and Eskenazi, 2016; Williams et al., 2017; Song et al., 2016; Chen et al., 2017) and communication games (Lazaridou et al., 2020).”
What Makes Good In-Context Examples for GPT-3? (2021, background): “, 2020), dialogue generation (Yan et al., 2016; Cai et al., 2018; Song et al., 2016; Pandey et al., 2018; Weston et al., 2018; Wu et al., 2019), text summarization (Cao et al.”
Augmenting Transformers with KNN-Based Composite Memory for Dialog (2020, neutral): “Some of this work has specialized to use both types of models to generate conversations in an ensemble (Song et al., 2016) or to specifically improve consistency (Song et al.”
A Survey on Dialogue Systems: Recent Advances and New Frontiers (2017, background): “Combing neural generative and retrieval based models can have significant effects on performance.[66] and [52] attempted to combine both methods.”
Response Generation by Context-aware Prototype Editing (2018, methodology): “Prior works have attempted to utilize prototype responses to guide the generation process (Song et al. 2016; Pandey et al. 2018), in which prototype responses are encoded into vectors and feed to a de-coder along with a context representation.”
Extract and Edit: An Alternative to Back-Translation for Unsupervised Neural Machine Translation (2019, neutral): “…related to the recent work on applying retrieval mechanisms to augment text generation, such as image captioning (Kuznetsova et al., 2013; Mason and Charniak, 2014), dialogue generation (Song et al., 2016; Yan et al., 2016; Wu et al., 2018) and style transfer (Lin et al., 2017; Li et al., 2018).”
Skeleton-to-Response: Dialogue Generation Guided by Retrieval Memory (2018, methodology): “Song et al. (2016) used an extra encoder to transform the retrieved response into a fixed-sized vector.”
Retrieval-guided Dialogue Response Generation via a Matching-to-Generation Framework (2019, methodology): “Experiments show our method signiﬁcantly outperforms several strong baselines.”
Text Generation with Exemplar-based Adaptive Decoding (2019, unknown): “…many recent efforts have been devoted to augmenting text generation models with retrieved exemplars (Ho-dosh et al., 2013; Mason and Charniak, 2014; Song et al., 2016; Lin et al., 2017; Li et al., 2018, Source : A Portuguese train derailed in the northern region of Oporto on Wednesday, killing…”
Alana: Social Dialogue using an Ensemble Model and a Ranker trained on User Feedback (2017, methodology): “We follow previous work by combing both paradigms into an ensemble-based approach (Song et al., 2016).”
Say What I Want: Towards the Dark Side of Neural Dialogue Models (2019, background): “These methods are able to provide natural, human-like responses, but fail to generate novel responses out of the range of the repository [8].”
Recent advances in deep learning based dialogue systems: a systematic survey (2021, methodology): “More recently, Wu et al. (2020a) pretrained Task-Oriented Dialogue BERT (TOD-BERT) and significantly improved the accuracy in the intent detection sub-task.”
RUBER: An Unsupervised Method for Automatic Evaluation of Open-Domain Dialog Systems (2017, background): “One is a feature-based retrieval-and-reranking system, which ﬁrst retrieves a coarse-grained candidate set by keyword matching and then reranks the candidates by human-engineered features; the top-ranked results are selected for evaluation (Song et al., 2016).”