Semantically Conditioned LSTM-based Natural Language Generation for Spoken Dialogue Systems

By Tsung-Hsien Wen, Milica Gasic, Nikola Mrksic, Pei-hao Su, David Vandyke, Susannah Young. 2015

974 citations recorded by the source corpus.

© 2015 Association for Computational Linguistics. Natural language generation (NLG) is a critical component of spoken dialogue and it has a significant impact both on usability and perceived quality. Most NLG systems in common use employ rules and heuristics and tend to generate rigid and stylised responses without the natural variation of human language. They are also not easily scaled to systems covering multiple domains and languages. This paper presents a statistical language generator based on a semantically controlled Long Short-term Memory (LSTM) structure. The LSTM generator can learn from unaligned data by jointly optimising sentence planning and surface realisation using a simple cross entropy training criterion, and language variation can be easily achieved by sampling from output candidates. With fewer heuristics, an objective evaluation in two differing test domains showed the proposed method improved performance compared to previous methods. Human judges scored the LSTM system higher on informativeness and naturalness and overall preferred it to the other systems..

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

Recurrent Neural Network Grammars (2016, neutral): “…state-ofthe-art baselines have been obtained using RNNs (Zaremba et al., 2015; Mikolov et al., 2010) as well as in various conditional language modeling tasks such as machine translation (Bahdanau et al., 2015), image caption generation (Xu et al., 2015), and dialogue generation (Wen et al., 2015).”
Semantic Noise Matters for Neural Natural Language Generation (2019, methodology): “The differences against the original MRs allow us to compute the semantic/slot error rate (SER; Wen et al., 2015; Reed et al., 2018; Dušek et al., 2019):”
Neural Generation for Czech: Data and Baselines (2019, methodology): “A prime example is the delexicalization technique used by most current generators (e.g., Oh and Rudnicky, 2000; Mairesse et al., 2010; Wen et al., 2015a,b; Juraska et al., 2018): It is generally assumed that attribute (slot) values from the input meaning representation (MR) can be replaced by…”
Creating Training Corpora for NLG Micro-Planners (2017, background): “Note that because datasets are typically delexicalised when training NLG models (cf. e.g., Wen et al. 2015 and Lampouras and Vlachos 2016), at training time, different instantiations of the same input pattern reduce to identical input.”
The E2E Dataset: New Challenges For End-to-End Generation (2017, background): “…development of NLG components in new domains: They jointly learn sentence planning and surface realisation from non-aligned data (Duˇsek and Jurˇc´ıˇcek, 2015; Wen et al., 2015; Mei et al., 2016; Wen et al., 2016; Sharma et al., 2016; Duˇsek and Jurˇc´ıˇcek, 2016a; Lampouras and Vlachos, 2016).”
Structure-to-Text Generation with Self-Training, Acceptability Classifiers and Context-Conditioning for the GEM Shared Task (2021, background): “In the research community, generative models (Wen et al., 2015; Duˇsek and Jurcıcek, 2016; Rao et al., 2019) have become popular for their data-driven scaling story and superior naturalness over typical template-based systems (Gatt and Krahmer, 2018; Dale, 2020).”
A Knowledge-Grounded Neural Conversation Model (2017, background): “This work should be seen as distinct from more goaldirected neural dialog modeling in which question-answer slots are explicitly learned from small amounts of crowdsourced data, customer support logs, or user data (Wen et al. 2015; 2017; Zhao et al. 2017).”
A Diversity-Promoting Objective Function for Neural Conversation Models (2015, methodology): “…of statistical machine translation (SMT), either phrase-based (Ritter et al., 2011), or using neural networks to rerank, or directly in the form of sequence-to-sequence (S EQ 2S EQ ) models (Sordoni et al., 2015; Shang et al., 2015; Vinyals and Le, 2015; Wen et al., 2015; Serban et al., 2016).”
Towards Coherent and Engaging Spoken Dialog Response Generation Using Automatic Conversation Evaluators (2019, background): “Recently, there has been signiﬁcant interest in building chatbots (Sordoni et al., 2015; Wen et al., 2015) fueled by the availability of popular dialog data sets such as Ubuntu, Twitter, and Movie dialogs (Lowe et al., 2015; Rit-ter et al., 2011; Danescu-Niculescu-Mizil and Lee, 2011).”
A Simple Language Model for Task-Oriented Dialogue (2020, methodology): “, 2017], and using dedicated response generation modules [Wen et al., 2015].”
Survey of Hallucination in Natural Language Generation (2022, neutral): “datasets: One way is employing annotators to write clean and faithful targets from scratch given the source [54, 204], which may lack diversity [67, 140, 143].”
DART: Open-Domain Structured Data Record to Text Generation (2020, background): “Such applications include explaining data records to non-experts in the healthcare domain (Cawsey et al., 1997), writing sports news (Chen and Mooney, 2008), summarizing information in multiple documents (Fan et al., 2019), and generating dialogue responses (Wen et al., 2015).”
GPTScore: Evaluate as You Desire (2023, methodology): “16 regarding NAT and FLU of the SFRES dataset, most of the 19 variants of GPTScore with instructions and demonstrations outperform fine-tuned BARTSCORE+CNN+Para.”
Deep Reinforcement Learning for Dialogue Generation (2016, background): “Neural response generation (Sordoni et al., 2015; Shang et al., 2015; Vinyals and Le, 2015; Li et al., 2016a; Wen et al., 2015; Yao et al., 2015; Luan et al., 2016; Xu et al., 2016; Wen et al., 2016; Li et al., 2016b; Su et al., 2016) is of growing interest.”
SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient (2016, background): “Recently, recurrent neural networks (RNNs) with long short-term memory (LSTM) cells (Hochreiter and Schmidhuber 1997) have shown excellent performance ranging from natural language generation to handwriting generation (Wen et al. 2015; Graves 2013).”
Neural Approaches to Conversational AI (2018, neutral): “It is in this sense that the gate rt plays the role of sentence planning (Wen et al., 2015).”