Towards Robust Online Dialogue Response Generation

Cornell University - arXiv, 2021

We present Viola, an open-domain dialogue system for spoken conversation that uses a topic-agnostic dialogue manager based on a simple generate-and-rank approach. Leveraging recent advances of generative dialogue systems powered by large language models, Viola fetches a batch of response candidates from various neural dialogue models trained with different datasets and knowledge-grounding inputs. Additional responses originating from template-based generators are also considered, depending on the user's input and detected entities. The hand-crafted generators build on a dynamic knowledge graph injected with rich content that is crawled from the web and automatically processed on a daily basis. Viola's response ranker is a fine-tuned polyencoder that chooses the best response given the dialogue history. While dedicated annotations for the polyencoder alone can indirectly steer it away from choosing problematic responses, we add rule-based safety nets to detect neural degeneration and a dedicated classifier to filter out offensive content. We analyze conversations that Viola took part in for the Alexa Prize Socialbot Grand Challenge 4 and discuss the strengths and weaknesses of our approach. Lastly, we suggest future work with a focus on curating conversation data specifcially for socialbots that will contribute towards a more robust data-driven socialbot. 4th Proceedings of Alexa Prize (Alexa Prize 2020).

2022

Natural Language Generation (NLG) represents a large collection of tasks in the field of NLP. While many of these tasks have been tackled well by the cross-entropy (CE) loss, the task of dialog generation poses a few unique challenges for this loss function. First, CE loss assumes that for any given input, the only possible output is the one available as the ground truth in the training dataset. In general, this is not true for any task, as there can be multiple semantically equivalent sentences, each with a different surface form. This problem gets exaggerated further for the dialog generation task, as there can be multiple valid responses (for a given context) that not only have different surface forms but are also not semantically equivalent. Second, CE loss does not take the context into consideration while processing the response and, hence, it treats all ground truths with equal importance irrespective of the context. But, we may want our final agent to avoid certain classes of responses (e.g. bland, non-informative or biased responses) and give relatively higher weightage for more context-specific responses. To circumvent these shortcomings of the CE loss, in this paper, we propose a novel loss function, CORAL, that directly optimizes recently proposed estimates of human preference for generated responses. Using CORAL, we can train dialog generation models without assuming non-existence of response other than the ground-truth. Also, the CORAL loss is computed based on both the context and the response. Our experiments show that, against various large and small scale baselines, our model is able to obtain significantly higher scores for the human preference estimate. While large-scale dialog models have shown much promise and aptitude for dialog generation, it is important to continue the search for more suitable loss functions for training dialog systems. Extensive comparisons on two benchmark datasets show that the proposed methods outperform strong state-of-the-art baseline models of different sizes. We will make the code and trained model checkpoints publicly available upon publication of this paper. Preprint. Under review.

Findings of the Association for Computational Linguistics: ACL 2023

Sample-and-rank is a key decoding strategy for modern generation-based dialogue systems. It helps achieve diverse and high-quality responses by selecting an answer from a small pool of generated candidates. The current stateof-the-art ranking methods mainly use an encoding paradigm called Cross-Encoder, which separately encodes each context-candidate pair and ranks the candidates according to their fitness scores. However, Cross-Encoder repeatedly encodes the same lengthy context for each candidate, resulting in high computational costs. Poly-Encoder addresses the above problems by reducing the interaction between context and candidates, but with a price of performance drop. In this work, we develop a new paradigm called Uni-Encoder 1 , that keeps the full attention over each pair as in Cross-Encoder while only encoding the context once, as in Poly-Encoder. Uni-Encoder encodes all the candidates with the context in one forward pass. We use the same positional embedding for all candidates to ensure they are treated equally and design a new attention mechanism to avoid confusion. Our Uni-Encoder can simulate other ranking paradigms using different attention and response concatenation methods. Extensive experiments show that our proposed paradigm achieves new state-of-theart results on four benchmark datasets with high computational efficiency. For instance, it improves R 10 @1 by 2.9% with an approximately 4× faster inference speed on the Ubuntu V2 dataset.

2020

Neural encoder-decoder models have shown promising performance for human-computer dialogue systems over the past few years. However, due to the maximum-likelihood objective for the decoder, the generated responses are often universal and safe to the point that they lack meaningful information and are no longer relevant to the post. To address this, in this paper, we propose semantic guidance using reinforcement learning to ensure that the generated responses indeed include the given or predicted semantics and that these semantics do not appear repeatedly in the response. Synsets, which comprise sets of manually defined synonyms, are used as the form of assigned semantics. For a given/assigned/predicted synset, only one of its synonyms should appear in the generated response; this constitutes a simple but effective semantic-control mechanism. We conduct both quantitative and qualitative evaluations, which show that the generated responses are not only higher-quality but also reflect ...

ArXiv, 2021

Pre-trained language models (PLM) have marked a huge leap in neural dialogue modeling. While PLMs are pre-trained on largescale text corpora, they are usually fine-tuned on scarce dialogue data with specific domain knowledge and dialogue styles. However, tailoring the language models while fully utilizing prior knowledge in large pre-trained models remains a challenge. In this paper, we present a novel approach for pre-trained dialogue modeling that casts the dialogue generation problem as a prompt-learning task. Instead of fine-tuning on limited dialogue data, our approach, DialogPrompt, learns continuous prompt embeddings optimized for dialogue contexts, which appropriately elicit knowledge from the large pre-trained model. To encourage the model to better utilize the prompt embeddings, the prompt encoders are designed to be conditioned on the input dialogue context. Experiments on popular conversation datasets show that our approach significantly outperforms the fine-tuning basel...

Proceedings of the 19th Annual SIGdial Meeting on Discourse and Dialogue, 2018

Natural language generation (NLG) is an important component in spoken dialog systems (SDSs). A model for NLG involves sequence to sequence learning. State-of-the-art NLG models are built using recurrent neural network (RNN) based sequence to sequence models (Dušek and Jurcicek, 2016a). Convolutional sequence to sequence based models have been used in the domain of machine translation but their application as natural language generators in dialogue systems is still unexplored. In this work, we propose a novel approach to NLG using convolutional neural network (CNN) based sequence to sequence learning. CNN-based approach allows to build a hierarchical model which encapsulates dependencies between words via shorter path unlike RNNs. In contrast to recurrent models, convolutional approach allows for efficient utilization of computational resources by parallelizing computations over all elements, and eases the learning process by applying constant number of nonlinearities. We also propose to use CNN-based reranker for obtaining responses having semantic correspondence with input dialogue acts. The proposed model is capable of entrainment. Studies using a standard dataset shows the effectiveness of the proposed CNN-based approach to NLG.

2020

Dialogue systems using deep learning have achieved generation of fluent response sentences to user utterances. Nevertheless, they tend to produce responses that are not diverse and which are less context-dependent. To address these shortcomings, we propose a new loss function, an Inverse N-gram loss (INF), which incorporates contextual fluency and diversity at the same time by a simple formula. Our INF loss can adjust its loss dynamically by a weight using the inverse frequency of the tokens’ n-gram applied to Softmax Cross-Entropy loss, so that rare tokens appear more likely while retaining the fluency of the generated sentences. We trained Transformer using English and Japanese Twitter replies as single-turn dialogues using different loss functions. Our INF loss model outperformed the baselines of SCE loss and ITF loss models in automatic evaluations such as DIST-N and ROUGE, and also achieved higher scores on our human evaluations of coherence and richness.

Proceedings of the First Workshop on NLP for Conversational AI, 2019

We propose an adversarial learning approach for generating multi-turn dialogue responses. Our proposed framework, hredGAN, is based on conditional generative adversarial networks (GANs). The GAN's generator is a modified hierarchical recurrent encoder-decoder network (HRED) and the discriminator is a word-level bidirectional RNN that shares context and word embeddings with the generator. During inference, noise samples conditioned on the dialogue history are used to perturb the generator's latent space to generate several possible responses. The final response is the one ranked best by the discriminator. The hredGAN shows improved performance over existing methods: (1) it generalizes better than networks trained using only the log-likelihood criterion, and (2) it generates longer, more informative and more diverse responses with high utterance and topic relevance even with limited training data. This improvement is demonstrated on the Movie triples and Ubuntu dialogue datasets using both automatic and human evaluations.

Personalized responses are essential for having an informative and human-like conversation. Because it is difficult to collect a large amount of dialogues involved with specific speakers , it is desirable that chatbot can learn to generate person-alized responses simply from monologues of individuals. In this paper, we propose a novel personalized dialogue generation method which reduces the training data requirement to dialogues without speaker information and monologues of every target speaker. In the proposed approach, a generative adver-sarial network ensures the responses containing recognizable personal characteristics of the target speaker, and a backward SEQ2SEQ model reconstructs the input message for keeping the coherence of the generated responses. The proposed model demonstrates its flexibility to respond to open-domain conversations , and the experimental results show that the proposed method performs favorably against prior work in coherence, personality classification, and human evaluation.

ACM Transactions on Information Systems

Recently, research on open domain dialogue systems have attracted extensive interests of academic and industrial researchers. The goal of an open domain dialogue system is to imitate humans in conversations. Previous works on single turn conversation generation have greatly promoted the research of open domain dialogue systems. However, understanding multiple single turn conversations is not equal to the understanding of multi turn dialogue due to the coherent and context dependent properties of human dialogue. Therefore, in open domain multi turn dialogue generation, it is essential to modeling the contextual semantics of the dialogue history rather than only according to the last utterance. Previous research had verified the effectiveness of the hierarchical recurrent encoder-decoder framework on open domain multi turn dialogue generation. However, using an RNN-based model to hierarchically encoding the utterances to obtain the representation of dialogue history still face the pro...