Filters








7,604 Hits in 3.5 sec

Exploring Dense Retrieval for Dialogue Response Selection [article]

Tian Lan, Deng Cai, Yan Wang, Yixuan Su, Heyan Huang, Xian-Ling Mao
2022 arXiv   pre-print
Recent progress in deep learning has continuously improved the accuracy of dialogue response selection.  ...  In this study, we present a solution to directly select proper responses from a large corpus or even a nonparallel corpus that only consists of unpaired sentences, using a dense retrieval model.  ...  However, the dense retrieval model has not been well explored in the dialogue response selection to the best of our knowledge.  ... 
arXiv:2110.06612v3 fatcat:bphgzlnyffcf3pkp6dokuguvyu

Sparse and Dense Approaches for the Full-rank Retrieval of Responses for Dialogues [article]

Gustavo Penha, Claudia Hauff
2022 arXiv   pre-print
We investigate both dialogue context and response expansion techniques for sparse retrieval, as well as zero-shot and fine-tuned dense retrieval approaches.  ...  Our findings based on three different information-seeking dialogue datasets reveal that a learned response expansion technique is a solid baseline for sparse retrieval.  ...  CONCLUSION We study here the problem of full-rank retrieval of responses for dialogues. We explore sparse and dense techniques that retrieve responses out of the entire collection available.  ... 
arXiv:2204.10558v1 fatcat:jbab4qy6rrdetisxxa6cuapbi4

Contextual Fine-to-Coarse Distillation for Coarse-grained Response Selection in Open-Domain Conversations [article]

Wei Chen, Yeyun Gong, Can Xu, Huang Hu, Bolun Yao, Zhongyu Wei, Zhihao Fan, Xiaowu Hu, Bartuer Zhou, Biao Cheng, Daxin Jiang, Nan Duan
2022 arXiv   pre-print
We study the problem of coarse-grained response selection in retrieval-based dialogue systems.  ...  The problem is equally important with fine-grained response selection, but is less explored in existing literature.  ...  Coarse-grained Response Selection On the other hand, coarse-grained dialogue retrieval is an important but rarely explored field.  ... 
arXiv:2109.13087v2 fatcat:5c6maij6lncodahocuaoptol4m

Fast and Light-Weight Answer Text Retrieval in Dialogue Systems [article]

Hui Wan, Siva Sankalp Patel, J. William Murdock, Saloni Potdar, Sachindra Joshi
2022 arXiv   pre-print
The state-of-the-art technology for neural dense retrieval or re-ranking involves deep learning models with hundreds of millions of parameters.  ...  Dialogue systems can benefit from being able to search through a corpus of text to find information relevant to user requests, especially when encountering a request for which no manually curated response  ...  For our use case on the small retrieval datasets, we explored using smaller dimensions for the vector representations in ColBERT.  ... 
arXiv:2205.14226v2 fatcat:y4tphpneabcjjeimvdenkvj5ky

Building an Efficient and Effective Retrieval-based Dialogue System via Mutual Learning [article]

Chongyang Tao, Jiazhan Feng, Chang Liu, Juntao Li, Xiubo Geng, Daxin Jiang
2021 arXiv   pre-print
Establishing retrieval-based dialogue systems that can select appropriate responses from the pre-built index has gained increasing attention from researchers.  ...  The former gives considerable improvements in accuracy but is often inapplicable in practice for large-scale retrieval given the cost of the full attention required for each sample at test time.  ...  tiveness of our framework, we train the pre-retrieval model and the re-ranking model at the same time via mutual learning, which enables two models to learn from each other throughout the training process  ... 
arXiv:2110.00159v1 fatcat:7bndfvl2ovc4di6t7ofneyfxku

A Survey on Retrieval-Augmented Text Generation [article]

Huayang Li and Yixuan Su and Deng Cai and Yan Wang and Lemao Liu
2022 arXiv   pre-print
It firstly highlights the generic paradigm of retrieval-augmented generation, and then it reviews notable approaches according to different tasks including dialogue response generation, machine translation  ...  This paper aims to conduct a survey about retrieval-augmented text generation.  ...  For instance, x and y could be the dialogue history and the corresponding response for dialogue response generation, the text in the source language and the translation in the target language for machine  ... 
arXiv:2202.01110v2 fatcat:vo6i5vq62raxtcry2xcbrm55be

DialFact: A Benchmark for Fact-Checking in Dialogue [article]

Prakhar Gupta, Chien-Sheng Wu, Wenhao Liu, Caiming Xiong
2022 arXiv   pre-print
There are three sub-tasks in DialFact: 1) Verifiable claim detection task distinguishes whether a response carries verifiable factual information; 2) Evidence retrieval task retrieves the most relevant  ...  Wikipedia snippets as evidence; 3) Claim verification task predicts a dialogue response to be supported, refuted, or not enough information.  ...  , and 3) DPR-Evidence, where we use DPR-WoWft-ctx for document retrieval and Ret-withcontext for evidence selection.  ... 
arXiv:2110.08222v2 fatcat:mv2khqyrvrhnna2v6nxqhdvuci

Not All Dialogues are Created Equal: Instance Weighting for Neural Conversational Models [article]

Pierre Lison, Serge Bibauw
2017 arXiv   pre-print
The weighting model, which is itself estimated from dialogue data, associates each training example to a numerical weight that reflects its intrinsic quality for dialogue modelling.  ...  Neural conversational models require substantial amounts of dialogue data for their parameter estimation and are therefore usually learned on large corpora such as chat forums or movie subtitles.  ...  Retrieval models are used to select the most relevant response for a given context amongst a (possibly large) set of predefined responses, such as the set of utterances extracted from a corpus (Lowe et  ... 
arXiv:1704.08966v2 fatcat:njrsiqtlozeq7hxm7d6xgdfaea

Not All Dialogues are Created Equal: Instance Weighting for Neural Conversational Models

Pierre Lison, Serge Bibauw
2017 Proceedings of the 18th Annual SIGdial Meeting on Discourse and Dialogue  
The weighting model, which is itself estimated from dialogue data, associates each training example to a numerical weight that reflects its intrinsic quality for dialogue modelling.  ...  Evaluation results on retrieval-based models trained on movie and TV subtitles demonstrate that the inclusion of such a weighting model improves the model performance on unsupervised metrics. * Also affiliated  ...  Retrieval models are used to select the most relevant response for a given context amongst a (possibly large) set of predefined responses, such as the set of utterances extracted from a corpus (Lowe et  ... 
doi:10.18653/v1/w17-5546 dblp:conf/sigdial/LisonB17 fatcat:ezthcs2pxzdbrpwrlnzldedsum

Improving Retrieval Modeling Using Cross Convolution Networks And Multi Frequency Word Embedding [article]

Guozhen An, Mehrnoosh Shafiee, Davood Shamsi
2018 arXiv   pre-print
In this paper we concentrate on the task of response selection for multi-turn human-computer conversation with a given context.  ...  We train several models using the Ubuntu Dialogue dataset which is the largest freely available multi-turn based dialogue corpus.  ...  Acknowledgments We thank Morten Pedersen and David Guy Brizan for their contributions to this study. We gratefully acknowledge financial support for this work by AOL of OATH.  ... 
arXiv:1802.05373v2 fatcat:jw3b2illanctphvgoxnb5m5xei

A Robotic Dating Coaching System Leveraging Online Communities Posts [article]

Sihyeon Jo, Donghwi Jung, Keonwoo Kim, Eun Gyo Joung, Giulia Nespoli, Seungryong Yoo, Minseob So, Seung-Woo Seo, Seong-Woo Kim
2020 arXiv   pre-print
We examine people's perceptions of the dating coaching robot with a dialogue module. 97 participants joined to have a conversation with the robot, and 30 of them evaluated the robot.  ...  Retrieval-based dialogue system conducts a dialogue by selecting an appropriate response for a given dialogue context, which differs from other conversation modeling paradigms such as generation-based  ...  Doc2vec embedding scheme is applied to reduce retrieval time with the dense representation of texts [11] . • Step 2 (Matching), in which the system selects candidate responses from by comparing matching  ... 
arXiv:2011.11855v1 fatcat:mpeloju76je2tllatc6nrb4bme

Multi-Stage Prompting for Knowledgeable Dialogue Generation [article]

Zihan Liu, Mostofa Patwary, Ryan Prenger, Shrimai Prabhumoye, Wei Ping, Mohammad Shoeybi, Bryan Catanzaro
2022 arXiv   pre-print
Then, we further prompt it to generate responses based on the dialogue context and the previously generated knowledge.  ...  In addition, our multi-stage prompting outperforms the finetuning-based dialogue model in terms of response knowledgeability and engagement by up to 10% and 5%, respectively.  ...  Baselines for Knowledge Generation DPR DPR, Dense Passage Retriever (Karpukhin et al., 2020) , is the state-of-the-art retrieval model.  ... 
arXiv:2203.08745v1 fatcat:wd3bkahaz5dr5kju32v56m73zm

Few-Shot Bot: Prompt-Based Learning for Dialogue Systems [article]

Andrea Madotto, Zhaojiang Lin, Genta Indra Winata, Pascale Fung
2021 arXiv   pre-print
In this paper, we explore prompt-based few-shot learning in dialogue tasks.  ...  skill, queries different knowledge bases or the internet, and uses the retrieved knowledge to generate a human-like response, all using only few dialogue examples per skill.  ...  As for the wiki retriever, we select the first sentence of the paragraph as knowledge for the next turn.  ... 
arXiv:2110.08118v1 fatcat:fhcmp7x34ndh5cr253w44vvide

Beyond Goldfish Memory: Long-Term Open-Domain Conversation [article]

Jing Xu, Arthur Szlam, Jason Weston
2021 arXiv   pre-print
Despite recent improvements in open-domain dialogue models, state of the art models are trained and evaluated on short conversations with little context.  ...  In particular, we find retrieval-augmented methods and methods with an ability to summarize and recall previous conversations outperform the standard encoder-decoder architectures currently considered  ...  For (2) we can use the same systems as presented in subsection 4.2 to both retrieve from the summarization memories, and to finally generate an appropriate response.  ... 
arXiv:2107.07567v1 fatcat:wrabh7xfcba67n2nj6jfbqlesq

Exemplars-guided Empathetic Response Generation Controlled by the Elements of Human Communication [article]

Navonil Majumder, Deepanway Ghosal, Devamanyu Hazarika, Alexander Gelbukh, Rada Mihalcea, Soujanya Poria
2021 arXiv   pre-print
To this end, we employ dense passage retrieval to extract relevant exemplary responses from the training set.  ...  The majority of existing methods for empathetic response generation rely on the emotion of the context to generate empathetic responses.  ...  In this context, previous work [32] has also explored retrieval-based response selection.  ... 
arXiv:2106.11791v3 fatcat:ob4n5ztu7ndavm6tizpz7dpowm
« Previous Showing results 1 — 15 out of 7,604 results