Filters








1,563 Hits in 2.4 sec

Lexical Knowledge Internalization for Neural Dialog Generation [article]

Zhiyong Wu, Wei Bi, Xiang Li, Lingpeng Kong, Ben Kao
2022 arXiv   pre-print
We propose knowledge internalization (KI), which aims to complement the lexical knowledge into neural dialog models.  ...  Instead of further conditioning the knowledge-grounded dialog (KGD) models on externally retrieved knowledge, we seek to integrate knowledge about each input token internally into the model's parameters  ...  Knowledge Internalization for Neural Dialog Models In this section, we illustrate how to train a dialog model with knowledge internalization.  ... 
arXiv:2205.01941v1 fatcat:dmxt56prungith4njknjwubjda

Lexical Knowledge Internalization for Neural Dialog Generation

Zhiyong Wu, Wei Bi, Xiang Li, Lingpeng Kong, Ben Kao
2022 Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)   unpublished
We propose knowledge internalization (KI), which aims to complement the lexical knowledge into neural dialog models.  ...  Instead of further conditioning the knowledge-grounded dialog (KGD) models on externally retrieved knowledge, we seek to integrate knowledge about each input token internally into the model's parameters  ...  Knowledge Internalization for Neural Dialog Models In this section, we illustrate how to train a dialog model with knowledge internalization.  ... 
doi:10.18653/v1/2022.acl-long.547 fatcat:f2tbhusi3bddxjb4nl3g2ppjte

A Neural Turn-Taking Model without RNN

Chaoran Liu, Carlos Ishi, Hiroshi Ishiguro
2019 Interspeech 2019  
In this paper, we propose a non-RNN model for the timing estimation of turn-taking in dialogs. The proposed model takes lexical and acoustic features as its input to predict a turn's end.  ...  Estimating the timing of turn-taking is a critical feature of dialog systems. Such tasks require knowledge about past dialog contexts and have been modeled using RNNs in several studies.  ...  Divesh Lala for providing the dataset and the baseline model. We also thank Mr. Taiken Shintani and Ms. Taeko Murase for their help with the data preprocessing.  ... 
doi:10.21437/interspeech.2019-2270 dblp:conf/interspeech/LiuII19 fatcat:5aythlx4c5edbguhayjv4udnpm

Modeling the intonation of discourse segments for improved online dialog ACT tagging

Vivek Kumar Rangarajan Sridhar, Shrikanth Narayanan, Srinivas Bangalore
2008 Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing  
Prosody is an important cue for identifying dialog acts.  ...  Using only lexical and prosodic cues from 3 previous utterances, we achieve a DA tagging accuracy of 72% compared to the best case scenario with accurate knowledge of previous DA tag, which results in  ...  Automatic cue-based identification of dialog acts exploits multiple knowledge sources in the form of lexical, syntactic, prosodic and discourse structure cues.  ... 
doi:10.1109/icassp.2008.4518789 pmid:19132136 pmcid:PMC2614672 fatcat:75cvjo5xvrezll4s4odxliexvu

Multi-domain Dialog State Tracking using Recurrent Neural Networks

Nikola Mrkšić, Diarmuid Ó Séaghdha, Blaise Thomson, Milica Gasic, Pei-Hao Su, David Vandyke, Tsung-Hsien Wen, Steve Young
2015 Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)  
This paper shows that dialog data drawn from different dialog domains can be used to train a general belief tracking model which can operate across all of these domains, exhibiting superior performance  ...  Dialog state tracking is a key component of many modern dialog systems, most of which are designed with a single, welldefined domain in mind.  ...  It also maintains a memory vector that stores internal information about the dialog context.  ... 
doi:10.3115/v1/p15-2130 dblp:conf/acl/MrksicSTGSVWY15 fatcat:zic6tg2c7jfpjdro347ybectxm

Conversational Document Prediction to Assist Customer Care Agents [article]

Jatin Ganhotra, Haggai Roitman, Doron Cohen, Nathaniel Mills, Chulaka Gunasekara, Yosi Mass, Sachindra Joshi, Luis Lastras, David Konopnicki
2020 arXiv   pre-print
Using this dataset and two others, we investigate state-of-the art deep learning (DL) and information retrieval (IR) models for the task.  ...  C Appendix: Extracting content from URL documents For the internal Mac-Support dataset, the document content for each URL was obtained by API calls to the customer service knowledge base.  ...  For our internal datasets, we filter out dialogs where: a) the agent doesn't provide a URL to the user, b) the URL is not in-domain (e.g.  ... 
arXiv:2010.02305v1 fatcat:7n2wzxmta5gzfmnwofj5zkrju4

Response Ranking with Deep Matching Networks and External Knowledge in Information-seeking Conversation Systems

Liu Yang, Minghui Qiu, Chen Qu, Jiafeng Guo, Yongfeng Zhang, W. Bruce Croft, Jun Huang, Haiqing Chen
2018 The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval - SIGIR '18  
Our models and research findings provide new insights on how to utilize external knowledge with deep neural models for response selection and have implications for the design of the next generation of  ...  More research is needed for information-seeking conversations. There is also a lack of modeling external knowledge beyond the dialog utterances among current conversational models.  ...  knowledge into deep neural matching networks for response ranking.  ... 
doi:10.1145/3209978.3210011 dblp:conf/sigir/YangQQGZCHC18 fatcat:xt6767facjezpd6umj3ojrrgh4

Information retrieval with semantic memory model

Julian Szymański, Włodzisław Duch
2012 Cognitive Systems Research  
The game facilitates lexical knowledge validation and acquisition through the interaction with humans via supervised dialog templates.  ...  Several similarity measures have been used to compare the completeness of knowledge retrieved automatically and corrected through active dialogs to a "golden standard".  ...  Acknowledgment This work has been supported by the Polish Committee for Scientific Research Grant N516 035 31/3499.  ... 
doi:10.1016/j.cogsys.2011.02.002 fatcat:nr2vdic72rgi5fahfmrlf6qahe

Multi-domain Dialog State Tracking using Recurrent Neural Networks [article]

Nikola Mrkšić, Diarmuid Ó Séaghdha, Blaise Thomson, Milica Gašić, Pei-Hao Su, David Vandyke, Tsung-Hsien Wen, Steve Young
2015 arXiv   pre-print
This paper shows that dialog data drawn from different dialog domains can be used to train a general belief tracking model which can operate across all of these domains, exhibiting superior performance  ...  Dialog state tracking is a key component of many modern dialog systems, most of which are designed with a single, well-defined domain in mind.  ...  It also maintains a memory vector that stores internal information about the dialog context.  ... 
arXiv:1506.07190v1 fatcat:oad6amofujeiffmqfjsqfcefsm

Robust Conversational AI with Grounded Text Generation [article]

Jianfeng Gao, Baolin Peng, Chunyuan Li, Jinchao Li, Shahin Shayandeh, Lars Liden, Heung-Yeung Shum
2020 arXiv   pre-print
GTG is a hybrid model which uses a large-scale Transformer neural network as its backbone, combined with symbol-manipulation modules for knowledge base inference and prior knowledge encoding, to generate  ...  responses grounded in dialog belief state and real-world knowledge for task completion.  ...  These language model based chatbots cannot reliably generate responses grounded in realworld knowledge, do not have internal cognitive models to keep track of dialog states, and cannot generate goal-directed  ... 
arXiv:2009.03457v1 fatcat:2462mlxn7fg3zgw5mwknoqngaa

Question detection from acoustic features using recurrent neural network with gated recurrent unit

Yaodong Tang, Yuchen Huang, Zhiyong Wu, Helen Meng, Mingxing Xu, Lianhong Cai
2016 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
Question detection is of importance for many speech applications. Only parts of the speech utterances can provide useful clues for question detection.  ...  Experimental results show that the features extracted within proper time scale make the classifier perform better than the baseline method with pre-designed lexical and acoustic feature set.  ...  However, there are two major problems for question detection using lexical features only. First, some questions share the same lexical representation (i.e. words) with its statement form.  ... 
doi:10.1109/icassp.2016.7472854 dblp:conf/icassp/TangHWMXC16 fatcat:2w44it2j3nccrkh5xvgvpjhtey

RubyStar: A Non-Task-Oriented Mixture Model Dialog System [article]

Huiting Liu, Tao Lin, Hanfei Sun, Weijian Lin, Chih-Wei Chang, Teng Zhong, Alexander Rudnicky
2017 arXiv   pre-print
RubyStar is a dialog system designed to create "human-like" conversation by combining different response generation strategies.  ...  We describe a rating scheme we developed for evaluating response generation.  ...  Neural Dialog Generation We used sequence-to-sequence variant approaches to generate candidate response.  ... 
arXiv:1711.02781v3 fatcat:orqlnnx46jamxngd2akuoun56y

Lexically Constrained Decoding for Sequence Generation Using Grid Beam Search

Chris Hokamp, Qun Liu
2017 Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)  
We present Grid Beam Search (GBS), an algorithm which extends beam search to allow the inclusion of pre-specified lexical constraints.  ...  The algorithm can be used with any model that generates a sequenceŷ = {y 0 . . . y T }, by maximizing p(y|x) = t p(y t |x; {y 0 . . . y t−1 }). Lex-  ...  We thank the anonymous reviewers, as well as Iacer Calixto, Peyman Passban, and Henry Elder for helpful feedback on early versions of this work.  ... 
doi:10.18653/v1/p17-1141 dblp:conf/acl/HokampL17 fatcat:b4u7j5jdonb6ji5tcufsqfwgnq

Cross-lingual Semantic Specialization via Lexical Relation Induction

Edoardo Maria Ponti, Ivan Vulić, Goran Glavaš, Roi Reichart, Anna Korhonen
2019 Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)  
Our results also suggest that the transfer method is effective even for lexically distant source-target language pairs.  ...  Semantic specialization integrates structured linguistic knowledge from external resources (such as lexical relations in WordNet) into pretrained distributional vectors in the form of constraints.  ...  We thank the three anonymous reviewers for their helpful comments and suggestions.  ... 
doi:10.18653/v1/d19-1226 dblp:conf/emnlp/PontiVGRK19 fatcat:iik7yixa7nhqloszlj7mhhjqky

Lexically Constrained Decoding for Sequence Generation Using Grid Beam Search [article]

Chris Hokamp, Qun Liu
2017 arXiv   pre-print
We demonstrate the feasibility and flexibility of Lexically Constrained Decoding by conducting experiments on Neural Interactive-Predictive Translation, as well as Domain Adaptation for Neural Machine  ...  This is a very general way to incorporate additional knowledge into a model's output without requiring any modification of the model parameters or training data.  ...  We thank the anonymous reviewers, as well as Iacer Calixto, Peyman Passban, and Henry Elder for helpful feedback on early versions of this work.  ... 
arXiv:1704.07138v2 fatcat:tfghmcy6orh6ncqgrkfd4btmxe
« Previous Showing results 1 — 15 out of 1,563 results