Диалоговый агент с обучаемым диалоговым менеджером
Журнал: Труды Московского физико-технического института @trudy-mipt
Рубрика: Информатика и управление
Статья в выпуске: 4 (48) т.12, 2020 года.
Бесплатный доступ
Представлен диалоговый агент bot#1337, cозданный в рамках участия в соревновании по разработке диалоговых агентов NIPS Conversational Intelligence Challenge 2017 (ConvAI). Этот агент способен вести диалог с человеком о заданном тексте. Диалог ведется с помощью навыков определения темы, реферирования, ответов на вопросы, генерации вопросов и разговора на общие темы. Система обучалась выбирать подходящий навык для выдачи ответа. Представленный бот произведен с использованием открытых инструментов и данных; он не зависит от внешних сервисов и может работать в закрытом контуре; его диалоговый менеджер является обучаемым. Последнее позволяет разработчику сфокусироваться на создании навыков вместо описания конечного автомата агента. Bot#1337 является победителем соревнования со средней оценкой качества диалога 2.78, которые были проставлены людьми. Исходный код и обученные модели представленного бота доступны на Github.
Похожие статьи в разделе Искусственный интеллект
Короткий адрес: https://sciup.org/142230090
IDS: 142230090 | УДК: 004.891.3
Dialogue system with a supervised dialog manager
We present bot#1337: a dialog system developed for the 1st NIPS Conversational Intelligence Challenge 2017 (ConvAI). The bot is capable of conversing with humans about a given text (e.g., a paragraph from Wikipedia article). The conversation is enabled by a set of skills, including chit-chat, topic detection, text summarization, question answering and question generation. The system has been trained in a supervised setting to select an appropriate skill for generating a response. Our bot is implemented with open-source software and open data; it is self-hosted and employs a supervised dialog manager with a linear hierarchy. The latter allows a developer to focus on skill implementation rather than the finite state machine based dialog manager. Bot#1337 was a winner of the competition with an average dialogue quality score of 2.78 given by human evaluators. Source code and trained models for the bot#1337 are available on GitHub.
Список литературы Диалоговый агент с обучаемым диалоговым менеджером
- Weizenbaum J. ELIZA^a computer program for the study of natural language communication between man and machine // Communications of the ACM. 1966. V. 9, N 1. P. 36-45.
- Graves A. Generating sequences with recurrent neural networks. arXiv preprint arXiv:1308.0850. 2013.
- Wallace R.S. The anatomy of ALICE. Parsing the Turing Test. Springer, Dordrecht, 2009. P. 181-210.
- Williams J.D. [et al.}. The dialog state tracking challenge series // AI Magazine. 2014. V. 35, N 4. P. 121-124.
- Ritter A., Cherry C., Dolan B. Unsupervised modeling of twitter conversations // Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics. Association for Computational Linguistics, 2010. P. 172-180.
- Lowe R. [et al.}. The ubuntu dialogue corpus: A large dataset for research in unstructured multi-turn dialogue systems. arXiv preprint arXiv:1506.08909. 2015.
- Sordoni A. [et al.}. A Neural Network Approach to Context-Sensitive Generation of Conversational Responses // Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2015. P. 196-205.
- Vinyals O., Le Q. A neural conversational model. arXiv preprint arXiv:1506.05869. 2015.
- Sutskever I., Vinyals O., Le Q. V. Sequence to sequence learning with neural networks // Advances in neural information processing systems. 2014. P. 3104-3112.
- Williams J.D., Atui K.A., Zweig G. Hybrid Code Networks: practical and efficient end-to-end dialog control with supervised and reinforcement learning // Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2017. P. 665-677.
- Serban I. V. [et al.}. A deep reinforcement learning chatbot. arXiv preprint arXiv:1709.02349. 2017.
- Klein G. [et al.}. OpenNMT: Open-Source Toolkit for Neural Machine Translation // Proceedings of ACL 2017, System Demonstrations. 2017. P. 67-72.
- Vorontsov K. [et al.}. Bigartm: Open source library for regularized multimodal topic modeling of large collections // International Conference on Analysis of Images, Social Networks and Texts. Springer, Cham, 2015. P. 370-381.
- Seo M. [et al.}. Bidirectional attention flow for machine comprehension. arXiv preprint arXiv:1611.01603. 2016.
- Zhou Q. [et al.}. Neural question generation from text: A preliminary study // National CCF Conference on Natural Language Processing and Chinese Computing. Springer, Cham, 2017. P. 662-671.
- Joulin A. [et al.}. Bag of Tricks for Efficient Text Classification // Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers. 2017. P. 427-431.
- Pennington J., Socher R., Manning C.D. Glove: Global vectors for word representation // Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP). 2014. P. 1532-1543.
- Tiedemann J. News from OPUS-A collection of multilingual parallel corpora with tools and interfaces / /Recent advances in natural language processing. 2009. V. 5. P. 237-248.
- Rajpurkar P. [et al.}. SQuAD: 100,000+ Questions for Machine Comprehension of Text // Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing. 2016. P. 2383-2392.
- Manning C.D. [et al.}. The Stanford CoreNLP natural language processing toolkit // Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations. 2014. P. 55-60.
- Liu C.W. [et al.}. How NOT To Evaluate Your Dialogue System: An Empirical Study of Unsupervised Evaluation Metrics for Dialogue Response Generation // Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing. 2016. P. 2122-2132.
- Mauldin M. L. Chatterbots, tinvmuds, and the turing test: Entering the loebner prize competition / /AAAI. 1994. V. 94. P. 16-21.
- Ram A. [et al.}. Conversational ai: The science behind the alexa prize. arXiv preprint arXiv:1801.03604. 2018.
- Fang H. [et al.}. Sounding board-university of Washington's alexa prize submission // Alexa prize proceedings. 2017.
- Serban I.V. [et al.}. Hierarchical neural network generative models for movie dialogues // arXiv preprint arXiv:1507.04808. 2015. V. 7, N 8.
- Lavie A., Agarwal A. METEOR: An automatic metric for MT evaluation with high levels of correlation with human judgments // Proceedings of the second workshop on statistical machine translation. 2007. P. 228-231.
- Papineni K. [et al.}. BLEU: a method for automatic evaluation of machine translation // Proceedings of the 40th annual meeting on association for computational linguistics. Association for Computational Linguistics. 2002. P. 311-318.
- Liu C.W. [et al.}. How NOT To Evaluate Your Dialogue System: An Empirical Study of Unsupervised Evaluation Metrics for Dialogue Response Generation // Proceedings of the
- 2016 Conference on Empirical Methods in Natural Language Processing. 2016. P. 2122-2132.
- Lowe R. [et al.}. On the Evaluation of Dialogue Systems with Next Utterance Classification // Proceedings of the 17th Annual Meeting of the Special Interest Group on Discourse and Dialogue. 2016. P. 264-269. MLA.
- Lowe R. [et al.}. Towards an Automatic Turing Test: Learning to Evaluate Dialogue Responses // Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2017. P. 1116-1126.
- Chorowski J. [et al.}. A Talker Ensemble: The University of Wroclaw's Entry to the NIPS
- 2017 Conversational Intelligence Challenge. The NIPS'17 Competition: Building Intelligent Systems. Springer, Cham, 2018. P. 59-77.
- Kingma D.P., Ba J. Adam: A Method for Stochastic Optimization. arXiv preprint arXiv:1412.6980. 2014.
