RNN (Recurrent Neural Network) & Chatbots

Notes:

Recurrent neural networks (RNNs) are a type of artificial neural network that is particularly well-suited for processing sequential data, such as natural language text or time series data. RNNs are able to process input sequences of any length by using feedback connections, which allow the network to incorporate information from previous time steps into the processing of the current time step.

In the context of chatbots, RNNs are often used to process user input and generate appropriate responses. For example, an RNN-based chatbot might be trained on a large dataset of conversational exchanges between humans, in which the input is a user’s message and the output is the chatbot’s response. The chatbot can then use this training to generate appropriate responses to new input it receives from users.

RNNs are typically trained using a variant of backpropagation called “backpropagation through time,” which allows the network to adjust the weights of its connections based on the error between its predicted output and the correct output. Once trained, an RNN-based chatbot can be used to handle a variety of tasks, including customer service, information retrieval, and language translation.

From 2014 to 2015, the number of academic papers in Google Scholar covering recurrent neural networks and dialog systems tripled, from around 50x in 2014 to 150x in 2015. In 2016, that number doubled again to over 300x.

  • Dialog state tracker is a component of a chatbot or conversational agent that is responsible for keeping track of the current state of the conversation. The dialog state tracker maintains a record of the conversation history and uses this information to infer the context and intent of the current user input. The dialog state tracker is typically used in conjunction with natural language processing (NLP) and machine learning algorithms to help the chatbot understand and respond appropriately to user input.
  • Dialog state tracking is the process of keeping track of the current state of a conversation between a chatbot and a user. Dialog state tracking involves maintaining a record of the conversation history and using this information to infer the context and intent of the current user input. This information is used by the chatbot to generate an appropriate response and to maintain a coherent and natural-sounding conversation.
  • Discourse relation recognition is the process of identifying the relationships between different units of text in a conversation or document. Discourse relations can include things like cause-and-effect, contrast, or elaboration. Discourse relation recognition is often used to improve the coherence and interpretability of natural language text, and can be used in applications such as text summarization and machine translation.

Resources:

References:

Wikipedia:

See also:

100 Best Recurrent Neural Network Videos

A knowledge-grounded neural conversation model
M Ghazvininejad, C Brockett, MW Chang… – Thirty-Second AAAI …, 2018 – aaai.org
… these models have been mostly applied to casual scenarios (eg, as “chatbots”) and have … Introduction Recent work has shown that conversational chatbot mod- els can be trained in … as distinct from more goal- directed neural dialog modeling in which question-answer slots are …

Response selection with topic clues for retrieval-based chatbots
Y Wu, Z Li, W Wu, M Zhou – Neurocomputing, 2018 – Elsevier
… (2) proposal of a topic aware attentive recurrent neural network for matching with topics … 3. Retrieval based chatbots overview. In this section, we introduce a general framework of a retrieval based chatbot, as illustrated in Fig …

Emotional chatting machine: Emotional conversation generation with internal and external memory
H Zhou, M Huang, T Zhang, X Zhu, B Liu – Thirty-Second AAAI Conference …, 2018 – aaai.org
… To cre- ate a chatbot capable of communicating with a user at the human level, it is necessary to equip the machine with the ability of perceiving and expressing emotions. Existing studies show that addressing affect and emo- tion in dialogue systems or conversational agents …

Dialog generation using multi-turn reasoning neural networks
X Wu, A Martinez, M Klyen – Proceedings of the 2018 Conference of the …, 2018 – aclweb.org
… Dialogue systems such as chatbots are a thriving topic that is attracting increasing attentions … To embed the previous word sequence into a fixed-size vector, recurrent neural networks (RNN) such … as a “document”, current user’s query as a “question” and the chatbot’s re- sponse …

Lingke: A fine-grained multi-turn chatbot for customer service
P Zhu, Z Zhang, J Li, Y Huang, H Zhao – arXiv preprint arXiv:1808.03430, 2018 – arxiv.org
… 2015) including TF-IDF, Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), LSTM … A new architecture for multi-turn response selection in retrieval-based chatbots … Docchat: An information retrieval approach for chatbot engines using unstructured documents …

From Eliza to XiaoIce: challenges and opportunities with social chatbots
HY Shum, X He, D Li – Frontiers of Information Technology & Electronic …, 2018 – Springer
… generates a text response as the output. It provides the communication capability of social chatbots. Fig … Fig. 9 Recurrent neural network (RNN) based encoder- decoder framework for response generation The user says, ‘hi dude’, and the chatbot replies ‘how are you’ …

Hierarchical variational memory network for dialogue generation
H Chen, Z Ren, J Tang, YE Zhao, D Yin – … of the 2018 World Wide Web …, 2018 – dl.acm.org
… models [20, 21, 33, 34, 38, 41] have been proved to be capable in multiple dialogue system applications with promising performance. Most of these end-to-end neural genera- tion models apply encoder-decoder architecture based on recurrent neural network, which directly …

Complex sequential question answering: Towards learning to converse over linked question answer pairs with a knowledge graph
A Saha, V Pahuja, MM Khapra… – Thirty-Second AAAI …, 2018 – aaai.org
… Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with … Abstract While conversing with chatbots, humans typically tend to ask many … While Question Answering (QA) and dialog systems have been studied independently, there is …

Response ranking with deep matching networks and external knowledge in information-seeking conversation systems
L Yang, M Qiu, C Qu, J Guo, Y Zhang… – The 41st International …, 2018 – dl.acm.org
… architecture for response generation that is both context-sensitive and data-driven utilizing the Recurrent Neural Network Language Model … Recently a number of neural rank- ing models have been proposed for information retrieval, question answering and conversation …

Multi-cast attention networks for retrieval-based question answering and response prediction
Y Tay, LA Tuan, SC Hui – arXiv preprint arXiv:1806.00778, 2018 – arxiv.org
… a wide assortment of real life applications, ranging from standard web search to automated chatbots … can be general- izable to different problem domains such as question-answering or message … Figure 1 illustrates the overall model architecture for question-answer retrieval …

Chitty-Chitty-Chat Bot: Deep Learning for Conversational AI.
R Yan – IJCAI, 2018 – ijcai.org
… A standard chatbot system presumes that only humans will take the initiative role in chit-chats … at the entrance of future success in more advanced conversational systems (social chatbots and/or … Empirical eval- uation of gated recurrent neural networks on sequence modeling …

Modeling multi-turn conversation with deep utterance aggregation
Z Zhang, J Li, P Zhu, H Zhao, G Liu – arXiv preprint arXiv:1806.09102, 2018 – arxiv.org
… Figure 2: Structure overview of the proposed dialogue system … the concerned commodities) and intentions, which makes our corpus more challenging than previous chitchat or question answering based corpora … Superagent: A customer service chatbot for e-commerce websites …

Conversational memory network for emotion recognition in dyadic dialogue videos
D Hazarika, S Poria, A Zadeh, E Cambria… – Proceedings of the …, 2018 – aclweb.org
… et al., 2018), and intelligent systems such as smart homes and chat- bots (Young et … mental in the progress of multiple research prob- lems, eg, question-answering (Weston et … Gated Recurrent Unit: GRUs are a gating mechanism in recurrent neural networks introduced by (Cho …

An Ensemble of Retrieval-Based and Generation-Based Human-Computer Conversation Systems.
Y Song, R Yan, CT Li, JY Nie, M Zhang, D Zhao – 2018 – openreview.net
… Recently, researchers have paid increasing attention to open-domain, chatbot-style human-computer conversations … the increasing popularity of on-line social media and community question-answering platforms, a … (2016a)), in which two recurrent neural networks (RNNs) are …

Emotional dialogue generation using image-grounded language models
B Huber, D McDuff, C Brockett, M Galley… – Proceedings of the 2018 …, 2018 – dl.acm.org
… Visual Question Answering CHI 2018 Paper … This model maps the input sequences to an output sequence (seq2seq model [45]) us- ing an encoder and a decoder recurrent neural network (RNN). The initial recurrent state is the 500-dimensional encoding of the textual context …

Knowledge-aware Multimodal Dialogue Systems
L Liao, Y Ma, X He, R Hong, T Chua – 2018 ACM Multimedia Conference …, 2018 – dl.acm.org
… Also known as chat bots, the non-task- oriented systems converse with human typically on … to- Language problems such as image captioning and visual question answering (VQA … through a combination of Convolutional Neural Networks (CNN) and recurrent neural networks [44] …

Topic-based evaluation for conversational bots
F Guo, A Metallinou, C Khatri, A Raju… – arXiv preprint arXiv …, 2018 – arxiv.org
… on natural language and dialog research, and researchers have experimented with Recurrent Neural Networks (RNNs)[11 … proven a state-of-the-art model for the Factoid Question Answering task [6 … RER correlates well negatively with the user ratings for 15 chatbots (? = -0.717) …

Alquist: The alexa prize socialbot
J Pichl, P Marek, J Konrád, M Matulík… – arXiv preprint arXiv …, 2018 – arxiv.org
… Eliza [2] is one of the most famous chat-bots from that era … language understanding [7] or the responses can be generated using a recurrent neural network with sequence-to … Question answering The question answering module is used whenever the user asks a factoid question …

Enhance word representation for out-of-vocabulary on ubuntu dialogue corpus
J Dong, J Huang – arXiv preprint arXiv:1802.02614, 2018 – arxiv.org
… for the exploration of deep neural network modeling in the context of dialogue systems … and sub-sequence levels and then aggregate the matching information by recurrent neural network … matching aggregation are effective in NLP tasks such as question/answering and natural …

Natural Language Processing for Industry
S Quarteroni – Informatik-Spektrum, 2018 – Springer
… a research perspective has greatly benefited from the recent generative models (recurrent neural networks), the “canned … of APIs or frameworks to support third-party companies in developing chatbots for their … These APIs and frameworks allow one to rapidly “bind” a chatbot to a …

Question Answering for Technical Customer Support
Y Li, Q Miao, J Geng, C Alt, R Schwarzenberg… – … Conference on Natural …, 2018 – Springer
… Chung, J., Gulcehre, C., Cho, KH: Empirical evaluation of gated recurrent neural networks on sequence … 2007)Google Scholar. 9. Yih, W., Chang, MW, Meek, C.: Question answering using enhanced … a new architecture for multi-turn response selection in retrieval-based Chatbots …

Conversational Modelling for ChatBots: Current Approaches and Future Directions
M McTear – 2018 – spokenlanguagetechnology.com
… to the identification of intents and the extraction of entities using Recurrent Neural Networks [5 … of the same skill (eg weath- er forecast) may behave differently on the same chatbot … se- quential mechanics of conversation is essential to make interactions with chatbots intuitive and …

Towards Automated Customer Support
M Hardalov, I Koychev, P Nakov – International Conference on Artificial …, 2018 – Springer
… Our encoder uses Open image in new window bidirectional recurrent neural network (RNN) based on long short-term memory (LSTM) [6 … However, it has been argued [10, 11] that such word-overlapping measures are not very suitable for evaluating chatbots … Chatbot responses …

Efficient dialog policy learning via positive memory retention
R Zhao, V Tresp – 2018 IEEE Spoken Language Technology …, 2018 – ieeexplore.ieee.org
… the collection of the required data in form of conversations between chat- bots and human … Index Terms— Goal-Oriented Dialog System, Deep Re- inforcement Learning, Recurrent Neural Network … of this game is to find out the target digit by a multi-round question-answering …

Coupled context modeling for deep chit-chat: towards conversations between human and computer
R Yan, D Zhao – Proceedings of the 24th ACM SIGKDD International …, 2018 – dl.acm.org
… values as personal assistants, agent systems, and social chatbots as real … forums, social medium (eg, Facebook, Twitter) and com- munity question-answering platforms (eg … chain-based matching through information propagated in the recurrent neural networks (RNNs) sequence …

Content-Oriented User Modeling for Personalized Response Ranking in Chatbots
B Liu, Z Xu, C Sun, B Wang, X Wang, DF Wong… – IEEE/ACM Transactions …, 2018 – dl.acm.org
… Being aware of the importance of the leading role of chatbots, Li et al … B. User Modeling The personalized chat is a newly emerging demand in the research on chatbot; thus, little work has been conducted in this field … 42] extracted user information from question-answer pairs and …

Towards Building Large Scale Multimodal Domain-Aware Conversation Systems
A Saha, MM Khapra, K Sankaranarayanan – Thirty-Second AAAI …, 2018 – aaai.org
… 2015), video question answering (Zeng et al … However, even though there is a growing demand for chat- bots that can converse using multiple modalities with hu- mans in several domains such as … 2, we use a standard recurrent neural net- work based decoder with GRU cells …

Empathetic Dialog Systems
P Fung, D Bertero, P Xu, JH Park, CS Wu… – The International …, 2018 – lrec-conf.org
… In this paper/talk, we propose that dialog systems, both task-oriented and chatbots, can benefit from a new paradigm of … Finally, we showed how to train an end-to-end chatbot with reinforce- ment deep learning that learns a … Sequence level training with recurrent neural networks …

CCG supertagging via Bidirectional LSTM-CRF neural architecture
R Kadari, Y Zhang, W Zhang, T Liu – Neurocomputing, 2018 – Elsevier
… The output from such systems can facilitate and benefit many applications such as information retrieval, question answering, and parsing … of deep neural networks based approaches have addressed the problem of CCG supertagging such as Recurrent Neural Networks (RNN) [5 …

Out-of-domain detection method based on sentence distance for dialogue systems
KJ Oh, DK Lee, C Park, YS Jeong… – … Conference on Big …, 2018 – ieeexplore.ieee.org
… sentence distances are measured by sentence embedding vectors using RNN(Recurrent Neural Network) encoders with … method, and we apply this method to develop a chatbot system for … Accordingly, we divide a dialogue into specific question- answer(Q&A) pairs based on …

Multitask learning for neural generative question answering
Y Huang, T Zhong – Machine Vision and Applications, 2018 – Springer
… Building chatbot in human–computer conversation via natural language is one of the … Gucehre, C., Cho, K., Bengio, Y.: Empirical evaluation of gated recurrent neural networks on sequence … network: a new architecture for multi-turn response selection in retrieval-based chatbots …

Learning goal-oriented visual dialog via tempered policy gradient
R Zhao, V Tresp – 2018 IEEE Spoken Language Technology …, 2018 – ieeexplore.ieee.org
… Index Terms— Goal-Oriented Dialog System, Deep Re- inforcement Learning, Recurrent Neural Network … [7] formulated a visual dialogue system which is about two chatbots asking and … 17] with At- tention Mechanism [18] to process the generated question- answer pairs …

Memory-based matching models for multi-turn response selection in retrieval-based chatbots
X Lu, M Lan, Y Wu – CCF International Conference on Natural Language …, 2018 – Springer
… new architecture for multi-turn response selection in retrieval-based chatbots (2017)Google … H., Radev, D.: Sentence ordering and coherence modeling using recurrent neural networks (2017)Google … Y., Phan, MC, Tuan, LA, Hui, SC: Learning to rank question answer pairs with …

Improving retrieval modeling using cross convolution networks and multi frequency word embedding
G An, M Shafiee, D Shamsi – arXiv preprint arXiv:1802.05373, 2018 – arxiv.org
… To build a satisfying chatbot that has the abil- ity of managing a goal … More specif- ically, Recurrent Neural Networks (RNN) built by Long Short-Term Memory (LSTM) units (Hochre- iter … in multi- turn conversation is more general than a traditional question answering (QA) problem …

Hierarchical hybrid code networks for task-oriented dialogue
W Liang, M Yang – International Conference on Intelligent Computing, 2018 – Springer
… Unlike open-domain chatbot, task-oriented dialogue system mainly focuses on a specific task … a pure neural network method, it’s one of the best model in question answering and dialog … in which domain knowledge is used to write codes, and recurrent neural network (RNN) is …

Why are Sequence-to-Sequence Models So Dull? Understanding the Low-Diversity Problem of Chatbots
S Jiang, M de Rijke – arXiv preprint arXiv:1809.01941, 2018 – arxiv.org
… Generally, a Seq2Seq model consists of two recurrent neural networks (RNN) as its encoder and … for tasks like question answering (Yin et al., 2015), dialogue response generation (Vinyals and Le … of Seq2Seq models is different from that of retrieval-based chatbots (Fedorenko et …

Convolutional neural networks for dialogue state tracking without pre-trained word vectors or semantic dictionaries
M Korpusik, J Glass – 2018 IEEE Spoken Language …, 2018 – ieeexplore.ieee.org
… These systems can be divided into two categories: chatbots that simply entertain the … In question answering, recent work showed improvements using deep CNN models for text … fed delexicalized user utterances into a recurrent neural network, which output a dis- tribution over …

Building Advanced Dialogue Managers for Goal-Oriented Dialogue Systems
V Ilievski – arXiv preprint arXiv:1806.00780, 2018 – arxiv.org
… The authors in [Vinyals and Le, 2015] used standard Recurrent Neural Networks (RNNs) and trained a Goal-Oriented Chatbot in a straightforward sequence-to sequence fashion … Thus, most of the chatbot research is on the closed-domain Chatbots, which is a case in this …

Mem2seq: Effectively incorporating knowledge bases into end-to-end task-oriented dialog systems
A Madotto, CS Wu, P Fung – arXiv preprint arXiv:1804.08217, 2018 – arxiv.org
… Recently, end-to-end approaches for dialog modeling, which use recurrent neural networks (RNN) encoder-decoder … Figure 1: The proposed Mem2Seq architecture for task-oriented dialog systems … sys- tems (Eric and Manning, 2017; Zhao et al., 2017) and chat-bots (Ritter et al …

Encoding emotional information for sequence-to-sequence response generation
YH Chan, AKF Lui – … on Artificial Intelligence and Big Data …, 2018 – ieeexplore.ieee.org
… The approach is essentially the same for response generation in chatbots that the … An emotional chatbot is a conversational agent that is conditioned to generate emotional … to Sequence Model Sequence to sequence model is built upon Recurrent Neural network (RNN) utilizing …

Personalized response generation by Dual-learning based domain adaptation
M Yang, W Tu, Q Qu, Z Zhao, X Chen, J Zhu – Neural Networks, 2018 – Elsevier
… applications, such as e-commerce, technical support services, entertaining chatbots, information retrieval … The decoding process is essentially a recurrent neural network language model … Tang, Duan, Qin, and Zhou (2017) treated the question answering and question generation …

Deep Learning in Spoken and Text-Based Dialog Systems
A Celikyilmaz, L Deng, D Hakkani-Tür – Deep Learning in Natural …, 2018 – Springer
… and Li 2012), and sequence-to-sequence models with different structures, such as, vanilla recurrent neural networks (Vinyals and … 19 a conversational dialog engine for creating chat bots, chatbot-rnn, 20 a toy chatbot powered by … In metaguide.com, 21 top 100 chatbots are listed …

Intent Detection System Based on Word Embeddings
K Balodis, D Deksne – International Conference on Artificial Intelligence …, 2018 – Springer
… we use the Wit.ai service 1 which is one of few popular chatbot creation services that … Liu, B., Lane, I.: Attention-based recurrent neural network models for joint intent detection and slot filling … Shawar, BA, Atwell, E.: Machine learning from dialogue corpora to generate chatbots …

Improving Matching Models with Contextualized Word Representations for Multi-turn Response Selection in Retrieval-based Chatbots
C Tao, W Wu, C Xu, Y Feng, D Zhao, R Yan – arXiv preprint arXiv …, 2018 – arxiv.org
… Existing approaches to building a chatbot include generation-based meth- ods (Shang et al … to im- prove matching models for multi-turn response selection in retrieval-based chatbots … for a word are naturally encoded by the utterance-level recurrent neural network (RNN) and the …

Multi-Cast Attention Networks
Y Tay, LA Tuan, SC Hui – Proceedings of the 24th ACM SIGKDD …, 2018 – dl.acm.org
… a wide assortment of real life applications, ranging from standard web search to automated chatbots … can be general- izable to different problem domains such as question-answering or message … Figure 1 illustrates the overall model architecture for question-answer retrieval …

Assigning Personality/Profile to a Chatting Machine for Coherent Conversation Generation.
Q Qian, M Huang, H Zhao, J Xu, X Zhu – IJCAI, 2018 – ijcai.org
… argued by [Vinyals and Le, 2015], it’s still quite im- possible for current chatbots to pass … our task re- quires to generate responses that are coherent to the chatbot’s pre-specified … Another related work is generative question answering (GenQA) [Yin et al., 2015] which generates a …

Conversational model adaptation via KL divergence regularization
J Li, P Luo, F Lin, B Chen – Thirty-Second AAAI Conference on Artificial …, 2018 – aaai.org
… Specifically, for the building of a new chatbot, it leverages not only a lim- ited amount of training data from this target domain, but also some existing conversational models … In other words, chat- bot users may ask similar questions to the two chatbots from related vertical …

A Survey on Chatbot Implementation in Customer Service Industry through Deep Neural Networks
M Nuruzzaman, OK Hussain – 2018 IEEE 15th International …, 2018 – ieeexplore.ieee.org
… the order of the utterances in the context with a recurrent neural network (RNN) … able to handle the more challenging issues that are normally obstacles for simpler chatbots … Chatbot Technical Specification Drawback Input/output Technique Eliza [9] Basic Pattern matching with …

PESUBot: An Empathetic Goal Oriented Chatbot
HV Kumar, J Nagaraj, M Irfan… – … on Advances in …, 2018 – ieeexplore.ieee.org
… The authors of this paper use Recurrent Neural Network (specifically, Long Short Term Memory Units (LSTM)) to build an end to end task-oriented dialog system and explicitly … This is because, the HRED model was built for general chatbots while our chatbot was more of a …

Supervised question answering system for technical support
S Shim, G Chodwadia, K Jain, C Patel… – 2018 IEEE 8th …, 2018 – ieeexplore.ieee.org
… A QA system can serve the chatbot with relevant answers in respect to a … B. Semi Supervised Question retrieval with Gated Convolutions Question answering forums have grown immensely … A. RNN Recurrent neural networks are a variant of neural networks that allows for time …

Neural Dialogue System with Emotion Embeddings
R Shantala, G Kyselov… – 2018 IEEE First …, 2018 – ieeexplore.ieee.org
… dialogue systems and shows how to create the simplest data- driven chat-bot using the … For example, such chatbots can be used for automatic user support or foreign language … training speed and perplexity by the use of more sophisticated recurrent neural network techniques or …

Improving Computer Generated Dialog with Auxiliary Loss Functions and Custom Evaluation Metrics
T Conley, JS Clair, J Kalita – cs.uccs.edu
… This research joins the quest by creating a dialog generating Recurrent Neural Network (RNN) and by enhancing … con- sist of 15 question and answer pairs generated by two different chatbots … With better evalua- tion models, a neural-network-based chatbot may be enhanced to …

Impact of Auxiliary Loss Functions on Dialogue Generation Using Mutual Information
JS Clair, T Conley, J Kalita – cs.uccs.edu
… Figure 1 illustrates a single time-step t in sequence pro- cessing by our recurrent neural network … Dialogue generation still has a long way to go in creating chatbots that can fluently converse in … in this paper, but it would shed more light on the competency of chatbot dialogue and …

A Manually Annotated Chinese Corpus for Non-task-oriented Dialogue Systems
J Li, Y Song, H Zhang, S Shi – arXiv preprint arXiv:1805.05542, 2018 – arxiv.org
… chatbots from yielding “one size fits all” replies, which is a major drawback of the existing chatbots (Li et … Therefore, it is more reliable than automatic collected data and thus potentially ben- eficial to chatbot training and … Speech Recognition with Deep Recurrent Neural Networks …

Modern Chatbot Systems: A Technical Review
AS Lokman, MA Ameedeen – Proceedings of the Future Technologies …, 2018 – Springer
… Referring to Table 1, it is clear that modern chatbots are 90% similar in term of architectural design and its implementation … Yin, Z., Chang, KH, Zhang, R.: DeepProbe: information directed sequence understanding and chatbot design via recurrent neural networks …

A Study on Dialogue Reward Prediction for Open-Ended Conversational Agents
H Cuayáhuitl, S Ryu, D Lee, J Kim – arXiv preprint arXiv:1812.00350, 2018 – arxiv.org
… 17] train a neural multiclass classifier and linear regressor from human-chatbot dialogues from … Figure 1: Illustration of our recurrent neural network for dialogue reward prediction … Future conversational agents or chatbots can optimise their behaviour using neural regressors with …

Implementing ChatBots using Neural Machine Translation techniques
A Nuez Ezquerra – 2018 – upcommons.upc.edu
… A great step forward to the area of generative-based chatbots was the implementation of a model using an encoder/decoder architecture with recurrent neural networks known as … The chatbot uses the Artificial Intelligence Markup Language (AIML) for the definition of …

Proceedings of the 19th Annual SIGdial Meeting on Discourse and Dialogue
K Komatani, D Litman, K Yu, A Papangelis… – Proceedings of the 19th …, 2018 – aclweb.org
… Changing the Level of Directness in Dialogue using Dialogue Vector Models and Recurrent Neural Networks Louisa Pragst and Stefan Ultes … Role play-based question-answering by real users for building chatbots with con- sistent personalities Ryuichiro Higashinaka …

Conversation Modeling with Neural Network
JY Patil, GP Potdar – Asian Journal of Research in Computer …, 2018 – journalajrcos.com
… Keywords: Natural language processing; deep learning; Chatbots; natural language understanding; artificial intelligence … Author then evaluates the model for question-answer task … Xiv160206291G, [19] Chung J, Ahn S, Bengio Y. Hierarchical multiscale recurrent neural networks …

A Systematic Approach to Implementing Chatbots in Organizations–RTU Leo Showcase
A Mislevics, J Grundspenkis, R Rollande – ceur-ws.org
… Production Ready Chatbots: Generate if not Retrieve, Nov. 2017. Available … Z. Yin, K. Chang, and R. Zhang. DeepProbe: Information Directed Sequence Understand- ing and Chatbot Design via Recurrent Neural Networks, Jul. 2017 …

Open-domain neural conversational agents: The step towards artificial general intelligence
S Arsovski, S Wong, AD Cheok – International Journal of …, 2018 – openaccess.city.ac.uk
… After almost 50 years since the introduction of chatbots and numerous surveys over the past several decades, chatbot … 2. Recurrent Neural Network (RNN) [14 … coherent and interesting dialogues by applying deep reinforcement learning to model future reward in chatbot dialogue …

Efficient Visual Dialog Policy Learning via Positive Memory Retention
R Zhao, V Tresp – nips2018vigil.github.io
… The training of chat-bots using on-policy policy gradient methods requires numerous training samples … game is to find out the target digit by a multi-round question-answering … Recurrent Language Models: The goal of a recurrent neural network (RNN) based language model in …

NIPS Conversational Intelligence Challenge 2017 Winner System: Skill-based Conversational Agent with Supervised Dialog Manager
I Yusupov, Y Kuratov – Proceedings of the 27th International Conference …, 2018 – aclweb.org
… This is not an optimal way of evaluating chatbots either, because a relevant answer can be different from an oracle (Liu et al., 2016) … 2013. Generating sequences with recurrent neural networks. arXiv preprint arXiv:1308.0850 … 2017. A deep reinforcement learning chatbot …

OurDirection: An Interactive Dialogue Framework For Chatting with Government Officials
S Abrishamkar, JX Huang – Proceedings of the ACM Symposium on …, 2018 – dl.acm.org
… This process resulted in total 27,731 question/answer pairs that are used in the training and testing of all models … 2013. Generating Sequences With Recurrent Neural Networks … DocChat: An Information Retrieval Approach for Chatbot Engines Using Unstructured Documents …

A Virtual Chatbot for ITSM Application
S Raut – Asian Journal For Convergence In Technology …, 2018 – asianssr.org
… Seq2Seq model consists of two Recurrent Neural Networks (RNNs) namely an Encoder and a Decoder … Chatbots that are developed using deep learning, mostly use a certain variant of sequence to sequence (Seq2Seq … [6] Bayu Setiaji, Ferry Wahyu Wibowo, “Chatbot Using A …

Context-Aware Dialog Re-Ranking for Task-Oriented Dialog Systems
J Ohmura, M Eskenazi – 2018 IEEE Spoken Language …, 2018 – ieeexplore.ieee.org
… dialog context from the Memory Networks, ie, the former uses recurrent neural networks, the latter … technique to other types of dialog systems (eg, chit-chat and question answering) … Hui Xue, “Ranking responses oriented to conversational relevance in chat- bots,” in Proceedings …

Fantom: A Crowdsourced Social Chatbot using an Evolving Dialog Graph
P Jonell, M Bystedt, FI Dogan, P Fallgren, J Ivarsson… – dex-microsites-prod.s3.amazonaws …
… Most chatbots of today are trained using example interactions … the system utterance on a letter-by-letter or word-by-word level often using variations of recurrent neural networks, such as … respect to the roles the both users had, while in our case, the role of the chatbot is different …

Sentiment classification using N-ary tree-structured gated recurrent unit networks
V Tsakalos, R Henriques – Proceedings of the 10th International Joint …, 2018 – run.unl.pt
… Artificial Neural Networks have achieved state-of-art perfor- mance at Question-Answering tasks (Berant and Li- ang, 2014), Dialogue agents(Chat-bots)(Young et al … The vanishing gradient problem du- ring learning recurrent neural nets and problem soluti- ons. Int …

Artificial Intelligence for Conversational Robo-Advisor
MY Day, JT Lin, YC Chen – 2018 IEEE/ACM International …, 2018 – ieeexplore.ieee.org
… in this study: • Seq2Seq: The sequence-to-sequence (Seq2Seq) model concatenates two recurrent neural networks (RNNs)/long … This type of chatbots is more suitable for the knowledge base and retrieval … Section 2 describes the literature on asset allocation, DL, and chatbot …

Natural Language Processing for Industrial Applications
S Quarteroni – Spektrum, 2018 – meine.gi.de
… a research perspective has greatly benefited from the recent generative models (recurrent neural networks), the “canned … of APIs or frameworks to support third-party companies in developing chatbots for their … These APIs and frameworks allow one to rapidly “bind” a chatbot to a …

Proceedings of the 22nd Conference on Computational Natural Language Learning
A Korhonen, I Titov – Proceedings of the 22nd Conference on …, 2018 – aclweb.org
… Comparing Attention-Based Convolutional and Recurrent Neural Networks: Suc- cess and Limitations in … Driven Ques- tion Generation and Neural Based Question Answering Deepak Gupta, Pabitra … Churn Intent Detection in Multilingual Chatbot Conversations and Social Media …

Using Deep Learning and an External Knowledge Base to Develop Human-Robot Dialogues
JY Huang, TA Lin, WP Lee – 2018 IEEE International …, 2018 – ieeexplore.ieee.org
… Recently, the demand on chatbots has begun to perform open-domain conversations … In a question-answering system, the main focus is on the selection of an appropriate response from a corpus given a … The model is a deep recurrent neural network with augmented memory …

Retrieval-Enhanced Adversarial Training for Neural Response Generation
Q Zhu, L Cui, W Zhang, F Wei, Y Chen, T Liu – arXiv preprint arXiv …, 2018 – arxiv.org
… in real-world applications, including customer service systems, personal assistants, and chatbots. Early dialogue systems are often built using the rule-based method (Weizenbaum, 1966 … representation repm by the discriminator using a recurrent neural network (RNN) (Mikolov …

Response selection from unstructured documents for human-computer conversation systems
Z Yan, N Duan, J Bao, P Chen, M Zhou, Z Li – Knowledge-Based Systems, 2018 – Elsevier
… Side-by-side evaluation between DocChat and a famous chatbot demonstrates that … Convolutional neural networks (CNN) [26], [27] and recurrent neural networks (RNN) [28] are used to … used in phrase table extraction which are crawled from community question answering sites …

Analysis, discovery and exploitation of open data for the creation of question-answering systems
G Molina Gallego – 2018 – rua.ua.es
… In this last decade, with the growth of Recurrent Neural Network (RNN) and Long Short-Term Memory … Chatbots have been become more popular from the last two decades, actually, many companies use … To better understand the potential of a chatbot, it is needed to know their …

A Knowledge-Grounded Multimodal Search-Based Conversational Agent
S Agarwal, O Dusek, I Konstas, V Rieser – arXiv preprint arXiv:1810.11954, 2018 – arxiv.org
… is a challenging new task: It extends visu- ally grounded question answering sys- tems … ubiquitous, with variants ranging from open-domain conversa- tional chit-chat bots (Ram et al … et al., 2016) addresses this limitation by using a context recurrent neural network (RNN), forming …

Concorde: Morphological Agreement in Conversational Models
D Polykovskiy, D Soloviev… – Asian Conference on …, 2018 – proceedings.mlr.press
… Neural conversational models are widely used in applications such as personal assistants and chat bots … appear in a wide range of applications, from simple rule-based chatbots to complex … We then evaluate the Q-Concorde model on a question answering task based on …

Towards Deep Conversational Recommendations
R Li, SE Kahou, H Schulz, V Michalski… – Advances in Neural …, 2018 – papers.nips.cc
… combine various elements of chit-chat, goal-directed dialogue, and even question answering in a … The chat bot is made of several components, including: a rule-based component, a matching … to output a message from the data, and a (generative) recurrent neural network (RNN) …

Advancing the State of the Art in Open Domain Dialog Systems through the Alexa Prize
C Khatri, B Hedayatnia, A Venkatesh, J Nunn… – arXiv preprint arXiv …, 2018 – arxiv.org
… Classification, Dialog Act Classification, Sensitive Content Detection, Question Answering and Conversation … first component a user interacts with in a spoken dialog system … incorporate context, involving adding contextual information for Recurrent Neural Network (RNN) models …

Why are Sequence-to-Sequence Models So Dull?
S Jiang, M de Rijke – EMNLP 2018, 2018 – aclweb.org
… Generally, a Seq2Seq model consists of two recurrent neural networks (RNN) as its encoder and decoder … Seq2Seq models also proved to be effective for tasks like question answering (Yin et al … of Seq2Seq models is different from that of retrieval-based chatbots (Fedorenko et al …

Creating an Emotion Responsive Dialogue System
A Vadehra – 2018 – uwspace.uwaterloo.ca
… 2.2 An un-rolled Recurrent Neural Network (RNN … Customer support or food ordering chatbots are examples of task spe- cific conversational agents where the … of the techniques for dialogue systems utilize the research in Machine Translation and Question Answering systems [79 …

Research Challenges in Building a Voice-based Artificial Personal Shopper-Position Paper
N Limsopatham, O Rokhlenko, D Carmel – Proceedings of the 2018 …, 2018 – aclweb.org
… Specifi- cally, in his pioneering work, Weizenbaum (1966) developed the Eliza chatbot agent for … helpdesk di- alog system using an encoder-decoder architec- ture based on recurrent neural networks … and Benton (2017) presents a literature review of quality issues with chatbots …

Evaluation of real-time deep learning turn-taking models for multiple dialogue scenarios
D Lala, K Inoue, T Kawahara – Proceedings of the 2018 on International …, 2018 – dl.acm.org
… a question-answering system and a conversational chatbot. Turn-taking is of particular importance to humanoid robots, because users will expect them to behave similar to a real human. However, human-like natural turn-taking is still a long way off in spoken dialogue systems …

Ask No More: Deciding when to guess in referential visual dialogue
R Shekhar, T Baumgartner, A Venkatesh… – arXiv preprint arXiv …, 2018 – arxiv.org
… et al., 2014) — which have shown some promise for modelling chatbot- style dialogue … Question Generator (QGen) This module is implemented as a Recurrent Neural Network (RNN) with a … since it pro- vides a simple setting with elementary question-answer sequences and is …

Improving Dialog Systems Using Knowledge Graph Embeddings
B Carignan – 2018 – curve.carleton.ca
… REM Relation Encoding Model RNN Recurrent Neural Network SGD Stochastic Gradient Descent … CHAPTER 2. BACKGROUND 6 2.1.2 Early Chatbots ELIZA [11] is an early chatbot program which uses a series of scripts to process user inputs and output pre-set responses …

Aspect-based question generation
W Hu, B Liu, J Ma, D Zhao, R Yan – 2018 – openreview.net
… skill, question generation plays an important role in both general-purpose chatbot systems and … Section 4.1 gave a method for extracting aspects from a raw question-answering training corpus … We use the Amazon question/answer corpus2 (AQAD) (Wan & McAuley, 2016) for …

Multi-Intent Hierarchical Natural Language Understanding for Chatbots
B Rychalska, H Glabska… – 2018 Fifth International …, 2018 – ieeexplore.ieee.org
… delivery chatbot, which breaks some commonly encountered limitations in datasets for goal-oriented chatbots … special characteristics of NLU in our real-life goal-oriented chatbot task, built … Attention-based recurrent neural network mo- dels for joint intent detection and slot filling …

Beyond” How may I help you?”: Assisting Customer Service Agents with Proactive Responses
M Wan, X Chen – arXiv preprint arXiv:1811.10686, 2018 – arxiv.org
… suggestion task as a candidate ranking problem; 2) different from the chatbot system (Cui … where the problem-solving task is regarded as a single-round question-answering problem, our … Model Architecture As shown in Figure 3, we apply a Recurrent Neural Network (RNN) with …

Topic-Based Question Generation
W Hu, B Liu, R Yan, D Zhao, J Ma – 2018 – openreview.net
… plays an important role in both general-purpose chatbot systems and goal … Generating factoid questions with recurrent neural networks: The 30m factoid question-answer corpus … Modeling ambiguity, subjectivity, and diverging viewpoints in opinion question answering systems …

Resolving Abstract Anaphora Implicitly in Conversational Assistants using a Hierarchically stacked RNN
P Khurana, P Agarwal, G Shroff, L Vig – Proceedings of the 24th ACM …, 2018 – dl.acm.org
… ABSTRACT Recent proliferation of conversational systems has resulted in an increased demand for more natural dialogue systems, capable of more … Dialogue and Question Answering: Broadly there are six types of conversational systems present in research litera- ture …

Conversational ai: The science behind the alexa prize
A Ram, R Prasad, C Khatri, A Venkatesh… – arXiv preprint arXiv …, 2018 – arxiv.org
… EC2 instances dedicated to a single task (eg news retrieval, facts, question-answer, weather, etc … goal of a user for a given utterance, and the dialog system needs to … Anaphora resolution is important for downstream tasks such as question answering and information extraction in …

KNADIA: Enterprise KNowledge Assisted DIAlogue Systems Using Deep Learning
M Singh, P Agarwal, A Chaudhary… – 2018 IEEE 34th …, 2018 – ieeexplore.ieee.org
… [1] proposed deep reinforcement learning based chatbot MILABOT which … Current advances in Question-Answering(QA) based on a knowledge graph, allow users to express their … 24] have used semantic parsing based model with and without use of question-answer pairs for …

Efficient Purely Convolutional Text Encoding
S Malik, A Lancucki, J Chorowski – arXiv preprint arXiv:1808.01160, 2018 – arxiv.org
… Such approaches have been applied by participants of recent chat- bot contests: The 2017 … Instead, Recurrent Neural Networks (RNNs), autoregressive models that can process input sequences of an … a method similar to the one used in Poetwannabe chatbot [Chorowski et al …

Alquist 2.0: Alexa Prize Socialbot Based on Sub-Dialogue Models
J Pichl, CTU FEE, P Marek, J Konrád, M Matulík… – dex-microsites-prod.s3.amazonaws …
… Code networks consists of following components: input convolutional neural networks, recurrent neural network, and domain … It is used to test end-to-end dialog systems in a way that … the only task which contains records of real-world conversations between humans and chatbot …

Can You be More Polite and Positive? Infusing Social Language into Task-Oriented Conversational Agents
YC Wang, R Wang, G Tur, H Williams – alborz-geramifard.com
… some researchers have studied how to incorporate social language into chatbots to generate … for natural language processing problems, such as machine translation [eg 32], question answering [eg 36 … Abstractive sentence summarization with attentive recurrent neural networks …

A Face-to-Face Neural Conversation Model
H Chu, D Li, S Fidler – … of the IEEE Conference on Computer …, 2018 – openaccess.thecvf.com
… Question- answering systems like Apple Siri and Amazon Alexa have also become a popular accessory … Dialogue systems have been explored since the 60′, with systems like ELIZA [34] and PARRY [5] already capable of engaging in relatively complex … The NeuralHank Chatbot …

Multimodal Differential Network for Visual Question Generation
BN Patro, S Kumar, VK Kurmi… – arXiv preprint arXiv …, 2018 – arxiv.org
… Current dialog systems as evaluated in (Chattopadhyay et al., 2017) show that when trained between … engaging question is an interesting and challenging task for a smart robot (like chat-bot) … natu- ral visual dialog instead of the widely prevalent vi- sual question answering bots …

Epilogue: Frontiers of NLP in the Deep Learning Era
L Deng, Y Liu – Deep Learning in Natural Language Processing, 2018 – Springer
… unsupervised deep learning approaches in NLP is language modeling using recurrent neural networks, as reviewed in … applied to all three types of dialogue systems or chatbots (intelligent assistants … in this research frontier is gaining greater urgency as chatbot conversations are …

Conversational query understanding using sequence to sequence modeling
G Ren, X Ni, M Malik, Q Ke – Proceedings of the 2018 World Wide Web …, 2018 – dl.acm.org
… 1 INTRODUCTION The recent rise of technologies such as chatbots, digital personal … However, information retrieval and question answering systems have traditionally been designed for stateless … form of the encoder and simply use a general recurrent neural network (RNN) to …

Building multi-domain conversational systems from single domain resources
D Griol, JM Molina – Neurocomputing, 2018 – Elsevier
… domains of dialog systems to complex information retrieval and question answering applications [5 … The use of Recurrent Neural Networks has been very recently proposed to complete this … information retrieval, dialogs between humans and a Japanese chatbot, dialogs between …

Natural answer generation with heterogeneous memory
Y Fu, Y Feng – Proceedings of the 2018 Conference of the North …, 2018 – aclweb.org
… Most previous question answering systems fo- cus on finding candidate words, phrases or sen- tence snippets from many resources, and ranking them for … In real-world applications like chatbots or personal assistants, users may want to know not only the exact answer word, but …

KITE: Building conversational bots from mobile apps
TJJ Li, O Riva – Proceedings of the 16th Annual International …, 2018 – dl.acm.org
… 1 INTRODUCTION The promise and excitement around conversational chatbots, or sim- ply bots, has rapidly … This approach has shown promise for non task-oriented “chit-chat” bots [11, 65, 79], where … During execution the chatbot tries to fill the slots for a target intent by asking …

Polite dialogue generation without parallel data
T Niu, M Bansal – Transactions of the Association of Computational …, 2018 – MIT Press
… Most current chatbots and conversational mod- els lack any such style, which can be a social … styles from such interactions, eg, kids learning to be rude be- cause the dialogue system encourages short … talk pages and the Stack Exchange (SE) question- answering communities …

Generating More Interesting Responses in Neural Conversation Models with Distributional Constraints
A Baheti, A Ritter, J Li, B Dolan – arXiv preprint arXiv:1809.01215, 2018 – arxiv.org
… Such re- sponse generation models could be combined with traditional dialogue systems to enable more natu- ral and adaptive conversation, in addition to new applications such as predictive response sugges- tion (Kannan et al., 2016), however many chal- lenges remain …

A taxonomy of attacks via the speech interface
MK Bispham, I Agrafiotis, M Goldsmith – 2018 – ora.ox.ac.uk
… perform an increasing range of tasks, including Web searching and question answering, diary management … Current task- based dialogue systems have some similarity with chatbots in that … Recurrent Neural Networks (RNNs), a particular type of DNN, have replaced n-grams to …

Emotional Human Machine Conversation Generation Based on SeqGAN
X Sun, X Chen, Z Pei, F Ren – 2018 First Asian Conference on …, 2018 – ieeexplore.ieee.org
… long-term dependencies, gradient disappearance and so on, and is a recurrent neural network in essence … The purpose of our emotional tags is to make the chatbot understood the … and M. Li, “Neural contextual conversation learning with labeled question-answering pairs,” 2016 …

Task-Oriented Dialog Agents Using Memory-Networks and Ensemble Learning
RFG Meléndez – 2018 – pdfs.semanticscholar.org
… than task-oriented or non task-oriented chatbots, Jurafsky and Martin (2018) use the term chatbot exclusively for … 2. task-oriented chatbots: these dialog agents are meant to deal with a small domain, specializing to perform just … (2014b) uses Recurrent Neural Networks (RNNs) to …

Computational Analysis and Understanding of Natural Languages: Principles, Methods and Applications
CR Rao, VN Gudivada – 2018 – books.google.com
… Modeling) 6.1 Count-Based Models or n-Grams 6.2 Recurrent Neural Networks Language Models 6.3 … textual entail- ment, language generation, semantic analysis, grammar correction, question- answering systems, spoken dialog systems, chatbots, passage retrieval …

Deep Context Resolution
J Chen – 2018 – uwspace.uwaterloo.ca
… To go beyond one-round con- versation, a chatbot must resolve contextual … including word embedding, convolution neural networks, recurrent neural networks, and their variances. 12 … Language data from Community Question Answering (cQA) websites fit our purpose perfectly …

Visual questioning agents
U Jain – 2018 – ideals.illinois.edu
… This helps AI systems such as driving assistants, chatbots, etc., to perform better on Turing tests … To address this issue, recurrent neural nets (RNNs) [18] decompose the space of a caption into a … While this largely resembles the visual question answering task, a variety …

Rich short text conversation using semantic-key-controlled sequence generation
K Yu, Z Zhao, X Wu, H Lin, X Liu – IEEE/ACM Transactions on Audio …, 2018 – dl.acm.org
… Index Terms—Question and answer, chatbot, short text conversation (STC), sequence to sequence … The method employs a recurrent neural network (RNN), usually long short term memory[12 … applied to many tasks, such as machine translation [7], question answering [15] and so …

A Review on Artificial Intelligence Decision Making Support System
SS Harnish Shah – j-asc.com
… Moreover, QA can be used to develop dialogue systems and chatbots … MM Arefin Zaman[1] proposed proposed a convolutional recurrent neural network model for question answering task … In the proposed model based on the Question answering model we are going to make a …

State-of-the-Art Approaches for German Language Chat-Bot Development
N Boisgard – 2018 – ec.tuwien.ac.at
… Figure 1.2: The number of publications containing the terms “chatbot”, “chatterbot”, “chat-bot” or “chat … literature analysis: the analysis of the definitions gathered for the term chat-bot has shown … 2.3: A set of additional search terms deduced from the features attributed to chat-bots …

Toward zero-shot entity recognition in task-oriented conversational agents
M Guerini, S Magnolini, V Balaraman… – Proceedings of the 19th …, 2018 – aclweb.org
… recognition. Task-oriented dialogue. E-commerce chat-bots are supposed to carry on a task-oriented dialogue whose goal is helping the user to select products presented in an online shop, and, ultimately, buy them. For the …

SlugBot: Developing a Computational Model and Framework of a Novel Dialogue Genre
KK Bowden, J Wu, W Cui, J Juraska, V Harrison… – dex-microsites-prod.s3.amazonaws …
… Other existing retrieval based chatbots also operate on large existing corpora such as … is a Neural Network model that uses a combination of Recurrent Neural Network and Convolutional … Question Answering: Our question answering mechanism is a three step inspection of the …

Policy learning for task-oriented dialogue systems via reinforcement learning techniques
C Yin – 2018 – minerva-access.unimelb.edu.au
… Aware Chatbot Action Selection via Parameterized Auxiliary Reinforcement Learning. In Proc … Unfortu- nately, most of existing commercial task-oriented dialogue systems still use handcrafted heuristic … [28] make a further progress by utilizing recurrent neural networks (RNNs) …

Review of State-of-the-Art in Deep Learning Artificial Intelligence
VV Shakirov, KP Solovyeva… – Optical Memory and …, 2018 – Springer
… When neural chat bots are trained to produce words with low perplexity in relation to the train- ing corpus of texts, they become capable to write coherent stories, to answer intelligibly, with common sense, to questions, related to the recently loaded information, to reason in a …

Jointly Learning to See, Ask, and GuessWhat
A Venkatesh, R Shekhar, T Baumgärtner… – arXiv preprint arXiv …, 2018 – arxiv.org
… Follow- ing approaches to non-goal-oriented chatbots (Vinyals and Le 2015; Sordoni … forward a hybrid system that learns state representation through a Recurrent Neural Network (RNN) and … is obtained by an LSTM (LSTMe) which processes each new question-answer pair in …

Response Selection and Automatic Message-Response Expansion in Retrieval-Based QA Systems using Semantic Dependency Pair Model
MH Su, CH Wu, KY Huang, WH Lin – ACM Transactions on Asian and …, 2018 – dl.acm.org
… between the post and the response [8]. The retrieval-based chatbots choose a … and it is difficult to develop an extensible open-domain chatbot system … response generation problem by using continuous representations based on the recurrent neural network language model [49 …

The RLLChatbot: a solution to the ConvAI challenge
N Gontier, K Sinha, P Henderson, I Serban… – arXiv preprint arXiv …, 2018 – arxiv.org
… Many chatbots have been proposed for this challenge, overall they all rely on modern deep … This model is exactly what the chatbot needs to ask questions about the news … to the previous generative model, NQG is a sequence-to-sequence recurrent neural network following an …

Intelligence Is Asking The Right Question: A Study On Japanese Question Generation
L Nio, K Murakami – 2018 IEEE Spoken Language Technology …, 2018 – ieeexplore.ieee.org
… Another future possibility is to utilize this gen- eration technique for another task such as a dialogue chatbot system or … [8] D. Tang, N. Duan, T. Qin, and M. Zhou, “Question answering and question … [22] M. Schuster and KK Paliwal, “Bidirectional recurrent neural net- works,” IEEE …

COBOTS-A Cognitive Multi-Bot Conversational Framework for Technical Support
S Subramaniam, P Aggarwal, GB Dasgupta… – Proceedings of the 17th …, 2018 – dl.acm.org
… formerly known as Watson Conversation Service is a cloud-based dialog management service for creating chat bots using IBM … Using Recurrent Neural Networks for Slot Filling in Spoken Language Understanding … A comparison between Alice and Elizabeth chatbot systems …

Interactive Question Answering Using Frame-Based Knowledge Representation
EG Boroujerdi – 2018 – yorkspace.library.yorku.ca
… A Collaborative Adversarial Network (CAN) architecture us- ing a common feature extractor [13] and a Recurrent Neural Network (RNN) with … 2.3 Community Question Answering … Dialogue systems (DS) and chatbots enable computer programs and applications to …

Natural Language Generation with Neural Variational Models
H Bahuleyan – arXiv preprint arXiv:1808.09012, 2018 – arxiv.org
… 10 2.3 Unrolling of a recurrent neural network (RNN) (Britz, 2015 … A few examples of NLP tasks include question answering, sentiment analysis, named- entity recognition and machine translation … Consider a conversational system such as a chatbot, where x is the line input by the …

A Neural Generation-based Conversation Model Using Fine-grained Emotion-guide Attention
Z Zhou, M Lan, Y Wu – 2018 International Joint Conference on …, 2018 – ieeexplore.ieee.org
… is to encode a source sentence (post) as a vector through a recurrent neural network (RNN) and to … information before RNN hidden representation) for answer selection in question answer … matching framework for multi-turn response selection in retrieval-based chatbots,” in ACL …

Analysing Seq-to-seq Models in Goal-oriented Dialogue: Generalising to Disfluencies.
S Bouwmeester – 2018 – esc.fnwi.uva.nl
… of applications, such as technical support services, digital personal assistants, chat bots, and home … networks were shown to outperform both LSTM’s and RNN’s(Recurrent Neural Networks) on question … A marked example of this is Microsoft’s AI-based chat-bot named Tay, who …

The Natural Auditor: How To Tell If Someone Used Your Words To Train Their Model
C Song, V Shmatikov – arXiv preprint arXiv:1811.00513, 2018 – arxiv.org
… Recurrent neural networks (RNNs) are a common neural- network architecture for modeling sequential input data … Dialog generation aims to automatically generate replies in a conversation, which is a common component of chatbots, question-answering services, automated …

Interactive bot to support the use of the UPTEC intranet
JFO e Silva – 2018 – 193.136.37.16
… turn-by-turn basis. Chat-bots can be transactional or conversational [Mct18]. A transactional chat-bot is used to achieve a specific goal, be it schedule a meeting, creating a memo or asking for the weather conditions. One of the …

Interactive bot to support the use of the UPTEC intranet
JF Oliveira – 2018 – repositorio-aberto.up.pt
… turn-by-turn basis. Chat-bots can be transactional or conversational [Mct18]. A transactional chat-bot is used to achieve a specific goal, be it schedule a meeting, creating a memo or asking for the weather conditions. One of the …

Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
I Gurevych, Y Miyao – Proceedings of the 56th Annual Meeting of the …, 2018 – aclweb.org
… The application of Deep Neural Networks to Visual Question Answering has achieved results that would have been thought impossible only a few years ago … Graham Neubig Hai Zhao Question Answering Lluís Màrquez Teruko Mitamura Zornitsa Kozareva Richard Socher …

Natural language generation for commercial applications
A van de Griend, W OOSTERHEERT, T HESKES – 2018 – ru.nl
… that, in terms of NLG, not much has changed since the first chatbot, Eliza (Weizenbaum … that it may be used to improve responses from retrieval-based chatbots (section 8.3 … Both styles use a different neural network architecture: a recurrent neural network (RNN) (section 3.1) and …

Deep Neural Language Generation with Emotional Intelligence and External Feedback
V Srinivasan – 2018 – search.proquest.com
… interpret in more human ways. Ebay has it’s own chatbot that acts as a shopping assistant and helps customers in … customer base in case of customer service chatbots. 1.3 Taxonomy of Models … something they struggle to learn!All recurrent neural networks have the form of a …

Hands-On Natural Language Processing with Python: A practical guide to applying deep learning architectures to your NLP applications
R Arumugam, R Shanmugamani – 2018 – books.google.com
… Chatbot’s are becoming an integrated part of any website; while virtual assistants … Google’s TensorFlow framework, and the building blocks of recurrent neural networks (RNNs), including … Coreference resolution 21 Searching 22 Question answering and chatbots 23 Converting …

Response Generation For An Open-Ended Conversational Agent
N Dziri – 2018 – era.library.ualberta.ca
… 25 2.4.1 Chatbot systems … 34 REI Response Echo Index. 63, 68 RNN Recurrent Neural Network. 17–21, 29, 30, 32, 43 ROUGE Recall-Oriented Understudy For Gisting Evaluation. 35, 36 … For example, chatbots can provide support to the elderly people. In fact, loneli …

Chatbot integration within Sitecore Experience Platform
G Albertengo, R Di Vittorio – webthesis.biblio.polito.it
… In the second one is presented the chatbot’s world: starting from the first bot developed until nowa- days, explaining their history, how they … Going on, the next chapter is about the Microsoft Bot Framework, a powerful tools for creating, developing and managing chatbots, with a ii …

A survey of available corpora for building data-driven dialogue systems: The journal version
IV Serban, R Lowe, P Henderson… – Dialogue & …, 2018 – dad.uni-bielefeld.de
… (1) where at is the dialogue system response action at time t, and ? is the set of parameters that defines f. While the goal of end-to-end … In this case, the dialogue history is projected into a Euclidean space using a recurrent neural network encoding the dialogue word-by-word …

Web forum retrieval and text analytics: A survey
D Hoogeveen, L Wang, T Baldwin… – … and Trends® in …, 2018 – nowpublishers.com
… Page 17. Preprint 1.3. Scope and outline 13 discussion sparking post, or a question-answer post … That includes data from both discussion forums (also called web user forums; see, for instance, [Wang et al., 2013b]) and community question-answering (cQA) archives …

A Bi-Encoder LSTM Model for Learning Unstructured Dialogs
D Shekhar – 2018 – digitalcommons.du.edu
… Introduction 1.1 Overview Recently statistical techniques based on recurrent neural networks (RNN) have achieved … growing rapidly. A Dialog System can communicate with human in text, speech or both and can be classified into – Task-oriented Systems and Chatbot Systems …

Cognitive agents and machine learning by example: Representation with conceptual graphs
A Gkiokas, AI Cristea – Computational Intelligence, 2018 – Wiley Online Library
… as used by SyntaxNet.23 Google relied mostly on part?of?speech (POS) tagging (the Parsey McParseface POS tagger) rather than on semantics, yet achieved the best F 1 scores to date (F 1 results range from 94.44% on news data to 95.40% on question?answer data and to …

Natural Language Processing and Chinese Computing: 7th CCF International Conference, NLPCC 2018, Hohhot, China, August 26–30, 2018, Proceedings
M Zhang, V Ng, D Zhao, S Li, H Zan – 2018 – books.google.com
… in eight key areas, including NLP Fundamentals (Syntax, Semantics, Discourse), NLP Applications, Text Mining, Machine Translation, Machine Learning for NLP, Information Extraction/Knowledge Graph, Conversational Bot/Question Answering/Information Retrieval, and NLP …

A Virtual Conversational Agent for Teens with Autism: Experimental Results and Design Lessons
MR Ali, Z Rasazi, AA Mamun, R Langevin… – arXiv preprint arXiv …, 2018 – arxiv.org
… based conversational agents can be impressive in their question-answering tasks but have been found to provide incomplete and inconsistent responses in areas like mental health, interpersonal violence, and physical health [37]. Woebot [16], a chat-bot recently developed at …

Cognitive architecture of multimodal multidimensional dialogue management
A Malchanau – 2018 – d-nb.info
… Hierarchical recurrent neural networks and memory networks have been proposed to generate open domain dialogues and build … Chat-oriented; Retail ‘chat commerce’ Question-answering skills … transfer tasks and end-to-end approaches handle well chatbot conversations, they …

CSIndexbr: Exploring the Brazilian Scientific Production in Computer Science
MT Valente, K Paixão, D Hafner, D Tran, A Irpan… – arXiv preprint arXiv …, 2018 – arxiv.org
… Geoinformation, 2018. Subjects: Computer Vision and Pattern Recognition (cs.CV). arXiv:1807.09556 [pdf, other] Title: Recurrent Neural Network-based Model Predictive Control for Continuous Pharmaceutical Manufacturing …

Deep Semantic Learning for Conversational Agents
M Morisio, M Mensio – 2018 – webthesis.biblio.polito.it
… introduction to the topic of Conversational Agents (also known as Chatbots or Bots … than Convolutional Networks, the best results are provided by Recurrent Neural Networks that are … The problem where this approach is applied is Question-Answering on synthesized images [45 …

Finding Good Representations of Emotions for Text Classification
JH Park – arXiv preprint arXiv:1808.07235, 2018 – arxiv.org
… human emotions. Popular NLP topics like task-oriented dialogue systems or question … When training NLP models, such as chatbots, things do not always go as intended. Famous incident of Microsoft chatbot Tay, which learned directly from users’ tweets with …

Natural Language Data Management and Interfaces
Y Li, D Rafiei – Synthesis Lectures on Data Management, 2018 – morganclaypool.com
… 2.3. This is followed by the problem of semantic parsing of natural language in Section 2.4. We then discuss the area of question answering in Section 2.6. Finally, we present an overview of dialog systems in Section 2.5. The …

Artificial Intelligence for All: An Abiding Destination
V Pathak, P Tiwari – 2018 – books.google.com
… Benefits of Learning Python Unsupervised Learning Introducing Unsupervised Learning Rule Association Cluster Analysis Anomaly Detection Dimensionality Reduction CHAPTER VI Convolutional and Recurrent Neural Networks Convolutional Neural Networks Recurrent …

Data-Driven Language Understanding for Spoken Dialogue Systems
N Mrkši? – 2018 – repository.cam.ac.uk
… Multi-domain Dialog State Tracking using Recurrent Neural Networks. In Proceedings of ACL 2015 … 1This term is used interchangeably with goal-oriented dialogue systems. 2Alternative chat-bot style systems do not make use of task ontologies or the pipeline model …

Design and development of Rexy: a virtual teaching assistant for on-site and online courses
M PARENTI – 2018 – politesi.polimi.it
… suggests that models based on recurrent neural networks and sequence-to-sequence models … lives. 1.1.2 Application of Current Chatbot Technologies The most popular and pervasive chatbots or digital assistants are the ones em …

Selecting and Generating Computational Meaning Representations for Short Texts
C Finegan-Dollak – 2018 – deepblue.lib.umich.edu
… viii Page 10. LIST OF FIGURES 1.1 An example recurrent neural network for part-of-speech tagging … of as communicating the most important parts of a text’s meaning. Question answering involves identifying a particular relationship between the meanings of two texts: one an …

Dynamic Search Models and Applications
J Luo – 2018 – repository.library.georgetown.edu
Page 1. Dynamic Search Models and Applications A Dissertation submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Computer Science By Jiyun Luo, MS …