Sequence-to-Sequence (seq2seq) & Chatbots

Notes:

Sequence-to-Sequence (seq2seq) is a type of model that is commonly used for tasks such as machine translation, summarization, and text generation. It is particularly well-suited for these tasks because it can handle variable-length input and output sequences.

At a high level, a seq2seq model consists of two main components: an encoder and a decoder. The encoder processes the input sequence and converts it into a fixed-length context vector, which is passed to the decoder. The decoder then uses this context vector to generate the output sequence.

Here’s a more detailed description of how seq2seq works:

The input sequence is passed through the encoder one element at a time. For each element in the input sequence, the encoder processes it and updates its internal state.
When the encoder has processed the entire input sequence, it outputs the final internal state as the context vector. This context vector is a fixed-length representation of the input sequence that captures the key information from it.
The decoder is then fed the context vector and begins generating the output sequence. It does this by predicting the next element in the sequence based on the context vector and the previously generated elements.
The decoder continues generating the output sequence until it has produced the desired number of elements, or until it generates a special end-of-sequence token.

Seq2seq models are often used in chatbots because they can handle the variable-length input and output that is common in conversations. For example, a chatbot using a seq2seq model might take a user’s question as input, process it using the encoder, and then generate a response using the decoder. The model can handle questions of varying length and complexity, and generate appropriate responses.

One important thing to note about seq2seq models is that they are trained using supervised learning, which means they require a large dataset of input-output pairs to learn from. This dataset is used to train the encoder and decoder to map input sequences to the corresponding output sequences.

A neural conversation model consists of a seq2seq model with an encoder and a decoder, as described above. The encoder processes the input (the user’s message) and converts it into a fixed-length context vector, which is passed to the decoder. The decoder then uses this context vector to generate a response to the user’s message.

The training data for a neural conversation model is a collection of input-output pairs, where the input is a message from a user and the output is a response from the chatbot. The seq2seq model is trained on this data to learn how to map user messages to appropriate responses.

One advantage of using a seq2seq model for chatbots is that it can handle variable-length input and output, which is common in conversations. Additionally, the use of a fixed-length context vector allows the model to effectively capture the key information from the input and use it to generate a relevant response.

In 2018, there was an explosion of papers published on seq2seq and chatbots.

Wikipedia:

TensorFlow

References:

How to Make an Amazing Tensorflow Chatbot Easily (2016)

Neural Dialogue System with Emotion Embeddings
R Shantala, G Kyselov… – 2018 IEEE First …, 2018 – ieeexplore.ieee.org
… A. Background: sequence-to-sequence models Seq2Seq models are the most commonly used architecture in … of the dialogue system language model compared to the basic seq2seq approach … of this work are also expanding and include the development of chatbots with an …

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
… The chit-chat model is an attention-based seq2seq model (Sutskever et al., 2014) achieved by a generic deep learning framework … Sequence to sequence learning with neural networks … Docchat: An information retrieval approach for chatbot engines using unstructured documents …

PESUBot: An Empathetic Goal Oriented Chatbot
HV Kumar, J Nagaraj, M Irfan… – … on Advances in …, 2018 – ieeexplore.ieee.org
… Firstly, vanilla sequence-to-sequence neural network is adopted and modification for sentiment level correlation is … The contextual model is architected to be a Seq2Seq Model using RNNs with GRUs as … the HRED model was built for general chatbots while our chatbot was more …

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 sequence-to-sequence (seq2seq) model in deep recurrent neural networks (DRNN) with attention mechanism … Cleverbot Cleverbot is one of the most popular entertainment chatbots that implement … IBM Watson Watson is rule-based AI chatbot developed by IBM’s DeepQA …

Empathetic Dialog Systems
P Fung, D Bertero, P Xu, JH Park, CS Wu… – The International …, 2018 – lrec-conf.org
… Figure 6: Seq2Seq + reinforce … We train a hierarchical sequence-to-sequence model in or- der to maximize a reward score based on funniness and rel- evance, and … Finally, we showed how to train an end-to-end chatbot with reinforce- ment deep learning that learns a sense of …

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
… The variability of Seq2Seq models is different from that of retrieval-based chatbots (Fedorenko et al … entropy maximizing methods, the self-repeating problem (Li et al., 2017) of Seq2Seq models may … Abstractive text summariza- tion using sequence-to-sequence rnns and beyond …

DeepPavlov: Open-Source Library for Dialogue Systems
M Burtsev, A Seliverstov, R Airapetyan… – Proceedings of ACL …, 2018 – aclweb.org
… development of production-ready chatbots and complex … There are currently three Skills implemented in the library, namely, modular and sequence-to-sequence goal-oriented skills as well … It can be a dialogue system that contains a goal-oriented and chatbot skills and chooses …

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
… like to focus on such real life settings encountered by chatbots which involve … pairs over a coherent conversation, we also review some related work on dialog systems … Here again, neural network based (hierarchical) sequence to sequence methods (?; ?; have become the de …

Artificial Intelligence for Conversational Robo-Advisor
MY Day, JT Lin, YC Chen – 2018 IEEE/ACM International …, 2018 – ieeexplore.ieee.org
… field of interaction between humans and machines [1]. The utilization of chatbots can be … data of generative model for training the deep learning sequence to sequence chatbot are obtained … Table 3. Parameters of Seq2Seq model Parameter Value Dropout 0.5 Loss Function …

Response selection with topic clues for retrieval-based chatbots
Y Wu, Z Li, W Wu, M Zhou – Neurocomputing, 2018 – Elsevier
… Generation based chatbots employ statistical machine translation techniques [18] or the sequence to sequence (S2S) framework [4], [5], [19 … 3. Retrieval based chatbots overview. In this section, we introduce a general framework of a retrieval based chatbot, as illustrated in Fig …

Creating an Emotion Responsive Dialogue System
A Vadehra – 2018 – uwspace.uwaterloo.ca
… 10 2.3 A Sequence-to-Sequence (Seq2Seq) Network with and without attention … Customer support or food ordering chatbots are examples of task spe- cific conversational agents where the success of the system is determined by the ability of the agent to … 1.2.1 Dialogue System …

Encoding emotional information for sequence-to-sequence response generation
YH Chan, AKF Lui – … on Artificial Intelligence and Big Data …, 2018 – ieeexplore.ieee.org
… An emotional chatbot is a conversational agent that is conditioned to generate emotional responses … [8] I. Sutskever, O. Vinyals, and QV Le, “Sequence to Sequence Learning with … [20] T. Luong, E. Brevdo and R. Zhao, “Neural Machine Translation (seq2seq) Tutorial,” https …

Towards Automated Customer Support
M Hardalov, I Koychev, P Nakov – International Conference on Artificial …, 2018 – Springer
… 4.3 Sequence-to-Sequence … Twitter using two types of models: (i) retrieval-based (IR with BM25), and (ii) based on generative neural networks (seq2seq with attention … Scholar. 4. Cui, L., Huang, S., Wei, F., Tan, C., Duan, C., Zhou, M.: SuperAgent: a customer service chatbot for e …

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 most … Sequence-to-sequence learning Seq2Seq model is an end-to-end learning with encoder–decoder framework where the … We define our main task as a generative QA via Seq2Seq learning …

Chitty-Chitty-Chat Bot: Deep Learning for Conversational AI.
R Yan – IJCAI, 2018 – ijcai.org
… We may stand at the entrance of future success in more advanced conversational systems (social chatbots and/or virtual assistants) … Alime chat: A sequence to sequence and rerank based chatbot engine … Sequence to sequence learning with neural networks …

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
… detection(OOD) method, and we apply this method to develop a chatbot system for … For training the encoder, we imple- ment a sequence-to-sequence model based on a bi … only in-domain sentences for out-of-domain sentence detection in dialog systems, Pattern Recognition …

Modern Chatbot Systems: A Technical Review
AS Lokman, MA Ameedeen – Proceedings of the Future Technologies …, 2018 – Springer
… Abbreviated from Sequence-to-Sequence phrase, Seq2Seq encoder takes input sentence as a sequence of … While Seq2Seq model is good for sequential data, LSTM application is more … Lokman, AS, Zain, JM: Chatbot enhanced algorithms: a case study on implementation in …

First Insights on a Passive Major Depressive Disorder Prediction System with Incorporated Conversational Chatbot
F Delahunty, ID Wood – ceur-ws.org
… heuristic that guides the development of neural baseline systems for the extractive conversational chatbot task is … the OpenNMT toolkit [18], which is a generic deep learning framework mainly specialising in sequence-to-sequence … (seq2seq) models and covers a variety of tasks …

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
… one size fits all” replies, which is a major drawback of the existing chatbots (Li et … reliable than automatic collected data and thus potentially ben- eficial to chatbot training and … Generating High-quality and Informative Conversa- tion Responses with Sequence-to-sequence Models …

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 … an encoder/decoder architecture with recurrent neural networks known as sequence to sequence (Seq2Seq) [12] used … The chatbot uses the Artificial Intelligence Markup Language (AIML) for the definition of …

DeepPavlov: An Open Source Library for Conversational AI
M Burtsev, A Seliverstov, R Airapetyan, M Arkhipov… – 2018 – openreview.net
… Sequence-to-sequence chit-chat, question answering or task-oriented skills can be assembled from components provided in the library … development of production-ready chatbots and complex conversational systems; • NLP and dialog systems research …

Deep Learning in Spoken and Text-Based Dialog Systems
A Celikyilmaz, L Deng, D Hakkani-Tür – Deep Learning in Natural …, 2018 – Springer
… This model is called sequence-to-sequence or seq2seq … 2011), retrieval-based response selection (Banchs and Li 2012), and sequence-to-sequence models with different structures … a conversational dialog engine for creating chat bots, chatbot-rnn, 20 a toy chatbot powered by …

First insights on a passive major depressive disorder prediction system with incorporated conversational chatbot
M Arcan, ID Wood, F Delahunty – Irish Conference on Artificial …, 2018 – library.nuigalway.ie
… of neural baseline systems for the extractive conversational chatbot task is … work has already explored the possibility of building conversational chatbots that emulate … is a generic deep learning framework mainly specialising in sequence-to-sequence (seq2seq) models covering …

A Next Generation Chatbot-Framework for the Public Administration
A Lommatzsch – International Conference on Innovations for Community …, 2018 – Springer
… We have found that currently foreigners seem to prefer this channel; but with the increased popularity of chatbots and the high … F.-L., Wang, S., Gao, X., Chen, Y., Zhao, W., Chen, H., Huang, J., Chu, W.: AliMe chat: a sequence to sequence and rerank based chatbot engine …

A Next Generation Chatbot-Framework for the Public Administration–DRAFT–
A Lommatzsch – dai-labor.de
… We have found that currently foreigners seem to prefer this channel; but with the increased popularity of chatbots and the high quality of the provided answers, the developed digital assistant will attract … Alime chat: A sequence to sequence and rerank based chatbot engine …

Impact of Auxiliary Loss Functions on Dialogue Generation Using Mutual Information
JS Clair, T Conley, J Kalita – cs.uccs.edu
… a chatbot, that learns from a corpus of conversations, using a basic sequence to sequence (Seq2Seq) model with … be to add a reinforcement learning and attention model to the Seq2Seq model from … paper, but it would shed more light on the competency of chatbot dialogue and …

Why are Sequence-to-Sequence Models So Dull?
S Jiang, M de Rijke – EMNLP 2018, 2018 – aclweb.org
… The variability of Seq2Seq models is different from that of retrieval-based chatbots (Fedorenko et al., 2017): in this study, we focus on the lack of variability of … (Li et al., 2017) of Seq2Seq models may … Abstractive text summariza- tion using sequence-to-sequence rnns and beyond …

The Design and Implementation of XiaoIce, an Empathetic Social Chatbot
L Zhou, J Gao, D Li, HY Shum – arXiv preprint arXiv:1812.08989, 2018 – arxiv.org
… The design objective of the chatbot is to find optimal policies and skills to … in the conversational AI community [4]. Its role in developing social chatbots is becoming … The neural response generator in XiaoIce follows the sequence-to-sequence (seq2seq) framework [14, 15] used …

End-to-End Task-Oriented Dialogue System with Distantly Supervised Knowledge Base Retriever
L Qin, Y Liu, W Che, H Wen, T Liu – Chinese Computational Linguistics …, 2018 – Springer
… has also been used in past literature for evaluating dialogue systems both of the chatbot and task … In the future, we would like to jointly model the KB retriever and seq2seq framework in an end … Sutskever, I., Vinyals, O., Le, QV: Sequence to sequence learning with neural networks …

Role play-based question-answering by real users for building chatbots with consistent personalities
R Higashinaka, M Mizukami, H Kawabata… – Proceedings of the 19th …, 2018 – aclweb.org
… char- acters, the next step is to determine if the col- lected pairs can be useful for creating chatbots that exhibit … as a query and the answer part of the most relevant pair is returned as a chatbot’s response … Translation model We trained a translation model by using a seq2seq model …

The First Conversational Intelligence Challenge
M Burtsev, V Logacheva, V Malykh, IV Serban… – The NIPS’17 …, 2018 – Springer
… For the seq2seq models, the team trained diverse neural networks with different data and network … The system uses four strategies to enhance existing sequence-to-sequence conversational models: 1 … defined by user success in the task, in a way that a good chatbot should be …

Dialog-to-Action: Conversational Question Answering Over a Large-Scale Knowledge Base
D Guo, D Tang, N Duan, M Zhou, J Yin – Advances in Neural …, 2018 – papers.nips.cc
… experiments on a large-scale dataset [17] for conversation question answering, which consists of … HRED+KVmem [17] is a sequence- to-sequence learning method, which uses a hierarchical … which explicitly manipulates the actions/functions and lets the Seq2Seq model learn …

A knowledge-grounded neural conversation model
M Ghazvininejad, C Brockett, MW Chang… – Thirty-Second AAAI …, 2018 – aaai.org
… 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 … The dialog encoder and response decoder form together a sequence-to-sequence (SEQ2SEQ model (Hochreiter …

Conversational Modelling for ChatBots: Current Approaches and Future Directions
M McTear – 2018 – spokenlanguagetechnology.com
… Sequence to sequence mapping between an input and a response means that the system does not … the same skill (eg weath- er forecast) may behave differently on the same chatbot … quential mechanics of conversation is essential to make interactions with chatbots intuitive and …

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
… Conversation Machine (PCCM) and compared it with several baselines: Seq2Seq: a general sequence to sequence model [Sutskever … Nevertheless, our model is still much better than the Seq2Seq model … Our work moves toward endowing a chatbot with control- lable personality …

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
… Ever since the success of Seq2Seq model in creating state of the art generative neural conversational agent … [10] J. Wakefield, “Microsoft chatbot is taught to … www.bbc.com/news/ technology-35890188 [11] I. Sutskever, O. Vinyals, and QV Le, “Sequence to sequence learning …

Information Discovery in E-commerce: Half-day SIGIR 2018 Tutorial
Z Ren, X He, D Yin, M de Rijke – … ACM SIGIR Conference on Research & …, 2018 – dl.acm.org
… We conclude with a summary of existing chatbots models applied in e-commerce, and outlook to future directions of e-commerce chatbots … Alime chat: A sequence to sequence and rerank based chatbot engine. In ACL, pages 498–503. ACL, 2017 …

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
… used 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 … of this category in recent years is the sequence to sequence (Seq2Seq) based model …

Knowledge-aware Multimodal Dialogue Systems
L Liao, Y Ma, X He, R Hong, T Chua – 2018 ACM Multimedia Conference …, 2018 – dl.acm.org
… in optimizing for better rewards with the power of hierarchical seq2seq models in … Also known as chat bots, the non-task- oriented systems converse with human typically on … model sequential utterances, we resort to the powerful hierarchi- cal sequence-to-sequence models such …

Tailored Sequence to Sequence Models to Different Conversation Scenarios
H Zhang, Y Lan, J Guo, J Xu, X Cheng – … of the 56th Annual Meeting of …, 2018 – aclweb.org
… While for the diverse- requirement scenario, the experiments on the pub- lic Chinese Weibo dataset (social chatbot) show that optimizing CVaR … 4 Tailored Sequence to Sequence Models … ?r(?) = inf{? ? R|P(?log P(Y |X) ? ?) ? r}, and ? are parameters of the Seq2Seq model …

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 from … and i<Td) which ensures the encoding and decoding processes in sequence-to-sequence models … using a reinforcement learning policy to determine when to let the chatbot to give …

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 … also achieved quite advanced performance, indicating the effectiveness of the seq2seq model on … Lu, Z., Li, H., Li, VOK: Incorporating copying mechanism in sequence-to-sequence learning …

Improving Dialog Systems Using Knowledge Graph Embeddings
B Carignan – 2018 – curve.carleton.ca
… The four embedding sets were trained with a sequence-to-sequence model from OpenNMT on a dialog dataset from OpenSubtitles … 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 …

Improving Computer Generated Dialog with Auxiliary Loss Functions and Custom Evaluation Metrics
T Conley, JS Clair, J Kalita – cs.uccs.edu
… the output of the SMT-based response generation system with a Seq2Seq model that … With better evalua- tion models, a neural-network-based chatbot may be enhanced to … Generating high- quality and informative conversation responses with sequence-to-sequence models …

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
… to im- prove matching models for multi-turn response selection in retrieval-based chatbots … Existing methods on building a chatbot are ei- ther generation-based or retrieval-based. Regard- ing to the former, on top of the basic sequence- to-sequence with attention architecture …

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 …
… On the other side of the scalability spectrum, sequence- to-sequence models based on recurrent neural networks have proven to be … Most chatbots of today are trained using example interactions … to the roles the both users had, while in our case, the role of the chatbot is different …

A Virtual Chatbot for ITSM Application
S Raut – Asian Journal For Convergence In Technology …, 2018 – asianssr.org
… Chatbots that are developed using deep learning, mostly use a certain variant of sequence to sequence (Seq2Seq) model … Thus, Seq2Seq model generates the potential response respective to the given input … [6] Bayu Setiaji, Ferry Wahyu Wibowo, “Chatbot Using A Knowledge …

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 … Existing studies show that addressing affect and emo- tion in dialogue systems or conversational … is based on the encoder-decoder framework of the general sequence-to-sequence (seq2seq for short …

Context-Aware Dialog Re-Ranking for Task-Oriented Dialog Systems
J Ohmura, M Eskenazi – 2018 IEEE Spoken Language …, 2018 – ieeexplore.ieee.org
… and Michiel Bacchiani, “State-of-the-art speech recognition with sequence-to-sequence models,” 2018 … taking a spoken dialog system to the real world,” in Ninth Euro- pean … and Hui Xue, “Ranking responses oriented to conversational relevance in chat- bots,” in Proceedings of …

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
… Lei Cui, Shaohan Huang, Furu Wei, Chuanqi Tan, Chaoqun Duan, and Ming Zhou. 2017. Superagent: A customer service chatbot for e-commerce websites … 2018. Seq2seq dependency parsing … 2017. Alime chat: A sequence to sequence and rerank based chatbot engine …

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
… The Facebook news seq2seq skill, in addition to the dialog context, may use a paragraph as an input … AIML-based Alice chit-chat skill covers a wide range of topics, while sequence-to- sequence based chit-chat skill trained on Open … 2017. A deep reinforcement learning chatbot …

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 … These neural systems mainly used sequence to sequence model as a backbone, to perform … In this work, we develop a human-robot dialogue system extended from our previous framework of …

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
… based method, the Okapi BM25 model, and the deep learning-based sequence-to-sequence with attention … alignment between the post and the response [8]. The retrieval-based chatbots choose a … methods, and it is difficult to develop an extensible open-domain chatbot system …

Improving goal-oriented visual dialog agents via advanced recurrent nets with tempered policy gradient
R Zhao, V Tresp – arXiv preprint arXiv:1807.00737, 2018 – arxiv.org
… Das et al. [2017] formulated a visual dialogue system which is about two chatbots asking and answering questions to identify a specific image … We utilize a Seq2Seq model [Sutskever et al., 2014] to process the image along with the historical dialogues for question generation …

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
… social chatbots as real world applications. Since the Web 2.0 is so popular nowadays, people are getting used to having conversations—in public—on a variety of websites, such as forums, social medium (eg, Facebook, Twitter) and com- munity question-answering platforms …

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
… Conversational Query Understanding Using Sequence to Sequence Modeling … 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 or …

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
… such as e-commerce, technical support services, entertaining chatbots, information retrieval … translation, most non-goal-oriented conversational systems employ sequence-to-sequence (seq2seq) framework to … individual embeddings and incorporating it into the seq2seq model …

Topic-Based Question Generation
W Hu, B Liu, R Yan, D Zhao, J Ma – 2018 – openreview.net
… Since questioning is an important communication skill, question generation plays an important role in both general-purpose chatbot systems and goal-oriented dialogue systems … Sequence to sequence learning with neural networks. In NIPS, 2014 …

From Eliza to XiaoIce: challenges and opportunities with social chatbots
HY Shum, X He, D Li – Frontiers of Information Technology & Electronic …, 2018 – Springer
… For this purpose, social chatbots must develop a set of skills to accommodate users’ requests … accomplishing tasks and ending the conversation as quickly as possible), social chat- bots take time … XiaoIce has been the most widely deployed so- cial chatbot since it was released …

Aspect-based question generation
W Hu, B Liu, J Ma, D Zhao, R Yan – 2018 – openreview.net
… is an important communication skill, question generation plays an important role in both general-purpose chatbot systems and goal-oriented dialogue systems … Multi-source sequence-to-sequence learning aims to integrate information from multiple sources to boost …

Concorde: Morphological Agreement in Conversational Models
D Polykovskiy, D Soloviev… – Asian Conference on …, 2018 – proceedings.mlr.press
… appear in a wide range of applications, from simple rule-based chatbots to complex … Dialogue and conversation modeling are a characteristic example of sequence-to-sequence problems: given a … model introduced by Vinyals and Le (2015) uses the seq2seq framework from …

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 … Firstly, by extending existing models with Seq2Seq and Memory Networks we could improve …

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
… (2013)). Recently, researchers have paid increasing attention to open-domain, chatbot-style human-computer conversations such as XiaoIce1 and Duer2 due to their important commercial values … We then propose a “multi sequence to sequence” (multi-seq2seq) model to …

MOOC-O-Bot: Using Cognitive Technologies to Extend Knowledge Support in MOOCs
A Mitral, L Vigentini, M Djatmiko… – … , and Learning for …, 2018 – ieeexplore.ieee.org
… These systems are generally modeled using sequence to sequence (seq2seq) models [27] and are based on … [28] proposes an ensemble of retrieval based and generative chatbot systems by … retrieval based and generation based model and uses an attentive Seq2Seq re-ranD …

Proceedings of the 22nd Conference on Computational Natural Language Learning
A Korhonen, I Titov – Proceedings of the 22nd Conference on …, 2018 – aclweb.org
… Pervasive Attention: 2D Convolutional Neural Networks for Sequence-to-Sequence Prediction Maha Elbayad … Churn Intent Detection in Multilingual Chatbot Conversations and Social Media … Improving Response Selection in Multi-Turn Dialogue Systems by Incorporating Domain …

A Trustworthy, Responsible and Interpretable System to Handle Chit-Chat in Conversational Bots
P Agrawal, A Suri, T Menon – arXiv preprint arXiv:1811.07600, 2018 – arxiv.org
… It also provides a signal to seq2seq auto- generation model to ensure that it only generates a … I Am A Sequence- to-Sequence Model trained on Reddit Data, Ask Me Any- thing … Docchat: An information re- trieval approach for chatbot engines using unstructured doc- uments …

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
… 38, 41] have been proved to be capable in multiple dialogue system applications with … 6]: (1) Meaningless responses: Given a wide range of contexts, dialogue systems trained via … Sequence to Sequence Models Our work is based on sequence-to-sequence (SEQ2SEQ) models …

Transfer Learning for Context-Aware Question Matching in Information-seeking Conversations in E-commerce
M Qiu, L Yang, F Ji, W Zhao, W Zhou, J Huang… – arXiv preprint arXiv …, 2018 – arxiv.org
… To form our data set, we concatenated utterances within three turns 7 to form a query, and used the chat- bot system to call back top 15 most similar candi- date questions as candidate “responses” … Alime chat: A sequence to sequence and rerank based chatbot engine …

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
… Figure 1: The proposed Mem2Seq architecture for task-oriented dialog systems … tems (Eric and Manning, 2017; Zhao et al., 2017) and chat-bots (Ritter et … We also implemented the following baseline models: standard sequence-to-sequence (Seq2Seq) models with and without …

Improving Neural Question Generation using Answer Separation
Y Kim, H Lee, J Shin, K Jung – arXiv preprint arXiv:1809.02393, 2018 – arxiv.org
… 2018) or to engage chatbots to start and continue a conversation (Mostafazadeh et al … When we evaluate our answer-separated seq2seq on the SQuAD dataset (Rajpurkar et al … learning with regard to the question generation problem using a sequence-to-sequence model with …

Epilogue: Frontiers of NLP in the Deep Learning Era
L Deng, Y Liu – Deep Learning in Natural Language Processing, 2018 – Springer
… application, document or text summarization, is also well suited to sequence-to-sequence learning and … been applied to all three types of dialogue systems or chatbots (intelligent assistants … progress in this research frontier is gaining greater urgency as chatbot conversations are …

Towards Understanding User Requests in AI Bots
OT Tran, TC Luong – Pacific Rim International Conference on Artificial …, 2018 – Springer
… and improved techniques, chatbots still face various challenges in understanding user requests, processing them to generate an appropriate response, especially in ordering chatbots … Qiu, M., et al.: AliMe chat: a sequence to sequence and rerank based chatbot engine …

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
… 1998), sequence-to-sequence models (eg Vinyals and Le 2015), reinforcement learning (eg Williams and Young 2007 … cally, in his pioneering work, Weizenbaum (1966) developed the Eliza chatbot agent for … Evaluating quality of chatbots and intelligent con- versational agents …

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 … The system combines machine learning (Sequence-to-Sequence Models) and rule based modules (Structured Topic … 10] Williams, JD; Young, S. Partially observable Markov decision processes for spoken dialog systems …

Personalizing Dialogue Agents: I have a dog, do you have pets too?
S Zhang, E Dinan, J Urbanek, A Szlam, D Kiela… – arXiv preprint arXiv …, 2018 – arxiv.org
… Our goal is to enable interesting direc- tions for future research, where chatbots can for instance … Generative PersonaChat Models Seq2Seq None 3.17(1.10) 3.18(1.41) 2.98(1.45) 0.51(0.50) … We also evaluate the scores of hu- man performance by replacing the chatbot with a …

A Goal-oriented Neural Conversation Model by Self-Play
W Wei, QV Le, AM Dai, LJ Li – 2018 – openreview.net
… information, we can potentially leverage knowledge of a much larger scale and train a much more reliable chatbot … The introduction of the seq2seq modeling (Sutskever et al., 2014) and the idea of transforming the dialogue … Sequence to sequence learning with neural networks …

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 …

Conversation Modeling with Neural Network
JY Patil, GP Potdar – Asian Journal of Research in Computer …, 2018 – journalajrcos.com
… seq-2-seq[1] illustrated ability of multi-layer LSTM RNN to achieve good performance on Machine Translation tasks … [h] Fig. 2. Example, use of seq2seq framework for conversational modeling … Sequence- to-Sequence Learning as Beam- Search Optimization …

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 DocChat … Vinyals and Le [18] propose a sequence-to-sequence framework where in the post utterance is … However, Seq2Seq based methods tend to generate safe but trivial responses, such as …

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 … Among all types of deep learning architectures, a sequence-to-sequence learning approach (seq2seq … by inserting a social language understanding layer into a typical seq2seq model …

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
… its applications involve a vast amount of domains such as a chatbot component in … And we are among the first to employ a deep sequence-to-sequence learning approach to … NLP tasks, ranging from sequence gen- eration for Twitter2 to chat-oriented dialog system response [38 …

Building Advanced Dialogue Managers for Goal-Oriented Dialogue Systems
V Ilievski – arXiv preprint arXiv:1806.00780, 2018 – arxiv.org
… These models are trained in a sequence-to-sequence fashion [Sutskever et al., 2014], that encode the user … On the other hand, we can model the GO Chatbots as a Partially Observable Markov … Figure 3.2 shows a typical composition of an RL-based GO Chatbot, as well as the …

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
… The purpose of our emotional tags is to make the chatbot understood the emotion of the input … Generative Model of our GAN network is a simple Seq2Seq model to generate the response of the post … [3] I. Sutskever, O. Vinyals, and QV Le, “Sequence to sequence learning with …

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 … As a starting point for our approach we leverage the Seq2Seq model (Sutskever et al., 2014; Bah- danau et al., 2014 …

A Knowledge-Grounded Multimodal Search-Based Conversational Agent
S Agarwal, O Dusek, I Konstas, V Rieser – arXiv preprint arXiv:1810.11954, 2018 – arxiv.org
… ubiquitous, with variants ranging from open-domain conversa- tional chit-chat bots (Ram et … Our model belongs to the encoder-decoder paradigm where sequence-to-sequence models (Cho … on the task of textual response generation in multimodal task-oriented dialogue system …

Policy learning for task-oriented dialogue systems via reinforcement learning techniques
C Yin – 2018 – minerva-access.unimelb.edu.au
… Context-Uncertainty- Aware Chatbot Action Selection via Parameterized Auxiliary Reinforcement Learning. In Proc … However, DST plays a determinant role in task-oriented dialogue systems. If dialogue states are not accurately tracked, the dialogue system …

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
… Anaphora and Co-reference Resolution: Multi-turn dialog systems need to resolve ambiguous phrases … to text, is the first component a user interacts with in a spoken dialog system … Evaluation of dialogue systems is a challenging research problem, which despite being studied …

Computational Analysis and Understanding of Natural Languages: Principles, Methods and Applications
CR Rao, VN Gudivada – 2018 – books.google.com
… Short-Term Memory and Gated RNNs 6.6 Encoder-Decoder Sequence-to-Sequence Architectures 6.7 … textual entail- ment, language generation, semantic analysis, grammar correction, question- answering systems, spoken dialog systems, chatbots, passage retrieval …

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
… 7]. [17] train a neural multiclass classifier and linear regressor from human-chatbot dialogues from … two dialogue turns, r1 and r3 refer to likelihood outputs of Seq2Seq models, r2 … Future conversational agents or chatbots can optimise their behaviour using neural regressors with …

Exemplar encoder-decoder for neural conversation generation
G Pandey, D Contractor, V Kumar, S Joshi – Proceedings of the 56th …, 2018 – aclweb.org
… With the availability of large datasets and the recent progress made by neural meth- ods, variants of sequence to sequence learning (seq2seq) (Sutskever et al., 2014) architectures have been successfully applied for building con- versational systems (Serban et al., 2016, 2017b …

Response Generation For An Open-Ended Conversational Agent
N Dziri – 2018 – era.library.ualberta.ca
… 25 2.4.1 Chatbot systems … ground for the data-driven approaches to become feasible and quite promis- ing. In particular, Sequence-to-Sequence (Seq2Seq) neural networks model [105] has witnessed substantial breakthroughs in enhancing the performance …

Natural answer generation with heterogeneous memory
Y Fu, Y Feng – Proceedings of the 2018 Conference of the North …, 2018 – aclweb.org
… In real-world applications like chatbots or personal assistants, users may want to … Natural Answer Generation with Sequence to Sequence Learning: Sequence to sequence mod- els (with attention … Memory is an effective way to equip seq2seq systems with external information …

Building a conversational agent overnight with dialogue self-play
P Shah, D Hakkani-Tür, G Tür, A Rastogi… – arXiv preprint arXiv …, 2018 – arxiv.org
… with just a task schema and an API client from the dialogue system developer, but it … is important when developing datasets to benchmark research aimed towards building human-level dialogue systems. However, we ar- gue that for consumer-facing chatbots, the primary aim is …

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 learning and … column) between a human user (Ui:) and the chatbot (Bi … The second generative sequence-to-sequence model used in the ensemble is the Neural Question Generator (NQG …

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
… $15.00 https://doi.org/10.1145/3209978.3210011 used1. These systems, with either text-based or voice-based con- versational interfaces, are capable of voice interaction, information search, question answering and voice control of smart devices …

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 CHI 2018, April 21–26, 2018, Montréal, QC, Canada … The previous conversation may consist of one or multiple turns. A widely adopted DNN approach to this problem is a sequence-to-sequence architecture (seq2seq) [41, 46, 25] …

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 … which we will use as our final response. One of the most important characteristics of sequence to sequence models is the … Seq2Seq models provide this flexibility …

Analysis, discovery and exploitation of open data for the creation of question-answering systems
G Molina Gallego – 2018 – rua.ua.es
… Chatbots have been become more popular from the last two decades, actually, many companies use these … To better understand the potential of a chatbot, it is needed to know their origin. A Dialogue System or CA aims to create comprehensive systems that can hold a real …

Conversational ai: The science behind the alexa prize
A Ram, R Prasad, C Khatri, A Venkatesh… – arXiv preprint arXiv …, 2018 – arxiv.org
… the goal of a user for a given utterance, and the dialog system needs to … There are four main types of approaches for response generation in dialog systems: template/rule … shown promising results in the past couple of years, usually with sequence-to-sequence approaches with …

Natural Language Generation with Neural Variational Models
H Bahuleyan – arXiv preprint arXiv:1808.09012, 2018 – arxiv.org
… 2.2.9 Dialog Systems … In this research, we work with deep neural network models known as sequence-to-sequence (Seq2Seq) models that take a sentence as input and … Consider a conversational system such as a chatbot, where x is the line input by the user and y is the line …

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 … Our base dialogue model is a simple sequence-to- sequence (Seq2seq) model that consists of a two- layer bi … fusion ratio ? is a hyperparameter that in- dicates how much Seq2seq output will …

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 learning … sequence and ct is final context vector at time t. Sequence-to-sequence learning with … applied to many tasks, such as machine translation [7], question answering [15] and …

Natural language generation for commercial applications
A van de Griend, W OOSTERHEERT, T HESKES – 2018 – ru.nl
… 4.1 The sequence-to-sequence model … of these applications (see section 6.1) and realised that, in terms of NLG, not much has changed since the first chatbot, Eliza (Weizenbaum … (2017) such that it may be used to improve responses from retrieval-based chatbots (section 8.3) …

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-oriented multi-turn di- alogue … 2016) found that the automated dialogue systems built using … This confirms that further investigation in re- trieval dialogue system using this dataset is worth- while and motivates us …

Ask No More: Deciding when to guess in referential visual dialogue
R Shekhar, T Baumgartner, A Venkatesh… – arXiv preprint arXiv …, 2018 – arxiv.org
… Vision community exploits encoder- decoder architectures (Sutskever et al., 2014) — which have shown some promise for modelling chatbot- style dialogue … 2014. Sequence to sequence learning with neural networks … POMDP-based statistical spoken dialog systems: A review …

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
… In recent years, the generation-based systems usu- ally adopt the sequence to sequence (Seq2seq) [3] models … diverse, content correlation and human-like emotion than the baseline Seq2seq model … framework for multi-turn response selection in retrieval-based chatbots,” in ACL …

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 et al … 2016b) that model dialogue as a sequence- to-sequence problem (Sutskever, Vinyals, and Le 2014 … an RNN that is jointly optimised with other components of the dialogue system — in our …

A Bi-Encoder LSTM Model for Learning Unstructured Dialogs
D Shekhar – 2018 – digitalcommons.du.edu
… 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. Task-oriented systems are designed for a particular task and set up to have short con …

Alexa Prize-State of the Art in Conversational AI
C Khatri, A Venkatesh, B Hedayatnia, A Ram… – AI …, 2018 – search.proquest.com
… the goal of a user for a given utter- ance, and the dialogue system needs to … WiseMacaw developed a two-layer chatbot framework that can be described as a metagame … have shown promising results in the past couple of years, usually with sequence-to-sequence approach- es …

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
… study the problem of providing recommended responses to customer service agents in live-chat dialogue systems … Most existing methods of ‘smart replies’ seek to build a sequence-to-sequence or a … task as a candidate ranking problem; 2) different from the chatbot system (Cui …

Gunrock: Building A Human-Like Social Bot By Leveraging Large Scale Real User Data
CY Chen, D Yu, W Wen, YM Yang, J Zhang, M Zhou… – m.media-amazon.com
… In comparison, social chatbots require in-depth communication skills with emotional support[21] … We trained a sequence to sequence model [24] using the Cornell Movie-Quotes Corpus [5 … Backstory and EVI to answer general facts and background questions about our chatbot …

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 … End-to-end optimization of goal-driven and visually grounded dialogue systems … Sequence to sequence learning with neural networks …

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 bots … question is an interesting and challenging task for a smart robot (like chat-bot) … Our method is based on a sequence to sequence network (Sutskever et al., 2014; Vinyals et …

Data Augmentation for Neural Online Chat Response Selection
W Du, AW Black – arXiv preprint arXiv:1809.00428, 2018 – arxiv.org
… (Yang et al., 2017) adopt a seq2seq model (Sutskever … 2014. Sequence to sequence learning with neural net- works … 2016. Sequential matching network: A new architecture for multi-turn response selec- tion in retrieval-based chatbots. arXiv preprint arXiv:1612.01627. Page 7 …

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 … During execution the chatbot tries to fill the slots for a target intent by … work has attempted to use neural networks to solve general sequence- to-sequence learning problems in …

Tartan: A retrieval-based socialbot powered by a dynamic finite-state machine architecture
G Larionov, Z Kaden, HV Dureddy… – arXiv preprint arXiv …, 2018 – arxiv.org
… in human-machine spoken language interaction had focused on goal-oriented dialog systems … of a more traditional sequence-to-sequence model was that most sequence-to- sequence models fail to … We picked question-answer pairs as the training dataset and used StarSpace …

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 Alexa … assess how the results for those tasks transfer to the actual dialogue system, we have … Wikiquotes, using a method similar to the one used in Poetwannabe chatbot [Chorowski et al …

A Face-to-Face Neural Conversation Model
H Chu, D Li, S Fidler – … of the IEEE Conference on Computer …, 2018 – openaccess.thecvf.com
… Dialogue systems have been explored since the 60′, with systems like ELIZA [34] and PARRY [5] already capable of engaging in relatively … Text(Face): The classic Seq2Seq [12, 27] method that uses single modality only (either text or face) and only the … The NeuralHank Chatbot …

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 …
… Discourse coherence in SCRIPT MODEL dialogue systems is created by the user interaction designer … Other existing retrieval based chatbots also operate on large existing corpora such as … Negative users, we would suggests some resources in our dialogue system to improve …

Deep Context Resolution
J Chen – 2018 – uwspace.uwaterloo.ca
… To go beyond one-round con- versation, a chatbot must resolve contextual information … It has been applied on Image Generation [18], Sequence-to-Sequence learning [43] and … Language data from Community Question Answering (cQA) websites fit our purpose perfectly since 1 …

Chatbot integration within Sitecore Experience Platform
G Albertengo, R Di Vittorio – webthesis.biblio.polito.it
… 30 2.2 How the chatbots work … 27 2.1 DeepQA Architecture . . . . . 34 2.2 Finite-state machine for task-oriented chatbot . . . . . 39 2.3 Sequence to sequence model for answers generation . . 54 3.1 Azure Bot Service …

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
… The personalized chat is a newly emerging demand in the research on chatbot; thus, little … For the scenario of chatbots, the situation is even worse: it is impossible to … that the sentence embedding component is fixed and pre-trained by sequence-to-sequence language model …

Selecting and Generating Computational Meaning Representations for Short Texts
C Finegan-Dollak – 2018 – deepblue.lib.umich.edu
… 136 6.2 Accuracy of seq2seq models with and without schema attention and copying. . 138 … 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 …

Community Regularization of Visually-Grounded Dialog
A Agarwal, S Gurumurthy, V Sharma, M Lewis… – arXiv preprint arXiv …, 2018 – ri.cmu.edu
… 24]), image segmentation ([18], [9], [22]), dialog ([28], [31], [5]), question answering ([38], [27 … different domains, it now seems plausible to build more advanced dialog systems capable of … prob- lem using supervised learning where, conditioned on the question – answer pair dialog …

The bottleneck simulator: a model-based deep reinforcement learning approach
IV Serban, C Sankar, M Pieper, J Pineau… – arXiv preprint arXiv …, 2018 – arxiv.org
Page 1. arXiv:1807.04723v1 [cs.LG] 12 Jul 2018 The Bottleneck Simulator: A Model-based Deep Reinforcement Learning Approach Iulian Vlad Serban1 Chinnadhurai Sankar 1 Michael Pieper1 Joelle Pineau2 Yoshua Bengio 1 Abstract …

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. An AI sys … question-answer pairs. While this largely resembles the visual question answering task, a variety of different approaches have been proposed recently. 6 Page 13 …

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 …

Analysing Seq-to-seq Models in Goal-oriented Dialogue: Generalising to Disfluencies.
S Bouwmeester – 2018 – esc.fnwi.uva.nl
… range of applications, such as technical support services, digital personal assistants, chat bots, and home … A normal sequence to sequence model with attention can be seen as a memory network with … A marked example of this is Microsoft’s AI-based chat-bot named Tay, who …

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 …

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 … 21 Coreference resolution 21 Searching 22 Question answering and chatbots 23 Converting … using GRU Data preparation Encoder network Decoder network Sequence to sequence Building the …

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 …

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 … Sequence to sequence learning with neural networks … POMDP-based statistical spoken dialog systems: A review …

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
… of chatbots, question-answering services, automated customer service, and other dialog-based systems. The input is a sentence in a conversation, the output is the next sentence in the same conversation. Dialog-generation models typically employ a sequence-to-sequence …

Emory IrisBot: An Open-Domain Conversational Bot for Personalized Information Access
A Ahmadvand, IJ Choi, H Sahijwani, J Schmidt, M Sun… – m.media-amazon.com
… employs proactive recommendation strategies, simulated emotion, and incorporate sequence to sequence learning to … all Chat component, which was based on the ALICE chatbot, implemented in … Candidate question-answer pairs are retrieved and re-ranked by the combination …

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 …

Adversarial Over-Sensitivity and Over-Stability Strategies for Dialogue Models
T Niu, M Bansal – arXiv preprint arXiv:1809.02079, 2018 – arxiv.org
… It consists of a dynamic knowledge graph, a graph embedding over the entity nodes, and a Seq2seq-based utterance generator … 1902; Marqués-Aguado, 2014); it is also frequently seen in blog posts.3 Thus, being robust to swapping adjacent words is useful for chatbots that take …

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 …

Task-Oriented Dialog Agents Using Memory-Networks and Ensemble Learning
RFG Meléndez – 2018 – pdfs.semanticscholar.org
… et al., 2015) make no further classification than task-oriented or non task-oriented chatbots, Jurafsky and Martin (2018) use the term chatbot exclusively for … This is a scenario commonly tackled with sequence-to-sequence models that build an entire phrase word by word …

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 …
… This paper presents the second version of the dialogue system named Alquist com- peting in Amazon … It is used to test end-to-end dialog systems in a way that favors reproducibility … is the only task which contains records of real-world conversations between humans and chatbot …

Amanuensis: The Programmer’s Apprentice
T Dean, M Chiang, M Gomez, N Gruver, Y Hindy… – arXiv preprint arXiv …, 2018 – arxiv.org
… Figure 2: The sequence-to-sequence encoder-decoder attentional model shown here uses a specialized mem- ory called a pointer network to construct a short summary of a source document by flexibly combining phrases from the source document with words from its existing …

BYU-EVE: Mixed Initiative Dialog via Structured Knowledge Graph Traversal and Conversational Scaffolding
N Fulda, T Etchart, W Myers, D Ricks, Z Brown… – dex-microsites-prod.s3.amazonaws …
… This type of interaction encompasses, but also goes beyond, database queries and question/answer models … Ultimately, a dialog system exists to generate text … Page 10. sentences. (In theory, this task could also be attempted using a sequence-to-sequence model …

Deep Semantic Learning for Conversational Agents
M Morisio, M Mensio – 2018 – webthesis.biblio.polito.it
… 37 2.2.6 Sequence-to-sequence models for slot tagging … has been conducted, an introduction to the topic of Conversational Agents (also known as Chatbots or Bots … The problem where this approach is applied is Question-Answering on synthesized images [45], and the focus is …

Cognitive architecture of multimodal multidimensional dialogue management
A Malchanau – 2018 – d-nb.info
… designs based on a reactive interlocutor paradigm and calls for dialog systems that can be … template-based Table 1.1: State-of-the-art techniques for task-oriented dialogue system … information-transfer tasks and end-to-end approaches handle well chatbot conversations, they …

Artificial Intelligence for All: An Abiding Destination
V Pathak, P Tiwari – 2018 – books.google.com
Page 1. ARTIFICIAL INTELLIGENCE FOR ALL AN ABIDING DESTINATION º Nº. -º- 2 º nº 2 wº º: DR. PANKAT TIWARI VIKAS PATHAIK Page 2. Artificial Intelligence For All Artificial Intelligence For All i Page 3. Artificial Intelligence …