Word2vec & Dialog Systems 2016


Resources:

Wikipedia:

See also:

100 Best GitHub: N-gram | 100 Best GitHub: Natural Language


Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models.
IV Serban, A Sordoni, Y Bengio, AC Courville, J Pineau – AAAI, 2016 – aaai.org
… form of common ground between speakers, eg to represent topics and concepts shared between the speakers using a dis- tributed vector representation, which we hypothesize to be important for building an effective dialogue system (Clark … 3http://code.google.com/p/word2vec …

How NOT to evaluate your dialogue system: An empirical study of unsupervised evaluation metrics for dialogue response generation
CW Liu, R Lowe, IV Serban, M Noseworthy… – arXiv preprint arXiv …, 2016 – arxiv.org
… Methods such as Word2Vec (Mikolov et al., 2013) calculate these embeddings using distributional semantics; that is, they approximate the … generated and actual responses, but still allows a quantitative comparison between responses generated by a dialogue system and the …

Strategy and Policy Learning for Non-Task-Oriented Conversational Systems.
Z Yu, Z Xu, AW Black, AI Rudnicky – SIGDIAL Conference, 2016 – aclweb.org
… TickTock is a non-task-oriented dialog system that takes typed text as the input and produces text as output … The utterance similarity fea- tures consist of a feature vector obtained from a word2vec model (Mikolov et al., 2013), the co- sine similarity score between the user …

The dialogue breakdown detection challenge: Task description, datasets, and evaluation metrics.
R Higashinaka, K Funakoshi, Y Kobayashi, M Inaba – LREC, 2016 – lrec-conf.org
… Abstract Dialogue breakdown detection is a promising technique in dialogue systems … a vector representing word and co-occurrence fre- quency vectors created by Sent2Vec (an extension of Word2Vec (Mikolov et al., 2013)). Run 1 used RNN and Run 2 used LSTM …

Sequential short-text classification with recurrent and convolutional neural networks
JY Lee, F Dernoncourt – arXiv preprint arXiv:1603.03827, 2016 – arxiv.org
… word vectors pretrained with GloVe on Twit- ter (Pennington et al., 2014) for MRDA and SwDA, as these choices yielded the best results among all publicly available word2vec, GloVe, SENNA (Collobert, 2011 … In 7th International Workshop on Spoken Dialogue Systems (IWSDS …

Dialogue State Tracking using Long Short Term Memory Neural Networks
K Yoshino, T Hiraoka, G Neubig, S Nakamura – 2016 – colips.org
… This change expands the variety of expressions of users, because the users will be free from limitations im- posed by dialogue systems … relieve this spar- sity, we compress the utterance to a 300 dimensional vector by using doc2vec [6]. doc2vec is a variation of word2vec [7], a …

Personalizing a Dialogue System with Transfer Learning
K Mo, S Li, Y Zhang, J Li, Q Yang – arXiv preprint arXiv:1610.02891, 2016 – arxiv.org
… In order to build a personalized dialogue system for the target user, we need to find a personalized Q-function Q?ut for this user … ?p u). All utterances and replies will be projected into state vec- tors with a state projection matrix M, where M is initial- ized with the word2vec and will …

An intelligent assistant for high-level task understanding
M Sun, YN Chen, AI Rudnicky – … of the 21st International Conference on …, 2016 – dl.acm.org
… Conventional multi-domain dialog systems passively select one domain from multiple domains according to a user in- put, ignoring relationships between domains and the … In this work, we used word2vec with the gensim toolkit1 on the model2 pre- trained on GoogleNews [20] …

Counter-fitting word vectors to linguistic constraints
N Mrkši?, DO Séaghdha, B Thomson, M Gaši?… – arXiv preprint arXiv …, 2016 – arxiv.org
… As far as we are aware, there is no previous work on exploiting antonymy in dialogue systems. The modelling work closest to ours are Liu et al. (2015), who use antonymy and WordNet hierarchy information to modify the heavyweight Word2Vec training objective; Yih et al …

Utterance Selection Based on Sentence Similarities and Dialogue Breakdown Detection on NTCIR-12 STC Task.
H Sugiyama – NTCIR, 2016 – research.nii.ac.jp
… [1], where some DBD systems tried to detect in- appropriate utterances generated by a dialogue system, it promises … Table 1 shows the features of the deep multilayer perceptron classifier. The word class features are calculated using k-means of vectors obtained from word2vec …

Dialogue session segmentation by embedding-enhanced texttiling
Y Song, L Mou, R Yan, L Yi, Z Zhu, X Hu… – arXiv preprint arXiv …, 2016 – arxiv.org
… 2.1. Dialogue Systems and Context Modeling Human-computer dialogue systems can be roughly divided into several categories. Template- and rule-based systems are mainly designed for certain domains [5, 6, 14] … To train the embeddings, we adopt the word2vec ap- proach …

A novel density-based clustering method using word embedding features for dialogue intention recognition
J Jang, Y Lee, S Lee, D Shin, D Kim, H Rim – Cluster Computing, 2016 – Springer
… Kim et al. [11] proposed an ensemble classification method for a Korean online messenger dialogue system; this method is composed of 37 different classification systems, and can analyze Korean texting language at the … This paper uses the Word2Vec method proposed in [23] …

Zara: A Virtual Interactive Dialogue System Incorporating Emotion, Sentiment and Personality Recognition.
P Fung, A Dey, FB Siddique, R Lin… – COLING …, 2016 – pdfs.semanticscholar.org
… As the availability of interactive dialogue systems is on a rise, people are getting more accustomed to talking to machines … We use word embedding vectors (Word2Vec) trained on the Google News corpus (Mikolov et al., 2013) of size 300, to train a CNN with one layer of …

Dialog state tracking, a machine reading approach using Memory Network
J Perez, F Liu – arXiv preprint arXiv:1606.04052, 2016 – arxiv.org
… In the DSTC-2 dialog corpus, a user queries a database of local restaurants by interacting with a dialog system … They are also subject to the same sharing constraints as A and C. The embedding matrix A and B are initialized using GoogleNews word2vec embedding model [16] …

Dialog state tracking with attention-based sequence-to-sequence learning
T Hori, H Wang, C Hori, S Watanabe… – … (SLT), 2016 IEEE, 2016 – ieeexplore.ieee.org
… A dialog system capable of responding in a human-like way must be capable of un- derstanding ambiguous and complicated user goals … LSTM and attention models, we initialized the projection layer Wpr with the 50 dimensional word vectors obtained by word2vec [23], where …

Distributional semantics for understanding spoken meal descriptions
M Korpusik, C Huang, M Price… – Acoustics, Speech and …, 2016 – ieeexplore.ieee.org
… SLT, 2014. [12] A. Celikyilmaz, D. Hakkani-Tur, P. Pasupat, and R. Sarikaya, “Enriching word embeddings using knowledge graph for se- mantic tagging in conversational dialog systems,” genre, 2010 … 3111–3119. [17] T. Mikolov, K. Chen, and J. Dean, “word2vec (2013),” …

Dialogue Act Classification in Domain-Independent Conversations Using a Deep Recurrent Neural Network.
H Khanpour, N Guntakandla, R Nielsen – COLING, 2016 – aclweb.org
… Many applications benefit from the use of automatic dialogue act classi- fication such as dialogue systems, machine translation, Automatic … other parameters constant (dropout = 0.5, decayrate = 0.5 and layersize = 2). Specifically, we tested the methods Word2vec using the …

Exploring Convolutional and Recurrent Neural Networks in Sequential Labelling for Dialogue Topic Tracking.
S Kim, RE Banchs, H Li – ACL (1), 2016 – aclweb.org
… Although they don’t aim at building components in dialogue systems di- rectly, the human behaviours learned from the con- versations can … first one learned the word embeddings from scratch with ran- dom parameters, while the other was initialized with word2vec (Mikolov et al …

Short Text Conversation (STC)
L Shanga, T Sakaib, Z Lua, H Lia, R Higashinakac… – research.nii.ac.jp
… The First Step at NTCIR-12 – Take it as an IR problem – Build a useful dialogue system that can interact naturally with humans 7 Page 8 … the similarity between two short texts – * the vector can be TF-IDF, word2vec, topic mode, etc. Feature Name #(Teams) The Teams …

Example-based spoken chat system which can be customized for each user
E Seto, N Kitaoka – The Journal of the Acoustical Society of …, 2016 – asa.scitation.org
… user. We then calculate the similarity between words related to the user and words in example phrases in the dialog system. Cosine similarity between distributed representations of words is calculated using Word2vec. We then …

A Deep Learning Methodology for Semantic Utterance Classification in Virtual Human Dialogue Systems
D Datta, V Brashers, J Owen, C White… – … Conference on Intelligent …, 2016 – Springer
… the development of a deep learning methodology for semantic utterance classification (SUC) for use in domain-specific dialogue systems … recurrent neural network (RNN) that uses domain-specific word embeddings which have been initialized using Word2Vec for determining …

Dialog state tracking for interview coaching using two-level LSTM
MH Su, CH Wu, KY Huang, TH Yang… – … (ISCSLP), 2016 10th …, 2016 – ieeexplore.ieee.org
… [9] K. Yoshino, T. Hiraoka, G. Neubig and S. Nakamura, “Dialog State Tracking using Long Short Term Memory Neural Networks,” in the Seventh International Workshop on Spoken Dialog Systems (IWSDS), Proceedings, 2016, pp. 1-8. [10] T. Mikolov, word2vec, accessed 2016 …

Adapting Spoken Dialog Systems Towards Domains and Users
M Sun – 2016 – lti.cs.cmu.edu
Page 1. Adapting Spoken Dialog Systems Towards Domains and Users Ming Sun CMU-LTI-16-006 … c 2016 , Ming Sun Page 2. Keywords: Lexicon learning, cloud speech recognition adaptation, high-level intention un- derstanding, spoken dialog systems Page 3. Abstract …

Backchanneling via Twitter Data for Conversational Dialogue Systems
M Inaba, K Takahashi – International Conference on Speech and …, 2016 – Springer
… Given the above issues, in this study, we propose a method for generating a rich variety of backchanneling to realize smooth communication in conversational dialogue systems … in [11], which was implemented in Word2Vec. We used 120 GB of Twitter data to train Word2Vec …

UT Dialogue System at NTCIR-12 STC.
S Sato, S Ishiwatari, N Yoshinaga, M Toyoda… – …, 2016 – pdfs.semanticscholar.org
… Our dialogue system is inspired by Yamamoto and Sumita’s work on domain adaptation for statistical machine transla- tion [6]. They showed that domain-specific models trained … We then use word2vec skip-gram model [3] to induce vector representations of words from ut tweets …

A multichannel convolutional neural network for cross-language dialog state tracking
H Shi, T Ushio, M Endo, K Yamagami… – … Workshop (SLT), 2016 …, 2016 – ieeexplore.ieee.org
… Index Terms- Convolutional neural networks, multi- channel architecture, dialog state tracking, dialog systems 1. INTRODUCTION … 3.3. Embedding models The word2vec [7] is one of the most COlmnon methods for producing word embeddings …

TEXT NORMALIZATION FOR AUTOMATIC SPEECH RECOGNITION SYSTEMS
AF VASILE, T BORO? – Editors: Maria Mitrofan Daniela Gîfu Dan Tufi? … – consilr.info.uaic.ro
… this, normalizing the text is extremely important for automatic machine translation (MT), speech-to-speech translation, information extraction, dialogue systems, etc … Before we trained our classifier we prepared our training data by running word2vec (Mikolov and Dean, 2013) on …

Dialogue act recognition for Chinese out-of-domain utterances using hybrid CNN-RF
J Wang, P Huang, Q Huang, Z Ke… – … Processing (IALP), 2016 …, 2016 – ieeexplore.ieee.org
… The log of the dialogue system lasts from September 2013 to June 2016, consisting of 1594 conversations with 2825 OOD utterances … The publicly available word2vec vectors he used were trained on 100 billion words from Google News …

Dialog State Tracking and action selection using deep learning mechanism for interview coaching
MH Su, KY Huang, TH Yang, KJ Lai… – Asian Language …, 2016 – ieeexplore.ieee.org
… [12] CH Wu, and GL Yan, “Speech act modeling and verification of spontaneous speech with disfluency in a spoken dialogue system,” IEEE transactions on speech and audio processing, vol. 13, no. 3, pp. 330-344, 2005. [13] T. Mikolov, word2vec, accessed 2016-03-20 …

On Dialogue Breakdown: Annotation and Detection
K Funakoshi, R Higashinaka, M Inaba, Y Kobayashi… – workshop.colips.org
… errors and limited capabilities of machines, and are one of the major issues to be addressed in dialogue systems research … team2 LSTM-RNN Word2Vec encoding of word frequencies team3 handcrafted rules Keywords extracted by a Japanese morphological analyzer team4 …

Adversarial Bandit for online interactive active learning of zero-shot spoken language understanding
E Ferreira, AR Masson, B Jabaian… – Acoustics, Speech and …, 2016 – ieeexplore.ieee.org
… Most of the time dedicated to create a dialogue system is for the data collection and annotation [6]. Some research works have focused on the … In [21, 22] a zero-shot learning method for SLU, the Zero-Shot Semantic Parser (ZSSP), based on word embeddings (word2vec [23]) is …

Neural dialog state tracker for large ontologies by attention mechanism
Y Jang, J Ham, BJ Lee, Y Chang… – … Workshop (SLT), 2016 …, 2016 – ieeexplore.ieee.org
… We project those words into a high-dimensional space maintaining relationship between them using Word2Vec … Convolutional neural networks for multi-topic dialog state tracking,” in Pro- ceedings of the 7th International Workshop on Spoken Dialogue Systems (IWSDS), 2016 …

Overview of the NTCIR-12 Short Text Conversation Task.
L Shang, T Sakai, Z Lu, H Li, R Higashinaka… – …, 2016 – pdfs.semanticscholar.org
… We review in this paper the task definition, evaluation measures, test collections, and the evaluation results of all teams. Keywords artificial intelligence, dialogue systems, evaluation, information re- trieval, natural language processing, social media, test collections …

CFGs-2-NLU: Sequence-to-sequence learning for mapping utterances to semantics and pragmatics
AJ Summerville, J Ryan, M Mateas… – arXiv preprint arXiv …, 2016 – arxiv.org
… Due to the highly structured nature of the interactions in service dialogue systems, rule-based systems for natural language understanding (NLU) can … the training data), we employ a short pipeline that utilizes a spellchecker and, if necessary, the Google News word2vec model [2 …

Optimizing neural network hyperparameters with gaussian processes for dialog act classification
F Dernoncourt, JY Lee – Spoken Language Technology …, 2016 – ieeexplore.ieee.org
… We initialize the word vectors with the 300-dimensional word vectors pretrained with word2vec on Google News [28, 29] for DSTC 4, and the 200-dimensional word vectors pretrained with GloVe on Twitter [30] for SwDA. 3.3. Hyperparameters …

Prediction of Prospective User Engagement with Intelligent Assistants.
S Sano, N Kaji, M Sassano – ACL (1), 2016 – aclweb.org
… Such user behaviors are rarely observed in conventional ex- perimental environments, where dialogue systems … To induce the word clusters, 100-dimensional word embeddings are first learned from the log data using WORD2VEC (Mikolov et al., 2013)5, and then K-means …

Context-Sensitive and Role-Dependent Spoken Language Understanding Using Bidirectional and Attention LSTMs.
C Hori, T Hori, S Watanabe, JR Hershey – INTERSPEECH, 2016 – merl.com
… Spoken language under- standing (SLU) technologies in dialog systems have been inten- sively investigated to estimate the intention of user utterances obtained from an automatic speech recognition (ASR) system [1, 2]. Conventional … LR + word2vec – 71.1 – 72.4 – 62.1 – 62.3 …

Stalematebreaker: A proactive content-introducing approach to automatic human-computer conversation
X Li, L Mou, R Yan, M Zhang – arXiv preprint arXiv:1604.04358, 2016 – arxiv.org
… 2 Related Work 2.1 Dialogue systems • Domain-specific systems … ?(·, ·) was learned via a learning-to-rank model similar to Burges et al. [2005] with rich features including textual similarity, translation models, as well as word2vec word embeddings [Mikolov et al., 2013] …

Deep Learning of Audio and Language Features for Humor Prediction.
D Bertero, P Fung – LREC, 2016 – lrec-conf.org
… Our first CNN takes as input a word vector for each token taken from Word2Vec (Mikolov et al., 2013) … Our ultimate goal is to integrate laughter response prediction in a machine dialog system, to allow it to understand and react to humor. 5. Acknowledgments …

Multimodal deep neural nets for detecting humor in TV sitcoms
D Bertero, P Fung – Spoken Language Technology Workshop …, 2016 – ieeexplore.ieee.org
… to a larger set of language-only features, which includes one-hot word vectors and character- trigram input vectors in addition to Word2Vec … no humorous intent is a very impolite behavior in human interaction, and is not really the desired outcome of an automatic dialog system …

RACAI Entry for the IWSLT 2016 Shared Task
S Pipa, AF Vasile, I Ionascu… – Proceedings of the …, 2016 – workshop2016.iwslt.org
… Text normalization is extremely important for automatic machine translation (MT), speech-to-speech translation, information extraction, dialog systems, etc … Before we trained our classifier, we prepared our training data by running word2vec (Mikolov and Dean, 2013) on a …

Multi-Domain Joint Semantic Frame Parsing Using Bi-Directional RNN-LSTM.
D Hakkani-Tür, G Tür, A Celikyilmaz… – …, 2016 – pdfs.semanticscholar.org
… In addition to 1-hot word vectors, we experimented with word2vec [37] and Senna [38] embeddings, and did not observe significant … [4] Y.-N. Chen, WY Wang, and AI Rudnicky, “Unsupervised in- duction and filling of semantic slots for spoken dialogue systems using frame …

Assisting discussion forum users using deep recurrent neural networks
JHP Suorra, O Mogren – Proceedings of the 1st Workshop on …, 2016 – aclweb.org
… Top: posts are represented as a sum of embed- dings from Word2Vec over the words in each post … Dialog systems, also known as conversational agents, typically focus on learning to produce a well-formed response, and put less emphasis on the message that they convey in …

Automatic Corpus Extension for Data-driven Natural Language Generation.
E Manishina, B Jabaian, S Huet, F Lefèvre – LREC, 2016 – lrec-conf.org
… The word2vec model (Mikolov et al., 2013) is trained on LDC Gigaword 5th edition, the Brown corpus and the English … International dialog systems, targeting non- native speakers, might consider employing basic simple phrases which use standard English vocabulary, with no …

A Study on Image Semantic Analysis Algorithm for Natural Language Understanding
J LUO, HUAJUN WANG, YANMEI LI… – Journal of Residuals …, 2016 – dpi-journals.com
… However, as an important research topic in this field, do we make full use of the great function of big data to build a data-driven natural language dialogue system … First, use the Word2Vec as a tool to collect the selected words …

Towards Empathetic Human-Robot Interactions
P Fung, D Bertero, Y Wan, A Dey, RHY Chan… – arXiv preprint arXiv …, 2016 – arxiv.org
… only to find human users disappointed by the lack of reciprocal empathy from these robots. It follows that we shall embody interactive dialog systems in simulated or robotic forms … In the current approach, we use a CNN-based classifier on Word2Vec …

Report on the Eighth Workshop on Exploiting Semantic Annotations in Information Retrieval (ESAIR’15)
K Balog, J Dalton, A Doucet, Y Ibrahim – ACM SIGIR Forum, 2016 – dl.acm.org
… The results showed that combining both ESA and Word2Vec led to slight improvements and indicated that ESA and Word2Vec tended to have different effects to retrieve different relevant sentences, suggesting that more adequate … Question answering and dialog systems …

Chinese poetry generation with planning based neural network
Z Wang, W He, H Wu, H Wu, W Li, H Wang… – arXiv preprint arXiv …, 2016 – arxiv.org
… In the future, we will investigate more methods for topic planning, such as PLSA, LDA or word2vec … 2016. How not to evaluate your dialogue system: An empirical study of unsupervised evaluation metrics for dialogue response generation. arXiv preprint arXiv:1603.08023 …

Assisting Discussion Forum Users using Deep Recurrent Neural Networks
J Suorra Hagstedt P, O Mogren – Proceedings of the …, 2016 – publications.lib.chalmers.se
… Top: posts are represented as a sum of embed- dings from Word2Vec over the words in each post … Dialog systems, also known as conversational agents, typically focus on learning to produce a well-formed response, and put less emphasis on the message that they convey in …

Recurrent convolutional neural networks for structured speech act tagging
T Ushio, H Shi, M Endo, K Yamagami… – … Workshop (SLT), 2016 …, 2016 – ieeexplore.ieee.org
… is 100 for each win- dow size. Word vectors are initialized to random vectors of length 100 and the pre-training procedure is performed by word2vec[20] using a large-scale Wikipedia corpus. The RCNN is truncated to a depth …

Weakly supervised user intent detection for multi-domain dialogues
M Sun, A Pappu, YN Chen… – … Workshop (SLT), 2016 …, 2016 – ieeexplore.ieee.org
… 2httpS : //code . google . com/archive/p/word2vec/ 95 X2 1089 22,546 0.138 0.147 0.145 … 83-86. Page 7. [4] Aasish Pappu and Alexander Rudnicky, “Predicting tasks in goal-oriented spoken dialog systems using se- mantic knowledge bases,” in SIGDIAL, 2013, pp. 242- 250 …

Globally Coherent Text Generation with Neural Checklist Models
C Kiddon, L Zettlemoyer, Y Choi – … of the 2016 Conference on Empirical …, 2016 – aclweb.org
… processing We used the hotel and restaurant dialogue system corpus and the same train-development-test split from Wen et al. (2015). We used the same pre-processing, sets 4Recipes and format at http://www.ffts.com/recipes.htm 5See https://code.google.com/p/word2vec …

Human-like Natural Language Generation Using Monte Carlo Tree Search
K Kumagai, I Kobayashi, D Mochihashi… – Proceedings of the …, 2016 – aclweb.org
… Brown Corpus Wikipedia Engilish (Wiki-En) Given words imilar ords word2vec (eg dog, dogs, run…) (eg puppy, passes…) … V. Rieser and O. Lemon. 2009. Natural language genera- tion as planning under uncertainty for spoken dialogue systems. In EACL 2009, pages 683–691 …

Overview of NTCIR-13
MP Kato, Y Liu, C Gurrin, H Joho… – Proceedings of the …, 2016 – research.nii.ac.jp
… KIT Dialogue System for NTCIR-13 STC Japanese Subtask Hiroshi Nakatani, Shigenori Nishiumi, Takahiro Maeda and Masahiro Araki [Pdf] [Table of Content] We … Method_1 is a retrieval-based method of scoring reply texts using TF-IDF, with relevance filtering using word2vec …

Human-like Natural Language Generation Using Monte Carlo Tree Search
KKI Kobayashi, D Mochihashi… – The INLG 2016 … – webprojects.eecs.qmul.ac.uk
… Brown Corpus Wikipedia Engilish (Wiki-En) Given words imilar ords word2vec (eg dog, dogs, run…)(eg puppy, passes…) using only non-terminal non-terminal using for holding enough … Natural language genera- tion as planning under uncertainty for spoken dialogue systems …

Sequential Match Network: A New Architecture for Multi-turn Response Selection in Retrieval-based Chatbots
Y Wu, W Wu, M Zhou, Z Li – arXiv preprint arXiv:1612.01627, 2016 – arxiv.org
Page 1. Sequential Match Network: A New Architecture for Multi-turn Response Selection in Retrieval-based Chatbots Yu Wu†? , Wei Wu‡ , Zhoujun Li† , Ming Zhou‡ †State Key Lab of Software Development Environment, Beihang …

Thematic fit evaluation: an aspect of selectional preferences
A Sayeed, C Greenberg, V Demberg – ACL 2016, 2016 – anthology.aclweb.org
… This is particularly important as dialog systems grow steadily less task-specific … BDK2014 Tests of word embedding spaces from Baroni et al.(2014), constructed via word2vec (Mikolov et al., 2013a). These are the best systems reported in their paper …

Neural Networks for Natural Language Processing
L Mou – sei.pku.edu.cn
… Page 21. CBOW, SkipGram (word2vec) [6] Mikolov T, Chen K, Corrado G, Dean J. Efficient estimation of word … Motivation: We don’t have the ground truth In a dialogue system, “The nature of of opendomain conversations shows that a variety of replies are plausible, but …

Sequence-to-sequence learning as beam-search optimization
S Wiseman, AM Rush – arXiv preprint arXiv:1606.02960, 2016 – arxiv.org
… addition to demon- strating impressive results for machine translation (Bahdanau et al., 2015), roughly the same model and training have also proven to be useful for sen- tence compression (Filippova et al., 2015), parsing (Vinyals et al., 2015), and dialogue systems (Ser- ban …

Overview of NTCIR-12.
K Kishida, MP Kato – NTCIR, 2016 – research.nii.ac.jp
… To estimate the similarity between words, we investigated the use of word2vec tool. For all these approaches, we discuss the obtained results during the experimental evaluation … We converted queries and iUnits to vectors with word2vec …

Affective analysis and Modeling of Spoken Dialogue Transcripts
E Palogiannidi – 2016 – researchgate.net
… Jose Lopes, Arodami Chorianopoulou, Elisavet Palogiannidi, Helena Moniz, Alberto Abad, Katerina Louka, Elias Iosif and Aleandros Potamianos “The SpeDial Datasets: Datasets for Spoken Dialogue Systems Analytics”, in Proceedings of the 10th edition of the Language …

The NITech text-to-speech system for the Blizzard Challenge 2016
K Sawada, C Asai, K Hashimoto, K Oura… – Blizzard Challenge 2016 …, 2016 – festvox.org
… the quality of synthetic speech has improved, and such systems are now used in various applica- tions, such as for in-car navigation, smartphones, and spoken dialogue systems … The word2vec [14] is used to measure word similarity of a test word and training corpus words. 2.2 …

DeepSoft: A vision for a deep model of software
HK Dam, T Tran, J Grundy, A Ghose – Proceedings of the 2016 24th ACM …, 2016 – dl.acm.org
… is treated in a sequential manner as natural languages, and existing deep learning-based NLP techniques such as word2vec, para- graph2vec or … Given the recent successes in NLP [5] (machine translation, question answering, and dialog systems) and vi- sion [4] (image/video …

Question answering in conversations: Query refinement using contextual and semantic information
M Habibi, P Mahdabi, A Popescu-Belis – Data & Knowledge Engineering, 2016 – Elsevier
… have with other users. To answer the questions, the proposed method leverages the local context of the conversation along with semantic resources, either WordNet or word embeddings from word2vec. The method first represents …

Neural Paraphrase Generation with Stacked Residual LSTM Networks
A Prakash, SA Hasan, K Lee, V Datla, A Qadir… – arXiv preprint arXiv …, 2016 – arxiv.org
… Cho et al., 2014; Bahdanau et al., 2015), speech recognition (Li and Wu, 2015), language modeling (Vinyals et al., 2015), and dialogue systems (Serban et al … In our experiments, we used Word2Vec embeddings pre-trained on the Google News Corpus (Mikolov et al., 2014) …

DocChat: An Information Retrieval Approach for Chatbot Engines Using Unstructured Documents.
Z Yan, N Duan, JW Bao, P Chen, M Zhou, Z Li, J Zhou – ACL (1), 2016 – aclweb.org
Page 1. Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, pages 516–525, Berlin, Germany, August 7-12, 2016. cO2016 Association for Computational Linguistics DocChat: An Information …

FBK-HLT-NLP at SemEval-2016 Task 2: A Multitask, Deep Learning Approach for Interpretable Semantic Textual Similarity.
S Magnolini, A Feltracco, B Magnini – SemEval@ NAACL-HLT, 2016 – aclweb.org
… Headlines), and a question-answer dataset collected and annotated during the evaluation of the BEETLE II tutorial dialogue system (Student Answers … We use Mikolov word2vec (Mikolov et al., 2013) with 100 dimensions using ukWaC, GigaWords (NYT), Europarl V.7, Training …

Deep reinforcement learning with an action space defined by natural language
J He, J Chen, X He, J Gao, L Li, L Deng, M Ostendorf – 2016 – openreview.net
… term rewards. network architecture that is trained jointly with multitask learning. Mikolov et al. (2013) introduced word2vec, which is an efficient estimation of continuous vector representations of words. They further explored …

Neural Document Embeddings for Intensive Care Patient Mortality Prediction
P Grnarova, F Schmidt, SL Hyland… – arXiv preprint arXiv …, 2016 – arxiv.org
… Following recent work in document classification [21] and dialogue systems [17], we adopt a two- layer architecture … 3.3 Parameters and Pretraining We pre-train 50-dimensional word vectors on the training data using the word2vec implementation of the gensim [16] toolbox …

Gaussian Attention Model and Its Application to Knowledgebase Embedding and Question Answering
L Zhang, J Winn, R Tomioka – arXiv preprint arXiv:1611.02266, 2016 – arxiv.org
… and entailment. However the work was presented in the word2vec (Mikolov et al., 2013)-style word embedding setting and the Gaussian embedding was used to capture the diversity in the meaning of a word. Our Gaussian …

Recent Improvements on Error Detection for Automatic Speech Recognition.
Y Estève, S Ghannay, N Camelin – MMDA@ ECAI, 2016 – pdfs.semanticscholar.org
… [13] proposed an extension of word2vec, called word2vecf and … Linar`es, Driss Matrouf, and Re- nato De Mori, ‘Integration of word and semantic features for theme identification in telephone conversations’, in 6th International Work- shop on Spoken Dialog Systems (IWSDS 2015 …

An empirical analysis of formality in online communication
E Pavlick, J Tetreault – … of the Association for Computational Linguistics, 2016 – transacl.org
… As a result, the ability to recognize for- mality is an integral part of dialogue systems (Mairesse, 2008; Mairesse and Walker, 2011; Battaglino and Bickmore, 2015), sociolinguistic … word2vec Average of word vectors using pre-trained word2vec embeddings, skipping OOV words …

Controlling the voice of a sentence in japanese-to-english neural machine translation
H Yamagishi, S Kanouchi, T Sato… – Proceedings of the 3rd …, 2016 – anthology.aclweb.org
… 3https://radimrehurek. com/gensim/models/word2vec. html 4We did not perform any processing of unknown words because we focused on the control of the voice … For example, one may prefer a polite expression for generating conversation in a dialog system …

Towards Building A Domain Agnostic Natural Language Interface to Real-World Relational Databases
SH Ramesh, J Jain, KS Sarath… – Proceedings of the 13th …, 2016 – aclweb.org
… The set of additional tokens is generated by making use of the above – Snowball stemming, word2vec similarity, and WordNet synsets … We are improving our system by adding support for complex SQL queries like nested queries and we also plan to make it a dialog system that is …

The comprehension of figurative language: what is the influence of irony and sarcasm on NLP techniques?
L Weitzel, RC Prati, RF Aguiar – Sentiment Analysis and Ontology …, 2016 – Springer
… or partially sighted users; automatic report generation (possibly multilingual); machine translation; plagiarism detection tools; email understanding and dialogue systems [5] … This model, called Word2Vec, has been attracting considerable attention in recent years [6, 43, 51, 52, 60 …

Relation schema induction using tensor factorization with side information
M Nimishakavi, US Saini, P Talukdar – arXiv preprint arXiv:1605.04227, 2016 – arxiv.org
… We define, Sij = { 1, if Similarity(Reli, Relj) ? ? 0, otherwise where ? is a threshold5. For the experiments in this paper, we use cosine similarity over word2vec (Mikolov et al., 2013) vector repre- sentations of the relational phrases …

Semi-supervised and unsupervised methods for categorizing posts in web discussion forums
K Perumal – arXiv preprint arXiv:1604.00119, 2016 – arxiv.org
… Other unsupervised techniques have been employed for the related tasks of dialogue act classification in spoken dialogue systems (Crook et al., 2009) and Twitter conversations (Ritter et al., 2010). Although they worked specifically on genres of text that are very …

Symbol emergence in robotics: a survey
T Taniguchi, T Nagai, T Nakamura, N Iwahashi… – Advanced …, 2016 – Taylor & Francis
Page 1. ADVANCED ROBOTICS, 2016 VOL. 30, NOS. 11–12, 706–728 http://dx.doi.org/10.1080/01691864.2016.1164622 SURVEY PAPER Symbol emergence in robotics: a survey Tadahiro Taniguchia, Takayuki Nagaib, Tomoaki …

Expanding science and technology thesauri from bibliographic datasets using word embedding
T Kawamura, K Kozaki, T Kushida… – Tools with Artificial …, 2016 – ieeexplore.ieee.org
… [7] Y. Nishio: “word2vec for NLP,” O’Reilly Japan, Inc., 2014 … [18] A. Celikyilmaz, D. Hakkani-Tur, P. Pasupat and R. Sarikaya: “En- riching Word Embeddings Using Knowledge Graph for Semantic Tagging in Conversational Dialog Systems,” In Proc …

PersoNER: Persian Named-Entity Recognition
H Poostchi, E Zare Borzeshi, M Abdous… – The 26th International …, 2016 – opus.lib.uts.edu.au
… Applications Tim Baldwin Maria Liakata Dialog Processing and Dialog Systems, Multimodal Interfaces Nina Dethlefs Simon Keizer Giuseppe Riccardi Speech Recognition, Text-To-Speech, Spoken Language Understanding Florian Metze Chung-Hsien Wu …

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers
Y Matsumoto, R Prasad – Proceedings of COLING 2016, the 26th …, 2016 – aclweb.org
… Applications Tim Baldwin Maria Liakata Dialog Processing and Dialog Systems, Multimodal Interfaces Nina Dethlefs Simon Keizer Giuseppe Riccardi Speech Recognition, Text-To-Speech, Spoken Language Understanding Florian Metze Chung-Hsien Wu …

Text analytics in industry: Challenges, desiderata and trends
A Ittoo, LM Nguyen, A van den Bosch – Computers in Industry, 2016 – Elsevier
The recent decades have witnessed an unprecedented expansion in the volume of unstructured data in digital textual formats. Companies are now starting to recogn.

Detecting Sarcasm in Multimodal Social Platforms
R Schifanella, P de Juan, J Tetreault… – Proceedings of the 2016 …, 2016 – dl.acm.org
… word2vec: average of word vectors using pre-trained word2vec embeddings [25]. OOV words are skipped. • combination: n-grams, word2vec and readability fea- tures (these include length of post in words and charac- ters, as well as the Flesch-Kincaid Grade level score [20]) …

Compressing neural language models by sparse word representations
Y Chen, L Mou, Y Xu, G Li, Z Jin – arXiv preprint arXiv:1610.03950, 2016 – arxiv.org
… et al., 2011); it is even possible to gen- erate new sentences from a neural LM, benefit- ing various downstream tasks like machine trans- lation, summarization, and dialogue systems (De- vlin et al … To learn the sparse codes, we first train the “true” embeddings by word2vec2 for …

Solving Verbal Questions in IQ Test by Knowledge-Powered Word Embedding.
H Wang, F Tian, B Gao, C Zhu, J Bian, TY Liu – EMNLP, 2016 – aclweb.org
… between pair (A, B) and pair (C, D). Such questions test the ability of identifying an implicit relation from word pair (A, B) and ap- ply it to compose word pair (C, D). Note that the Analogy-I questions are also used as a major eval- uation task in the word2vec models (Mikolov et al …

Review of state-of-the-arts in artificial intelligence. Present and future of AI.
V Shakirov – alpha.sinp.msu.ru
… It reminds an unsupervised objective function somewhat similar to what is used for example in word2vec … ”Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models” http:// arxiv.org/abs/1507.04808 …

A Corpus for Event Localization
C Ward – 2016 – bir.brandeis.edu
… 2015). Word embeddings. Every model has its first layer of weights initialized with pretrained word embeddings available at https://code.google.com/archive/p/word2vec/. The embeddings are 300-dimensional and are trained using the skip-gram architecture on a …

Deep learning for sentiment analysis
LM Rojas?Barahona – Language and Linguistics Compass, 2016 – Wiley Online Library
… Different word embeddings have been made public. They usually encode syntactic/semantic similarities7 (Collobert & Weston, 2008 or Turian et al., 2010); contextual information, Word2Vec (Mikolov et al., 2013); or correlation between words, Glove (Pennington et al., 2014) …

Learning to Interpret and Generate Instructional Recipes
C Kiddon – 2016 – digital.lib.washington.edu
… 69 4.5 Dialogue System Results … Laroche et al. (2013) proposed Cooking Coach, a spoken dialogue system to help a user search for recipes and prepare the recipe. A similar, if not more futuristic, application is the construction of robotic cooking assistants (Beetz …

Music Predictions Using Deep Learning. Could LSTM Networks be the New Standard for Collaborative Filtering?
E Keski-Seppälä, M Snellman – 2016 – diva-portal.org
Page 1. INOM EXAMENSARBETE TEKNIK, GRUNDNIVÅ, 15 HP , STOCKHOLM SVERIGE 2016 Music Predictions Using Deep Learning. Could LSTM Networks be the New Standard for Collaborative Filtering? EMIL KESKI-SEPPÄLÄ AND MICHAEL SNELLMAN …

Review of state-of-the-arts in artificial intelligence with application to AI safety problem
V Shakirov – arXiv preprint arXiv:1605.04232, 2016 – arxiv.org
… It’s like an unsupervised objective function somewhat similar to what is used for example in word2vec … ”Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models” http:// arxiv.org/abs/1507.04808 …

An Exploratory Study on Process Representations
CN Naik – 2016 – search.proquest.com
… and have been shown to benet question answering [7, 8], textual entailment [9], machine translation [1012], and dialogue systems [13, 14 … We used an approach that combined WordNet-based phrase similarity method, and word2vec vector similarity, where the vectors where …

Computer Vision and Natural Language Processing: Recent Approaches in Multimedia and Robotics
P Wiriyathammabhum, D Summers-Stay… – ACM Computing …, 2016 – dl.acm.org
… Compositional semantics are related to parsing and grammars and formal semantics to predicate and ?-calculus. Finally, distributional semantics concern the use of latent variables (word2vec, embeddings, deep learning etc.) …

Automated fictional ideation via knowledge base manipulation
MT Llano, S Colton, R Hepworth, J Gow – Cognitive computation, 2016 – Springer

Word representation using a deep neural network
Y Li – 2016 – search.proquest.com
… The embedding model is based on the word vectors obtained using word2vec as described in [41, 42] … The vectors can be used in various scenarios in NLP, such as machine translation, knowledge extraction, information retrieval, and dialogue systems …

An Iterative Transfer Learning Based Ensemble Technique for Automatic Short Answer Grading
S Roy, HS Bhatt, Y Narahari – arXiv preprint arXiv:1609.04909, 2016 – arxiv.org
… (LSA) [38] trained on a Wikipedia dump. We also use the recently popular word2vec tool (W2V) [39] to obtain vector representation of words which are trained on 100 billion words of Google news dataset and are of length 300 …

Linguistic Knowledge in Data-Driven Natural Language Processing
Y Tsvetkov – 2016 – cs.cmu.edu
… driven models underperform in low-resource settings: they are inadequate, for example, to translate African languages, to detect metaphors in Russian and Persian, to grammatically parse Cantonese, to model Latin or Hebrew morphology, to build dialog systems for indigenous …

Linguistic Linked Open Data: 12th EUROLAN 2015 Summer School and RUMOUR 2015 Workshop, Sibiu, Romania, July 13-25, 2015, Revised Selected …
D Trandab??, D Gîfu – 2016 – books.google.com
… For instance, we have shown that improved word2vec-style distributed vector representations of words can be acquired if explicit pattern matches bear additional weight during training [3]. The second major challenge is that of knowledge integration …

Computational methods in semantics
G Recski – 2016 – nytud.hu
Page 1. Computational Methods in Semantics Gábor Recski Ph.D. Dissertation Supervisor: András Kornai D.Sc. Ph.D. School of Linguistics Gábor Tolcsvai Nagy MHAS Theoretical Linguistics Ph.D Program Zoltán Bánréti C.Sc. Department of Theoretical Linguistics …

1st International Workshop on Multimodal Media Data Analytics (MMDA 2016)
S Vrochidis, M Melero, L Wanner, J Grivolla, Y Estève… – ecai2016.org
Page 1. ECAI 2016, MMDA 2016 workshop, August 2016 1st International Workshop on Multimodal Media Data Analytics (MMDA 2016) The rapid advancements of digital technologies, as well as the penetration of internet and …

Generation of textual summaries at different target reading levels: summarizing line graphs for visually impaired users
PS Moraes – 2016 – search.proquest.com
Generation of textual summaries at different target reading levels: Summarizing line graphs for visually impaired users. Abstract. This work is concerned with the generation of text at different reading levels by tailoring the generated …

Large-scale affective computing for visual multimedia
B Jou – 2016 – search.proquest.com
Large-scale affective computing for visual multimedia. Abstract. In recent years, Affective Computing has arisen as a prolific interdisciplinary field for engineering systems that integrate human affections. While human-computer …

(Visited 265 times, 1 visits today)