Cosine Similarity, Semantic Similarity & Dialog Systems

Notes:

Cosine similarity and semantic similarity are two measures of similarity that can be used in dialog systems to determine how closely related two pieces of text are. These measures can be used to help the system understand the meaning of user input and generate appropriate responses.

Cosine similarity is a measure of similarity based on the angle between two vectors in a vector space. It is often used in natural language processing to measure the similarity between two pieces of text by representing each piece of text as a vector and calculating the angle between the vectors. The cosine similarity between two pieces of text is a number between 0 and 1, with 0 indicating that the text is completely dissimilar and 1 indicating that the text is identical.

Semantic similarity, on the other hand, is a measure of similarity based on the meaning of the words in a piece of text. It is often used in natural language processing to measure the similarity between two pieces of text by analyzing the meanings of the words and phrases used in the text, and by comparing the overall meanings of the texts. The semantic similarity between two pieces of text is also a number between 0 and 1, with 0 indicating that the texts have completely different meanings and 1 indicating that the texts have the same meaning.

In a dialog system, cosine similarity and semantic similarity can be used to determine how similar the user’s input is to the system’s previous responses, or to other pieces of text in the system’s knowledge base. This can help the system understand the meaning of the user’s input and generate a response that is appropriate and relevant. For example, if the user asks a question, the system can use cosine similarity or semantic similarity to compare the user’s question to previous questions and answers in its knowledge base, and to generate a response that is similar to the previous answers. This can help the system provide more accurate and relevant responses, and improve the overall quality of the dialogue.

Resources:

Wikipedia:

References:

See also:

Language Computer Corporation (LCC) | Lexical Chain & Dialog Systems | LSM (Latent Semantic Mapping) | NLP Tools & Dialog Systems | NLTK & Dialog Systems | OpenEphyra (Ephyra Question Answering System) | Porter Stemmer | Quora & Natural Language Processing | TweetNLP

Eliciting positive emotional impact in dialogue response selection
N Lubis, S Sakti, K Yoshino, S Nakamura – Advanced Social Interaction …, 2019 – Springer
… end{aligned}$$. (2). Given a query, this cosine similarity is computed over all example queries in the database and treated as the … 4 Proposed Dialogue System … The baseline system will output the response of the tri-turn example with the highest semantic similarity score (Eq …

Transferable multi-domain state generator for task-oriented dialogue systems
CS Wu, A Madotto, E Hosseini-Asl, C Xiong… – arXiv preprint arXiv …, 2019 – arxiv.org
… state tracking (DST) is a core component in task-oriented dialogue systems, such as … actions and the current user utterance as input, and com- putes semantic similarity with predefined … tracking performance, as mentioned in Section 1. We visualize the cosine similarity matrix for …

Training neural response selection for task-oriented dialogue systems
M Henderson, I Vuli?, D Gerz, I Casanueva… – arXiv preprint arXiv …, 2019 – arxiv.org
… and retrieves the most relevant response from a large re- sponse collection by computing semantic similarity between the … Although conceptually attractive, retrieval-based dialogue systems still suffer from data scarcity, as deployment to a new … Scaled Cosine Similarity Scoring …

Approximating interactive human evaluation with self-play for open-domain dialog systems
A Ghandeharioun, JH Shen, N Jaques… – Advances in Neural …, 2019 – papers.nips.cc
… We estimate Semantic Similarity by calculating the cosine similarity between the … Self-play as an approximation for interactive evaluation Since interactive human evaluation is costly, we propose a self-play scenario where the dialog system talks to itself, ie the bot …

Sentence Similarity Techniques for Short vs Variable Length Text using Word Embeddings
D Shashavali, V Vishwjeet, R Kumar, G Mathur… – Computación y …, 2019 – cys.cic.ipn.mx
… 1. The primary goal of the dialogue systems is to understand the user’s input or … The recent approach [14] to sentence level semantic similarity technique is based on unsupervised … Here, we propose methods which are based on Cosine similarity calculation along with Sliding …

Automatic evaluation of end-to-end dialog systems with adequacy-fluency metrics
LF D’Haro, RE Banchs, C Hori, H Li – Computer Speech & Language, 2019 – Elsevier
… objective metrics such as word-overlap (eg BLEU or ROUGE) or semantic similarity over word … into the evaluation of generative answers in the context of dialog systems, we borrow … then computed in the low dimensional embedding by calculating the cosine similarity between the …

Better automatic evaluation of open-domain dialogue systems with contextualized embeddings
S Ghazarian, JTZ Wei, A Galstyan, N Peng – arXiv preprint arXiv …, 2019 – arxiv.org
… Model Unreferenced Referenced Pooling Pearson Spearman Cosine Similarity … The main drawback of mentioned n-gram overlap metrics, which makes them inapplicable in dialogue system evaluation is that they don’t con- sider the semantic similarity between sentences (Liu …

The feasibility of embedding based automatic evaluation for single document summarization
S Sun, A Nenkova – Proceedings of the 2019 Conference on Empirical …, 2019 – aclweb.org
… We build upon this line of work and show that cosine similarity between the reference and … Unlike prior work (Ng and Abrecht, 2015), we thoroughly abandon ROUGE and n-gram co-occurrences in the computation of semantic similarity …

Boosting dialog response generation
W Du, AW Black – Proceedings of the 57th Annual Meeting of the …, 2019 – aclweb.org
… has become one of the most popular ap- proaches to dialog systems, for it … averaging approach by (Arora et al., 2016), and measure the cosine similarity of the … Boosting can improve lexical-level semantic similarity between generate responses and true responses, measured by …

Follow-Up Question Generation Using Neural Tensor Network-Based Domain Ontology Population in an Interview Coaching System.
MH Su, CH Wu, Y Chang – INTERSPEECH, 2019 – isca-speech.org
… oriented dialog systems [2]-[3]. There have been many task-oriented dialog systems constructed in … They used the topic model to generate concepts and used semantic similarity measure to … the performance of CNTN model and the traditional TFIDF with cosine similarity method …

Cross-lingual semantic specialization via lexical relation induction
EM Ponti, I Vuli?, G Glavaš, R Reichart… – Proceedings of the 2019 …, 2019 – aclweb.org
… property of distributional word vectors; semantic specialization of such spaces for a particular lexico- semantic relation (eg, semantic similarity or lex … RE- PEL) pairs are chosen by retrieving the nearest (or farthest) neighbours to xl and xr, respectively, in terms of cosine similarity …

Quizbot: A dialogue-based adaptive learning system for factual knowledge
S Ruan, L Jiang, J Xu, BJK Tham, Z Qiu, Y Zhu… – Proceedings of the …, 2019 – dl.acm.org
… We first describe QuizBot and its three key components: a dialogue system, a semantic similarity … on a rule-based chat system combined with a supervised sentence semantic similarity model … The model will return the cosine similarity between the correct answer and the user’s …

Positive Emotion Elicitation in Chat-Based Dialogue Systems
N Lubis, S Sakti, K Yoshino… – IEEE/ACM Transactions …, 2019 – ieeexplore.ieee.org
… given time, the system will return the best response according to semantic similarity between user … Given a query, this cosine similarity is computed over all example queries in the database … LUBIS et al.: POSITIVE EMOTION ELICITATION IN CHAT-BASED DIALOGUE SYSTEMS …

Measuring interpretable semantic similarity of sentences using a multi chunk aligner
G Majumder, P Pakray, D Pinto – Journal of Intelligent & Fuzzy …, 2019 – content.iospress.com
… like lowercased token overlap, stemmed or lemmatized token overlap, cosine similarity between Mikolov’s … Jiang JJ and Conrath DW , Semantic similarity based on corpus statistics and lexical … VanLehn K. and Albacete PL , A natural language tutorial dialogue system for physics …

Augmenting dialogue response generation with unstructured textual knowledge
Y Wang, W Rong, Y Ouyang, Z Xiong – IEEE Access, 2019 – ieeexplore.ieee.org
… Furthermore, the retrieval method based dialogue system also requires many prepared answers to the questions [25 … One is the Embedding Average metric (EACosine) defined as the cosine similarity between the embedding vectors … It is a metric to encourage semantic similarity …

Short Text Clustering Enhanced by Semantic Matching Model
Z Peng, G Xin, Y Wei, W Wang… – 2019 2nd International …, 2019 – ieeexplore.ieee.org
… traditional NLP tasks, such as machine translation, emotional analysis, dialogue system, reading comprehension … a real number between 0 and 1, representing the semantic similarity between the … nonlinear layer should be close to fitting the cosine similarity calculation function …

A Deep Semantic Matching Network for Answer Selection
Y Li, H Li, J Li, N Zhang, G Yuan – 2019 Chinese Control …, 2019 – ieeexplore.ieee.org
… From the functional point of view, the conversation system includes task-oriented dialogue system aiming at … al.[9] utilized the relationship between the words to calculate the semantic similarity of the … The methods include indicator function, cosine similarity, dot-product and so on …

Word similarity datasets for Thai: Construction and evaluation
P Netisopakul, G Wohlgenannt, A Pulich – IEEE Access, 2019 – ieeexplore.ieee.org
… However, the performance in NLP downstream tasks such as dialogue systems or document classification depends on many factors … [28] study semantic similarity between Turkish … The similarity of word pairs is computed as the cosine similarity of the corresponding word vectors …

Towards best experiment design for evaluating dialogue system output
S Santhanam, S Shaikh – arXiv preprint arXiv:1909.10122, 2019 – arxiv.org
… Coherence measures the ability of the dialogue system to produce responses consistent with the topic of … This metric computes the cosine similarity on embedding vectors of gener- ated response and target while … The penalized semantic similarity (SS) score is then calculated as …

Investigating the effects of word substitution errors on sentence embeddings
R Voleti, JM Liss, V Berisha – ICASSP 2019-2019 IEEE …, 2019 – ieeexplore.ieee.org
… goal is to develop sentence embeddings that can successfully model the semantic similarity between two … simi- larity, closely related to dS ij in Equation 2. Cosine similarity is given … Lee, “An integrated dialog simulation technique for evaluating spoken dialog systems,” in Coling …

Hierarchical attention and knowledge matching networks with information enhancement for end-to-end task-oriented dialog systems
J He, B Wang, M Fu, T Yang, X Zhao – IEEE Access, 2019 – ieeexplore.ieee.org
… The task-oriented dialog systems were equipped with chatting capability by inserting … It can be formulated as a ranking problem based on semantic matching given a pre- defined entity list E. If a word is an entity, it must have high cosine similarity (close to 1.0) between itself and …

Sentence mover’s similarity: Automatic evaluation for multi-sentence texts
E Clark, A Celikyilmaz, NA Smith – … of the 57th Annual Meeting of the …, 2019 – aclweb.org
Page 1. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 2748–2760 Florence, Italy, July 28 – August 2, 2019. c 2019 Association for Computational Linguistics 2748 Sentence …

Chatbol, a chatbot for the Spanish “La Liga”
C Segura, À Palau, J Luque, MR Costa-Jussà… – … Dialogue System …, 2019 – Springer
… Interactive System (IRIS) approach [3]. In this type of systems, semantic similarity scores at … The chat-oriented dialogue system is inspired on the vector space model framework [20, 24 … Cosine similarity is employed to compute scores among both previous and current state vectors …

Towards Simple but Efficient Next Utterance Ranking
BEA Boussaha, N Hernandez, C Jacquin… – Computación y …, 2019 – cys.cic.ipn.mx
… 2. Architecture of our multi-level context response matching dialogue system … which learns a representation of the chronological dependency and the semantic similarity between the … Then, a cosine similarity is computed between the context and the response vectors and used …

Car FAQ Assistant Based on BILSTM-Siamese Network
B Jin, Z Jin – Open Access Library Journal, 2019 – scirp.org
… China’s first human-computer dialogue system was invented in 1980 by the Research Institute of the Chinese Academy of Sciences … 4. Semantic Similarity Model Based on BILSTM-Siamese. 4.1 … Finally, we use cosine similarity to calculate the similarity of advanced features …

Adversarial training and decoding strategies for end-to-end neural conversation models
T Hori, W Wang, Y Koji, C Hori, B Harsham… – Computer Speech & …, 2019 – Elsevier
… An advanced end to end conversation system for the 6-th edition of Dialog System Technology Challenge (DSTC6). • … In our prior work, we built several end-to-end conversation systems for the 6th Dialog System Technology Challenges (DSTC6) Twitter help-desk dialog task …

MOLI: Smart Conversation Agent for Mobile Customer Service
G Zhao, J Zhao, Y Li, C Alt, R Schwarzenberg… – Information, 2019 – mdpi.com
… [19] to handle the ambiguity and variability of linguistic expression when modeling sentence semantic similarity … Although question answering and dialog systems recently received a lot of attention, work combining both is still sparse. 3. Conversational Question Answering. 3.1 …

Distributional Models with Syntactic Contexts for the Measurement of Word Similarity in Brazilian Portuguese
EE Berlitz, DA Araujo, AB Silva, RR Righi… – Journal of Computer …, 2019 – researchgate.net
… identifying word similarity can help in many NLP tasks such as dialogue systems, question answering … up using the PT65 as a gold-standard for evaluating semantic similarity and relatedness … each word pair of PT65 dataset, the expected result with the Cosine similarity given by …

Transfer Hierarchical Attention Network for Generative Dialog System
X Zhang, Q Yang – International Journal of Automation and Computing, 2019 – Springer
Page 1. Transfer Hierarchical Attention Network for Generative Dialog System Xiang Zhang Qiang Yang … Abstract: In generative dialog systems, learning representations for the dialog context is a crucial step in generating high quality re- sponses …

Machine Reading Comprehension for Answer Re-Ranking in Customer Support Chatbots
M Hardalov, I Koychev, P Nakov – Information, 2019 – mdpi.com
… We evaluate the model using both semantic similarity measures, as well as word-overlap ones such as BLEU … (5). The vector extrema [46] was proposed for dialogue systems … vectors for the chatbot output and for the gold human answer are compared using the cosine similarity …

Sigmoid similarity-a new feature-based similarity measure
S Likavec, I Lombardi, F Cena – Information Sciences, 2019 – Elsevier
… Semantic similarity can be employed in many areas, such as text mining, dialogue systems, Web page retrieval, image retrieval from the Web, machine translation, ontology mapping, word-sense disambiguation, and item recommendation, to name just a few …

Model comparison for semantic grouping
F Vargas, K Brestnichki, N Hammerla – arXiv preprint arXiv:1904.13323, 2019 – arxiv.org
… cosine similarity — a choice stemming only from the fact that cosine similarity gives good … a new research direction that could help improve the understanding of semantic similarity by allowing … practical and key to use-cases that involve information retrieval in dialogue systems …

Deep learning for spoken dialogue systems: application to nutrition
MB Korpusik – 2019 – dspace.mit.edu
Page 1. Deep Learning for Spoken Dialogue Systems: MAACrHUSETTSZ 11NT1TUTE: Application to Nutrition r ECHNLOGY by JUN 13 201. Mandy Barrett Korpusik jA E … 2 Page 3. Deep Learning for Spoken Dialogue Systems: Application to Nutrition by Mandy Barrett Korpusik …

Zero Shot Intent Classification Using Long-Short Term Memory Networks.
K Williams – INTERSPEECH, 2019 – pdfs.semanticscholar.org
… an utter- ance U and an intent Ij , where the cosine similarity is given by … to produce meaningful features that can be used for computing semantic similarity and we are … tion of Language Understanding Impact for Reinforcement Learn- ing Based Dialogue Systems,” arXiv preprint …

Generating Dialogue Responses From A Semantic Latent Space
WJ Ko, A Ray, Y Shen, H Jin – 2019 – openreview.net
… Each sentence is treated independently, and the model is unable to use semantic similarity between different utterances (Qiu et al., 2019 … the following automatic evaluation metrics: (1) BLEU-1 and BLEU-2 (Papineni et al., 2002) (2) Embedding Average cosine similarity (Foltz et …

Towards coherent and engaging spoken dialog response generation using automatic conversation evaluators
S Yi, R Goel, C Khatri, A Cervone, T Chung… – arXiv preprint arXiv …, 2019 – arxiv.org
… architectures serve as the basis of state-of-the-art ap- proaches in end-to-end open domain dialog systems … In MT, given a source sentence, the correctness of the target sentence can be measured by semantic similarity to the … Response Similarity: Cosine similarity be- tween …

Hierarchical reinforcement learning for open-domain dialog
A Saleh, N Jaques, A Ghandeharioun, JH Shen… – arXiv preprint arXiv …, 2019 – arxiv.org
… Semantic Similarity: Paraphrasing and style matching are important in facilitating good conversation … Therefore, we reward the cosine similarity between the simulated user and bot … Toxicity: Open-domain dialog systems generate mali- cious, offensive, and biased language …

Machine learning based review on Development and Classification of Question-Answering Systems
S Uttarwar, S Gambani, T Thakkar… – 2019 3rd International …, 2019 – ieeexplore.ieee.org
… For OBTM (ontology based text mining) tools like Chimacra, KAON, Protégé were used to get semantic similarity … B. Su et. al [23] proposes improved TF-IDF weights and use of cosine similarity with significantly improved accuracy compared to other algorithms …

Conversational Help for Task Completion and Feature Discovery in Personal Assistants
MG Jhawar, V Vangala, N Sharma… – arXiv preprint arXiv …, 2019 – arxiv.org
… environ- ments and acquiring tourist information by combining spoken dialogue system, question-answering … of fetching appropriate response, we used DSSM based semantic similarity between queries … We use cosine similarity between the DSSM features (DSSM model used is …

Multi-Granularity Representations of Dialog
S Mehri, M Eskenazi – arXiv preprint arXiv:1908.09890, 2019 – arxiv.org
… Given the LSTM response encoder, fr, the mea- sure of semantic similarity is defined as: ri = fr(Ri,j) j ? [1,Mi] (9) rk = fr(Rk,j) j ? [1,Mk] (10) d(Ri,Rk) = ri T rk ||ri|| · ||rk|| (11) This approach relies on a cosine-similarity as a measure of semantic distance between dialog ut- terances …

Financial News Generation Based on Artificial Intelligence Technology
S Cao, Y Yue – 2019 2nd International Conference on Safety …, 2019 – ieeexplore.ieee.org
… try different method, such as the reference to the latest research related areas such as dialogue system, machine translation … ?? as a long sequence, we measure the similarity between ?? and ?? via a semantic similarity algorithm called Cosine similarity measurement …

Unsupervised paraphrasing by simulated annealing
X Liu, L Mou, F Meng, H Zhou, J Zhou… – arXiv preprint arXiv …, 2019 – arxiv.org
… Mck- eown 1983), information retrieval (Knight and Marcu 2000), and dialogue systems (Shah et … In addition, we define our searching objective involving not only semantic similarity and language … find the closest word in the candidate para- phrase x? in terms of cosine similarity …

Extending a model for ontology-based Arabic-English machine translation
NA Dahan, FM Ba-Alwi – arXiv preprint arXiv:1902.02326, 2019 – arxiv.org
… In this way, a semantic similarity function will be used to evaluate the resulted translation by NAN, Google, and Bing in … Moussallem, et al [10] developed a model for Portuguese-English MT to be used in the dialogue systems … Equation 2: the Cosine Similarity Equation …

Statement Analysis using NLP
MS Vinu, P Mohan, SG Moorthy, M Gobinath – 2019 – ijsret.com
… At a given time, the system will return the best response according to semantic similarity between user input and the dialogue examples in … (2) Given a query, this cosine similarity is computed … Lubis Et Al.: Positive Emotion Elicitation In Chat- Based Dialogue Systems 869 scores …

Semantic representations for under-resourced languages
J Mazarura, A de Waal, P de Villiers – … of the South African Institute of …, 2019 – dl.acm.org
… Cosine similarity [28] provides a measure of simi- larity for word2vec embeddings … One would expect a very high semantic similarity between these 2000 sentences and a lower … Intent classification, which is used in task-driven dialogue systems and sentiment classification …

Constructing Uyghur Commonsense Knowledge Base by Knowledge Projection
A Anwar, X Li, Y Yang, Y Wang – Applied Sciences, 2019 – mdpi.com
… Finally, the method encodes the source and target sentence in the same semantic space using the bidimensional attention network, and calculates the distance between them to get the semantic similarity score … 3.4. Bilingual Semantic Similarity Scoring Model …

Enhancing generative conversational service agents with dialog history and external knowledge
Z Wang, Z Wang, Y Long, J Wang, Z Xu… – Computer Speech & …, 2019 – Elsevier
… The proposed dialog systems are first evaluated automatically using several word-overlap based metrics … Then, metrics based on word embeddings that compute the semantic similarity of words … based on word embeddings, and then calculate the cosine similarity between the …

ProductQnA: Answering user questions on e-commerce product pages
A Kulkarni, K Mehta, S Garg, V Bansal… – … Proceedings of The …, 2019 – dl.acm.org
… [22] recently presented a task-oriented dialog system that leverages … applied to the question answering task, some of the existing work is based on the semantic similarity of a … an }, the similarity-based rank- ing model fdeep ranks the answers based on their cosine similarity cos …

Learning to Memorize in Neural Task-Oriented Dialogue Systems
CS Wu – arXiv preprint arXiv:1905.07687, 2019 – arxiv.org
… by the user for all the slots in all the domains. 23 3.3 The embeddings cosine similarity visualization for (a) slots and (b) (domain, slot) pairs. 26 … 38 5.1 The proposed Mem2Seq architecture for task-oriented dialogue systems. (a) …

Improving the Processing of Question Answer Based Legal Documents.
S Chakravarty, M Mehrotra, RVSP Chava, H Liu… – JURIX, 2019 – books.google.com
… a pair of sentences to their vector space representations and measuring their cosine-similarity can be … in 8 of the 10 DA combinations and had the best semantic similarity scores in … J. Williams, A belief tracking challenge task for spoken dialog systems, in: NAACL-HLT Workshop …

Processing Open Text Input in a Scripted Communication Scenario
FPM Heemskerk – 2019 – dspace.library.uu.nl
… We compare each player statement option vector with a player input statement by calculating a cosine similarity score.The highest cosine similarity score of a player statement option vector and a player input statement is considered as an possible match … 14 Cosine similarity …

Driven answer generation for product-related questions in e-commerce
S Chen, C Li, F Ji, W Zhou, H Chen – … on Web Search and Data Mining, 2019 – dl.acm.org
… Many applications such as dialogue systems [2, 11, 13, 25, 26, 32, 35], doc- ument … Mover’s Distance (WMD) [8] is an effective metric to measure the semantic similarity between two … Instead of relying on keyword matching like cosine similarity, WMD attempts to find an optimal …

Open Domain Conversational Chatbot
V Deshmukh, SJ Nirmala – International Conference on Information …, 2019 – Springer
… The dataset is then searched for answer for the query using cosine similarity … of natural data and vector representations are build to capture the semantic similarity between the … 2016)Google Scholar. 3. Bart, A., Spanakis, G.: A retrieval-based dialogue system utilizing utterance …

A study of incorrect paraphrases in crowdsourced user utterances
MA Yaghoub-Zadeh-Fard, B Benatallah… – Proceedings of the …, 2019 – aclweb.org
… Also known as dialogue systems, virtual assistants, chatbots or simply bots (Campagna et al., 2017; Su et al., 2017), some advanced bots are now designed to perform complex tasks (eg, flight booking), many of which are built using machine learning tech- niques …

Asking the Right Question: Inferring Advice-Seeking Intentions from Personal Narratives
L Fu, JP Chang, C Danescu-Niculescu-Mizil – arXiv preprint arXiv …, 2019 – arxiv.org
… Second, semantic similarity is ar- guably a different concept from plausibility with respect to a narrative … To operationalize this intu- ition, we first find pairs of similar narratives based on the cosine similarity of their tf-idf representa- tions.11 A greedy search based on this similarity …

Incremental Domain Adaptation for Neural Machine Translation in Low-Resource Settings
M Kalimuthu, M Barz, D Sonntag – … of the Fourth Arabic Natural Language …, 2019 – aclweb.org
… Calculating the cosine similarity between all samples of a new domain Snew src ? Ssrc and this Page 5. 5 reference point, results in a distribution of dis- tances indicating the semantic similarity or dis- similarity of samples from the new domain to the generic domain …

CoaCor: code annotation for code retrieval with reinforcement learning
Z Yao, JR Peddamail, H Sun – The World Wide Web Conference, 2019 – dl.acm.org
… For example, the code retrieval model proposed by Gu et al. [13] uti- lized two deep neural networks to learn the vector representation of a natural language query and that of a code snippet respectively, and adopted cosine similarity to measure their matching degree …

Towards the Automatic Classification of Student Answers to Open-ended Questions
JG Alvarado Mantecon – 2019 – ruor.uottawa.ca
… entity mention embeddings based on GloVe and entity URI embeddings extracted from Wikipedia. For the reference-based classifier, we explored fourteen features: cosine similarity between … 36 3.3.2.2 Semantic Similarity Features …

Feature Engineering and Ensemble-Based Approach for Improving Automatic Short-Answer Grading Performance
A Sahu, PK Bhowmick – IEEE Transactions on Learning …, 2019 – ieeexplore.ieee.org
… The mean of the word vectors in each of the answers in a pair is obtained and cosine similarity between the pair of mean vectors is used as the semantic similarity score between the pair of student answer and model answer …

DRCoVe: An Augmented Word Representation Approach using Distributional and Relational Context
MA Parwez, M Abulaish, M Fazil – abulaish.com
… 0) = ??. To avoid this situation, positive pointwise mutual information (PPMI) has been used in which negative PMI values are mapped to zero as given in equation 4. In addition, Bullinaria and Levy (2007) showed that PPMI performs better than PMI in finding semantic similarity …

Controllable paraphrase generation with a syntactic exemplar
M Chen, Q Tang, S Wiseman, K Gimpel – arXiv preprint arXiv:1906.00565, 2019 – arxiv.org
… It can also bear on dialogue systems that seek to generate utterances that fit particular functional categories (Ke et al., 2018; Li et al., 2019 … Since word ordering is relatively unimportant for semantic similarity (Wieting et al., 2016), we as- sume it is more relevant to the syntax of a …

Nlsc: Unrestricted natural language-based service composition through sentence embeddings
OJ Romero, A Dangi, SA Akoju – 2019 IEEE International …, 2019 – ieeexplore.ieee.org
… data. Word embeddings capture the idea that is possible to express “meaning” of words using a vector, so that the cosine of the angle between the vectors captures semantic similarity (“cosine similarity” property). Sentence embeddings …

A Readiness Evaluation of Applying e-Government in the Society: Shall Citizens begin to Use it?
LT Khrais, MA Mahmoud… – Editorial Preface From …, 2019 – researchgate.net
… and apply evaluation metrics such as Bilingual Evaluation Understudy and cosine similarity … the reference response and the generated response and their semantic similarity … Modern intelligent conversational and dialogue systems draw principles from many disciplines …

Keyphrase extraction from disaster-related tweets
J Ray Chowdhury, C Caragea, D Caragea – The world wide web …, 2019 – dl.acm.org
… Often, embedding algo- rithms encode contextual information into the vector representa- tions, such that when clustering word-vectors using the Euclidean distance or Cosine-similarity, vectors corresponding to words that appear in similar contexts tend to cluster together …

Learning Outcomes and Their Relatedness in a Medical Curriculum
S Mondal, T Dhamecha, S Godbole, S Pathak… – Proceedings of the …, 2019 – aclweb.org
… RQ3: Are there any latent factors beyond curricu- lum and semantic similarity establishing related- ness among LOs … They employ a random-walk paradigm on a graph whose edges are weighed by tf-idf based cosine similarity …

Cgmh: Constrained sentence generation by metropolis-hastings sampling
N Miao, H Zhou, L Mou, R Yan, L Li – … of the AAAI Conference on Artificial …, 2019 – aaai.org
… For soft constraints, the matching function could be, for example, a measure of semantic similarity … 2016) and dialog systems (Mou et al … For any word w in a sentence x, we first find the closest word in the input sentence x? by computing their cosine similarity (Pennington, Socher …

Automated Approaches to Community Question Answering
A Uva – 2019 – eprints-phd.biblio.unitn.it
… The latter can be used to quickly bootstrap Natural Language Understanding pipelines for dialog systems. To conclude, we study advantages and disadvantages of neural networks and tree kernel models when applied to cQA tasks …

Automatic summarization of medical conversations, a review
J Lopez – TALN-RECITAL 2019-PFIA 2019, 2019 – hal.archives-ouvertes.fr
… methods for multidocument update summarization, improving it by taking into semantic similarity and document … vector space and the documents similarity is measure by cosine similarity to generate … to help the health field, such as intelligence agents and health dialog systems …

Investigating Semantic Properties of Images Generated from Natural Language Using Neural Networks
SW Schrader – 2019 – scholarworks.boisestate.edu
… v Page 6. investigate if the semantic similarity is preserved during the transformation process … term frequency-inverse document frequency, cosine similarity, word-to-vec distance, etc … tasks such as dialog systems, text analysis, automated speech-to-text transcription, and more …

Learning from Fact-checkers: Analysis and Generation of Fact-checking Language
N Vo, K Lee – Proceedings of the 42nd International ACM SIGIR …, 2019 – dl.acm.org
… Then, we measured cosine similarity between a FC-tweet and the fact-checking article embedded in the FC-tweet as shown in Figure 4(a). Interestingly, most FC-tweets do not have high similar- ity with FC-articles, suggesting that fact-checkers rarely include details from fact …

Distantly Supervised Question Parsing
H Zafar, M Tavakol, J Lehmann – 2019 – researchgate.net
… The usage of RL in various question answering as well as dialog systems has been exploited [48, 14 … Second, to support semantic similarity, we utilize GloVe vectors [28] to initial- ize the word embedding (EMB) vectors, where we decide for cosine similarity over euclidean …

Semantically aligned sentence-level embeddings for agent autonomy and natural language understanding
NE Fulda – 2019 – scholarsarchive.byu.edu
… The resulting distances correlate with many human estimations of semantic similarity … can be found directly within linguistic embeddings, and that this knowledge can be extracted and utilized for real-world applications like robot navigation and dialog systems. We further …

Learning dense representations for entity retrieval
D Gillick, S Kulkarni, L Lansing, A Presta… – arXiv preprint arXiv …, 2019 – arxiv.org
… Cosine Similarity (d) Compound Encoder … Here, g(x)=1/(1 + e?x) is the logistic function, and ah and bh are learned scalar parameters to transform the cosine similarity into a logit.3 For the hard negatives task, we track Area Under the ROC curve (AUC) on a held out set of pairs …

ScratchThat: Supporting Command-Agnostic Speech Repair in Voice-Driven Assistants
J Wu, K Ahuja, R Li, V Chen, J Bigham – Proceedings of the ACM on …, 2019 – dl.acm.org
… We select this encoder due to its ability to accept and generate representations for entity chunks, which are often greater than word length. The semantic similarity measure is computed with the commonly used cosine similarity metric for natural language processing tasks …

Generating Counterfactual Explanations using Reinforcement Learning Methods for Tabular and Text data
D Garcia-Olano, A Jain – pdfs.semanticscholar.org
… such as active learn- ing (Fang et al., 2017), improving dialog system models by … a procedure of synonym extraction, equal part of speech enforcement and semantic similarity checking via … and perturbed sentence into high dimen- sional vectors whose cosine similarity score can …

Emoji prediction for Hebrew political domain
C Liebeskind, S Liebeskind – … Proceedings of The 2019 World Wide Web …, 2019 – dl.acm.org
… available dataset called EmoSim508 [43], which assigns human-annotated semantic similarity scores to … [47] investigated the task of emoji recommendation in multi-turn dialogue systems … words are described by similar vectors (with respect for instance to their cosine similarity) …

Submodular optimization-based diverse paraphrasing and its effectiveness in data augmentation
A Kumar, S Bhattamishra, M Bhandari… – Proceedings of the 2019 …, 2019 – aclweb.org
… to design functions that exploit the decoder search space to maximize the semantic similarity between the … 1We find gaussian rbf to work better than other similarity metrics such as cosine similarity … have focused on the task of producing diverse responses in dialog systems (Li et …

FAQ-based Question Answering via Knowledge Anchors
R Xie, Y Lu, F Lin, L Lin – arXiv preprint arXiv:1911.05930, 2019 – arxiv.org
… cal compared to those of information retrieval or dialogue systems … Finally, with helps of knowl- edge anchors, query-document matching module measures the semantic similarity between queries and documents … We use the cosine similarity to build the interaction matrix …

Unsupervised Common Question Generation from Multiple Documents using Reinforced Contrastive Coordinator
WS Cho, Y Zhang, S Rao, A Celikyilmaz… – arXiv preprint arXiv …, 2019 – arxiv.org
… The generated questions are also evaluated with respect to the human reference questions, in terms of BLEU (Papineni et al., 2002), ROUGE (Lin and Hovy, 2003), METEOR (Banerjee and Lavie, 2005), CIDEr (Vedantam et al., 2015), Skip- Thought cosine similarity (Kiros et al …

Deep learning based recommender system: A survey and new perspectives
S Zhang, L Yao, A Sun, Y Tay – ACM Computing Surveys (CSUR), 2019 – dl.acm.org
Page 1. 5 Deep Learning Based Recommender System: A Survey and New Perspectives SHUAI ZHANG and LINA YAO, University of New South Wales AIXIN SUN and YI TAY, Nanyang Technological University With the growing …

Text Classification With Deep Neural Networks
T Huynh – 2019 – oro.open.ac.uk
… word spans without explicit training labels. In the future I propose the learned representations to be used with the discussed Deep Neural Net- works in different NLP tasks such as Dialog Systems, Machine Translation or Natural Language Inference. Page 4. Contents …

Learning Dense Representations for Entity Retrieval
A Presta, D Gillick, D Garcia-Olano, E Ie, J Baldridge… – 2019 – research.google
… Cosine Similarity (d) Compound Encoder … Here, g(x)=1/(1 + e?x) is the logistic function, and ah and bh are learned scalar parameters to transform the cosine similarity into a logit.3 For the hard negatives task, we track Area Under the ROC curve (AUC) on a held out set of pairs …

ELG: An Event Logic Graph
X Ding, Z Li, T Liu, K Liao – arXiv preprint arXiv:1907.08015, 2019 – arxiv.org
… To this end, we pro- pose learning distributed representations for each event, and utilize the cosine similarity to measure the semantic similarity between two event … We can also incorporate ELG into dialog systems to ensure that the auto-reply answers are more logical …

Exploring improvisational approaches to social knowledge acquisition
D Feng, E Carstensdottir, MS El-Nasr, S Marsella – 2019 – eprints.gla.ac.uk
… Semantic continuity is indispens- able for coherent text. In this work, the semantic similarity was evaluated using the average cosine similarity of adjacent sentence vectors. The vectors were obtained using GloVe word embedding trained on Wikipedia 2014, and Gigaword 5[34] …

Machine Learning from Casual Conversation
A Mohammed Ali – 2019 – stars.library.ucf.edu
Page 1. University of Central Florida STARS Electronic Theses and Dissertations, 2004-2019 2019 Machine Learning from Casual Conversation Awrad Mohammed Ali University of Central Florida Part of the Computer Sciences …

Topological Representation of Text for Entailment
K Savle – 2019 – search.proquest.com
… on specialized tasks such as, legal data, dialogue systems and larger documents etc. In this thesis … 41 TABLE 5-5 Cosine similarity vs Topological representation … 1.3 Contributions Similarity and Relevance: Semantic similarity and relevance of two texts are used to establish …

Learning to Annotate: Modularizing Data Augmentation for TextClassifiers with Natural Language Explanations
Z Wang, Y Qin, W Zhou, J Yan, Q Ye, L Neves… – arXiv preprint arXiv …, 2019 – arxiv.org
… (1) CBOW-GloVe uses bag-of-words (Mikolov et al., 2013) on GloVe embeddings (Pennington et al., 2014) to represent an instance, or surface patterns in NL explanation. It then annotates the sentence with the label of its most similar surface pattern (by cosine similarity) …

A Home Service-Oriented Question Answering System With High Accuracy and Stability
M Zhang, G Tian, Y Zhang – IEEE Access, 2019 – ieeexplore.ieee.org
… have been taken to penalize semantic similarity between consecutive sentences from answers. Both hpi and hpi+1 are hidden vectors representing consecutive sentences generated from the proposed model, so the reward is the negative log of the cosine similarity between two …

Learning Representations for Text Classification of Indian Languages
N CHOUDHARY – 2019 – web2py.iiit.ac.in
… Dialogue Systems: To narrow down the possible answers to a sentence based on the previous dialogues … The contrastive loss function is based on a similarity metric (cosine similarity). The architecture takes a pair of sentences as input …

Semantic and Discursive Representation for Natural Language Understanding
D Sileo – 2019 – tel.archives-ouvertes.fr
Page 1. HAL Id: tel-02619733 https://tel.archives-ouvertes.fr/tel-02619733 Submitted on 25 May 2020 HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not …

Cross-lingual word embeddings
A Søgaard, I Vuli?, S Ruder… – Synthesis Lectures on …, 2019 – morganclaypool.com
… 2015 Semantic Similarity from Natural Language and Ontology Analysis Sébastien Harispe, Sylvie Ranwez, Stefan Janaqi, and Jacky Montmain 2015 … 2010 Spoken Dialogue Systems Kristiina Jokinen and Michael McTear 2009 …

Enriching Scientific Paper Embeddings with Citation Context
K Henner – 2019 – digital.lib.washington.edu
… (2015). semantic similarity across the citing and cited papers, the cited paper’s overall citation count … for example, by simple vector operations such as dot product or cosine similarity. Early methods such as Latent Semantic Analysis (LSA) (Landauer and Dumais, 1997) …

Deep Recommendation Systems
S Chiusano, L Likforman-Sulem, D Gallitelli – webthesis.biblio.polito.it
Page 1. POLITECNICO DI TORINO Corso di Laurea Magistrale in Ingegneria Informatica (Computer Engineering) Tesi di Laurea Magistrale Deep Recommendation Systems Relatori: prof. Silvia Chiusano (Politecnico di Torino) prof …

Text mining in education
R Ferreira?Mello, M André, A Pinheiro… – … : Data Mining and …, 2019 – Wiley Online Library
Abstract The explosive growth of online education environments is generating a massive volume of data, specially in text format from forums, chats, social networks, assessments, essays, among other…

Emotion-Aware and Human-Like Autonomous Agents
N Asghar – 2019 – uwspace.uwaterloo.ca
… A prime example is a dialogue system, where the agent should converse fluently and coher- ently with a user and connect with them emotionally … 19 2.4.9 Conditional Variational Autoencoders . . . . . 21 2.5 A Brief History of Dialogue Systems …

Hybrid Deep Question Answering
A Aghaebrahimian – 2019 – dspace.cuni.cz
… not a trivial task. They also offer an opportunity, because they pave the way for developing a wide array of useful applications such as dialogue systems, tutoring systems, scientist’s assistants, etc … measures which are used as semantic similarity metrics in further chapters …

Recognition and extraction of named entities in online medical diagnosis data based on a deep neural network
X Liu, Y Zhou, Z Wang – Journal of Visual Communication and Image …, 2019 – Elsevier
… It has also been found that word vectors obtained by a trained language model contain the semantic information of the vocabulary, and some algorithms can also be used to acquire semantic similarity and other data to some extent …

Towards a Self-aware Intelligent Agent
M Morisio, O Isabeau – 2019 – webthesis.biblio.polito.it
… Remembering that we are talking about conversational agent, the similar- ity function that is reported below in the formulas is a similarity function based on semantic similarity, that in our case exploits the power of word … 1.1 Dialogue system architecture … 22 2.5 Cosine similarity …

An automatic literature knowledge graph and reasoning network modeling framework based on ontology and natural language processing
H Chen, X Luo – Advanced Engineering Informatics, 2019 – Elsevier
… Along with the rapid development of information and computation technology, NLP has played significant roles in various applications, including conversational agent or dialogue systems [29], machine translation [30], knowledge mining and reasoning [31], search engine [32 …

Stylistic Control for Neural Natural Language Generation
S Oraby – 2019 – escholarship.org
Page 1. UC Santa Cruz UC Santa Cruz Electronic Theses and Dissertations Title Stylistic Control for Neural Natural Language Generation Permalink https://escholarship.org/uc/item/54p9r87q Author Oraby, Shereen Publication Date 2019 Peer reviewed|Thesis/dissertation …

Response Retrieval in Information-seeking Conversations
L Yang – 2019 – scholarworks.umass.edu
Page 1. University of Massachusetts Amherst ScholarWorks@UMass Amherst Doctoral Dissertations Dissertations and Theses 2019 Response Retrieval in Information-seeking Conversations Liu Yang College of Information and Computer Sciences, UMass Amherst …

A sequential matching framework for multi-turn response selection in retrieval-based chatbots
Y Wu, W Wu, C Xing, C Xu, Z Li, M Zhou – Computational Linguistics, 2019 – MIT Press
… academia. Existing conversational agents can be categorized into task-oriented dialog systems and non–task-oriented chatbots. Dialog systems focus on helping people complete specific tasks in vertical domains (Young et al …

Study of Argument Structure in Parliamentary Debates
SVK Rohit – 2019 – web2py.iiit.ac.in
Page 1. Study of Argument Structure in Parliamentary Debates Thesis submitted in partial fulfillment of the requirements for the degree of Masters of Science in Exact Humanities by Research by Sakala Venkata Krishna Rohit 201356204 rohit.sakala@research.iiit.ac.in …

Neural machine reading comprehension: Methods and trends
S Liu, X Zhang, S Zhang, H Wang, W Zhang – Applied Sciences, 2019 – mdpi.com
Machine reading comprehension (MRC), which requires a machine to answer questions based on a given context, has attracted increasing attention with the incorporation of various deep-learning techniques over the past few years. Although research on MRC based on deep …

Intelligent Asset Management
F Xing, E Cambria, R Welsch – 2019 – Springer
… 115 7.2 Robo-Advisory and Dialog System … L lag operator L sentiment lexicon M iteration times O time complexity P asset mentioning matrix Q subjective expected returns r probability distribution for returns R asset return R the set of real numbers s cosine similarity S semantic …

Natural Language Processing and Chinese Computing: 8th CCF International Conference, NLPCC 2019, Dunhuang, China, October 9–14, 2019 …
J Tang, MY Kan, D Zhao, S Li, H Zan – 2019 – books.google.com
Page 1. Jie Tang· Min-Yen Kan · Dongyan Zhao · Sujian Li· Hongying Zan (Eds.) Natural Language Processing and Chinese Computing 8th CCF International Conference, NLPCC 2019 Dunhuang, China, October 9–14, 2019 Proceedings, Part I 123 Page 2 …

Representation Learning for Information Extraction
E Amjadian – 2019 – curve.carleton.ca
Page 1. Representation Learning for Information Extraction by Ehsan Amjadian A thesis submitted to the Faculty of Graduate and Postdoctoral Affairs in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Cognitive Science Carleton University …

Deep learning based chatbot models
R Csaky – arXiv preprint arXiv:1908.08835, 2019 – arxiv.org
… and the goal is to build vector representations that capture the semantic similarity between words … a neural network [Cho et al., 2014] or by simply computing the cosine similarity between the … have also been approaches that integrate the two types of dialog systems by comparing …

Improving Software Defect Assignment Accuracy with the LSTM and Rule Engine Model
R Zhu – 2019 – search.proquest.com
Page 1. Improving Software Defect Assignment Accuracy With the LSTM and Rule Engine Model by Robert Zhu, BE, MS, MAS Submitted in partial fulfillment of the requirements for the degree of Doctor of Professional Studies in Computing at …

Rich and Scalable Models for Text
J Boyd-Graber, P Resnik – 2019 – drum.lib.umd.edu
Page 1. ABSTRACT Title of dissertation: RICH AND SCALABLE MODELS FOR TEXT Thang Dai Nguyen, Doctor of Philosophy, 2019 Dissertation directed by: Professor Jordan Boyd-Graber Department of Computer Science and Institute for Advanced Computer Studies …

Rich and Scalable Models for Text
T Dai Nguyen – 2019 – search.proquest.com
Page 1. ABSTRACT Title of dissertation: RICH AND SCALABLE MODELS FOR TEXT Thang Dai Nguyen, Doctor of Philosophy, 2019 Dissertation directed by: Professor Jordan Boyd-Graber Department of Computer Science and Institute for Advanced Computer Studies …

Recurrent neural networks for structured data
TTM Pham – 2019 – dro.deakin.edu.au
… color) . . . . 124 6.5 (Left) Pairwise cosine similarity matrix of label embedded vectors, (Right) Pairwise correlation matrix of 101 labels for MediaMill dataset. Values are normalized so both matrices are in the same range [0, 1]. . 125 …