Natural Language & Word Spans

Notes:

Word spans are continuous sequences of words in a text. In natural language processing, word spans are often used to represent the entities or concepts that are mentioned in a text. For example, in the sentence “I went to the store to buy some milk and bread,” the word span “store” would represent the entity of a store, and the word span “milk and bread” would represent the entities of milk and bread.

Word spans are commonly used in natural language processing tasks such as named entity recognition and coreference resolution. In named entity recognition, the goal is to identify and classify the entities mentioned in a text, such as people, locations, and organizations. By representing these entities as word spans, algorithms can more easily identify and classify them. In coreference resolution, the goal is to determine which entities in a text refer to the same real-world entities. By representing entities as word spans, algorithms can more easily identify when multiple entities refer to the same real-world entity.

  • Corpora processing refers to the process of analyzing and manipulating a corpus, which is a large collection of text data. This can include tasks such as tokenization, stemming, and part-of-speech tagging.
  • Neural NLP model refers to a natural language processing model that is based on neural networks. These models are trained on large amounts of text data and are able to perform tasks such as language translation, text summarization, and sentiment analysis.
  • Modeling word spans refers to the process of identifying and analyzing the relationships between words in a text. This can include tasks such as identifying named entities, analyzing syntactic structure, and identifying the roles of different words in a sentence.

Resources:

Wikipedia:

References:

See also:

ConcGrams | Sentence Processing

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 …

Exploring the direction of collocations in eight languages
RW Todd – Canadian Journal of Linguistics/Revue canadienne de …, 2019 – cambridge.org
… collocations may be required to have a non-literal meaning, word spans to identify … meaning or word types are made in identifying collocations, and the word span is usually … English is the most researched lan- guage in reading research, natural language processing, pragmatics …

Segtree transformer: Iterative refinement of hierarchical features
Z Ye, Q Guo, Q Gan, Z Zhang – ICLR 2019 Workshop on” …, 2019 – assets.amazon.science
… Once we map the nodes to word spans, we reconnect the nodes in two directions … Glove: Global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), pp. 1532–1543, 2014 …

EANet: Enhanced Attention Network for Question Answering System
I Song – pdfs.semanticscholar.org
… Reading comprehension is one of the important NLP (natural language processing) tasks that can assist human in many applications areas … We inspect failed predictions manually, and here are two cases: i) Imprecise word span and ii) ambiguous answer in contextual meaning …

A k-Nearest Neighbor Approach towards Multi-level Sequence Labeling
Y Chen, J Chen – Proceedings of the 2019 Conference of the North …, 2019 – aclweb.org
… For over fifteen years TiMBL has been mostly used in natural language processing as a machine learning classifier component … We assume any consecutive N- word span in one utterance can be labeled as one Page 3. 151 Figure 2: Annotation on the Three Levels …

Improving neural entity disambiguation with graph embeddings
Ö Sevgili, A Panchenko, C Biemann – … of the 57th Annual Meeting of the …, 2019 – aclweb.org
… entity explanations (long abstracts), and multi- word spans … The second bar presents the results of the input combination, context, word/span, and long abstract … In Proceed- ings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Com …

A Syntax-aware Multi-task Learning Framework for Chinese Semantic Role Labeling
Q Xia, Z Li, M Zhang – arXiv preprint arXiv:1911.04641, 2019 – arxiv.org
… Semantic role labeling (SRL) is a fundamental and important task in natural language processing (NLP), which aims to identify the semantic struc … The first is called span-based SRL, which employs a continuous word span as a semantic role and follows the manual annotations …

An improved TextRank keywords extraction algorithm
S Pan, Z Li, J Dai – Proceedings of the ACM Turing Celebration …, 2019 – dl.acm.org
… 2012. A Method of Automatic Keyword Extraction Based on Word Span. Modern Property Management 11, 04 (2012), 108–111. [7] Zhang Jin … 2004. Textrank: Bringing order into text. In Proceedings of the 2004 conference on empirical methods in natural language processing …

Identifying the structure of students’ explanatory essays
MA Britt – reed.cs.depaul.edu
… In this paper, we attempt to determine the optimal natural language processing (NLP) techniques for identifying conceptual information and causal relations in … The brat tool [23, 24] was used to annotate word spans as concepts and explicit connections between them as causal …

BiPaR: A Bilingual Parallel Dataset for Multilingual and Cross-lingual Reading Comprehension on Novels
Y Jing, D Xiong, Y Zhen – arXiv preprint arXiv:1910.05040, 2019 – arxiv.org
… reading comprehension is to evaluate how well computer systems understand natural language texts, where … Fill in entity Fill in word Span of words Span of words Manual summary … from nearly 4K bilingual parallel novel passages with consecutive word spans from these …

Identifying the Structure of Students’ Explanatory Essays
S Hughes, P Hastings, MA Britt – International Conference on Artificial …, 2019 – Springer
… In this paper, we attempt to determine the optimal natural language processing (NLP) techniques for identifying conceptual information and causal relations in … The brat tool [23, 24] was used to annotate word spans as concepts and explicit connections between them as causal …

cognitive and Behavioural Weaknesses in children with Reading Disorder and AD (H) D
S Turker, A Seither-Preisler, SM Reiterer… – Scientific Reports, 2019 – nature.com
… While children with AD(H)D performed similar to typically developing children on all tasks, RD children performed weakly on various language learning and working memory tasks, with major deficits in non-word span, phonetic memory and vocabulary learning …

Auditory cortex morphology predicts language learning potential in children and teenagers
S Turker, SM Reiterer, P Schneider… – Frontiers in …, 2019 – frontiersin.org
… a syllable database developed according to German phonotactic rules (eg, “knoll,” “pflax,” “bamp”) at the Institute of Natural Language Processing, University … The Hindi score, considered a measure of non-word span, was only linked to vocabulary learning (LLAMA B, r = 0.323 …

Story ending selection by finding hints from pairwise candidate endings
M Zhou, M Huang, X Zhu – IEEE/ACM Transactions on Audio …, 2019 – ieeexplore.ieee.org
… Though attention mechanism is effective to highlight supporting evidence in the context, it suffers from the evidence bias issue because a wrong ending can also match word spans in the story context. Aakanksha [31] proposed stress tests for natural language inference (NLI or …

Better modeling of incomplete annotations for named entity recognition
Z Jie, P Xie, W Lu, R Ding, L Li – Proceedings of the 2019 Conference of …, 2019 – aclweb.org
… Kim Sang, 2002; Tjong Kim Sang and De Meul- der, 2003) as one of the most fundamental tasks within natural language processing (NLP … In practice, annotators are typically in- structed to annotate named entities for complete word spans only (Settles et al., 2008; Surdeanu et al …

Complex Word Identification as a Sequence Labelling Task
S Gooding, E Kochmar – Proceedings of the 57th Annual Meeting of the …, 2019 – aclweb.org
… The original data includes the annotation for a selected set of content words, which is provided alongside the full sentence and the word span … In Pro- ceedings of the 2018 Conference on Empirical Meth- ods in Natural Language Processing (EMNLP), pages 3749–3760 …

Neural Semantic Role Labeling using Verb Sense Disambiguation.
D Alfano, R Abbruzzese, D Cappetta – CLiC-it, 2019 – eustema.it
… Therefore, there are many sub-tasks for natural language applications that have already been stud- ied … The model makes independent decisions about what relationship, if any, holds between ev- ery possible word-span pair, and learns contextual- ized span representations …

IS THERE A DICHOTOMY BETWEEN SYNTHETIC COMPOUNDS AND PHRASES IN THAI?
K Hongthong, K Thepkanjana… – Taiwan Journal of …, 2019 – tjl.nccu.edu.tw
… Group A consists of seven NV(P) strings with the strongest collocations, each of which exhibits a semantically different relation. Group B is a similar set of NV(P)s to Group A, but they feature interventions, coordinations, and modifications or alterations within a 5-word span …

Resolving Gendered Ambiguous Pronouns with BERT
M Ionita, Y Kashnitsky, K Krige, V Larin… – arXiv preprint arXiv …, 2019 – arxiv.org
… This is an important task for natural language understanding and a nec- essary component of machine translation sys- tems, chat bots and assistants … she” refers to Julia, it also correctly clusters together two men- tions of “John” and detects that Mary Hendriks is a two-word span …

Improving generalization in coreference resolution via adversarial training
S Subramanian, D Roth – arXiv preprint arXiv:1908.04728, 2019 – arxiv.org
… In coreference resolution, the goal is to find and cluster phrases that refer to entities. We use the word “span” to mean a series of consecutive words. A span that refers to an entity is called a mention … 2018. Generating natural language adversarial ex- amples …

Automated organ-level classification of free-text pathology reports to support a radiology follow-up tracking engine
JM Steinkamp, CM Chambers, D Lalevic… – Radiology: Artificial …, 2019 – pubs.rsna.org
… The two general families of natural language processing algorithms are rule based and statistical ( 2 ). Many existing clinical systems rely on … might incorporate the overall interpretability algorithms into the overall system by auto-populating the most salient word spans from new …

Multidocument Abstractive Summarization using Abstract Meaning Representation for Indonesian Language
V Severina, ML Khodra – 2019 International Conference of …, 2019 – ieeexplore.ieee.org
… Guided Natural Language Generation has not been applied to this paper because it is more focused on Indonesian AMR graph constructions … The concept from each node will be arranged from the most frequent aligned word span of the corresponding concept …

Ambiguity in Explicit Discourse Connectives
B Webber, R Prasad, A Lee – … of the 13th International Conference on …, 2019 – aclweb.org
… As with both usage and sense ambiguity, it would be useful to determine whether syntactic features might help distinguish whether a particular multi-word span should be analyzed as a single connective or … In Proceedings, Empirical Methods in Natural Language Processing …

ÚFAL-Oslo at MRP 2019: Garage Sale Semantic Parsing
K Droganova, A Kutuzov, N Mediankin… – … on Natural Language …, 2019 – aclweb.org
… The parser was further improved for the SemEval 2016 Shared Task 8 (Flanigan et al., 2016). JAMR parser uti- lizes a rule-based aligner to match word spans in a sentence to concepts they evoke, which is applied in a pipeline before training the parser. Damonte et al …

AmazonQA: a review-based question answering task
M Gupta, N Kulkarni, R Chanda, A Rayasam… – arXiv preprint arXiv …, 2019 – arxiv.org
… of design- ing such systems, we introduce the review-based community question answering task: Given a set of product reviews and a question con- cerning a specific product, generate an informative natural language answer … The an- swers are multi-word spans from the context …

Mitigating the impact of speech recognition errors on spoken question answering by adversarial domain adaptation
CH Lee, YN Chen, HY Lee – ICASSP 2019-2019 IEEE …, 2019 – ieeexplore.ieee.org
… The task of this work is extractive SQA; that means a is a word span from the reference transcription of d. An overview … of rule-based annotators for named- entity recognition tasks,” in Proceedings of the 2010 conference on empirical methods in natural language processing …

Building a Flexible Knowledge Graph to Capture Real-World Events
L Burdick, M Wang, O Ignat, S Wilson, Y Zhang, Y Wei… – 2019 – researchgate.net
… We do not consider spans that are too long (we define this as a 15+ word span) or those that contain end-of-sentence markers, since these likely do not capture a meaningful connection between the … [2] M. Honnibal and I. Montani, “spaCy 2: Natural language understanding with …

Text Generation from Abstract Meaning Representation
L Jin – pdfs.semanticscholar.org
… 1 Introduction Tasks transforming between semantic representations and natural language usu- ally include text generation as a subroutine … Rules follow the general form of LHS fragments and argument slots yielding RHS word spans punctuated by nonterminals …

AMR-to-Text Generation with Cache Transition Systems
L Jin, D Gildea – arXiv preprint arXiv:1912.01682, 2019 – arxiv.org
… Page 3. stack cache buffer edges word span preceding action [] [$, o, c, d, f, 2} {A, t, m, y} — — [1, $] [$ c] {o, d, f, 2} {A, t, m, y} the center will Push(c, 1) … As previously mentioned we use JAMR on the gold data to align word spans and AMR concepts …

Zero-shot entity linking with dense entity retrieval
L Wu, F Petroni, M Josifoski, S Riedel… – arXiv preprint arXiv …, 2019 – arxiv.org
… (2018) have established state-of-the-art results using neu- ral networks to model context word, span and en- tity … In Proceedings of the 2017 Confer- ence on Empirical Methods in Natural Language Processing, pages 2681–2690 …

Data-to-text generation with content selection and planning
R Puduppully, L Dong, M Lapata – Proceedings of the AAAI Conference on …, 2019 – aaai.org
… 20, PTS). Wiseman et al. (2017) train an IE system on RO- TOWIRE by determining word spans which could represent entities (ie, by matching them against players, teams or cities in the database) and numbers. They then consider …

Words can shift: Dynamically adjusting word representations using nonverbal behaviors
Y Wang, Y Shen, Z Liu, PP Liang, A Zadeh… – Proceedings of the AAAI …, 2019 – aaai.org
… In addition, modeling subword information has become essen- tial for various tasks in natural language processing (Faruqui et al … Instead, they summarize the subword information during each word span using the simple averaging strategies (Liang et al. 2018; Liu et al …

Behind the dystopian sentiment: a sentiment analysis of George Orwell’s 1984
M Pavlovski – 2019 42nd International Convention on …, 2019 – ieeexplore.ieee.org
… But by and large, in order to provide more adequate conclusions on whether this applied natural language processing technique is suitable … x analysing some of the sentences was very demanding, as the necessary word span for correctly identifying different semantic concepts …

Joint type inference on entities and relations via graph convolutional networks
C Sun, Y Gong, Y Wu, M Gong, D Jiang, M Lan… – Proceedings of the 57th …, 2019 – aclweb.org
… 3.1 Entity Span Detection To extract entity spans from a sentence (Figure 2), we adopt the BILOU sequence tagging scheme: B, I, L and O denote the begin, inside, last and out- side of a target span, U denotes a single word span …

Impact of OCR Quality on Named Entity Linking
EL Pontes, A Hamdi, N Sidere, A Doucet – International Conference on …, 2019 – Springer
… These methods combine context-aware word, span and entity embeddings with neural similarity functions to analyze the context of … In: Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language …

Still a pain in the neck: Evaluating text representations on lexical composition
V Shwartz, I Dagan – Transactions of the Association for Computational …, 2019 – MIT Press
Create a new account. Email. Returning user. Can’t sign in? Forgot your password? Enter your email address below and we will send you the reset instructions. Email. Cancel. If the address matches an existing account you will …

Interactive machine comprehension with information seeking agents
X Yuan, J Fu, MA Cote, Y Tay, C Pal… – arXiv preprint arXiv …, 2019 – arxiv.org
… 2018) to benchmark a system’s ability to understand and reason over natural language … The answer is a word span defined by head and tail positions in p. NewsQA is more difficult than SQuAD because it has a larger vocabulary, more difficult questions, and longer source …

REAL MEN, TOUGH MEN, BAD MEN’: CONSTRUCTIONS OF MASCULINITY ON/R/THE_DONALD.
R Lawson – 2019 – academia.edu
… Page 25. MEN: COLLOCATIONS < Collocational analysis conducted in #LancsBox. < Mutual Information score, five-word span (L&R), 0.001 significance score. Sub-corpus Frequency Collocates men 203 times (43), create (26), trans (23), weak (19) …

Correlating neural and symbolic representations of language
G Chrupa?a, A Alishahi – arXiv preprint arXiv:1905.06401, 2019 – arxiv.org
… of neu- ral models of language are increasingly needed as deep learning becomes the dominant approach to natural language processing … where classifiers are trained to predict various lexical, syntactic and semantic relations between representation of word spans within a …

Reasoning over semantic-level graph for fact checking
W Zhong, J Xu, D Tang, Z Xu, N Duan, M Zhou… – arXiv preprint arXiv …, 2019 – arxiv.org
… Existing approaches are dom- inated by natural language inference models (Angeli and Manning 2014) because the task essentially requires match- ing between the claim and the evidence … Each node in the graph is a word span in the input text …

Capturing Greater Context for Question Generation
LA Tuan, DJ Shah, R Barzilay – arXiv preprint arXiv:1910.10274, 2019 – arxiv.org
… 5.4 Baselines As baselines, we compare our proposed model against several prior work on question generation. These include: • PCFG-Trans (Heilman, 2011): a rule-based system that generates a question based on a given answer word span …

Syntax-aware neural semantic role labeling
Q Xia, Z Li, M Zhang, M Zhang, G Fu, R Wang… – Proceedings of the AAAI …, 2019 – aaai.org
… 2004) and uses a continuous word span to be a se- mantic role … We denote the whole role set as R. Each role corre- sponds to a word span of wj…wk (1 ? j ? k ? n). Taking Figure 1 as an example,“Ms. Hag” is the A0 role of the pred- icate “plays” …

Clinical Concept Extraction for Document-Level Coding
S Wiegreffe, E Choi, S Yan, J Sun… – arXiv preprint arXiv …, 2019 – arxiv.org
… where BCE(yi, ˆyi) is the standard (binary cross- entropy) loss from the baseline for the clinical coding task, p(ci,m | ai,m) is the probability as- signed by the auxiliary model to the true cTAKES- annotated concept given word span ai,m as input, ? is the hyperparameter to tradeoff …

TWEETQA: A Social Media Focused Question Answering Dataset
W Xiong, J Wu, H Wang, V Kulkarni, M Yu… – arXiv preprint arXiv …, 2019 – arxiv.org
… Thus we consider the task of answer genera- tion for TWEETQA and we use several stan- dard metrics for natural language generation to evaluate QA systems on our dataset, namely we consider BLEU-15 (Papineni et al., 2002), Me- teor (Denkowski and Lavie, 2011) and …

Distantly supervised entity relation extraction with adapted manual annotations
C Sun, Y Wu – Proceedings of the AAAI Conference on Artificial …, 2019 – aaai.org
… into structured knowledge, the entity and re- lation extraction task attracts long lasting interests in both researches and applications of natural language processing … scheme: B, I, L and O denote the begin, inside, last and outside of a target span, U denotes a single word span …

Syntax in SMT
O Bojar – 2019 – ufal.mff.cuni.cz
… 26/69 Page 30. Treelet Alignments: Heuristics • Similar to common phrase-extraction techniques given word alignments. • Basic units are little trees instead of word spans. 1. Parse both sides of the parallel corpus. 2. Obtain node-to-node alignments (GIZA++ on linearized trees) …

RACAI’s System at PharmaCoNER 2019
R Ion, VF P?i?, M Mitrofan – Proceedings of The 5th Workshop on …, 2019 – aclweb.org
… word receiving the best label by the softmax output, accumulating labels as the window passes by. The label with the highest accumulated score wins for each word. Spans of consecutive tokens having the same non-NONE labels are the new de- tected named entities …

Multi-task learning for target-dependent sentiment classification
D Gupta, K Singh, S Chakrabarti… – Pacific-Asia Conference …, 2019 – Springer
… MTL has shown significant improvements in many fields of Natural Language Processing and Computer Vision … we do not necessarily expect our auxiliary learner to outperform more direct approaches for the auxiliary task—its goal is to supply better word/span representations to …

Extracting symptoms and their status from clinical conversations
N Du, K Chen, A Kannan, L Tran, Y Chen… – arXiv preprint arXiv …, 2019 – arxiv.org
… Speech and natural language processing are now sufficiently mature that there has been con- siderable interest, both in academia and industry, to investigate how these technologies can be ex- ploited to simplify the task of documentation, and to allow physicians to dedicate …

Improved sentiment detection via label transfer from monolingual to synthetic code-switched text
B Samanta, N Ganguly, S Chakrabarti – arXiv preprint arXiv:1906.05725, 2019 – arxiv.org
… NP and VP: We allow as candidates all subtrees rooted at NP (noun phrase) and VP (verb phrase) nonterminals, which may cover multiple words. Translating single-word spans is more likely to result in ungrammatical output [30] …

Learning to Read Academic Literature
J Wang, Y Hong, R Zhou, X Wang – 2019 – evelinehong.github.io
… Next, we use MLP to select the sentence with the highest rank as the evidence snippet. Finally, the query-aware context sequence is passed into a biLSTM-CRF model in order to extract the specific word span. with the similarity function f(h, u) = wS [h; u; h ? u] (2) …

Amazon at MRP 2019: Parsing meaning representations with lexical and phrasal anchoring
J Cao, Y Zhang, A Youssef, V Srikumar – … on Natural Language Learning, 2019 – aclweb.org
… The design and implementation of broad-coverage and linguistically motivated meaning representa- tion frameworks for natural language is attracting growing … in EDS and UCCA may align to larger overlapped word spans which involves syntactic or semantic pharsal structure …

Replicating “Extracting Action and Event Semantics from Web Text”
L Carlin – pdfs.semanticscholar.org
Page 1. Replicating “Extracting Action and Event Semantics from Web Text” Louis Carlin Abstract An important challenge in AI research is to build models which can encapsulate the behaviour of agents in a dynamically changing world …

Jointly extracting and compressing documents with summary state representations
A Mendes, S Narayan, S Miranda, Z Marinho… – arXiv preprint arXiv …, 2019 – arxiv.org
… In contrast, abstractive systems require natural language generation and semantic repre- sentation, problems that are inherently harder to solve than just … learns a system that selects textual units to include in the summary and compresses them by deleting word spans guided by …

Conceptualisation and Annotation of Drug Nonadherence Information for Knowledge Extraction from Patient-Generated Texts
A Belz, R Hoile, E Ford, A Mullick – Proceedings of the 5th Workshop on …, 2019 – aclweb.org
… (3) In order to produce the above we have to have cre- ated a suitable conceptualisation (concept model), a KE template, a KE task construal and methods for implementing it, here detecting word spans corresponding to the above concepts and for map- ping the word spans to …

SberQuAD–Russian Reading Comprehension Dataset: Description and Analysis
P Efimov, L Boytsov, P Braslavski – arXiv preprint arXiv:1912.09723, 2019 – arxiv.org
… Baselines. Contest organizers made two baselines18 avail- able. Simple baseline: The model returns a sentence with the maximum word overlap with the question. ML base- line generates features for all word spans in the sentence returned by the simple baseline …

Transductive Parsing for Universal Decompositional Semantics
E Stengel-Eskin, AS White, S Zhang… – arXiv preprint arXiv …, 2019 – arxiv.org
… 2017).1 We present the first joint UDS parser, which learns to extract both UDS graph structures and attributes from natural language input … Because both predicates and arguments can consist of multi- word spans, there can be multiple instance edges leaving a semantic node …

Discovering Classification Dimensions for Managing Scientific Resources
B Ma, H Zhuge – … on Semantics, Knowledge and Grids (SKG), 2019 – ieeexplore.ieee.org
… The word span of a word reflects the scope of influence of the word within a paper … 99 Page 4. journal published from 1994 to 2018, and the ACL-data contains 173 conference papers about Natural Language Processing from the proceedings of ACL2014 conference …

Exploiting long-term temporal dynamics for video captioning
Y Guo, J Zhang, L Gao – World Wide Web, 2019 – Springer
… However, all of these methods are highly dependent on the templates of sentences, which is insufficient to model the richness of natural language … The first LSTM encodes the visual features from pre-trained CNNs and the second LSTM generates words. Pan et al …

End-to-end neural opinion extraction with a transition-based model
M Zhang, Q Wang, G Fu – Information Systems, 2019 – Elsevier
… Fine-grained opinion extraction has received increasing interests in the natural language processing community … Assuming that the position of the next incoming (also the first) word in the queue is q 1 , and the word span (sub-entity) in the buffer is w b 1 ? w q 1 ? 1 ( b 1 is …

Encode, tag, realize: High-precision text editing
E Malmi, S Krause, S Rothe, D Mirylenka… – arXiv preprint arXiv …, 2019 – arxiv.org
Page 1. Encode, Tag, Realize: High-Precision Text Editing Eric Malmi Google Research emalmi@google.com Sebastian Krause Google Research bastik@google. com Sascha Rothe Google Research rothe@google.com Daniil …

Impairments of auditory-verbal short-term memory: Do selective deficits of the input phonological buffer exist?
T Shallice, C Papagno – Cortex, 2019 – Elsevier
JavaScript is disabled on your browser. Please enable JavaScript to use all the features on this page. Skip to main content Skip to article …

Towards information extraction from ISR reports for decision support using a two-stage learning-based approach
D Mühlenberg, A Kuwertz, P Schenkel… – … Open Business Model …, 2019 – spiedigitallibrary.org
… the relevant information from such reports, we require natural language understanding techniques (NLU – a subfield of natural language processing (NLP … Proposition Bank entries, in the first phase is done by optimizing a feature function that connects word spans (partitions of …

Personal Knowledge Base Construction from Text-based Lifelogs
AZ Yen, HH Huang, HH Chen – … of the 42nd International ACM SIGIR …, 2019 – dl.acm.org
… The issues to be tackled include (1) not all text descriptions are related to life events, (2) life events in a text description can be expressed explicitly or implicitly, (3) the predicates in the implicit events are often absent, and (4) the mapping from natural language predicates to …

Exploring the Use of Collocation in the Writing of Foundation-Year Students at King Abdulaziz University
HYY Khoja – 2019 – etheses.whiterose.ac.uk
… Page 16. 5 in this study. Nation and Shin (2008:340) indicate that the use of natural language for EFL learners is problematic, especially when language teaching focuses primarily on grammar. They illustrate this argument with examples of Korean students who …

Phoneme concatenation method considering half vowel sound for the Myanmar speech synthesis system
CS Hlaing, A Thida – International Journal of Advanced …, 2019 – search.proquest.com
… For Myanmar language, there has been considerable effort on speech processing in Myanmar natural language processing research works … The Myanmar words, -/pan/ (flower) and -/ban/ (tray) are different in the first position of sound as /p/ and /b/. Likewise, for the word -/pan …

Cognitive language aptitude
CJ Doughty – Language learning, 2019 – Wiley Online Library
… of this special issue, it is also interesting to observe that in the fields of first language (L1) acquisition and simultaneous bilingual- ism, researchers claim that natural language development, a … Sufficient mental workspace Short-term memory r Non-word Span Attention Control …

Translate and label! An encoder-decoder approach for cross-lingual semantic role labeling
A Daza, A Frank – arXiv preprint arXiv:1908.11326, 2019 – arxiv.org
… st = LSTM([st?1;yt?1;lt?1;ct]) (4) 3 Data 3.1 SRL Monolingual Datasets Two labeling schemes have been established for PropBank SRL: span-based and dependency- based. In the former, arguments are characterized as word-spans …

Learning to generate questions by learningwhat not to generate
B Liu, M Zhao, D Niu, K Lai, Y He, H Wei… – The World Wide Web …, 2019 – dl.acm.org
… performance of question generation and out-performs all pre- vious state-of-the-art neural question generation models by a substantial margin. CCS CONCEPTS • Computing methodologies ? Natural language pro- cessing; Natural language generation; Machine transla- tion …

Learning to Generate Diverse and Authentic Reviews via an Encoder-Decoder Model with Transformer and GRU
K Jin, X Zhang, J Zhang – … Conference on Big Data (Big Data), 2019 – ieeexplore.ieee.org
… II. RELATED WORK A. Fake Review Generation Review generation is an important part of natural language generation and information content security … f? is an attenuation function with parameter ?, and n ? gram means one n-word span in a sentence …

MusTIC: Research and Innovation Group on Music, Technology, Interactivity and Creativity
F Calegario, G Cabral, G Ramalho – Anais do XVII Simpósio …, 2019 – sol.sbc.org.br
… Principal working on the defini- tion of a Natural Language to Human-Machine Interaction with a many of gesture contexts (like to body, hand … way of alter- ing the pitch using a trombone slide [5]. The instrument was called Pandivá (reduction of Portuguese words “pan- deiro de …

Phonological short-term memory as a predictor for the uptake of collocations
G Boone, J Eyckmans – KONI?SKIE STUDIA J?ZYKOWE, 2019 – researchgate.net
… number of studies on formulaic sequences (FS) and collocations has steadily increased, due to their ubiquity in natural language (Er- man & … ways to assess the phonological store of working memory, and tasks such as nonword repetition, rhyme detection, word span (eg, Avons …

Interactions between item set and vocoding in serial recall
AK Bosen, MC Luckasen – Ear and hearing, 2019 – cdn.journals.lww.com
Objectives: Serial recall of digits is frequently used to measure short-term memory span in variou.

Development Considerations for Implementing a Voice-Controlled Spacecraft System
G Salazar – … Symposium on Measurement and Control in …, 2019 – ieeexplore.ieee.org
… In short, the many human intrinsic speech variations pose challenges for spacecraft applications using speech control and natural language processing … the task actions the system must do, eg, command a camera pan/tilt unit should have the vocabulary words “pan” and “tilt …

An individual?differences framework for comparing nonnative with native speakers: Perspectives from BLC theory
JH Hulstijn – Language Learning, 2019 – Wiley Online Library
… in a variety of tasks, Mulder and Hulstijn (2011) administered four speed tasks (word association, auditory and visual lexical decision, and picture naming), a paper-and-pencil productive vocabulary test, two short-term memory tasks (auditory and visual word span), and four …

The undergraduate learner translator corpus: a new resource for translation studies and computational linguistics
RF Alfuraih – Language Resources and Evaluation, 2019 – Springer
Around the world, a growing interest has been seen in learner translator corpora, which are invaluable resources for teaching and research. This paper intr.

Formulaicity of affixes in Turkish
H Badrulhisham – 2019 – summit.sfu.ca
… the ability function. Natural language processing Formulaic sequences pose a challenge for various tasks in natural language processing (NLP), where they are commonly referred to as ‘multiword expressions’. General methods …

Extraction and analysis of fictional character networks: A survey
V Labatut, X Bost – ACM Computing Surveys (CSUR), 2019 – dl.acm.org
Page 1. 89 Extraction and Analysis of Fictional Character Networks: A Survey VINCENT LABATUT, Laboratoire Informatique d’Avignon – LIA EA 4128 XAVIER BOST, Orkis and Laboratoire Informatique d’Avignon – LIA EA 4128 …

Identification and Prediction of Interdisciplinary Research Topics: A Study Based on the Concept Lattice Theory
H Xu, C Wang, K Dong, Z Yue – Journal of Data and …, 2019 – content.sciendo.com
Jump to Content Jump to Main Navigation …

We can (‘t) do this: A corpus-assisted critical discourse analysis of migration in Germany
T Griebel, E Vollmann – Journal of Language and Politics, 2019 – jbe-platform.com
… A collocation is “a combination of two words that exhibit a tendency to occur near each other in natural language, ie to cooccur” (Evert 2009, 1214, italics in original). Collocation analyses uncover the collocates of a node (like refugee) and thereby hint at its meaning …

OpBerg: Discovering causal sentences using optimal alignments
J Wood, NJ Matiasz, AJ Silva, W Hsu, A Abyzov… – arXiv preprint arXiv …, 2019 – arxiv.org
… Language Processing, EMNLP 2011, 27-31 July 2011, John McIntyre Conference Centre, Edinburgh, UK, A meeting of SIGDAT, a Special Interest Group of the ACL, 2011, pp. 294–303. [12] F. Huang, A. Yates, Open-domain semantic role labeling by modeling word spans, in …

Students’ Perceptions Of Corpora And Online Tools In Teaching Collocations To Improve Academic Written …
D NIZONKIZA – The Description, Measurement and Pedagogy of …, 2019 – books.google.com
… If the user chooses the Icollocates” option, s/he may specify the part of speech (POS) of both the Ney word and collocate. The user also has the option to determine slots before or after the Ney word (span) with the default set at four. Page 302 …

We can (‘t) do this
T Griebel, E Vollmann – academia.edu
… A collocation is “a combination of two words that exhibit a tendency to occur near each other in natural language, ie to cooccur” (Evert 2009, 1214, italics in original) … For the collocation analyses the word span has been set to five words to the left side and right side of the node …

Towards Narrative Understanding with Deep Neural Networks and Hidden Markov Models
JW Orr – 2019 – ir.library.oregonstate.edu
… 1 1.2 Natural Language Processing and Machine Reading … 39 3.3 An example of the first layer of GTNN for event sequence extraction. A GRU model produces a vector per word. Spans of words are averaged into a single vectors represented by the blue and red nodes …

Multimodal joint learning for personal knowledge base construction from Twitter-based lifelogs
AZ Yen, HH Huang, HH Chen – Information Processing & Management, 2019 – Elsevier
… The issues to be tackled include (1) not all text descriptions are related to life events, (2) life events in a text description can be expressed explicitly or implicitly, (3) the predicates in the implicit life events are often absent, and (4) the mapping from natural language predicates to …

A Natural Language Processing Approach to Predicting the Persuasiveness of Marketing Communications
S Atalay, S El Kihal, F Ellsaesser – Available at SSRN 3410351, 2019 – papers.ssrn.com
1 A NATURAL LANGUAGE PROCESSING APPROACH TO PREDICTING … unresolved question: What is the role of language in predicting how persuasive a message will be? We propose a natural language processing approach to measure language complexity and predict …

Constructing a corpus-informed list of Arabic formulaic sequences (ArFSs) for language pedagogy and technology
A Alghamdi, E Atwell – International Journal of Corpus Linguistics, 2019 – jbe-platform.com
… the attention of researchers in various language-related disciplines eg linguistics, psychology, language pedagogy (LP) and Natural Language Processing (NLP) … most relevant word pairs which co-occur with moderate to high frequency within a four-word span across common …

Analyzing Meaning in Big Data: Performing a Map Analysis Using Grammatical Parsing and Topic Modeling
J Goldenstein, P Poschmann – Sociological Methodology, 2019 – journals.sagepub.com
Social scientists have recently started discussing the utilization of text-mining tools as being fruitful for scaling inductively grounded close reading. We aim to progress in this direction and pr…

Neural Text Generation from Structured and Unstructured Data
H Shahidi – 2019 – uwspace.uwaterloo.ca
… Their model takes a row from the table and generates a natural language sentence describing that row by leveraging the semantics of the table … For instance, PCFG-Trans proposed by Heilman [22] is a rule-based system that can generate question based on a given word span …

Adversarial Domain Adaptation Network for Semantic Role Classification
H Yang, G Zhou, T He, M Li – IEICE TRANSACTIONS on …, 2019 – search.ieice.org
… Semantic Role Labeling (SRL) is an important fundamen- tal task in Natural Language Processing (NLP) community and its goal is to assign a formal semantic structure for each predicate of a given sentence, like WHO did WHAT to WHOM, WHEN, WHERE, WHY, HOW …

Media Bias: A Corpus-based Contrastive Study of the Online News Coverage on the Syrian Revolution: a Critical Discourse Analysis Perspective
A Algamde – 2019 – research.bangor.ac.uk
Page 1. Bangor University DOCTOR OF PHILOSOPHY Media Bias: a corpus-based contrastive study of the online news coverage on the Syrian revolution – a cxritical discourse analysis perspective Algamde, Amaal Award date: 2019 Awarding institution: Bangor University …

Theory of Mind and referring expressions after Traumatic Brain Injury
N Balaban, M Biran, Y Sacher – Aphasiology, 2019 – Taylor & Francis
ABSTRACTBackground: This study focused on the linguistic consequences of damage to Theory of Mind (TOM) in patients after Traumatic Brain Injury (TBI). It was designed to extend a previous study th…

Injecting constraints into neural NLP models
JY Lee – 2019 – andrew.cmu.edu
… have set new state-of-the-art performances in many tasks across different applications such as vision and Natural Language Processing (NLP … For example, in semantic role labeling (SRL), the model has to identify the relationship between a verb predicate and word span …

Gaze analysis of user characteristics in magazine style narrative visualizations
D Toker, C Conati, G Carenini – User Modeling and User-Adapted …, 2019 – Springer
… to corresponding datapoints in the accompanying visualization(s) of existing documents via either crowdsourcing or natural language processing techniques … OSPAN (Operation-word span) (Turner and Engle 1989), a short computer-based test where users are briefly shown a …

Methods for taking semantic graphs apart and putting them back together again
J Groschwitz – 2019 – coli.uni-saarland.de
… Neural networks have proven to be enormously effective machine learning tools for natural language processing … This makes semantic parsing an important part of the natural language processing toolkit. At the same time, semantic parsing is a very challenging task …

Individual Differences in Comprehending Japanese Scrambled Sentences
Y Eshita – 2019 – drum.lib.umd.edu
… Japlish were using a semi-artificial language and participants were only exposed to this language for less than an hour before the test, results using natural language and … Page 24. 15 word-span tasks, non-word repetition tasks, and sentence repetition tasks; while a …

Neural Entity Linking For Company Names
Z Chen, IS Möller, L Hennig – 2019 – researchgate.net
… and knowledge base population. A large percentage of the web data is in the form of natural language, which is highly ambiguous, primarily the named entities. To make ambiguously … Page 17. xvii List of Abbreviations NLP Natural Language Processing EL Entity Linking …

In search of meaning: Lessons, resources and next steps for computational analysis of financial discourse
M El?Haj, P Rayson, M Walker… – Journal of Business …, 2019 – Wiley Online Library
… replicability and objectivity (see Section 4.4 for further discussion). 3.3 Natural language processing Natural language processing (NLP) sits at the core of computational linguistics. Liddy (2001) defines NLP as a suite of computational …

The role of individual variability in tests of functional hearing
M Courtland – 2019 – maury.science
… (like digit and word span tasks) only involved simple rehearsal and retrieval of common lexical items, and … Given the parallel nature and purpose of the task to digit and word span tasks, but its emphasis on predicting reading ability, they named the task the “reading span” task …

Neural attentions for natural language understanding and modeling
H Luo – 2019 – dspace.mit.edu
Page 1. Neural Attentions for Natural Language ARCHIVES Understanding and Modeling MASSF ! LTITUTE by JUN 13 2019 Hongyin Luo LIBRARIES … Page 2. 2 Page 3. Neural Attentions for Natural Language Understanding and Modeling by Hongyin Luo …

Short-term memory based on activated long-term memory: A review in response to Norris (2017).
N Cowan – 2019 – psycnet.apa.org
Short-term memory (STM), the limited information temporarily in a state of heightened accessibility, includes just-presented events and recently retrieved information. Norris (2017) argued for a prominent class of theories in which STM depends on the brain keeping a separate copy …

Semantic and Discursive Representation for Natural Language Understanding
D Sileo – 2019 – tel.archives-ouvertes.fr
… Damien Sileo To cite this version: Damien Sileo. Semantic and Discursive Representation for Natural Language Understanding … as opposed to pragmatics (meaning as use) is preponderant in the current training and evaluation data of natural language understanding models …

Semantic Role Labeling for Indian languages
A Gupta – 2019 – web2py.iiit.ac.in
… It was shown that the knowledge from these semantic frames can be utilized to improve other fields in natural language understanding such as Information Extraction [62, 18, 4], Question Answering [45, 60, 48] and … Similar to previous work, we call a word-span/constituent as a …

The lexicon
J Pustejovsky, O Batiukova – 2019 – books.google.com
… What principles determine the functioning of the lexicon as a component of natural language grammar? What role does lexical information play in linguistic theory … (2) what principles determine the functioning of the lexicon as a component of natural language grammar …

Hybrid Deep Question Answering
A Aghaebrahimian – 2019 – dspace.cuni.cz
… Martin Holub, Ph.D., Institute of Formal and Applied Lin- guistics Abstract: As one of the oldest tasks of Natural Language Processing, Question … is concerned with building systems that automatically retrieve answers to ques- tions posed by humans in a natural language …

Word Importance Modeling to Enhance Captions Generated by Automatic Speech Recognition for Deaf and Hard of Hearing Users
S Kafle – 2019 – scholarworks.rit.edu
Page 1. Rochester Institute of Technology RIT Scholar Works Theses 11-2019 Word Importance Modeling to Enhance Captions Generated by Automatic Speech Recognition for Deaf and Hard of Hearing Users Sushant Kafle sxk5664@rit.edu …

Collocations in a Learner English Corpus: analysis of Yoruba-speaking Nigerian English learners’ use of collocations
P Obukadeta – 2019 – eprints.kingston.ac.uk
… Research on L2 collocational competence and production has increased tremendously in the field of Applied and Corpus linguistic as well as Natural Language Processing (NLP) from the 1990s to date. By Natural language processing, I mean the field of computer science …

Machine Learning in Translation Corpora Processing
K Wolk – 2019 – books.google.com
… References Index Page 10. Abbreviations and Definitions Bilingual Evaluation Understudy (BLEU): An algorithm for evaluating the quality of text that has been machine-translated from one natural language to another. Scores lower …