Filters








3,350 Hits in 4.4 sec

Contextualised Graph Attention for Improved Relation Extraction [article]

Angrosh Mandya, Danushka Bollegala, Frans Coenen
2020 arXiv   pre-print
This paper presents a contextualized graph attention network that combines edge features and multiple sub-graphs for improving relation extraction.  ...  Two types of edge features are proposed, which are effectively combined with GAT and GCN models to apply for relation extraction.  ...  To the best knowledge of the authors, this is the first study that proposes a contextualised graph attention network with edge features, to learn from multiple sub-graphs for improved relation extraction  ... 
arXiv:2004.10624v1 fatcat:zbbc23564zallbaf2u73myjq7u

Representing Syntax and Composition with Geometric Transformations [article]

Lorenzo Bertolini, Julie Weeds, David Weir, Qiwei Peng
2021 arXiv   pre-print
Furthermore, we investigate which GT better encodes syntactic relations, so that these representations can be used to enhance phrase-level composition via syntactic contextualisation.  ...  The exploitation of syntactic graphs (SyGs) as a word's context has been shown to be beneficial for distributional semantic models (DSMs), both at the level of individual word representations and in deriving  ...  We also thank the anonymous reviewers for their helpful comments, and NVIDIA for the donation of the GPU that supported our work.  ... 
arXiv:2106.01904v1 fatcat:msgq6aj2v5hzjlbrcghjdwfrgy

New trends on digitisation of complex engineering drawings

Carlos Francisco Moreno-García, Eyad Elyan, Chrisina Jayne
2018 Neural computing & applications (Print)  
This paper presents a general framework for complex engineering drawing digitisation.  ...  Real-life industrial scenario on how to contextualise the digitised information from specific type of these drawings, namely piping and instrumentation diagrams, is discussed in details.  ...  Brian Bain from DNV-GL Aberdeen for his feedback and collaboration in the project. This work is supported by a Scottish national project granted by the Data Lab Innovation Centre.  ... 
doi:10.1007/s00521-018-3583-1 fatcat:uel7divnvnb7fipqlelijrhmmq

Automatic Expansion of Domain-Specific Affective Models for Web Intelligence Applications

Albert Weichselbraun, Jakob Steixner, Adrian M.P. Braşoveanu, Arno Scharl, Max Göbel, Lyndon J. B. Nixon
2021 Cognitive Computation  
This article introduces expansion techniques for affective models, combining common and commonsense knowledge available in knowledge graphs with language models and affective reasoning, improving coverage  ...  For instance, the brand manager of a television show may consider fear or sadness to be desired emotions for its audience.  ...  Dependency parsing and rule-based syntactic processing can yield improved results, even for sophisticated language models that extract contextualised embeddings for documents in the pipeline.  ... 
doi:10.1007/s12559-021-09839-4 pmid:33552304 pmcid:PMC7846919 fatcat:c4x4upqz6bdjni2uuwt6tdzohe

Bridging the Gap: Attending to Discontinuity in Identification of Multiword Expressions [article]

Omid Rohanian, Shiva Taslimipoor, Samaneh Kouchaki, Le An Ha, Ruslan Mitkov
2019 arXiv   pre-print
Two neural architectures are explored: Graph Convolutional Network (GCN) and multi-head self-attention. GCN leverages dependency parse information, and self-attention attends to long-range relations.  ...  The experiments on a standard multilingual dataset for verbal MWEs show that our model outperforms the baselines not only in the case of discontinuous MWEs but also in overall F-score.  ...  Graph Convolution as Feature Extraction Standard convolutional filters act as sequential ngram detectors (Kim, 2014) .  ... 
arXiv:1902.10667v2 fatcat:j3gvtea26fg7jlihlgslxt75lq

Structured Multimodal Attentions for TextVQA [article]

Chenyu Gao and Qi Zhu and Peng Wang and Hui Li and Yuliang Liu and Anton van den Hengel and Qi Wu
2021 arXiv   pre-print
SMA first uses a structural graph representation to encode the object-object, object-text and text-text relationships appearing in the image, and then designs a multimodal graph attention network to reason  ...  With better OCR results, different models share dramatic improvement over the VQA accuracy, but our model benefits most blessed by strong textual-visual reasoning ability.  ...  The experimental results reveal that each of the four modeled relations has improved the accuracy. In particular, the to relation attention leads to the largest improvement than others.  ... 
arXiv:2006.00753v2 fatcat:4lk4yloglnhdxftkzrrlcs3ztq

A Discourse Aware Sequence Learning Approach for Emotion Recognition in Conversations [article]

Sreyan Ghosh, Harshvardhan Srivastava, S. Umesh
2022 arXiv   pre-print
literature, with a conversational graph that is both sparse and avoids complicated edge relations like much of previous work.  ...  We conduct experiments on four benchmark datasets for ERC and show that our model achieves performance competitive to state-of-the-art and at times performs better than other graph-based approaches in  ...  Temporal Information Flow in Graph Layers For encoding discourse relations in a conversation, we use a Graph Attention Network (GAT) [30] with the information flow through layers inspired by [31] .  ... 
arXiv:2203.16799v2 fatcat:bh5uqrtbyzgrhkwxqfoyv5cghm

Introducing Self-Attention to Target Attentive Graph Neural Networks [article]

Sai Mitheran, Abhinav Java, Surya Kant Sahu, Arshad Shaikh
2022 arXiv   pre-print
Such graph-based architectures have representational limits, as a single sub-graph is susceptible to overfit the sequential dependencies instead of accounting for complex transitions between items in different  ...  Our experimental results and ablation show that our proposed method is competitive with the existing methods on real-world benchmark datasets, improving on graph-based hypotheses.  ...  Despite the convenient representation of sessions offered by GNNs, it lacks the ability to model long-range dependencies and intricate interactions as substantiated in Graph-Contextualised Self-Attention  ... 
arXiv:2107.01516v3 fatcat:3fwwcnkc7vavtbe3wenumx6xgm

Semantic Disambiguation and Contextualisation of Social Tags [chapter]

Ignacio Fernández-Tobías, Iván Cantador, Alejandro Bellogín
2012 Lecture Notes in Computer Science  
The framework is used for building contextualised tag-based user and item profiles.  ...  The analysis includes a comparison between semantic similarities obtained for pairs of tags in Delicious folksonomy, and their semantic distances in the whole Web, according to cooccurrence based metrics  ...  With all the obtained tags (at most ), we create a new graph , whose edges are extracted from the global graph .  ... 
doi:10.1007/978-3-642-28509-7_18 fatcat:nuyzvvc7wrgkbmoh34fbb6g4eq

Presentation of laboratory test results in patient portals: influence of interface design on risk interpretation and visual search behaviour

Paolo Fraccaro, Markel Vigo, Panagiotis Balatsoukas, Sabine N. van der Veer, Lamiece Hassan, Richard Williams, Grahame Wood, Smeeta Sinha, Iain Buchan, Niels Peek
2018 BMC Medical Informatics and Decision Making  
We hypothesized that visual cues improve patients' abilities to correctly interpret laboratory test results presented through patient portals.  ...  Results: Misinterpretation of risk was common, with 65% of participants underestimating the need for action across all presentations at least once.  ...  Acknowledgements We thank: the NIHR Clinical Research Network Portfolio for the support in the recruitment of participants; and Mr.  ... 
doi:10.1186/s12911-018-0589-7 pmid:29433495 pmcid:PMC5809992 fatcat:j4qameagirff7mvtgu7w7h6zou

Connecting the Dots: Document-level Neural Relation Extraction with Edge-oriented Graphs [article]

Fenia Christopoulou, Makoto Miwa, Sophia Ananiadou
2019 arXiv   pre-print
We thus propose an edge-oriented graph neural model for document-level relation extraction. The model utilises different types of nodes and edges to create a document-level graph.  ...  Document-level relation extraction is a complex human process that requires logical inference to extract relationships between named entities in text.  ...  Acknowledgments The authors would like to thank the anonymous reviewers for their comments and AIST/AIRC for providing the computational resources for this study.  ... 
arXiv:1909.00228v1 fatcat:nccn74lgvzcpbietudvl24crgy

Connecting the Dots: Document-level Neural Relation Extraction with Edge-oriented Graphs

Fenia Christopoulou, Makoto Miwa, Sophia Ananiadou
2019 Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)  
We thus propose an edge-oriented graph neural model for document-level relation extraction. The model utilises different types of nodes and edges to create a document-level graph.  ...  Document-level relation extraction is a complex human process that requires logical inference to extract relationships between named entities in text.  ...  Acknowledgments The authors would like to thank the anonymous reviewers for their comments and AIST/AIRC for providing the computational resources for this study.  ... 
doi:10.18653/v1/d19-1498 dblp:conf/emnlp/ChristopoulouMA19 fatcat:yyvvuwrl7nchne5euoetaiixlq

QMUL-SDS @ SardiStance: Leveraging Network Interactions to Boost Performance on Stance Detection using Knowledge Graphs [article]

Rabab Alkhalifa, Arkaitz Zubiaga
2020 arXiv   pre-print
For our submission for Task B, we ranked 6th (f-avg 0.709).  ...  embedding for Task A.  ...  Related Work In social media, classical features can be extracted by using stylistic signals from text such as bag of n-grams, char-grams, part-of-speech labels, and lemmas (Sobhani et al., 2019) , structural  ... 
arXiv:2011.01181v3 fatcat:esdshyetrjcqlfznzlywcl6lpe

On Incorporating Structural Information to improve Dialogue Response Generation [article]

Nikita Moghe, Priyesh Vijayan, Balaraman Ravindran, Mitesh M. Khapra
2020 arXiv   pre-print
More specifically, we use (i) Graph Convolutional Networks (GCNs) to capture structural information, (ii) LSTMs to capture sequential information and (iii) BERT for the deep contextualized representations  ...  This is a new task and has not received much attention from the community.  ...  Structure Layer: To capture structural information we propose multi-graph GCN, M-GCN, a simple extension of GCN to extract relevant multihop multi-relational dependencies from multiple structures/graphs  ... 
arXiv:2005.14315v1 fatcat:kcgkgr2y2ffzdngoikkf7psvsi

A Dataset Independent Set of Baselines for Relation Prediction in Argument Mining [article]

Oana Cocarascu, Elena Cabrio, Serena Villata, Francesca Toni
2020 arXiv   pre-print
Argument Mining is the research area which aims at extracting argument components and predicting argumentative relations (i.e.,support and attack) from text.  ...  In particular, numerous approaches have been proposed in the literature to predict the relations holding between the arguments, and application-specific annotated resources were built for this purpose.  ...  ., 2018) do not bring any improvements compared to non-contextualised word embeddings for the relation prediction task in AM.  ... 
arXiv:2003.04970v1 fatcat:6hw432njg5ggrm3vf5ybo3oqfu
« Previous Showing results 1 — 15 out of 3,350 results