16,167 Hits in 6.4 sec

Inductive Entity Representations from Text via Link Prediction [article]

Daniel Daza, Michael Cochez, Paul Groth
2021 arXiv   pre-print
In this work, we propose a holistic evaluation protocol for entity representations learned via a link prediction objective.  ...  We consider the inductive link prediction and entity classification tasks, which involve entities not seen during training. We also consider an information retrieval task for entity-oriented search.  ...  In practice, a fixed value of can be selected (such as 32 or 64 in our experiments), so if we consider it equal for all entities, Algorithm 1: Learning inductive entity representations via link prediction  ... 
arXiv:2010.03496v2 fatcat:a5yzkw3ijjhzrhhuaqwrilpppa

Reasoning Through Memorization: Nearest Neighbor Knowledge Graph Embeddings [article]

Ningyu Zhang, Xin Xie, Xiang Chen, Shumin Deng, Chuanqi Tan, Fei Huang, Xu Cheng, Huajun Chen
2022 arXiv   pre-print
Previous knowledge graph embedding approaches usually map entities to representations and utilize score functions to predict the target entities, yet they struggle to reason rare or emerging unseen entities  ...  Experimental results demonstrate that our approach can improve inductive and transductive link prediction results and yield better performance for low-resource settings with only a few triples, which might  ...  Contextualized KG Representation Masked Entity Modeling For link prediction, given an incomplete triple (e i , r j , ?)  ... 
arXiv:2201.05575v1 fatcat:umhtd3wp75h2doegoe7q3yd3ve

Semantic Triple Encoder for Fast Open-Set Link Prediction [article]

Bo Wang, Tao Shen, Guodong Long, Tianyi Zhou, Yi Chang
2020 arXiv   pre-print
We improve both the open-set generalization and efficiency of link prediction on knowledge graphs by leveraging the contexts of entities and relations in a novel semantic triple encoder.  ...  We train this semantic triple encoder by optimizing two objectives specifically designed for link prediction.  ...  to inductively generate the embedding for the unseen entities via translation formula, such as "h + r = t" in TransE (Bordes et al., 2013) .  ... 
arXiv:2004.14781v1 fatcat:rujdhefl2jhptjxxffvcwjxkci

CoLAKE: Contextualized Language and Knowledge Embedding [article]

Tianxiang Sun, Yunfan Shao, Xipeng Qiu, Qipeng Guo, Yaru Hu, Xuanjing Huang, Zheng Zhang
2020 arXiv   pre-print
Instead of injecting only entity embeddings, CoLAKE extracts the knowledge context of an entity from large-scale knowledge bases.  ...  Few works explore the potential of deep contextualized knowledge representation when injecting knowledge.  ...  For inductive setting, we encourage CoLAKE to infer the unseen entity by aggregating messages from its neighbors. We take several well-known models for link prediction as our baselines 5 .  ... 
arXiv:2010.00309v1 fatcat:mbygoyqswzhtndjorvpkealk7q

Biomedical Knowledge Graph Refinement and Completion using Graph Representation Learning and Top-K Similarity Measure [article]

Islam Akef Ebeid, Majdi Hassan, Tingyi Wanyan, Jack Roper, Abhik Seal, Ying Ding
2020 arXiv   pre-print
We show that this simple procedure can be used alternatively to binary classifiers in link prediction.  ...  We perform a knowledge graph completion and refinement task using a simple top-K cosine similarity measure between the learned embedding vectors to predict missing links between drugs and targets present  ...  KG construction can vary from manual curation to automatic construction via extracting entities and relations from unstructured text using Natural Language Processing [16] .  ... 
arXiv:2012.10540v1 fatcat:2sztiyp2wvea7bfnffmzzfspwu

Unsupervised, Knowledge-Free, and Interpretable Word Sense Disambiguation

Alexander Panchenko, Fide Marten, Eugen Ruppert, Stefano Faralli, Dmitry Ustalov, Simone Paolo Ponzetto, Chris Biemann
2017 Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing: System Demonstrations  
The presented tool features a Web interface for all-word disambiguation of texts that makes the sense predictions human readable by providing interpretable word sense inventories, sense representations  ...  Interpretability of a predictive model is a powerful feature that gains the trust of users in the correctness of the predictions.  ...  Acknowledgments We acknowledge the support of the DFG under the "JOIN-T" project, the RFBR under project no. 16-37-00354 mol a, Amazon via the "AWS Research Grants" and Microsoft via the "Azure for Research  ... 
doi:10.18653/v1/d17-2016 dblp:conf/emnlp/PanchenkoMRFUPB17 fatcat:cu2jbnitdnep3g5xrwnee7v7zu

Scientific Language Models for Biomedical Knowledge Base Completion: An Empirical Study [article]

Rahul Nadkarni, David Wadden, Iz Beltagy, Noah A. Smith, Hannaneh Hajishirzi, Tom Hope
2021 arXiv   pre-print
Predicting missing links in these graphs can boost many important applications, such as drug design and repurposing.  ...  Finally, we demonstrate the advantage of LM models in the inductive setting with novel scientific entities. Our datasets and code are made publicly available.  ...  Due to their ability to form compositional representations from entity text, LMs are well-suited to this setting.  ... 
arXiv:2106.09700v2 fatcat:l3d6kofthzfbjngvpfemqghpra

Prix-LM: Pretraining for Multilingual Knowledge Base Construction [article]

Wenxuan Zhou, Fangyu Liu, Ivan Vulić, Nigel Collier, Muhao Chen
2021 arXiv   pre-print
Experiments on standard entity-related tasks, such as link prediction in multiple languages, cross-lingual entity linking and bilingual lexicon induction, demonstrate its effectiveness, with gains reported  ...  We leverage two types of knowledge, monolingual triples and cross-lingual links, extracted from existing multilingual KBs, and tune a multilingual language encoder XLM-R via a causal language modeling  ...  First, since XEL links mentions extracted from free text to KBs, it can be leveraged to enrich KBs withTable 1: Link prediction statistics and results.  ... 
arXiv:2110.08443v1 fatcat:7kdeih45nbhpvmq2bouzlre5ja

Low-resource Learning with Knowledge Graphs: A Comprehensive Survey [article]

Jiaoyan Chen and Yuxia Geng and Zhuo Chen and Jeff Z. Pan and Yuan He and Wen Zhang and Ian Horrocks and Huajun Chen
2021 arXiv   pre-print
), but also tasks for KG curation (e.g., inductive KG completion), and some typical evaluation resources for each task.  ...  ., dynamic contexts with emerging prediction targets and costly sample annotation.  ...  Inductive Entity Representations from Text via Link Prediction. In Proceedings of the Web Conference 2021. 798–808.  ... 
arXiv:2112.10006v3 fatcat:wkz6gjx4r5gvlhh673p3rqsmgi

A novel framework for biomedical entity sense induction

J.A. Lossio-Ventura, J. Bian, C. Jonquet, M. Roche, M. Teisseire
2018 Journal of Biomedical Informatics  
Each ambiguous entity was linked to on average 180 titles/abstracts obtained from MEDLINE.  ...  that proved to perform well for text data, and used the predicted number of senses "k" from the previous step.  ... 
doi:10.1016/j.jbi.2018.06.007 pmid:29935347 fatcat:2hxgg5g5qjgdfnbo3scrwf2wzy

Inductive Relation Prediction by BERT [article]

Hanwen Zha, Zhiyu Chen, Xifeng Yan
2021 arXiv   pre-print
Unfortunately, they are not able to handle inductive learning where unseen entities and relations are present and cannot take advantage of prior knowledge.  ...  Relation prediction in knowledge graphs is dominated by embedding based methods which mainly focus on the transductive setting.  ...  inductive ability for predicting missing links in KG.  ... 
arXiv:2103.07102v1 fatcat:a4ntvdq7bvedtbwfzqmss7yxcy

OTNEL: A Distributed Online Deep Learning Semantic Annotation Methodology

Christos Makris, Michael Angelos Simos
2020 Big Data and Cognitive Computing  
The semantic information extraction process from an unstructured text fragment to a corresponding representation from a concept ontology is known as named entity disambiguation.  ...  Semantic representation of unstructured text is crucial in modern artificial intelligence and information retrieval applications.  ...  In [18] , some innovative approaches for text annotation and entity linking were contributed.  ... 
doi:10.3390/bdcc4040031 fatcat:vftpewqo7vfzfai4x37bm2xyym


Roman Vitaliyovych Shaptala, Gennadiy Dmytrovych Kyselev
2019 Bulletin of National Technical University KhPI Series System Analysis Control and Information Technologies  
Wikipedia link prediction tries to find pages that should be interlinked due to some semantic relation.  ...  After that a binary classifier given a pair of embeddings predicts the probability of the existence of a link between the encoded nodes.  ...  Here, we obtain referent entities and their neighboring words from links contained in a Wikipedia page, and the model learns embeddings by predicting neighboring words given each entity.  ... 
doi:10.20998/2079-0023.2019.01.09 fatcat:okncwmfjund5fjl6jeckd6iu2m

Node Classification Meets Link Prediction on Knowledge Graphs [article]

Ralph Abboud, İsmail İlkan Ceylan
2021 arXiv   pre-print
Node classification and link prediction are widely studied in graph representation learning.  ...  While both transductive node classification and link prediction operate over a single input graph, they have so far been studied separately.  ...  Both components contribute, via the weighted summation, to an overall entity representation which holistically captures node properties. Relational inductive bias.  ... 
arXiv:2106.07297v2 fatcat:hqayfccwa5dphalsvac527xnti

Relational World Knowledge Representation in Contextual Language Models: A Review [article]

Tara Safavi, Danai Koutra
2021 arXiv   pre-print
We propose to organize knowledge representation strategies in LMs by the level of KB supervision provided, from no KB supervision at all to entity- and relation-level supervision.  ...  as knowledge representations.  ...  IIS 1845491, the Advanced Machine Learning Collaborative Grant from Procter & Gamble, and an Amazon faculty award.  ... 
arXiv:2104.05837v2 fatcat:bzi7rawbgzdchmxxga7dqgdfie
« Previous Showing results 1 — 15 out of 16,167 results