Cargando…

Negative Sampling for Hyperlink Prediction in Networks

While graphs capture pairwise relations between entities, hypergraphs deal with higher-order ones, thereby ensuring losslessness. However, in hyperlink (i.e., higher-order link) prediction, where hyperlinks and non-hyperlinks are treated as “positive” and “negative” classes respectively, hypergraphs...

Descripción completa

Detalles Bibliográficos
Autores principales: Patil, Prasanna, Sharma, Govind, Murty, M. Narasimha
Formato: Online Artículo Texto
Lenguaje:English
Publicado: 2020
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7206280/
http://dx.doi.org/10.1007/978-3-030-47436-2_46
_version_ 1783530384821059584
author Patil, Prasanna
Sharma, Govind
Murty, M. Narasimha
author_facet Patil, Prasanna
Sharma, Govind
Murty, M. Narasimha
author_sort Patil, Prasanna
collection PubMed
description While graphs capture pairwise relations between entities, hypergraphs deal with higher-order ones, thereby ensuring losslessness. However, in hyperlink (i.e., higher-order link) prediction, where hyperlinks and non-hyperlinks are treated as “positive” and “negative” classes respectively, hypergraphs suffer from the problem of extreme class imbalance. Given this context, “negative sampling”—under-sampling the negative class of non-hyperlinks—becomes mandatory for performing hyperlink prediction. No prior work on hyperlink prediction deals with this problem. In this work, which is the first of its kind, we deal with this problem in the context of hyperlink prediction. More specifically, we leverage graph sampling techniques for sampling non-hyperlinks in hyperlink prediction. Our analysis clearly establishes the effect of random sampling, which is the norm in both link- as well as hyperlink-prediction. Further, we formalize the notion of “hardness” of non-hyperlinks via a measure of density, and analyze its distribution over various negative sampling techniques. We experiment with some real-world hypergraph datasets and provide both qualitative and quantitative results on the effects of negative sampling. We also establish its importance in evaluating hyperlink prediction algorithms.
format Online
Article
Text
id pubmed-7206280
institution National Center for Biotechnology Information
language English
publishDate 2020
record_format MEDLINE/PubMed
spelling pubmed-72062802020-05-08 Negative Sampling for Hyperlink Prediction in Networks Patil, Prasanna Sharma, Govind Murty, M. Narasimha Advances in Knowledge Discovery and Data Mining Article While graphs capture pairwise relations between entities, hypergraphs deal with higher-order ones, thereby ensuring losslessness. However, in hyperlink (i.e., higher-order link) prediction, where hyperlinks and non-hyperlinks are treated as “positive” and “negative” classes respectively, hypergraphs suffer from the problem of extreme class imbalance. Given this context, “negative sampling”—under-sampling the negative class of non-hyperlinks—becomes mandatory for performing hyperlink prediction. No prior work on hyperlink prediction deals with this problem. In this work, which is the first of its kind, we deal with this problem in the context of hyperlink prediction. More specifically, we leverage graph sampling techniques for sampling non-hyperlinks in hyperlink prediction. Our analysis clearly establishes the effect of random sampling, which is the norm in both link- as well as hyperlink-prediction. Further, we formalize the notion of “hardness” of non-hyperlinks via a measure of density, and analyze its distribution over various negative sampling techniques. We experiment with some real-world hypergraph datasets and provide both qualitative and quantitative results on the effects of negative sampling. We also establish its importance in evaluating hyperlink prediction algorithms. 2020-04-17 /pmc/articles/PMC7206280/ http://dx.doi.org/10.1007/978-3-030-47436-2_46 Text en © Springer Nature Switzerland AG 2020 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic.
spellingShingle Article
Patil, Prasanna
Sharma, Govind
Murty, M. Narasimha
Negative Sampling for Hyperlink Prediction in Networks
title Negative Sampling for Hyperlink Prediction in Networks
title_full Negative Sampling for Hyperlink Prediction in Networks
title_fullStr Negative Sampling for Hyperlink Prediction in Networks
title_full_unstemmed Negative Sampling for Hyperlink Prediction in Networks
title_short Negative Sampling for Hyperlink Prediction in Networks
title_sort negative sampling for hyperlink prediction in networks
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7206280/
http://dx.doi.org/10.1007/978-3-030-47436-2_46
work_keys_str_mv AT patilprasanna negativesamplingforhyperlinkpredictioninnetworks
AT sharmagovind negativesamplingforhyperlinkpredictioninnetworks
AT murtymnarasimha negativesamplingforhyperlinkpredictioninnetworks