Joint Use of Node Attributes and Proximity for Node Classification

Arpit Merchant; Michael Mathioudakis

doi:10.1007/978-3-030-93413-2_43

Joint Use of Node Attributes and Proximity for Node Classification

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

Abstract

Node classification aims to infer unknown node labels from known labels and other node attributes. Standard approaches for this task assume homophily, whereby a node’s label is predicted from the labels of other nodes nearby in the network. However, there are also cases of networks where labels are better predicted from the individual attributes of each node rather than the labels of nearby nodes. Ideally, node classification methods should flexibly adapt to a range of settings wherein unknown labels are predicted either from labels of nearby nodes, or individual node attributes, or partly both. In this paper, we propose a principled approach, JANE, based on a generative probabilistic model that jointly weighs the role of attributes and node proximity via embeddings in predicting labels. Experiments on multiple network datasets demonstrate that JANE exhibits the desired combination of versatility and competitive performance compared to baselines.

Original language	English
Title of host publication	COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2 : Complex Networks & Their Applications X
Editors	R.M. Benito, C. Cherifi, H. Cherifi, E. Moro, L.M. Rocha, M. Sales-Pardo
Number of pages	12
Publisher	Springer, Cham
Publication date	1 Jan 2022
Pages	511-522
ISBN (Print)	978-3-030-93412-5
ISBN (Electronic)	978-3-030-93413-2
DOIs	https://doi.org/10.1007/978-3-030-93413-2_43
Publication status	Published - 1 Jan 2022
MoE publication type	A4 Article in conference proceedings
Event	The International Conference on Complex Networks and their Applications (2021) - Madrid, Spain Duration: 30 Nov 2021 → 2 Dec 2021 Conference number: 10th https://2021.complexnetworks.org/

Publication series

Name	Studies in Computational Intelligence
Publisher	SPRINGER INTERNATIONAL PUBLISHING AG
Volume	1016
ISSN (Print)	1860-949X

Fields of Science

113 Computer and information sciences
Node classification
Graph embeddings
Spectral graph analysis
NETWORK

Access to Document

10.1007/978-3-030-93413-2_43

nodeAccepted author manuscript, 801 KBLicence: CC BY-NC-SA

https://arxiv.org/abs/2010.11536Licence: CC BY-NC-SA

Cite this

Merchant, A., & Mathioudakis, M. (2022). Joint Use of Node Attributes and Proximity for Node Classification. In R. M. Benito, C. Cherifi, H. Cherifi, E. Moro, L. M. Rocha, & M. Sales-Pardo (Eds.), COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2: Complex Networks & Their Applications X (pp. 511-522). (Studies in Computational Intelligence; Vol. 1016). Springer, Cham. https://doi.org/10.1007/978-3-030-93413-2_43

@inproceedings{ed1d5d03140f41d789def8d26737065d,

title = "Joint Use of Node Attributes and Proximity for Node Classification",

abstract = "Node classification aims to infer unknown node labels from known labels and other node attributes. Standard approaches for this task assume homophily, whereby a node{\textquoteright}s label is predicted from the labels of other nodes nearby in the network. However, there are also cases of networks where labels are better predicted from the individual attributes of each node rather than the labels of nearby nodes. Ideally, node classification methods should flexibly adapt to a range of settings wherein unknown labels are predicted either from labels of nearby nodes, or individual node attributes, or partly both. In this paper, we propose a principled approach, JANE, based on a generative probabilistic model that jointly weighs the role of attributes and node proximity via embeddings in predicting labels. Experiments on multiple network datasets demonstrate that JANE exhibits the desired combination of versatility and competitive performance compared to baselines.",

keywords = "113 Computer and information sciences, Node classification, Graph embeddings, Spectral graph analysis, NETWORK",

author = "Arpit Merchant and Michael Mathioudakis",

year = "2022",

month = jan,

day = "1",

doi = "10.1007/978-3-030-93413-2_43",

language = "English",

isbn = "978-3-030-93412-5",

series = "Studies in Computational Intelligence",

publisher = "Springer, Cham",

pages = "511--522",

editor = "R.M. Benito and C. Cherifi and H. Cherifi and Moro, { E.} and L.M. Rocha and M. Sales-Pardo",

booktitle = "COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2",

address = "Switzerland",

note = "The International Conference on Complex Networks and their Applications (2021), Complex Networks 2021 ; Conference date: 30-11-2021 Through 02-12-2021",

url = "https://2021.complexnetworks.org/",

}

Merchant, A & Mathioudakis, M 2022, Joint Use of Node Attributes and Proximity for Node Classification. in RM Benito, C Cherifi, H Cherifi, E Moro, LM Rocha & M Sales-Pardo (eds), COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2: Complex Networks & Their Applications X. Studies in Computational Intelligence, vol. 1016, Springer, Cham, pp. 511-522, The International Conference on Complex Networks and their Applications (2021), Madrid, Spain, 30/11/2021. https://doi.org/10.1007/978-3-030-93413-2_43

Joint Use of Node Attributes and Proximity for Node Classification. / Merchant, Arpit ; Mathioudakis, Michael.

COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2: Complex Networks & Their Applications X. ed. / R.M. Benito; C. Cherifi; H. Cherifi; E. Moro; L.M. Rocha; M. Sales-Pardo. Springer, Cham, 2022. p. 511-522 (Studies in Computational Intelligence; Vol. 1016).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Joint Use of Node Attributes and Proximity for Node Classification

AU - Merchant, Arpit

AU - Mathioudakis, Michael

N1 - Conference code: 10th

PY - 2022/1/1

Y1 - 2022/1/1

N2 - Node classification aims to infer unknown node labels from known labels and other node attributes. Standard approaches for this task assume homophily, whereby a node’s label is predicted from the labels of other nodes nearby in the network. However, there are also cases of networks where labels are better predicted from the individual attributes of each node rather than the labels of nearby nodes. Ideally, node classification methods should flexibly adapt to a range of settings wherein unknown labels are predicted either from labels of nearby nodes, or individual node attributes, or partly both. In this paper, we propose a principled approach, JANE, based on a generative probabilistic model that jointly weighs the role of attributes and node proximity via embeddings in predicting labels. Experiments on multiple network datasets demonstrate that JANE exhibits the desired combination of versatility and competitive performance compared to baselines.

AB - Node classification aims to infer unknown node labels from known labels and other node attributes. Standard approaches for this task assume homophily, whereby a node’s label is predicted from the labels of other nodes nearby in the network. However, there are also cases of networks where labels are better predicted from the individual attributes of each node rather than the labels of nearby nodes. Ideally, node classification methods should flexibly adapt to a range of settings wherein unknown labels are predicted either from labels of nearby nodes, or individual node attributes, or partly both. In this paper, we propose a principled approach, JANE, based on a generative probabilistic model that jointly weighs the role of attributes and node proximity via embeddings in predicting labels. Experiments on multiple network datasets demonstrate that JANE exhibits the desired combination of versatility and competitive performance compared to baselines.

KW - 113 Computer and information sciences

KW - Node classification

KW - Graph embeddings

KW - Spectral graph analysis

KW - NETWORK

U2 - 10.1007/978-3-030-93413-2_43

DO - 10.1007/978-3-030-93413-2_43

M3 - Conference contribution

SN - 978-3-030-93412-5

T3 - Studies in Computational Intelligence

SP - 511

EP - 522

BT - COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2

A2 - Benito, R.M.

A2 - Cherifi, C.

A2 - Cherifi, H.

A2 - Moro, E.

A2 - Rocha, L.M.

A2 - Sales-Pardo, M.

PB - Springer, Cham

T2 - The International Conference on Complex Networks and their Applications (2021)

Y2 - 30 November 2021 through 2 December 2021

ER -

Merchant A , Mathioudakis M. Joint Use of Node Attributes and Proximity for Node Classification. In Benito RM, Cherifi C, Cherifi H, Moro E, Rocha LM, Sales-Pardo M, editors, COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2: Complex Networks & Their Applications X. Springer, Cham. 2022. p. 511-522. (Studies in Computational Intelligence). doi: 10.1007/978-3-030-93413-2_43