Back to search

Cornell University

Principled and Scalable Methods for Analysis of Higher-Order Networks

Abstract

dc:description.abstract

Graphs have enjoyed immense success as a model for analyzing pairwise interactions, such as friendships on Facebook, where people are encoded as nodes and friendships as edges that link nodes. However, many datasets exhibit higher-order interactions that go beyond the scope of the pairwise graph model, such as product co-purchasing data on Amazon. We lose important information by ignoring this higher-order structure. Currently, few methods for directly analyzing higher-order data exist. In this dissertation, we present principled and scalable methods for analyzing higher-order information. We first focus on the hypergraph perspective, where the hyperedges representing an interaction can connect two or more nodes at once. In the first part of the thesis, we introduce methods for analyzing hypergraphs that have multiple types of interactions. In particular, in Chapter 2 we develop a framework for detecting communities of nodes in hypergraphs that have different types or categories of hyperedges. In Chapter 3, we use the theory from Chapter 2 to construct a method for diverse recommender systems that scales to large datasets. In the next part of the thesis, we present some novel methods for dealing with hypergraphs that have special structure. In particular, in Chapter 3 we present a principled method for recovering planted "core" nodes. In the last part of the thesis, we present higher-order methods for graphs that come with edge flow metadata. In particular, in Chapter 4 we show that their associated Laplacian harmonic vectors exhibit spacial localization, and leverage this insight to create a principled algorithm that allows us to efficiently localize holes in these complexes.

Degree

thesis:*
Name thesis:degree_name
Ph. D., Applied Mathematics
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Applied Mathematics
Grantor
Cornell University
Year dc:date.issued
2022

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Amburg, Ilya
Chair dc:contributor.chair
  • Benson, Austin Reilley
Committee members dc:contributor.committeemember
  • Townsend, Alex John
  • Damle, Anil

Rights

Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
15529907
Dc Identifier Other
ProQuest Submission ID: 13080
ProQuest Publication ID: 29168882

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Amburg, Ilya. Principled and Scalable Methods for Analysis of Higher-Order Networks. Doctor of Philosophy thesis, Cornell University, 2022. https://hdl.handle.net/1813/111669