A comprehensive and efficient framework for subgraph matching
Cao, Hongtai
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/127353
Description
Title
A comprehensive and efficient framework for subgraph matching
Author(s)
Cao, Hongtai
Issue Date
2024-11-20
Director of Research (if dissertation) or Advisor (if thesis)
Chang, Kevin Chen-Chuan
Doctoral Committee Chair(s)
Chang, Kevin Chen-Chuan
Committee Member(s)
Tong, Hanghang
Park, Yongjoo
Cheng, Reynold
Department of Study
Siebel School Comp & Data Sci
Discipline
Computer Science
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Subgraph Matching
Heterogeneous Graphs
Edge Induce
Vertex Induce
Homomorphism
Sharing
Symmetry Breaking
Worst-case Optimal Join
Trie Structural Order
Clustering
Equivalence
Efficiency
Scalability
Language
eng
Abstract
My doctoral research aims to develop a comprehensive and efficient subgraph matching framework to support graph analytic tasks. In the era of information technology, graphs, consisting of vertices and edges, are widely used to model real-world entities and their relationships, offering new opportunities to understand the world through graph data analysis. To navigate graph data, retrieve interesting structures, and enable knowledge discovery, subgraph matching (SM), which identifies all instances of a given pattern within a graph, is fundamental to graph analysis and consistently attracts intense research interest. Meanwhile, the massive scale and heterogeneous nature of graph data, as well as the complexity of SM problems pose significant challenges to developing efficient algorithms. However, traditional subgraph matching works ignore the above characteristics of real-world data and over-simplify subgraph matching, resulting in ineffective and inefficient solutions. Consequently, graph analysis is extremely time-consuming or impossible on heterogeneous graphs at a large scale. To uncover knowledge hidden behind the data, my research fills in the gap between subgraph matching performance and graph analysis requirements, by directly addressing challenges imposed by graph data characteristics. To consolidate the foundation of graph analysis, my research systematically studies subgraph matching on heterogeneous graphs. First, I analyze the data characteristics in graph analysis and reveal the limitations of existing SM works when they face these characteristics. To develop a comprehensive framework, I focus on large heterogeneous graphs to support different types of vertices and edges, as well as edge directions. To be efficient, I observe the limitations of existing works and identify new concepts that are previously overlooked. Due to the graph heterogeneity, existing algorithms show repetitive computation in checking vertex and edge types, motivating the idea of indexing candidates by heterogeneity characteristics such that repetition can be avoided. Due to the complex edge connections in patterns, existing works perform repetitive matching, indicating a novel concept of candidate dependency to reuse candidates and avoid repetition. As real-world graph analysis often relies on frequent patterns, such patterns are often symmetric and contain subpatterns whose matching results can be shared. This motivates me to improve subgraph matching performance by sharing intermediate results, a novel idea not yet studied in existing works. Second, to systematically develop a subgraph matching framework, I divide the problem into cases of small patterns and large patterns because pattern sizes determine the research direction. In the small pattern case, I design a multilevel sharing solution to improve subgraph matching computation and achieve the goal through systematic multi-objective optimization. In contrast, to handle the complexity imposed by large patterns, I develop another solution using heuristics that scales better with pattern sizes. Finally, my work is systematically evaluated through experiments, and it outperforms state-of-the-art algorithms, showing its effectiveness and efficiency in subgraph matching.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
