Harnessing Graph Changes: Challenges and Benefits

PhD Thesis Proposal Defence


Title: "Harnessing Graph Changes: Challenges and Benefits"

by

Mr. Xun JIAN


Abstract:

In many real-world applications, the underlying data can be modeled as graphs. 
Querying graph data involves four main building blocks: the query semantics, 
query parameters, the underlying graph data, and the output. When handling 
graph queries, changes can happen on each of them. For example, modern graphs 
are dynamically changing, and bring challenges to the efficiency of querying 
algorithms. On the other hand, query rewriting techniques are developed to 
modify query parameters, so that the output matches the user’s intent.

In this proposal, we first consider the challenge of community search on 
dynamic heterogeneous information networks (HINs). We propose a relational 
community model that captures connection constraints between different types of 
nodes. We then propose a local search method for searching communities on 
dynamic HINs. Extensive experiments are conducted to show the effectiveness and 
efficiency of the proposed methods.

Then we consider the benefit of rewriting SPARQL queries. Due to the complex 
structure of knowledge graphs, it is not easy to write a correct SPARQL query. 
Thus, it would be valuable if we can automatically fix the wrong queries. 
Specifically, given an initial query, and a part of intended (or undesired) 
output, we propose several methods to relax or restrict the query, so that the 
actual output is close to the intention. We conduct extensive experiments and a 
user study to test the efficiency and effectiveness of the proposed methods.


Date:			Monday, 12 October 2020

Time:                  	2:00pm - 4:00pm

Zoom Meeting: 
https://hkust.zoom.us/j/98733596381?pwd=MHRPb0k2R2xJTDlpMHZOQmlWRFk4QT09

Committee Members:	Prof. Lei Chen (Supervisor)
  			Prof. Ke Yi (Chairperson)
 			Dr. Xiaojuan Ma
 			Dr. Wei Wang


**** ALL are Welcome ****