Data Analytics for Graphs COMP9312 26T2

COMP9312 Project Released

Posted by Dong Wen Sunday 05 July 2026, 02:18:55 PM.

Hope you are enjoying your holiday. This email is to inform you that the project has been released. You may check this when you come back in Week 7 if you do not want any course matters bothering you. We will briefly introduce the project problems in Week 7 Monday Lecture.

The project specification link can be found on webcms .

The deadline is 9pm Friday 24 July (Week 8). Feel free to let us know if you have any questions. We may update the project page over the next few days to fix unclear descriptions, typos, or other mistakes. Any updates will be listed at the top of the entry page .

LiC

Dong

Assignment 1 Release

Posted by Yifan Yang Friday 12 June 2026, 05:17:37 PM.

Dear students,

The specification for the first assignment has been released on Webcms. The deadline is 9pm Friday Week 4. We may have some minor updates in the next few days to fix typos or any unclear presentations. Feel free to ask/clarify any details. I will quickly discuss questions in the assignment in Monday lecture.

Regards,
Dong

Issue of lecture recording

Posted by Dong Wen Thursday 11 June 2026, 03:59:10 PM, last modified Thursday 11 June 2026, 04:08:40 PM.

Dear students,

We identified that the slides were not correctly captured in the recording of yestersday's lecture (thanks for several students reminding us). That is because the cable in the classroom was replaced during last few days, which is not compatible with all Mac devices. We have contacted the IT service team to solve the problem. It will be fixed by the next lecture.

Below is a summary of the Wednesday lecture:

I first introduce the adjacency list and CSR. They have the same theoretical time complexity in scanning neighbors and checking if an edge exists or not. CSR is terrible to handle updates, because all neigbhors of nodes are placed in a flat array. However, CSR is often used for static data. It has better practical performance than adjacency list because of the CPU cache. In most graph problems, we often need to vist neighbors of nodes agagin and again. When visit neighbors of a node in CSR, some nearby neighbors may also be cached. In that way, if we visit some cached neighbors, the time for reading RAM can be saved.

Then, we start to fill a form to study the performance of different graph data structure for scaning neighbors, check edge existence, inserting edge, and deleting edge.

The results of all contect discussed in the lecture can be summarized by the following table. We completed structures related to RDBMS in the Wednesday lecture. We will go through the rest in the next lecture.

https://cgi.cse.unsw.edu.au/~cs9312/26T2/lectures/basic_graph_structures.txt

Regards,

Dong