Interplay of Visual Analytics and Topic Modeling in Gameplay Analysis
No Thumbnail Available
Date
2024
Journal Title
Journal ISSN
Volume Title
Publisher
The Eurographics Association
Abstract
Spatio-temporal event sequences consist of activities or occurrences involving various interconnected elements in space and time. Exploring these sequences with topic modeling is a relatively new and evolving research area. We use topic modeling to analyze football games, as an example of complex and under-explored spatio-temporal event data. A key challenge in topic modeling is selecting the most suitable number of topics for the downstream application. Selecting too few topics oversimplifies the data, merging distinct patterns, whereas selecting too many can fragment coherent themes into overlapping categories. We propose a visual analytics technique that uses dimensionality reduction on topics derived from multiple topic modeling runs, each with a different number of topics. Our technique organizes all the topics in a hierarchical layout based on their spatial similarity, making it easier to make an informed decision about selecting the most expressive set of topics that represent distinctive spatial patterns. We apply our visual analytics technique to a football dataset, illustrating how it can be used to select an appropriate set of topics for this data. We then use these topics to represent game episodes, which help us summarize game dynamics and uncover insights into the games.
Description
CCS Concepts: Human-centered computing → Visual Analytics
@inproceedings{10.2312:cgvc.20241213,
booktitle = {Computer Graphics and Visual Computing (CGVC)},
editor = {Hunter, David and Slingsby, Aidan},
title = {{Interplay of Visual Analytics and Topic Modeling in Gameplay Analysis}},
author = {Moussavi, Laleh and Andrienko, Gennady and Andrienko, Natalia and Slingsby, Aidan},
year = {2024},
publisher = {The Eurographics Association},
ISBN = {978-3-03868-249-3},
DOI = {10.2312/cgvc.20241213}
}