Video games use several summarization techniques in the form of gameplay trailers, reveals, and demo levels. These techniques mostly create summaries that are non-interactive and one-dimensional in medium (either audio, video, or text) which cannot portray the full video game on their own. We propose an approach to create a complete media summary that is also interactive and generate a playable summary level from a collection of Super Mario Bros. levels using user-obtained salient features based on patterns of memorability. We use the Mario AI Framework and the Occupancy-Regulated Extension generator to create the playable levels. We perform an analysis of the obtained features and perform curation on the generated summary levels. For evaluation, we compare the play experiences of the summary levels to those of the selected collection of levels. Although not yet complete, our preliminary evaluation suggests that players input similar adjectives to describe both the play experience of the Super Mario bros. levels and the play experience of the summary levels. With further testing in the future, it should be possible to gauge how much the summary level portrays the original level in terms of memorability.

Creator

• Bhat, Varun P.

Contributors

• Smith, Gillian Margaret
• Roberts, Charles Davis
• deWinter, Jennifer
• Claypool, Mark L.

Degree

• MS

Unit

• Interactive Media & Game Development

Publisher

• Worcester Polytechnic Institute

Identifier

• etd-4166

Keyword

• mario
• automatic summarization
• procedural content generation
• summarization

Advisor

• Claypool, Mark L.
• Smith, Gillian Margaret

Defense date

• 2020-08-01

Year

• 2020

Date created

• 2020-08-14

Resource type

• Thesis

Rights statement

• In Copyright

Last modified

• 2023-11-07