Zamorano López, María del Mar; (2025) Automated Software Transplantation for Procedural Content Generation. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Mar___Thesis.pdf - Accepted Version Download (8MB) | Preview

Abstract

Game Software Engineering (GSE) is a specialized branch of Software Engineering dedicated to the development of video games. Unlike traditional software, video game development presents unique challenges, including the management of diverse artefacts, differing development paces, limited code reuse, and difficulties in automated testing. Research in GSE has grown significantly since 2010, with contributions appearing in top Software Engineering venues. Video games are complex products that blend art and programming, requiring input from designers, programmers, and audio engineers. The industry has rapidly expanded, with a wide variety of titles driving economic growth and cultural influence. However, all games share a common challenge: the immense need for content. Large-scale productions require vast amounts of high-quality assets, leading to costly and time-intensive development cycles. Smaller teams, constrained by limited resources, must carefully balance ambition and feasibility, while games designed for specialized purposes are restricted by domain-specific content requirements. To address these challenges, researchers have explored Procedural Content Generation (PCG), an approach that automates or semi-automates the creation of game assets. This thesis introduces a novel technique, Procedural Content Transplantation (PCT), which applies Software Transplantation principles to PCG. In software, transplantation involves extracting functional components from one system and integrating them into another. Traditionally, it has been used for program repair, testing, security, and functionality enhancements. Applied to content generation, PCT extracts functional content from a game to integrate it into another content of a game. PCT employs a search-based algorithm to identify and integrate suitable content fragments from existing games into new contexts, mitigating the challenges of manual content creation. To validate this approach, we designed and implemented the first search-based transplantation algorithm for video game content. Our evaluation, conducted in collaboration with a commercial game studio, assessed the effectiveness of PCT in generating non-playable characters (NPCs) and compared its results with a standard PCG method from the literature.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item