Visual Novel Metadata Schema
Class Project, Autumn 2023
Class Project, Autumn 2023
This was the quarter-long project in Professor Jin Ha Lee's Gaming & Libraries course at the University of Washington.
On a team of 4, I worked to create a taxonomy, metadata schema, and knowledge graph to facilitate the search and indexing of visual novel games on itch.io, a website to play and download indie games.
Timeline: 3 months (October - December 2023)
Collaborators: 4 researchers - myself and my awesome friends Doreen Chen, Emilie Hoy, and Alex Heller.
Audience: itch.io developers and users, as well as gamers/visual novel fans in general.
Responsibilities:
User Research
Metadata Design & Curation
Content Classification
Ontology & Knowledge Graph Design
Technologies Used: PoolParty, RDF, Protege, Qualtrics, Excel, Canva, Miro.
After defining our scope with the taxonomy to specifically be visual novel games, the team developed a survey that was administered over Qualtrics and shared with itch.io users in our immediate networks - friends of ours and members of Discord servers we were in.
We saw 29 responses, which showed many common themes:
100% of respondents identified as gamers. There is some potential bias here, because it was shared in spaces that had many self-identified "gamers."
~65.5% of users played visual novels, with some common themes for how they defined them including: story-driven games, first-person games, and games centered around player interactions with non-player characters and building relationships with them (like dating sims).
~79.3% of users were dissatisfied with itch.io's tagging system, particularly for how itch's tags represent game genres and content warnings.
Using this new knowledge of pain points and different types of gamers attracted to itch.io, we then developed a user persona representing an itch.io user that may be interested in finding visual novel games.
Gathering information from gamers on the best possible elements led to us creating a metadata schema that could help developers better catalog their games on itch.io, with a specific emphasis on user needs related to visual novels. We used the software PoolParty, creating 11 metadata attributes and over 50 corresponding values representing entities and their relationships.
We then attempted content classification by cataloging games on itch.io using the metadata elements we created, and wrote a reflection analyzing how well our elements worked to catalog 24 visual novel games. This whole process was documented in our metadata application profile, shown below.
After testing our metadata elements, we developed an ontology, linking different entities and relationships as well as providing users with a starting point to understand attributes and values. The goal was to ensure that all game developers and users could easily understand how to implement metadata elements.
After test cataloging some visual novels, we have found that our taxonomy enhances itch.io's semantic search and improved descriptive metadata for games.
We wanted to create as many pathways as possible for users to access visual novels, so we also used Japanese terms such as "isekai" (or "portal game") and "otome" (or "maiden game") in the Game Genre tag, as well as "Fangan" to describe a specific type of fan game based on the popular visual novel Danganronpa.
Our taxonomy is stored in an RDF file format, allowing for easy file transfer and editing with any metadata software.
To maintain the taxonomy, we recommended the following actions:
Quarterly meetings between company stakeholders to give feedback on the taxonomy and create an actionable list of terms and relationships to add, edit, or remove.
User feedback via website forms and a contact email.
Website analytics to see which terms are being used to support search, and which aren't.
With these actions in mind, the team will return to the taxonomy and remove any unused terms, update all terms to be more relevant, and edit terms and relationships based on user and company feedback.