Following the success of the first JVMG Lab in Stuttgart we are happy to report that a second JVMG Lab has now also been held at the end of February in Kyoto at Ritsumeikan University with participants from various universities and institutions in Japan.
The concept of the lab event remained unchanged with each participant working on a custom mini-TUC (Tiny Use Case) related to their own research interests and building on data from the JVMG knowledge graph during this two-day hackathon-like gathering. What made this JVMG Lab extra special was the presence of the community representatives, whose data the participants were working with. We had on-site – to answer questions and see first-hand the many different TUCs undertaken by the participants – Maria Pino from AnimeClick, Stephen and Sabrina Goral from Anime Characters Database (ACDB), and Yoran Heling from The Visual Novel Database (VNDB), who is since then tragically no longer with us. The two-days flew by even faster than last time, and similar to the first lab we finished the event with all participants presenting their results to the group, which were once again above and beyond our expectations, and served for lively discussion following each presentation. In order of the final presentations here is a very brief couple of sentences highlighting the topic of each TUC once again showcasing the breadth and diversity of topics engaged with.
Alexandra Maiuga from the Graduate School of International Relations at Ritsumeikan University explored the differences between the production of doujin and non-doujin games based on the data available in VNDB, comparing among other things the propensity to use free software and the choice of game engines by developers. Interestingly, for example, doujin games were found to be just as likely to be published as freeware as not.
Miguel Arturo González Sáenz also from the Graduate School of International Relations at Ritsumeikan University unpacked the available data on VTubers in ACDB, which features information on 355 characters from various VTuber agencies. Beyond offering a comparison of the distribution by gender, age, and aesthetic traits of the characters belonging to different agencies, he also highlighted some of the limits of the data: namely the absence of certain key agencies, the fact that the data seems to be limited to corporate VTubers only, and most importantly the lack of information on the mamas (designer) and papas (animator) of the characters.
Jill Ashley Hagen from the Graduate School of International Culture and Communication Studies (GSICCS) at Waseda University attempted to use the data in the JVMG knowledge graph to explore anime paratexts (most notably openings and endings). The creator data available in AnimeClick was able to provide a large number of opening and ending credits, most of which were for opening and ending songs. The analysis of these data points highlighted the prominence of musicals among the genre of works for which these paratextual information were recorded, which ties into further questions regarding the affective engagement that the voice actors of characters in the anime performing opening or ending credits can ellicit.
Andrea Mariucci from the Asian Studies Department at Cornell University and currently on a research stay at the Waseda University Game Lab at the GSICCS worked on two separate datasets, data from VNDB on platforms on the one hand, and data from the Game Preservation Society on games released between 1979-85 on cassette and often sold via postal service distribution. The first dataset helped identify the presence of communities that are working to keep games alive, evidenced by the constant trail of releases including ports and remakes of games on certain platforms. The second dataset offered a view of the four-five main platforms of the era as well as the prevalence of copyright infringements characterizing it.
Daiki Hanabata from the College of Image Arts and Science at Ritsumeikan University explored the differences between the heroines of three major visual novel brands/companies, namely Key, NitroOrigin, Leaf/Aquaplus as documented in the VNDB data. The findings seem to indicate that NitroOrigin heroines tend to be less characteristic (meaning that there is a wider and more even spread in traits found among them), while Leaf/Aquaplus and Key heroines exhibit a stronger tendency towards an on-brand style with this tendency seeming to be the strongest for Key main heroine characters (for example often characterized by the spiky bangs hairstyle).
Hinata Tomita from the University of Tsukuba analyzed the presence of mimikko characters over time and according to genre in the ACDB data (with genre information from AnimeClick). In ACDB the Mimikko trait refers to (mostly) antropomorphic characters having special ears (e.g. animal or elf) or horns. Beyond an observable upward trend in the number and ratio of mimikko characters from the 2000s onward, there also seems to be a tendency for these characters to appear more often in action, adventure, comedy and fantasy works. One interesting hypothesis that was explored in this TUC is the potential connection between the increase in mimikko characters and the rise in popularity of isekai works, which show a weak connection based on the available data.
Seiichi Kuwabara from the University of Tsukuba explored the change over time of the way heroines of romantic comedy anime are depicted in one of the most well motivated TUCs that once again relied on ACDB character data coupled with genre information from AnimeClick. Arguing that simplification and symbolism are a key aspect of Japanese visual media hair color was offered as an especially important signifier. Highlighting the changes over time of the dominant hair color of these heroines (from black in the 70s to Blue in the 80s to then various hues of red-pink-purple taking over in the 90s) among other interesting findings, the presentation ended on the note that orange or brown hair are the safest pick for creating a successful love comedy heroine.
Sujin Park from the Graduate School of Core Ethics and Frontier Sciences at Ritsumeikan University examined the changes in visual novel creators from the 2010s onward using the data from VNDB and focusing in on directors with at least ten different game credits. Two of his interesting findings included the lack of a monopoly-like concentration for creators, with a number of directors being active alongside each other, as well as the relative short span of activity found for these directors with their output often published over the course of a couple of years.
Yuanxue Jing from the Asian Studies and Media Studies Departments at Cornell University — and currently a visiting research fellow at Waseda University — was looking for the traces of keitai culture (a development in mobile phones unique to Japan that has since been relegated to media history with the advent of smart phones) in the JVMG knowledge graph. First, she examined all media from ACDB, AnimeClick, VNDB and MADB that had one of the three words/spellings used to refer to mobile phones in Japanese in their titles: ケータイ (ke-tai), スマホ (sumaho), 携帯 (keitai). Then, she found an entry point to trace potential keitai games in VNDB thanks to the “other mobile” platform category. This would be a very valuable discovery, if these games do indeed turn out to be at least in-part keitai games, as any kind of information on these games is now almost impossible to find.
Tamako Kitaoka and Akihiro Kameda from the National Institutes for the Humanities wanted to explore the intersection between new visual media content creation and the legacy of literary works. The AnimeClick tag “based on a literary classic” provided 72 works to analyse as a starting point for this question. Tamako Kitaoka used these works to analyze the distribution of genres over time of these adaptations of literary classics, highlighting the trends in what type of literary works were taken up to be adapted into works of manga (from history and romance in the 70-80s through fantasy and adventure in the 90-00s to mystery and thriller in the 10s), as well as the way seinen adaptations become more numerous from the late 2000s onward. At the same time Akihiro Kameda demonstrated how to quickly assemble a pipeline that could enrich these works — and also works from VNDB tagged with “literary adaptation” — by connecting them with even more adaptations and further information on the source works drawn from Wikidata, as well as provide a custom browser to explore these relationships of adaptation.
This second JVMG Lab once again proved how engrossing data work can become even for researchers who are new to this type of approach, and it also again directed the spotlight on a number of topics we had never even considered that the JVMG knowledge graph could contribute to researching. All in all, we learned so much again from these Tiny Use Cases with their breadth of unique questions, topics and frameworks. Thus, we would like to sincerely thank all participants for their preparation prior to and hard work during the lab event itself, it is thanks to them that this event was such a success once again. It is with the enthusiasm generated by this second lab event that we look forward to the opportunity to hopefully host further JVMG Labs in the future and to help introduce the JVMG knowledge graph and data work to even more researchers coming from the arts and humanities and beyond.