Last semester, the initial goal of the horror_games_feminist_themes project was to create a curated dataset by scraping the “Category: Horror video games” tree on Wikipedia to classify keywords from horror video games that feature a female or LGBTQ+ protagonist. Here’s a spreadsheet of the output .CSV file for convenience. This was my first time constructing a dataset, and my hope was to create one that could eventually help identify and analyze recurring feminist themes, patterns, and harmful tropes within the horror video game genre. I chose to scrape, curate, and then manually review Wikipedia pages because video games involve a wide range of elements, from gameplay mechanics to visual design; it would be nearly impossible to begin an analysis on gameplay alone. Starting with Wikipedia seemed like a feasible place to start. 

So now what?

The project I propose aims to build a website that brings the dataset to life. The creation of a public-facing website that translates a dataset into an accessible and interactive experience, one that can make invisible structures visible by dissecting feminist themes from a medium that is not often analyzed in such a way. I envision that there will be playful design choices to lighten the load of this sort of gruesome topic–something fun and feminine as an entry point to make sense of the genre and medium where patterns can be analyzed and challenged.

Here are some data viz examples from https://pudding.cool/ that I was inspired by while doing a brainstorm scan: 

• Is the love song dying?
• She Giggles, He Gallops
• Film or Digital?

What question or problem will this project answer? Horror studies is typically centered around film, and video games are often underexamined as cultural artifacts. Scholars like Barbara Creed highlight themes of female monstrosity, embodiment, and patriarchal structures in horror films. For example, the horror genre historically frames female bodies and the reproductive system as something monstrous or abject (Creed, 1993), and this is one of the themes that I have also noticed when sifting through reviewing keywords for my dataset. I’d like to build from these frameworks to examine horror video games as cultural artifacts that expose similar structures and themes, such as patriarchy and embodiment. 

What audience will this serve?

• People who are interested in games, horror, and feminism.
• Students who are interested in media/game studies and/or gender studies
• The (female and LGBTQ+) gaming community(?)
• This project contributes to DH by applying feminist principles of DH not only to interpret “data as capta” (Drucker, 2011) but also to design a project that analyzes the medium of video games, which, as mentioned previously, is often difficult to dissect and comprehensively analyze compared to other media.  

Tentatively…

the final product is some sort of data viz website, but we might need to refine the data a bit more. This may include refining the keywords and classifiers, or scraping more titles. We might need to narrow down the scope. One thing I found particularly interesting while constructing the dataset was the differences between horror games published in the 90s to early 2000s versus more recent titles. This could be one of the ways to help centralize and narrow the scope of this project. 

Tools, skillsets, and various roles I envision us needing:

Some of these roles may be merged, shared, or rotated amongst other team members (based on individual preferences) through the different stages of the project!

• Web-developer:
  • HTML/CSS/JS 
  • Build out the site
• UX & visual designer
  • Design the site’s color, typography, and layout
  • Tools tbd.
• Data curator & researcher
  • Python (for pulling data)
  • Review Wikipedia pages and validate keyword entries
  • Refine thematic keywords/classifiers
  • Help document interpretative decisions
• Writer & content development
  • Draft and edit website textual content
  • Help shape the project’s voice and tone
  • Content creation for the chosen social media platform

Potential barriers and questions:

• Wikipedia bias: I want to acknowledge that creating a dataset using keywords pulled from Wikipedia can have its limitations and biases. 
• How can the interface be used to display data and invite engagement without minimizing gendered-violence, trauma, or harm?

NB: When I saw all the great proposals at the end of last semester, I wasn’t sure I would pitch mine this semester, not because I don’t believe in my proposal, but because I wanted to work on everyone else’s projects too! I’m pitching it now both because I still want to do it, and perhaps to prompt others to pitch their projects.

This is an extract, cleaned up to incorporate the feedback I received, of my Fall Semester proposal, comprising the Abstract, the Enhancing the Humanities portion of the Narrative, a brief Environmental Scan, and the Final Product.

My project seeks to compile and exhibit a digital intertextual comparative concordance of themes that occur in epics authored by or attributed to women authors. The initial phase will focus on a small corpus comprising three themes (death, love, and vengeance) across each of three female epics, with later phases covering more themes and epics. Leveraging the work in Approaches to the Anglo and American female epic, 1621-1982, edited by Bernard Schweizer, it analyzes Telemachus by Anna Seward (unfinished as of 1809, officially published in 2016), Psyche by Mary Tighe (1805), and Aurora Leigh by Elizabeth Barrett Browning (1856) and facilitates questions of if or how the language used by women writing in traditionally male spaces to express these themes reflects feminist perspectives, as well as any insights that can be gained by comparative digital analysis.

The project has three main digital outputs: a toolset for extracting and documenting specified themes, a dataset comprising the extracted themes, and a Web-accessible display of those themes. The creation of this toolset, dataset, and Web-accessible presentation layer will further allow for future expansion, via, for example, text selections in other languages, additional translations, and selection of additional epics. While it aims to be neither a comprehensive collection of female epics nor a primary source for the epics it does include, the project nevertheless highlights the relative absence of such digital collections and serves as a thematic reference for scholars of epic literature, especially those interested in female epic literature.

Enhancing the Humanities

Historically, concordances have been laborious creations made for intense scholarship of works that, because of their cultural importance, were read and re-read, such as religious texts. Father Busa’s Index Thomisticus, created with the assistance of digital computers, is generally regarded as the beginning of digital humanities as a discipline. In the intervening years, more powerful computing technologies have made the creation of concordances per se easier, and at the same time, the rise of natural language chatbots and fuzzy searches based on statistical sampling presents us with the foregone conclusion that concordances lack comparative value in the face of powerful modern search technologies. Looking beyond the marketing terminology, however, we see that mere statistical correlations yield decontextualized results arranged according to internal algorithmic relevance. Concordances, as tools positioned specifically for textual and intertextual comparison and interpretation, remain vital parts of the landscape for examinations of the use of language to convey concepts, and digital concordances offer a chance to be more deliberate in building human-scale searches. The central questions afforded by this concordance are exploratory, focusing as it does on what we should automate, but the relative scarcity of scholarship focused here underscores the importance of conducting the scholarship in the first place.

In his introduction to Approaches to the Anglo and American Female Epic,1621-1982, Bernard Schweizer suggests that the epic is perhaps the most male-coded genre of literature, “so much so that epic and masculinity appear to be almost coterminous” (Schweizer et al 1). This gender assumption is apparent from several standpoints: first, of who has historically produced epics; second, who defined and formed the body of the genre’s critics; and third, the genre’s main characters. A fourth standpoint could be the themes of epics, but this question is afforded in part by the outputs of this project. And yet, as Schweizer and his contributors demonstrate, British and American women have been producing epics at least since the 17th century. Production of epics by women authors is not limited to modern American and the United Kingdom, however. At his blog, Interesting Literature, Dr. Oliver Tearle lists the Sumerian poem The Descent of Inanna, attributed to the high priestess Enheduanna, as a particularly early example of the female epic, suggesting that “if Enheduanna was the author of this poem, … that makes it the oldest work of poetry written by any named poet, male or female” (Tearle). His article goes on to list six other epic poems, half of which, had they all been published at the time of their writing, all should be in the public domain. There are likely to be others that have either been misattributed or forgotten, awaiting rediscovery.

Centering women-authored and attributed texts in Digital Humanities scholarship will bring more attention to these works, elevating them in the public consciousness, as well as the other non-epic works by the included authors. Additionally, it allows scholars interested in the use of figurative language a ready platform to explore how or even if the use of such language in female epics may differ from that used in male epics. 

This project focuses on three of the epics identified by Schweizer. The first, and most problematic from a sourcing standpoint, is Anna Seward’s epic poem Telemachus. In her introduction to The Collected Poems of Anna Seward, editor Lisa L. Moore writes that Seward had arranged with Walter Scott to publish a complete collection of her poems. Among the collection was one unpublished poem, an epic she considered her masterpiece, and which she “took special care to recommend … to Scott’s attention.” (Seward 37). When he published the collection in 1811, two years after Seward’s death, he had excluded Telemachus with no explanation. The poem would not appear in print until Moore laboriously transcribed it from the original manuscript and included it in her 2016 collection of Seward’s poems. What this means for scholars is that unlike the other epics in the selection, there is a question about copyright for Telemachus, and there is no public domain source from which one can acquire it. Careful attention is paid in this project to avoid full replication of the text, arguing that extractions of the text for concordancing and other search purposes constitutes a Fair Use claim.

The second and third epics included, having been published in 1805 and 1856 respectively, are firmly in the public domain. Mary Tighe’s Psyche, a six canto allegorical poem written in Spenserian stanzas, and Elizabeth Barrett Browning’s Aurora Leigh. As these were chosen specifically because of their treatment in Approaches to the Anglo and American Female Epic,1621-1982, they are naturally limited in their linguistic and temporal representation. Further, they were chosen because, with the exception of Telemachus, of their publication dates. This places some constraints on the broad applicability of the project and its tools and outputs in their initial phases, but an output of the project includes support to locate and process additional texts from beyond this immediate set.

This project naturally raises questions about what could potentially be included in the future. While it proposes eventually to encompass all female epics, a main question is what counts as an epic? Is it limited to poetry, or are prose works also applicable? If so, the field expands again, but a key consideration is to maintain some boundaries on the genre to keep the overall scope constrained. As the project grows, other types may be added, but the continuing mission will be to maintain sufficient constraint to demonstrate the related concepts. Additionally, the issue of copyright vis-a-vis translations of older, especially ancient and antique works, may impact selection even while the existence of multiple translation offers interesting opportunities to compare interpretations of figurative language from translator to translator. To account for this, the project will review concepts in the concordance annually, new works will be added when possible, and the project will maintain a registry of desired works that are unavailable because of copyright constraints. Additionally, the project is committed to obtaining permission from translators for less available texts.

A final consideration is the methodology. Existing digital concordances facilitate keyword searches for words, word forms, or phrases that occur in the works included. In most cases, the searches are limited to a single work. While a broad keyword search is possible and perhaps desirable, this project proposes both a curated approach focused on thematic subjects that are known to occur in the included works, and a fully intertextual display of those occurrences across all included works. By curating the theme selection, the intent is not to limit the possible explorations afforded by the texts and interface, but rather to help orient users to examine major themes that occur in the works. The selection of three initial themes (love, death, and vengeance) speaks to some of the timeless aspects of epic literature, but in no way is it asserted that these epics are limited to or even mainly about these subjects. They are mere starting points for additional analysis, and it is the intention of this project to set the basis for additional curated theme selections even as more works are identified for inclusion. During the execution of the project, the thematic selection may be adjusted to accommodate what the project team discover, but only if this has no impact on delivery.

Brief Environmental Scan

In the interest of brevity from here, I will just list the projects that informed mine.

• The Index Thomisticus
• Open Source Shakespeare
• The Electronic Dictionary of Armenian Bibliography
• The Chinese Text Project
• The Hyper-Concordance at the Victorian Literary Studies Archive
• Skovoroda Online Concordance
• PHI Latin Texts
• Intertextual Dante at the Digital Dante Project
• The Women Writers Project at Women Writers Online

Final Product and Dissemination

The final output of the project will live in GitHub as a code repository containing the extracted themes properly annotated with the designed metadata schema, the tools and scripts used to generate the extractions, and the website, including all narrative content, that exhibits the Concordance. Additionally, the site itself will be hosted online at a location to be determined and secured by the project team. In addition, the project lead and co-lead will share links to the final project via their various social media accounts, namely BlueSky, as well as the DH program lighting talks, the CUNY IT Conference, and potentially other conferences. 

Brief Project Overview: This project will build a public-facing, interactive web archive for the Humanities AI Hallucination Database. We are starting with a significant advantage: a pre-generated seed dataset of 100 records documenting AI hallucinations related to Black histories. However, our goal this semester would be to expand this scope to document algorithmic erasures in Latinx, Asian-American, and white American histories as well. We aim to create a comparative dataset that reveals how generative AI distorts knowledge across different identities.

The Problem: AI chatbots are fast becoming de facto historians for students. However, these models often invent figures, misattribute theories, or erase narratives. Failures that disproportionately affect marginalized groups. The gap: There is no centralized, standardized repository that educators can use to show students specifically how these models fail across different cultural contexts. We need empirical data to verify if, for example, AI “hallucinates” Asian-American historical figures differently than it does Black feminist scholars.

Intended Audience

• Educators & Librarians: Seeking diverse examples for teaching source evaluation and information literacy.

• Ethnic Studies Scholars: Researchers analyzing comparative patterns in algorithmic bias.

• Students: CUNY undergraduates learning to navigate generative AI tools critically.

Contribution to DH & Potential Impact This project contributes to Intersectional Digital Humanities and Critical Data/AI Studies. By moving this data from a spreadsheet to a public web interface, and expanding it to include Latinx, Asian-American, and white American histories, we are creating a vital Open Educational Resource (OER). Team members will help build a tool that operationalizes “epistemic justice” for multiple communities, giving educators the concrete evidence they need to challenge algorithmic authority in the classroom.

Final Product (What we will build) We will build a searchable, scalable web archive.

• The Web Archive: A website where users can filter hallucinations by demographic (e.g., “Black History,” “Latinx History,” “Asian-American History”) and error type.

• The Expanded Dataset: We will clean the existing 100 records and generate/verify 20-30 new records for Latinx and Asian-American topics to demonstrate the database’s capacity for growth.

• Data Visualizations: Charts comparing error rates or types across different demographic categories.

Feasibility Assessment

• Current Status: High feasibility with room for growth. We have a “Day 1” dataset (100 records) to start building the site immediately. The expansion into new histories provides a meaningful research task for the semester without overwhelming the team.

• Skills Needed:

  • Frontend Dev: To build the site structure and search interface.

  • Research Leads (Crucial): To generate and verify new prompts regarding Latinx, Asian-American, and white American history (using JSTOR/Library of Congress, etc.).

  • Data Curator: To ensure metadata standards match across different historical categories.

• Barriers: The main challenge is verifying new data accurately.