Jen Kim
Independent researcher building MindCap — a permission-first knowledge graph that reveals the story your attention has been telling.
Currently: drafting the first MindCap paper for CHI 2027 workshops. Based in Kansas City.
Read the researchAbout
I’m an independent researcher working on personal knowledge graphs — specifically, how to build them so consent is structural rather than a policy layer bolted on after.
My current project, MindCap, is a knowledge graph that reveals the narrative arc of feed-mediated attention — the story your curiosity has been telling across browsing, video, and social platforms, at a scale working memory can’t hold. Rabbit holes look like chaos from the inside; from enough distance, they have a shape. MindCap is the vantage point. It is a working Firefox extension, a cross-platform app in development, and a research artifact — simultaneously. The research agenda is a three-paper arc: a foundational ontology (Year 1, in progress), a privacy-preserving architecture for cross-user comparison (Year 2), and empirical findings from deployment (Year 3).
I came to research sideways. I’m a self-taught developer with a business background, and the kind of pattern-thinker who notices small structural irregularities and pulls on them until something gives. That turns out to be useful for ontology work.
I’m open to conversations with researchers in personal information management, knowledge graphs, algorithmic accountability, and consent-first design — whether for collaboration, peer review, or shared reading. Reach me at admin@linuxgrrrl.com.
Research
Current focus
Permission-first behavioral knowledge graphs — ontologies in which user consent is first-class structural data rather than a policy layer applied after capture. The motivating system, MindCap, models feed-mediated attention across browsing, video, audio, and social platforms as a single behavioral graph governed by per-source, per-data-type, per-recipient permission grants.
Three-paper agenda
A permission-first behavioral knowledge graph for feed-mediated attention
An ontology and system architecture in which consent is first-class structural data, applied to a behavioral KG that captures feed-mediated attention across multiple platforms. Target: CHI 2027 workshops.
Privacy-preserving comparison of behavioral knowledge graphs
Cross-user curiosity comparison built from federated computation, differential privacy, and private set intersection — never centralizing raw behavioral data. Architecturally anticipated in the Year 1 ontology.
Empirical findings from N-user deployment
What curiosity patterns emerge at scale, what archetypes exist, how they vary across users, what communities form. Predicated on a real user base built through the MindCap product roadmap.
Artifacts
- · Competency questions document — forthcoming May 2026
- · MindCap ontology (Turtle) + Zenodo DOI — forthcoming Jul 2026
- · arXiv preprint: position paper — forthcoming Jul 2026
- · Year 1 paper (full) — submission Nov 2026, target CHI 2027 workshops
Methods & substrate
Ontology engineering grounded in established upper ontologies and patterns: PROV-O for provenance, SOSA/SSN for observation, OWL-Time for temporal structure, schema.org for content alignment, and ontology design patterns from Gangemi & Presutti. Reasoning via HermiT / OpenLLET. Graph substrate: Neo4j with weighted PageRank and Louvain via the Graph Data Science library, evaluated against an RDF-native baseline. Validation through competency questions, METHONTOLOGY audits, and queries over real captured data.
Recent writing
The Reading List Behind MindCap
The bibliography got its own page—curiosity research, information foraging, knowledge graphs, and the books that keep me honest while I build.
Dev Diary: Bridges Are Paths, Not Debates
Scrapped the entire bridge system and rebuilt it around entities, lineage, and visual echoes—24,000 bridges that can actually explain why two garments are connected.
The Great Reorganization (and Teaching an AI to Argue Better)
Thirty files moved, twenty-three tests fixed, and seven changes to the narrative engine—all so a Dockerfile can be five lines instead of fifty.
A Dress Doesn’t Just Look Like Another Dress. It Argues With It.
Six-dimensional bridge classification, contrast detection between garments that share structural DNA but argue opposite aesthetics, and a social function explorer for cross-cultural comparisons.
Research projects
MindCap
Behavioral KGA permission-first knowledge graph that reveals the narrative arc of a person’s attention and curiosity. Captures browsing, video, audio, and social activity into a single graph governed by per-source consent, then surfaces the story underneath — the questions a person has been circling, the turning points, the arc that working memory can’t hold.
Substrate: Firefox extension (Plasmo / TypeScript), FastAPI backend, Supabase, Neo4j analytical layer. Cross-platform Expo app in development.
Vintage Vestige
Cultural KGA cultural knowledge graph for vintage clothing — semantic search across centuries of fashion via vector embeddings, controlled vocabularies, and Getty AAT mappings. Sibling project to MindCap: same KG/ontology methodology, different domain.
Utilities
Firefox History Crawler
Async Python pipeline that enriches a Firefox places.sqlite export with content extraction and metadata. Supporting tool for MindCap research data.
Methods & tools
Knowledge representation
Graph systems
Languages & frameworks
Data & ML
Research methods
Education
Bachelor of Business Administration
University of Missouri-Kansas City
Contact
Open to conversations with researchers in personal information management, knowledge graphs, ontology engineering, algorithmic accountability, and consent-first system design. Also interested in peer review, collaboration, and shared reading lists.
Get in touchadmin@linuxgrrrl.com