Artifacts

Standalone working objects from the Acevedology project

A showcase of self-contained artifacts from my research and teaching practice: curated datasets, navigational graphs, and methodological prototypes. Each one has its own page and lives as a working object, not a finished product. More will be added over time.

Chilean Spanish Micro-Archive (Prototype)

A situated lexical curation project for culturally grounded NLP evaluation

Mapudungún · Quechua Chilean Spanish Bilingual EN/ES NLPgap 50 entries · prototype

A bilingual annotated lexicon documenting Chilean Spanish terms with indigenous substrate, where meaning, class coding, and colonial history are compressed into everyday words. Built as a proof-of-concept that rigorous, community-grounded curation produces data that large-scale corpus collection cannot replicate. Each entry carries a verification-as-claims layer separating what is known from what is assumed.

Open artifact →

Mapa de Compatibilidad LLM

A task-first navigational graph for situated LLM evaluation

LLM evaluation task-first navigation workshop artifact Spanish-language graph

A Spanish-language navigational graph for evaluating language models by task, interaction dynamic, risk, and situated platform fit. Created for the Multiverse workshop as a public pedagogical artifact, it helps users move away from universal model rankings and toward practical questions of use, context, verification, and compatibility.

Open artifact →

Contact acevedology