Experimentation with the IEML artificial language for tagging articles in Stylo
The project aims to explore and experiment with the semantic possibilities offered by IEML, the artificial language created by Pierre Lévy, in the context of writing and tagging scientific articles.
Issues
The semantics of texts are increasingly being approached through inductive methods that attempt to extract information from poorly tagged texts based on probabilistic models. This growing reliance on such models exposes us to an increased risk of losing control over the information we produce and its meaning. In the era of LLMs, it becomes even more urgent to reintroduce semantic layers into the texts we produce—layers that we can control. IEML allows us to build concepts whose syntax makes their semantics explicit and computable. This is a very promising avenue for reclaiming the meaning of our intellectual outputs. We will start with simple experiments using article keywords, with the intention of iterating on more complex cases as the Revue 3.0 partnership progresses.
Technical challenges
The production of concepts in IEML is very complex and difficult to master. We will identify and specify lists of words and concepts in IEML that could be useful for journals and develop a selection interface that allows for relatively easy handling.
Research activities
- Study of the semantic fields of partner journals
- Identification of keywords
- Translation of keywords into IEML
Deliverables
Prototype of keyword alignment with IEML in Stylo.
