APPROX-IA 2027

Deadlines
AI/CORE Unranked

APPROX-IA 2027

APPROX-IA: Approximation, Human Judgments and LLMs

Mar 25 2027Maison des Sciences de l’Homme de Bordeaux, Bordeaux, FranceOfficial conference site Site reachable

The APPROX-IA 2027 workshop focuses on approximation, human pragmatic judgments, and large language model (LLM) behavior, using approximative expressions as a test case. It brings together researchers from semantics, pragmatics, psycholinguistics, typology, cognitive modeling, and LLM evaluation to explore theoretical, experimental, and computational perspectives.

Official CFP Back to deadlines Verified August 30, 2026

Key deadlines

Verified August 30, 2026

Full paper

January 15, 2027

AoE

Conference timeline

Submission and decisions

Full paperKey deadline

January 15, 2027 · AoE

Paper fit

Contribution paths

A strong submission should clearly identify its contribution and evaluate it appropriately.

Long abstracts

2–4 pages, including references; oral presentation

Short abstracts

1 page; poster or lightning talk

Demos / resources

1–2 pages; tools, datasets, protocols

Research areas in scope

01

Relevant topics

human judgments and LLM behavior on approximative expressionsscalar approximators such as almost, nearly, barely and hardly and related words across languagesverbal approximatives and constructions expressing near-attainment, near-avoidance, unrealized outcomes or frustrated expectationsconative, frustrative, avertive and apprehensional constructionsprospective or proximal-future expressions, and related phenomena involving thresholds, partial realization, failed outcomes or contextual marginsapproximation and granularity: thresholds, margins, contextual modulationexperimental pragmatics of approximation: acceptability, forced choice, graded ratings, response variabilityLLM pragmatics and evaluation: pragmatic reasoning, robustness, calibration, non-literal meaninghuman–LLM comparison protocols: prompt design, reproducibility, stability across runs, agreement/disagreement analysesquantitative measures: surprisal, semantic similarity, alignment/misalignment indices, modeling of response profilescross-linguistic variation in approximative expressions and human–LLM alignmentresources, datasets, guidelines, code and reproducible releases

Policies worth checking twice

  • Submissions must be anonymized and uploaded as PDF files through OpenReview
  • The review process is double‑blind
  • Abstracts must not include author names, affiliations, acknowledgements or any other self‑identifying information
  • References to the authors’ own previous work must be written in the third person
  • Submissions must fall into one of the defined categories (long abstract, short abstract, demo/resource) with the specified page limits

Official sources

Compiled from the official call for papers. The organizers’ pages remain authoritative.

Official conference site

Last verified August 30, 2026