Work

Generating Activity Definitions with Large Language Models

Year: 2025

Type: conference-paper

Abstract: Composite activity recognition systems reason over streams of low-level, symbolic events in order to detect instances of composite activities, based on their formal definitions. Composite activity definition construction is a highly involved task, as these definitions often involve numerous spatio-temporal constraints, while labels that may be used for learning them automatically are hard to obtain. To address this issue, we propose a method that generates composite activity definitions from natural language descriptions using pre-trained Large Language Models (LLMs). In order to assess the quality of LLM-generated definitions, we propose a novel similarity metric, which reflects the human effort required to correct them. We present a thorough experimental evaluation of our approach on the maritime domain, demonstrating its effectiveness.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.14881949

Authors: Kouvaras, Andreas, Mantenoglou, Periklis, Artikis, Alexander

Institutions: University of Piraeus, National Centre of Scientific Research "Demokritos"

Language: English

Topic: Context-Aware Activity Recognition Systems

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funder: European Commission

Award: Critical Action Planning over Extreme-Scale Data