Research map / Video / Video segmentation and tracking
Temporal Specification Optimisation for the Event Calculus
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Identifies the class of RTEC simple fluent definitions translatable into equivalent statically determined fluents, and presents a compiler performing the translation with an empirical evaluation on real applications.
Problem
Simple fluent definitions in the Run-Time Event Calculus for Boolean combinations of situations are exponentially larger and slower to reason with than equivalent statically determined fluent definitions, and knowledge engineers often do not choose the efficient form.
Approach
Formally characterises, via inertial condition symmetry, guard condition symmetry and Boolean representation symmetry, which simple FVP definitions are equivalent to Boolean FVP definitions, and gives a compiler that rewrites them as statically determined FVP rules.
Key results
- Compiling the human activity recognition event description reduces it from 131 rules and 569 conditions to 17 rules and 71 conditions.
- Reasoning with optimised event descriptions yields efficiency gains of about 1-2 orders of magnitude over the input simple-fluent descriptions in RTEC.
- The compiler finishes in under half a second per event description and produces identical FVP intervals to the original descriptions.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Translate translatable simple FVP definitions into statically determined FVP definitions | RTEC event descriptions · Human activity recognition (CAVIAR) | number of rules: 131 -> 17 | Helped |
| Translate translatable simple FVP definitions into statically determined FVP definitions | RTEC event descriptions · Human activity recognition (CAVIAR) | rule conditions: 569 -> 71 | Helped |
| Use an optimised event description instead of the input event description | RTEC · Composite event recognition | reasoning time: about 1-2 orders of magnitude | Helped |
| Use an optimised event description instead of the input event description | RTEC · Composite event recognition | computed FVP intervals | No clear effect |
Limitations
- Only translatable simple fluent definitions (those satisfying the symmetry conditions) can be optimised
- Legal, clinical and policy applications were only compared by size, with no reasoning experiments due to missing datasets
Experimental setup
Datasets: CAVIAR Brest AIS maritime dataset (18M signals) Helsinki public transport streams
Benchmarks: CAVIAR human activity recognition maritime situational awareness (AIS) city transport management (Helsinki)
Metrics: number of rules number of rule conditions reasoning time per window (ms)
Baselines: original simple-fluent event descriptions (unoptimised RTEC)
Builds on: Run-Time Event Calculus RTEC windowing and caching
Closest papers by meaning
- A Comprehensive Evaluation on Event Reasoning of Large Language Models (AAAI 2025 · similarity 0.82)
- Foundations of Reactive Synthesis for Declarative Process Specifications (AAAI 2024 · similarity 0.81)
- TimE: A Multi-level Benchmark for Temporal Reasoning of LLMs in Real-World Scenarios (NeurIPS 2025 · similarity 0.81)
- A Syntactic Approach to Computing Complete and Sound Abstraction in the Situation Calculus (AAAI 2025 · similarity 0.81)
- Temporal Numeric Planning with Patterns (AAAI 2025 · similarity 0.80)
- First-Order Automata (AAAI 2025 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.