Incremental Event Calculus for Run-Time Reasoning

Authors: Efthimis Tsilionis, Alexander Artikis, Georgios Paliouras

JAIR 2022 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We examine RTECinc theoretically, presenting a complexity analysis, and show the conditions in which it outperforms RTEC. Moreover, we compare RTECinc and RTEC experimentally using real-world and synthetic datasets. The results are compatible with our theoretical analysis and show that RTECinc outperforms RTEC in many practical cases.
Researcher Affiliation Academia Efthimis Tsilionis EMAIL, Department of Informatics & Telecommunications National and Kapodistrian University of Athens, Greece Alexander Artikis EMAIL Department of Maritime Studies University of Piraeus, Greece Georgios Paliouras EMAIL Institute of Informatics & Telecommunications NCSR Demokritos , Greece
Pseudocode Yes Algorithm 1 recognise Simple Fluent 1: retrieve(F=V, se Qi 1, IQi 1) 2: delete SEpoints(F=V, se Qi 1) 3: compute SEpoints(F=V, se Qi 1, se Qi) 4: make Intervals(se Qi, IQi) 5: symmetric Difference(IQi, IQi 1, I+, I ) 6: assert(F=V, se Qi, IQi, I+, I )
Open Source Code Yes The code of RTECinc is publicly available1. Moreover, some of the employed datasets are available2, allowing the reproducibility of our results. 1. https://github.com/eftsilio/Incremental_RTEC
Open Datasets Yes The code of RTECinc is publicly available1. Moreover, some of the employed datasets are available2, allowing the reproducibility of our results. 2. The datasets concerning maritime activity in the area of Brest, France, are public and can be downloaded through the link provided in the github repository of RTECinc. ... The first dataset concerns approx. 5K vessels sailing in the Atlantic Ocean around the port of Brest, France, and spans a period from 1 October 2015 to 31 March 2016 (Ray et al., 2019).
Dataset Splits No The paper mentions generating synthetic delays and varying the percentage of delayed events (e.g., 5%, 10%, 20%). It also states, "We simulated a streaming behavior by consuming the input events little by little, i.e. reading small chunks periodically from the CSV files according to slide step specifications." This describes how experimental scenarios were set up by modifying event arrival, rather than providing standard training/validation/test splits of a dataset.
Hardware Specification Yes The experiments were performed on a computer with 24 cores (Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz) and 252 GB of RAM, running Debian GNU/Linux 9 with Kernel 4.9.0-7-amd64 and YAP Prolog 6.2.2.
Software Dependencies Yes The experiments were performed on a computer with 24 cores (Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz) and 252 GB of RAM, running Debian GNU/Linux 9 with Kernel 4.9.0-7-amd64 and YAP Prolog 6.2.2.
Experiment Setup Yes We employed two datasets for our empirical analysis; the first is a publicly available dataset6 concerning the area of Brest, France... The composite event description used in our empirical analysis includes sixteen simple fluents and five statically determined fluents, forming a hierarchy displayed in Figure 1 on page 973. ... We employed large temporal windows, that usually are not found in practice, in order to stress test the CER systems. Figure 9 presents the results where the sliding window varies from 12 to 168 hours while the slide step is constant and equal to 12 hours. ... We selected uniformly 5%, 10%, 20%, 40% and 80% of the total events to be delayed. ... we used a Gamma distribution with shape parameter k = 2 and scale parameter θ = 2.