This experiment was based on a sentence chunking dataset from:
Agirre, E.; Gonzalez Agirre, A.; Lopez-Gazpio, I.; Maritxalar, M.; Rigau Claramunt, G.; and Uria, L. 2016. Semeval-2016 task 2: Interpretable semantic textual similarity. In SemEval-2016. 10th International Workshop on Semantic Evaluation. ACL
The goal is to learn to split a sentence into short phrases called chunks. The paper:
Kazmi, M.; Schüller, P.; and Saygın, Y. 2017. Improving scalability of inductive logic programming via pruning and best-effort optimisation. Expert Systems with Applications 87:291–303.
describes how to transform each sentence into a set of facts consisting of part of speech (POS) tags, forming a pre-processing step. In the above paper, these facts were then used as part of the input to the INSPIRE system. We have run FastLAS on this pre-processed version of the dataset (which we have translated into FastLAS input format) using each of these sets of facts as an example, and learning rules for whether to split the sentence between each pair of tags.
In our AAAI 2020 paper, we also ran the ILASP (Inductive Learning of Answer Set Programs) system on the same dataset. The ILASP and FastLAS input files, together with scripts for running the experiments are available here.
To run the learning, run:
mkdir results
./run_learning.rb *system_name*
Where *system_name* should be replaced with either FastLAS or
ILASP. To run the ILASP experiments, you will need to install ILASP.
You can view a summary of the results by running:
./print_results.rb *system_name*
Note that if the experiments are interrupted for any reason, when you rerun the learning command, it will resume from the last experiment. For a full restart, you need to empty the results folder.