SBVR2ASP

A tool for converting Semantic of Business Vocabulary and Rules specifications into Answer Set Programming.

Translation

The translation of the main SBVR constructs is reported below:

SBVR	ASP
it is prohibited that p	ASP constraint with p
it is obligatory that p	ASP constraint with p negated
it is necessary that p	ASP constraint with p negated
it is impossible that p	ASP constraint with p
only if (combined with another modal operator)	it inverts the modality of the modal operator (positive becomes negative)
each	generate the corresponding atom
at least n	#count{...} >= n / arithmetic operation when combined with before or after
at most n	#count{...} <= n / arithmetic operation when combined with before or after
exactly n	#count{...} = n / arithmetic operation when combined with before or after
at least n and at most m	n <= #count{...} <= m / arithmetic operation when combined with before or after
if p then q	generate a constraint with p and q, depending on the modal operator p and q might be negated
q if p	generate a constraint with p and q, depending on the modal operator p and might be negated
p of q	generate a relation between p and q adding the corresponding atom
p before q	p < q
p after q	p > q

Dependencies

Install the required Python package:

pip install lark

Usage

You can run sbvr2asp using the following command:

python3 src/main.py VOCABULARY RULES

where

• VOCABULARY is the path to the text file containing the vocabulary
• RULES is the path to the text file containing the business rules.

Examples

Example datasets are available in the examples directory. These include:

• eu_rent
• loan
• photo_equipment

Each dataset has its own folder containing:

• vocabulary.txt: the SBVR vocabulary.
• rules.txt: the SBVR business rules.
• encoding.lp: the conding generated by SBVR2ASP

You can generate the same encoding by running:

python3 src/main.py examples/DATASET_NAME/vocabulary.txt examples/DATASET_NAME/rules.txt

where DATASET_NAME is one of: eu_rent, loan, photo_equipment.

Running Benchmarks

Each dataset also includes benchmark instances. To run them, you need clingo. To install it, you can follow the instruction available here: https://github.com/potassco/clingo .

Then, you can run an instance using:

clingo examples/DATASET_NAME/encoding.lp examples/DATASET_NAME/data/CONSISTENCY/INSTANCE

where DATASET_NAME is one of: eu_rent, loan, photo_equipment; while CONSISTENCY is one of: sat (instances without conflicts), unsat (instances with conflicts). A working example is:

clingo examples/loan/encoding.lp examples/loan/data/sat/instance_5_100.lp

Note that for the eu_rent dataset, due to GitHub's file size limitations, some eu_rent input files are split into multiple parts and stored into a folder. You need to pass all parts to Clingo:

  clingo examples/eu_rent/encoding.lp examples/eu_rent/data/sat/instance_500_400_5_5/*