The legal fiction principle requires all regulations to be publicly available. Yet text-based search often fails when a question depends on how documents relate, such as which regulation amends or revokes another, because those relationships are not represented explicitly.
LexID addresses this with a knowledge graph holding both legal metadata and the semantic content of articles, modelled with the LexID ontology. Because Indonesian regulations follow a fixed drafting procedure, the graph is built with a rule-based approach from thousands of legal documents.
Evaluated on question answering, LexID reaches a macro-average F1 score of about 0.91. The paper is published in the Jurnal Ilmu Komputer dan Informasi (JIKI).