Open Information Extraction Using Dependency Parser for Business Rule Mining in SBVR Format

Business Rules exists at the core of any Business Organization. For efficient execution of the business system, all the business rules must be in machine-interpretable format. There is an absence of such a system that can convert the business rule sentences into corresponding structured format automatically. We present BRMiner, a system which automatically converts business rules represented as Natural Language sentences to the corresponding SBVR format which is a structured representation that can be further converted to the machine-interpretable format. BRMiner is based on the idea of Open Information Extraction (OIE). We have shown that existing OIE systems are not suitable for SBVR rule formation that leads to the development of a new OIE system BRMiner, with more accurate prediction and additional capabilities. BRMiner uses the state of the art dependency parser to convert an unstructured business rule to the corresponding structured format. We have used internal as well as publically available datasets for our system evaluation and the results are encouraging which we have shown in the paper.