Dependency-Based Extraction of Conditional Statements for Understanding Business Rules

Abstract

SUMMARY For the maintenance of a business system, developers must understand the business rules implemented in the system. One type of business rules defines computational business rules; they represent how an output value of a feature is computed from the valid inputs. Unfortunately, understanding business rules is a tedious and error-prone activity. We propose a program-dependence analysis technique tailored to understanding computational business rules. Given a variable representing an output, the proposed technique extracts the conditional statements that may affect the computation of the output. To evaluate the usefulness of the technique, we conducted an experiment with eight developers in one company. The results confirm that the proposed technique enables developers to accurately identify conditional statements corresponding to computational business rules. Furthermore, we compare the number of conditional statements extracted by the proposed technique and program slicing. We conclude that the proposed technique, in general, is more effective than program slicing.

On the design of an advanced business rule engine

“…Unfortunately, the need to introduce machine-learnt components to perform POS tagging or dependency analysis results in a proposal that cannot attain perfect parsing accuracy, which is a strong requirement. The proposals by Aiello et al, 16 Hatano et al, 17 Chittimalli et al, 21 Gallego and Corchuelo, 22,23 and Haj et al 24 used similar approaches and have the same problem. Only the proposals by Zámečníková and Kreslíková 18 and Hnatkowska and Gaweda 19 can achieve perfect parsing accuracy because they rely on grammars that can either parse a piece of text perfectly or report an error so that the user can correct it; unfortunately, none of them provide full support for SBVR-SE and the authors did not explore languages other than English; furthermore, it is not clear whether Zámečníková and Kreslíková's 18 proposal is open-domain or not and it produces non-executable decision tables.…”

“…Aiello et al 16 presented a proposal that differs in the translation approach, which is replaced by a standard parser for a grammar that severely restricts SBVR‐SE, but can produce executable Drools rules. Hatano et al 17 presented an approach that uses dependency‐parsing to identify conditionals in free‐text requirements documents, which leverages the idea that business rules must be expressed using if‐then structures. Zámečníková and Kreslíková 18 presented a proposal that builds on so‐called matrix grammars, which allow to parse a subset of context‐aware grammars using context‐free grammars only; they illustrated their proposal with a case study in which they generated decision tables for event‐driven business rules in a stock exchange domain.…”

“…Chittimalli et al 21 presented a recent follow‐up that uses a pipeline with three stages: first, two three‐gram statistical models are used to compute the probability that each sentence in the input text contains a rule or noise; second, the sentences that contain rules are POS tagged and transformed into dependency trees to which several heuristics are applied in order to extract the business domain (entities and relationships); third, several additional heuristics are used to extract the rules; note that the result consists in SBVR‐compliant rules in natural language. Gallego and Corchuelo 22,23 followed up on Hatano et al's 17 or Chittimalli et al's 21 proposals, but they used a neural‐based deep‐learning approach instead of using dependency parsing; their results proved to be promising since they were far more resilient than the competitors, but they did not put their focus on generating rules, but on parsing unusual forms of conditionals that might introduce business rules. Haj et al 24 presented the most recent approach of which we are aware; they use a pipeline in which the input document is first lemmatized, POS tagged, named entities are identified, and dependencies are identified; next, the business model is identified by using a number of pre‐defined patterns; finally, the system outputs SBVR‐compliant rules.…”