Towards ontology based BPMN Implementation

Natural language is understandable by human and not machine. None technical persons can only use natural language to specify their business requirements. However, the current version of Business process management and notation (BPMN) tools do not allow business analysts to implement their business processes without having technical skills. BPMN tool is a tool that allows users to design and implement the business processes by connecting different business tasks and rules together. The tools do not provide automatic implementation of business tasks from users’ specifications in natural language (NL). Therefore, this research aims to propose a framework to automatically implement the business processes that are expressed in NL requirements. Ontology is used as a mechanism to solve this problem by comparing between users’ requirements and web services’ descriptions. Web service is a software module that performs a specific task and ontology is a concept that defines the relationships between different terms.

💡 Research Summary

The paper addresses a critical gap in current Business Process Model and Notation (BPMN) tools: while business analysts can graphically design processes, they lack the technical means to automatically bind those designs to executable services without programming expertise. The authors propose an ontology‑driven framework that translates natural‑language (NL) business requirements directly into implementable BPMN artifacts by matching them with existing web services.

The framework starts with a BPMN design file (XML) generated by any standard BPMN editor. An “Orchestra” module parses this file, extracts each task’s textual description, and forwards it to a keyword extraction pipeline. The pipeline employs a tokenizer, sentence splitter, and part‑of‑speech (POS) tagger, followed by JAPE rules that retain only verbs and noun phrases, producing a concise set of domain‑relevant keywords for each task.

These keywords are then fed into a semantic matching component that searches a pre‑built service ontology. The ontology is constructed from the descriptions of all registered web services, using DAML+OIL (a predecessor of OWL‑S) for its compatibility with XML/RDFS and its object‑oriented modeling capabilities. During ontology creation, noun phrases from service descriptions are extracted, low‑frequency terms are pruned, and domain experts validate the initial hierarchy. This results in a hierarchical tree where concepts (services) are linked to their functional capabilities.

Semantic matching yields a list of candidate web services whose ontological concepts overlap with the extracted keywords. If a single candidate is found, it is returned to the Orchestra module. When multiple candidates exist, the framework evaluates Quality of Service (QoS) attributes—availability (number of successful responses), average response time, and popularity (call count)—derived from runtime logs. A weighted QoS score selects the most suitable service.

If no existing service matches the requirement, the framework invokes a Service Composition module. This module attempts to synthesize a new composite service by chaining existing services whose input/output types align with the task’s needs. It uses graph‑based techniques, such as the Bellman‑Ford shortest‑path algorithm, to find an optimal composition path. Successful composition registers a new service in the repository; failure generates a user‑facing error message.

The selected or newly composed service is then automatically bound to the corresponding BPMN task, producing an enriched BPMN XML that includes executable service invocation code. This file can be imported back into any BPMN suite (the authors demonstrate with Oracle BPM Suite) for deployment, eliminating the need for manual coding.

Key contributions include:

1. A full pipeline from NL task description to executable BPMN, requiring no additional language learning by business analysts.
2. Integration of POS‑based keyword extraction with JAPE rule filtering to focus on actionable verbs and noun phrases.
3. Use of a DAML+OIL‑based service ontology for semantic matching, enabling domain‑agnostic reasoning.
4. QoS‑driven service selection to prefer high‑performance, reliable services among multiple matches.
5. Dynamic service composition as a fallback mechanism, employing graph algorithms for optimal chaining.

The paper also surveys related work on service orchestration vs. choreography, static vs. dynamic composition, and existing ontologies (OWL‑S, WS‑MO, etc.), highlighting how prior approaches either require formal specification languages (e.g., SBVR) or lack automatic NL handling.

Limitations acknowledged by the authors include the overhead of building and maintaining the service ontology, the dependency on accurate keyword extraction (which may suffer from ambiguous NL), and the need for up‑to‑date service registries to ensure composition feasibility. Future work is suggested in the direction of machine‑learning‑enhanced semantic parsing, automated ontology evolution, and cross‑domain service interoperability.

In summary, the research presents a practical, ontology‑centric solution that bridges the gap between business analysts’ natural‑language requirements and the technical implementation of BPMN processes, paving the way for more accessible, rapid, and automated business process deployment.