Pengembangan Backend dengan Pendekatan Rule-Based untuk BPMN Multi-Format Parser, Petri Nets Converter, dan LTL Analysis

Wadagraprana, Lalu Aldo (2026) Pengembangan Backend dengan Pendekatan Rule-Based untuk BPMN Multi-Format Parser, Petri Nets Converter, dan LTL Analysis. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025221101-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Penggunaan Business Process Model and Notation (BPMN) sebagai standar pemetaan proses bisnis sering kali terkendala oleh interoperabilitas format data antar tools pemodelan, sehingga berkas dari berbagai tools seperti Bizagi, Camunda, Bonita, dan bpmn.io perlu distandardisasi menjadi format JSON yang seragam. Sehingga, BPMN rentan kesalahan logika seperti deadlock dan livelock. sehingga diperlukan transformasi model ke Petri Net, namun pustaka umum seperti PM4Py memiliki perbedaan logika konversi dengan landasan teoretis yang diacu. Sehingga diperlukan algoritma konversi khusus untuk menghasilkan format Petri Net Markup Language (PNML) yang sesuai.
Penelitian ini berfokus pada pengembangan backend yang menjembatani kesenjangan tersebut melalui modul Multi-Format Parser, Petri Net Converter, dan LTL Analysis dengan pendekatan Rule-Based. Sistem dibangun menggunakan bahasa pemrograman Python dengan arsitektur Clean Architecture. Modul parser menstandardisasi input BPMN menjadi JSON baku, converter mentransformasikannya menjadi PNML yang disesuaikan dengan rujukan literatur, dan modul LTL Analysis menerapkan aturan berbasis Linear Temporal Logic (LTL) untuk mendeteksi empat kategori kesalahan, yaitu Loop Deadlock, Resource Deadlock, Improper Structuring, dan Conditional Livelock.
Kontribusi utama penelitian adalah perbandingan hasil analisis LTL pada dua representasi formal BPMN yaitu JSON dan PNML dari model yang sama melalui empat kuadran konsistensi. Pengujian terhadap 1.795 model BPMN menunjukkan tingkat keberhasilan parsing dan konversi 100% dengan integritas referensi arc 100%, namun validitas struktural Petri Net yang dihasilkan tercatat 74,76%. Nilai ini bukan menandakan kegagalan konversi, melainkan konsekuensi converter yang sengaja dibuat toleran terhadap model malformed sehingga place atau transition menggantung dipertahankan sebagai sinyal cacat pada LTL.
========================================================================================================================================
The use of Business Process Model and Notation (BPMN) as a standard for business process mapping is often hindered by data-format interoperability issues across modeling tools, so that files from various tools such as Bizagi, Camunda, Bonita, and bpmn.io must first be standardized into a uniform JSON format. Moreover, BPMN lacks adequate formal semantics, making logical errors such as deadlocks and livelocks difficult to detect at design time. Addressing this requires transforming the model into a Petri Net, however, common libraries such as PM4Py exhibit conversion logic discrepancies compared to the referenced theoretical foundations, necessitating a custom conversion algorithm to produce an appropriate Petri Net Markup Language (PNML) format.
This research focuses on developing a backend that bridges these gaps through a Multi-Format Parser, a Petri Net Converter, and LTL Analysis modules. The system is built in the Python programming language using a Clean Architecture. The parser standardizes BPMN input into canonical JSON, the converter transforms it into PNML through eleven literature-based transformation rules, and the LTL Analysis module applies rule-based LTL formulas to detect four error categories: Loop Deadlock, Resource Deadlock, Improper Structuring, and Conditional Livelock.
The main contribution is the comparison of LTL analysis results on two formal representations BPMN JSON and PNML of the same model through four consistency quadrants, together with a comparison of the conversion logic against the PM4Py library. Testing on 1,795 BPMN models achieved 100% parsing and conversion success with 100% arc reference integrity, the structural validity of the resulting Petri Nets, however, was 74.76%. This figure does not indicate a conversion failure but is a consequence of the deliberately tolerant converter design: malformed source models are still converted, so that dangling places or transitions are preserved as defect signals that are in fact exploited during the LTL analysis stage. Dual analysis over both representations proved to provide more complete error-detection coverage than relying on the PNML representation alone.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	BPMN, LTL Analysis, Multi-Format Parser, Petri Net, Python.
Subjects:	Q Science > QA Mathematics > QA76.758 Software engineering Q Science > QA Mathematics > QA9.58 Algorithms T Technology > T Technology (General) T Technology > T Technology (General) > T57.6 Operations research--Mathematics. Goal programming
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Lalu Aldo Wadagraprana
Date Deposited:	25 Jul 2026 07:22
Last Modified:	25 Jul 2026 07:22
URI:	http://repository.its.ac.id/id/eprint/137776

Actions (login required)

View Item