Penggalian Proses Berdasarkan Event Log Dan Pesan Dari Lintas Organisasi Menggunakan Data-Aware Heuristic Miner

Model proses bisnis organisasi dapat diperoleh dengan cara menggali data proses. Sumber data proses berasal dari event log (log kejadian) dalam sistem informasi organisasi. Sistem informasi yang digunakan, misalnya: WMS – workflow management system, ERP – enterprise resource planning, CRM – customer relationship management, dan SCM – suply chain management. Tujuan dari penelitian disertasi ini adalah untuk menggali proses lintas organisasi berdasarkan event log dan nilai data atribut (material flow) sebuah organisasi yang di-asosiasikan dengan pesan yaitu nilai data atribut dokumen (document flow) dari organisasi yang sesuai dengan kapasitasnya (berwenang). Sebuah method Data-driven Cross-organization (DCO) diajukan untuk melakukan tahapan-tahapan dalam membentuk model proses lintas organisasi: pengumpulan data, praproses data tidak terstruktur, melakukan ekstraksi data flat menjadi event log, pembersihan dan pemeriksaan, pelabelan data atribute dan proses, klasifikasi data atribute, process discovery event log lintas organisasi dengan cara menggabungkan perspektif control-flow dan perspektif data-flow menggunakan metode Data-aware Heuristic Miner (DHM). Penelitian menunjukkan bahwa model proses lintas organisasi seperti pada proses supply chain network pada terminal pelabuhan petikemas dapat dideteksi dengan menggabungkan nilai data atribut document-flow dan nilai data atribut material-flow. Hasil pengukuran Key Performance Indicator Import Dwell-time menghasilkan durasi median 5.5 hari dan durasi rata-rata 6.07 hari. Sebuah novelty pada penelitian disertasi ini adalah membuat metode Mapping Messages Position (M2P) yaitu koreografi data kejadian menjadi orkestrasi data kejadian dalam struktur data event log berdasarkan pesan dari lintas organisasi berbagi tanggung jawab. Sinkronisasi dihitung berdasarkan bobot frequency tertinggi dari pesan yang dikirim menuju data kejadian tertentu. Dengan demikian model proses dapat dibuat berdasarkan struktur event log yang sinkron berdasarkan data kejadian dan data pesan (data-driven cross-organization) ================================================================================================================================= The organizational business process model can be properly obtained by discovering data processes. The historical source of process data typically comes from the event log in the private organization's automated information system. Automated information systems used, for a practical example, Workflow Management System (WMS), Enterprise Resource Planning (ERP), Customer Relationship Management (CRM), and Supply Chain Management (SCM). The specific purpose of this dissertation study is precisely to mining of the cross-organization process model scientifically based on the event log and the possible value of a private organization's material flow that is typically associated with the document messages attribute of data values of the independent organization in accordance with its functional capacity (authorized). A Data-driven Cross-organization (DCO) method was merely proposed to properly carry out the logical steps in instantly forming a process cross-organization model. The logical steps carefully collecting historical data, pre-processing unstructured data, extracting flat data into event logs, properly cleaning and carefully inspecting, attribute data labeling and process, specific classification of attribute data, process discovery of cross-organization event logs by typically combining perspective control-flow and data-flow perspectives properly using the Data-aware Heuristic Miner (DHM) method. Extensive research sufficiently shows that a cross-organization process model such as the Supply Chain Network process at container port terminals can be reliably detected by uniquely combining the historical value of the document flow attribute data and the nominal value of the material flow attribute data. The measurement results of the Key Performance Indicator Import Dwell-time efficiently produce a median duration of 5.5 days and an average duration of 6.07 days. A novelty in this dissertation research is to develop properly the Mapping Messages Position (M2P) method that the complex choreography of event data into orchestration of event data in the event log data structure based on direct messages from cross-organizations capacity sharing. Synchronization is reasonably calculated based on the highest frequency weights of critical messages typicallyx sent to specific event data. Thus process model cross-organizational capacity sharing can be produced synchronous event log structures based on events data and messages data (data-driven cross-organization