Parallel process discovery using a new Time-Based Alpha++ Miner

Yutika Amelia Effendi, Riyanarto Sarno

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)


A lot of services in business processes lead information systems to build huge amounts of event logs that are difficult to observe. The event log will be analysed using a process discovery technique to mine the process model by implementing some well-known algorithms such as deterministic algorithms and heuristic algorithms. All of the algorithms have their own benefits and limitations in analysing and discovering the event log into process models. This research proposed a new Time-based Alpha++ Miner with an improvement of the Alpha++ Miner and Modified Time-based Alpha Miner algorithm. The proposed miner is able to consider noise traces, loop, and non-free choice when modelling a process model where both of original algorithms cannot override those issues. A new Time-based Alpha++ Miner utilizing Time Interval Pattern can mine the process model using new rules defined by the time interval pattern using a double-time stamp event log and define sequence and parallel (AND, OR, and XOR) relation. The original miners are only able to discover sequence and parallel (AND and XOR) relation. To know the differences between the original Alpha++ Miner and the new one including the process model and its relations, the evaluation using fitness and precision was done in this research. The results presented that the process model obtained by a new Timebased Alpha++ Miner was better than that of the original Alpha++ Miner algorithm in terms of parallel OR, handling noise, fitness value, and precision value.

Original languageEnglish
Pages (from-to)126-141
Number of pages16
JournalIIUM Engineering Journal
Issue number1
Publication statusPublished - 2020


  • Alpha++ miner
  • Business process model
  • Parallel process
  • Process discovery
  • Process mining
  • Time interval pattern


Dive into the research topics of 'Parallel process discovery using a new Time-Based Alpha++ Miner'. Together they form a unique fingerprint.

Cite this