Olemite leidmine protsessi läbiviimise logidest
Date
2012
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Tartu Ülikool
Abstract
Töö on kirjutatud protsessikaeve valdkonnas Artefaktikeskse teenuste koosvõime
projekti (ACSI) raames. Töö eesmärgiks oli luua meetod sündmuste logidest olemite
avastamiseks ja seda meetodit rakendada.
Loodud meetod on kirjutatud Javas ning kujutab endast pluginat ProM
raamistikule. ProM on geneeriline avatud lähtekoodiga Java raamistik protsessikaeve
algoritmide rakendamiseks pluginatena.
Olemite leidmise protsessi saab jaotada järgmisteks sammudeks:
1. Integreerimine ProM-iga.
2. Sisendandmetest (XES formaadis logifailidest) sündmuste tüüpide relatsioonide
koostamine.
3. Funktsionaalsete sõltuvuste leidmine sündmuste logide relatsioonilisest esitusest.
Funktsionaalsete sõltuvuste leidmiseks kasutatakse algoritmi TANE.
4. Funktsionaalsete sõltuvuste alusel kandidaatvõtmete leidmine. Kui relatsioonil on
mitu kandidaatvõtit, palutakse kasutajal valida neist üks primaarseks võtmeks.
5. Sama primaarse võtmega sündmustest moodustatakse üks olem.
6. Kasutajale esitatakse töö käigus moodustatud olemid väljundina või saadetakse
need järgmisele algoritmile töötlemiseks.
Meetodit testiti kahe logifaili puhul, milles olid andmed CD-poe näitel. Meetod töötas
mõlema logifaili puhul korrektselt.
The thesis is written in the field of process mining and in the frames of Artifact-Centric Service Interoperation (ACSI) project. The goal of the thesis was to create a method for discovering entities in process execution logs and to implement this method. The method is implemented as plugin for ProM open source process mining framework and is written in Java. This implementation can be divided into the following steps: 1. Integration with ProM. 2. Extracting the event type tables from the raw log input. 3. Finding functional dependencies from relational representation of event logs. The functional dependencies are found using an algorithm called TANE. 4. Finding the candidate keys from the functional dependencies. In case a relation has multiple candidate keys, the user is prompted to select one as primary key. 5. Grouping together the event types that have the same primary keys and integrating them into one entity. 6. The output is shown to the user or the entities are sent to another algorithm. Two different event log files were used to test this method. Both of these logs are based on the example of online CD-shop. The method was working correclty for the both event logs.
The thesis is written in the field of process mining and in the frames of Artifact-Centric Service Interoperation (ACSI) project. The goal of the thesis was to create a method for discovering entities in process execution logs and to implement this method. The method is implemented as plugin for ProM open source process mining framework and is written in Java. This implementation can be divided into the following steps: 1. Integration with ProM. 2. Extracting the event type tables from the raw log input. 3. Finding functional dependencies from relational representation of event logs. The functional dependencies are found using an algorithm called TANE. 4. Finding the candidate keys from the functional dependencies. In case a relation has multiple candidate keys, the user is prompted to select one as primary key. 5. Grouping together the event types that have the same primary keys and integrating them into one entity. 6. The output is shown to the user or the entities are sent to another algorithm. Two different event log files were used to test this method. Both of these logs are based on the example of online CD-shop. The method was working correclty for the both event logs.