Cursustekst DMN - uitleg over decision modeling & notation PDF

Preview

Summary

uitleg over decision modeling & notation...

Description

Beleidsinformatica (3512) Werkzittingen 7-8: Decision modeling & Notation 1ste bachelor Handelsingenieur/Handelsingenieur in de beleidsinformatica - academiejaar 2016-2017

Decision Modeling & Notation (DMN) Klassieke beslissingstabellen zijn beperkt wanneer we bedrijfsbeslissingen willen visualiseren en analyseren:  Ze worden erg uitgebreid als we complexe beslissingslogica met veel verschillende condities willen voorstellen en zijn vervolgens moeilijk op te stellen en interpreteren.  Ze bieden geen houvast om afhankelijkheden tussen beslissingen weer te geven (zie voorbeeld verder in de tekst).  Ze mixen beslissingslogica met hun bijbehorende acties (proceslogica). Nochtans verandert de beslissingslogica met een ander ritme dan de proceslogica. Hierdoor is het aangeraden om de beslissingen en de acties van elkaar los te koppelen. Daarom werd Decision Modeling & Notation (DMN) ontworpen als beslissingsmodelleertaal. In de lecture notes werd ingegaan op DMN en de basisconcepten. Hier focussen we op het modelleren van de logica met behulp van Decision Requirement Diagrams (DRD’s) en DMN beslissingstabellen.

DMN beslissingstabellen Lees de tutorial ‘DMN 1.1 tutorial’ (zie blackboard). Bij deze tutorial gelden de volgende opmerkingen/aanvullingen. Onderdelen beslissingstabellen De onderstaande figuur benoemt alle onderdelen van een DMN beslissingstabel.

1

Hit policy Deze cursus beperkt zich tot de ‘unique’ (U) hit policy. Dit betekent dat de beslissingstabel geen overlappende regels mag bevatten. Types   

van de inputs (condities) Boolean: ‘true’ of ‘false’ Opsomming van mogelijke waarden: {employed, unemployed, …} Intervallen: [0..10) betekent 0 ≤ x < 10

Data-input en BPMN Data-input (ovaal in DMN) kan verwijzen naar de attributen van data-objecten in BPMN, zoals weergegeven in de figuur op volgende pagina. Deze data-input vinden we terug als de inputwaarden in een beslissingstabel. In volgende beslissingstabel zijn de inputwaarden:  Claim value  Claim theft time  Customer contract type Voor iedere combinatie van deze inputwaarden, kan beslist worden over de output “Claim Coverage”.

2

3

Klassieke beslissingstabellen vs DMN Voorbeeld “Toekennen loonsverhoging en aantal extra verlofdagen”: Een firma in schoonheidsproducten heeft een aantal vertegenwoordigers actief in het verkopen van hun producten in 2 sectoren nl. speciaalzaken (parfumeries en schoonheidsinstituten) en warenhuizen. Jaarlijks worden de vertegenwoordigers beoordeeld aan de hand van al dan niet behaalde objectieven en krijgen ze als beoordeling: goed of slecht. Op basis van een aantal factoren worden loonsverhogingen toegekend. Deze loonsverhogingen zijn onderhevig aan het beleid van loonsverhogingen. Als dit beleid verandert, moeten de beslissingen ook geüpdatet worden. Vertegenwoordigers die een loonsverhoging krijgen, hebben ook recht op een aantal extra verlofdagen. Loonsverhoging:  Een loonsverhoging van 0.5% wordt toegekend aan een vertegenwoordiger met een goede beoordeling, welke actief is in de sector speciaalzaken en een anciënniteit bezit van minstens 5 jaar.  Een loonsverhoging van 0.2% wordt gegeven aan een vertegenwoordiger met een goede beoordeling en een anciënniteit van minstens 3 jaar en minder dan 5 jaar.  Een loonsverhoging van 0.3% wordt gegeven aan vertegenwoordigers actief in de sector warenhuizen met een anciënniteit van minstens 5 jaar en een goede beoordeling.  Alle andere vertegenwoordigers krijgen geen loonsverhoging. Extra verlofdagen:  Naast een loonsverhoging, kunnen de vertegenwoordigers nog aanspraak maken op een aantal extra verlofdagen, namelijk:  Twee extra verlofdagen indien ze een loonsverhoging van 0.2 % krijgen en niet afwezig waren het afgelopen jaar.  Vier extra verlofdagen indien de loonsverhoging 0.3 % of 0. 5% bedraagt en ze niet afwezig waren het afgelopen jaar.  De niet vernoemde vertegenwoordigers krijgen geen extra verlofdagen. Als we van bovenstaand voorbeeld de acties willen oplijsten, komen we aan volgend overzicht van 7 verschillende acties:       

loonsverhoging van 0.5 % loonsverhoging van 0.2 % loonsverhoging van 0.3 % geen loonsverhoging 2 extra verlofdagen 4 extra verlofdagen 0 extra verlofdagen

Vier condities waarvan bovenstaande acties afhankelijk zijn, samen met hun mogelijke waarden, zijn de volgende:    

sector: speciaalzaken of warenhuizen beoordeling: goed of slecht anciënniteit: < 3 jaar, >= 3 jaar en < 5 jaar of >= 5 jaar afwezigheid afgelopen jaar: nooit afwezig of afwezig

Een klassieke beslissingstabel op basis van bovenstaande condities en acties ziet er als volgt uit: 4

Maar, in de opgave over de extra verlofdagen zien we dat deze actie niet enkel afhankelijk is van de afwezigheid gedurende het afgelopen jaar, maar ook van de loonsverhoging. 

Conditie: loonsverhoging: geen, 0.2%, 0.3% of 0.5%

Deze loonsverhoging is een actie in het eerste deel van de opgave. Bovenstaande beslissingstabel zal dus niet ingevuld kunnen worden voor deze opgave aangezien er verschillende beslissingen afhankelijk zijn van elkaar. Dit wil zeggen dat één klassieke beslissingstabel niet volstaat. Visueel kan bovenstaand probleem als volgt voorgesteld worden met een DMN-model:

5

Hierbij kan iedere beslissing met een individuele beslissingstabel gevisualiseerd worden.

Beslis over type loonsverhoging:

Beslis over extra verlofdagen:

Vervolgens kunnen opnieuw onvolledigheden, tegenstrijdigheden en indifferenties aangeduid en opgelost worden.

6

Oefeningen 1.

Purchase process: reduction decision

A customer that purchases some products can get a reduction on the purchase price depending on the purchase volume and loyalty of that customer. This decision is based on two company policies: the policy regarding volume reductions and the policy regarding customer loyalty. The inputs of the purchase reduction decision are the volume reduction and the customer loyalty reduction. The output is the percentage of reduction on the purchase price. If both volume and loyalty reductions are ‘high’, then the purchase price reduction is 10%. If one of the volume and loyalty reductions is ‘high’ and the other ‘low’, then the purchase price reduction is 5%. The purchase price reduction is 2% when both volume and loyalty reductions are ‘low’. If one of the volume and loyalty reductions is ‘none’, then the purchase price reduction is 0%. The decision is made each time a purchase occurs. If a policy changes, then the decision must be updated. The volume reduction is decided based on the purchase order information: the product type and the volume. Possible product types are ‘accessories’ and ‘electronics’. For accessory products the volume reduction will always be ‘none’ no matter the volume ordered. Purchase orders of more than 50 ‘electronics’ get a ‘low’ volume reduction, an order size of more than 100 hundred leads to a ‘high’ volume reduction. Orders of less than 50 products automatically get a ‘none’ volume reduction. This is all based on the volume reduction policy which can change monthly. The loyalty reduction depends on the loyalty score. A customer with a ‘regular’ loyalty score gets no loyalty reduction, i.e. loyalty reduction is ‘none’. A ‘gold’ customer receives a ‘low’ loyalty reduction and ‘platinum’ customers get a ‘high’ loyalty reduction. The loyalty score is based on a client’s number of years loyalty to the company. Customers that have less than a year of loyalty are regarded as ‘regular’ customers. More than a year and up to 5 years of loyalty earns a customer a ‘gold’ label. Customers with more than 5 years of loyalty get a ‘platinum’ label. The loyalty reductions change when the policy regarding customer loyalty changes. Question:  

Visualise these decisions with a DMN model. Visualise also the decision tables for each of the decisions in the DMN model.

7

2.

Rent a car process – ‘Validate documents’ decision

A customer who wants to rent a car is obligated to have a valid ID and driving license in order to use the services of the company. The inputs for this decision are the ID status and the driving license status. The output is the document status. If the status of both ID and the driving license is valid, then the document status is also considered valid. If the status of any of the two inputs is not valid, then the document status is also listed as not valid. The validation is executed every time a customer requests to rent a car. It is the branch manager who has to make sure that the decision is made. If a regulation related to the validation of documents is changed (e.g. minimum age required for an individual to rent a car), the decision must be updated. Thus, the final decision is based on two subdecisions: a. Validate ID The personal information of each customer who requests a car should be validated (name, PIN, date of birth, address) as well as the validity of his ID card (ID card number, expiry date). There are no exceptions from this rule. To do this, the company checks whether the ID information of the client corresponds the official ID information from the Civil Registry. If the information matches, then the ID status (output for the decision) is also Valid. If the information stated in the ID of the client is different than the one returned by the Official Institution database, then the ID status is set to Not Valid. b. Validate Driving license When a customer requests a car, his input driving license information (Name, PIN, Driving License number, Driving category, Expiry date) must be validated. The validation relies on the driver’s license on the one hand, and the license data from an Official Institution external database on the other hand. If the official information corresponds to the one in the document of the customer, then the output Driving license status is Valid. If not, the status is Not Valid. Question:  

Visualise these decisions with a DMN model. Visualise also the decision tables for each of the decisions in the DMN model.

Extra information: 

 

ID – a personal identification document, issued by the government and proving the identity of a person; includes name, date of birth, nationality, sex, ID number, validity, PIN, remarks, place of issue Driving license – a personal document, proving the driving skills of an individual; includes personal information, validity of the document and driving category Civil Register – a governmental registry that contains different information for the citizens and residents in a region, including identity information

8

3.

Hiring process: ‘Validate Job Application’ decision

The ‘Validate Job Application’ decision defines whether a job applicant is eligible to apply for a job in the company. If the CV, additional documents and the criminal record status of a job applicant are validated (the documents are received, checked, the authenticity is confirmed), the applicant is considered eligible by the HR department, and can be sent for review to the responsible potential direct manager of the job applicant. This decision is driven by an internal policy stating that only documents of eligible job applicants – i.e. with validated documents – are reviewed by managers. Update of the decision may occur in case some regulations in the HR field or some intercompany policies are changed. The inputs for this decision are CV validation status, appendices validation status and criminal record status. The output is the eligibility status which will define if the job application is valid. Thus, the eligibility status of a job applicant is based on three checks (two subdecisions, and one check in an external database): 1. The CV of the applicant is checked 2. The appendix documents of the applicant are checked 3. The criminal record of the applicant is checked For this reason, the applicant is explicitly asked for all of these documents when registering a job application. a. Validate CV In this step the validity of the CV of a job applicant is checked by a HR specialist. Validation of a CV of a job applicant is an obligatory condition for the job applicant eligibility check. The input data is the CV information where two main parts are distinguished: Personal information and work experience information. The output of the decision is a validation status of the CV. If both personal and work experience information are included in the CV, then it is considered valid. If any of this information is missing, the CV is considered not valid. b. Validate appendices In this step the validity of appendices of the job applicant are checked. The motivation for this subdecision is that validation of additional documents of a job applicant is an obligatory condition for job applicant eligibility check. The inputs here are the appendices, provided by the job applicant and the output is the appendices validation status, set by the HR specialist after review of these documents. Two types of documents are obligatory to be presented: Valid basic health check and work experience file and/or a diploma for completed education. The latter two are checked in a first subdecision, if one of them is present and authentic, then the work experience/diploma validation is ok, if the two are provided, both need to be authentic for the work experience/diploma validation to be ok. The second subdecision concerns the valid basic health check. Each applicant should provide one which is proven to be not illegal and not made by an unregistered doctor. Then, the work experience/diploma decision is combined with the valid basic health check. If both of them are valid then the status for ‘Valid appendices’ is ‘valid’. If any of them is missing or not valid, the conclusion is that the appendices are not valid.

9

c. Criminal record status The Criminal record status of a job applicant is ‘called’ from an external database. The record might be clear or not. An internal policy for employees criminality states that ‘job applicants who do not have a clear criminal file are not hired in the company’. Question:  

Visualise these decisions with a DMN model. Visualise also the decision tables for each of the decisions in the DMN model.

Extra information:       

  

  



 

Job Applicant – a person who applies for a job and registers/is being registered for a job application Eligible – a Job Candidate is considered Eligible to apply for a job if the necessary documents that he presents are validated Potential Direct Manager – the potential direct manager of the Job Applicant (if approved), defined by the HR Department in consistence with the character of the job Internal policy – a statement, defined within a company, which triggers the execution of certain rules Cycle time of eligibility check – the time for which the validation of documents is executed by the HR Specialist and the system Rightness of the Eligibility Status – check of the rightness of the Eligibility Status of a Job Applicant; an error might be made by a person or the system Core system speed of 'connection to external sources' – the speed with which the Core system of the company is connecting to external database, exchanges/receives information from it Criminal Record – an external database that contains the criminality status of people from a specific region (town, country, region) CV – Curriculum Vitae – a document that includes Personal Information and Work Experience Information for the Job Applicant Personal Information – name, birth date, gender, Personal Identification Number, nationality, marital status, picture (optional), professional skills, hobbies & interests (optional), strengths Work Experience Information – work experience: previous employer/s (company name), period of working, position, responsibilities; certifications Appendices – additional documents to a CV of a Job Applicants Valid Basic Health Check – a document, issued by a legally exercising activity doctor, which proves that the Job Applicant is in good health and shape to exercise the job that he is applying for Work Experience File – an official document that contains a list with details for all employments of an individual; the document is issued at the first work place of a person and is added information by every other employer of a worker Diploma for Completed Education – an official school or university document that proves the successful completion of a certain level of education of an individual Clear Criminal File – a criminal file of an individual that states that he has never committed a crime or the prescription of the crime has expired

10...