Ontology modeling quick guide Experiences on Taxonomy and Ontology development work heimo hanninen Tieto, [email_address]
Why do we need better concepts for organizing information? 2011-02-09 [email_address]
Information is Constantly in Motion – - a need for better management 30% of people’s time   is spent searching for relevant information Only one-third of CFOs  believe that the information is easy to use, tailored, cost-effective or integrated 17% of IT budgets   for storage hardware  and storage management  software and people More than 60% of CEOs   believe their business needs to access and understand information faster to make swift decisions 30–50% of design time  is copy management 85% of information   is unstructured 37% growth of disk storage  in 2005 40% of IT budgets   may be spent on integration Source: IBM [email_address] Documents Transactions Customers Partners Employees Organizations Financials Products E-mails Databases Media Web content Reports
Corporate information in silos –  a need for connecting layer 2011-05-11 [email_address] Partners R&D Production ICT Sales & Marketing HR
What is  ontology  in IT system scope? Enterprise Architect Dmitry Bogachev – Topic Maps 2007 -metadata
IM Layers (a’la Heimo) 2011-05-03 project technology service staff solution customer Business Ontology/Domain model : -Business entities -With associations and -Properties -Across domains  Taxonomy : -Categorize -Hierarchical -Domain specific Information : -with metadata -identified -content and data Direct link Indirect link Value chain : -Processes document
Phased modelling work 2011-06-17 Preparation [email_address] iterative iterative
Recursive ontology development [email_address] 2011-02-09 Visualize Formalize Populate Analyze & Refine Biz reqs Value  chain User study Conceptual model  Cmap Logical model  UML class diagram Implementation model  RDF(S), OWL, Topic Maps, XML schema test Use case Prototype Pilot Use case Process walk
About the design work A kind of data modeling and knowledge engineering 80% of communication, 20% of design Top down: business needs the information Bottom up: inventorying where to get the data Middle out: start with a motivated gang, a small system Skills required: SMEs, often a guy who has been around long enough (find them) Communication and facilitation skills Data modeling (visualization) skills Team work: to create understanding on situation and common goal Deliver something tangible to review (PoC approach) Focus on the business not on the technology “ Keep stakeholders close and enemies closer” 2011-02-09 [email_address]
System development with semantic technology approach Understanding semantics of biz entities and taking good care of managing identities of entities and information related. Understand business (value proposition & process) Extract info objects needed in process: Entities,  their relationship,  properties and  info resources (links to docs) Model the formal ontology in W3C: RDF/OWL or ISO: Topic Maps Analyze and build adapters to sources (mapping) Analyze biz app interface and provide methods and Create ontology queries and package results for methods 2011-03-04 [email_address]
Finding critical information assets and mapping them to business objects 1. Biz Process  Walk through 2. Describe biz entities, properties and relations in  common  biz language 3. Identify sources. Understand  local glossaries 4. Create  mapping  from local to global glossary Business ontology Heimo Hänninen [email_address] R&D Logistics Manufac- turing  Marketing Sales Distri- bution Customer Support Product Design PDM, ERP CMS BI  Sales & Marketing Partner’s data CRM
Faced problems in the real world Poor data quality in source systems missing metadata, no identifiers, ad hoc field naming etc. Security mgt.  do you have to copy ACL or request it while user runs a query? Performance if running complex queries Governance difficult to define, who owns centralized knowledge service (it's not CMS, nor DW, nor CRM, nor PDM but can benefit from those all).  Usage: everyone want to ride on a bus but nobody wants to run the bus company. Ontology is redundant metadata about data in other systems. Not good for real time critical systems. Virtual ontology engine  which reads data directly from source is not here yet. Difficult to calculate the business case.  Not in ICT main stream. Educate and lobby managers, train developers to use  sem tech [email_address] 2011-03-04

Ontology And Taxonomy Modeling Quick Guide

  • 1.
    Ontology modeling quickguide Experiences on Taxonomy and Ontology development work heimo hanninen Tieto, [email_address]
  • 2.
    Why do weneed better concepts for organizing information? 2011-02-09 [email_address]
  • 3.
    Information is Constantlyin Motion – - a need for better management 30% of people’s time is spent searching for relevant information Only one-third of CFOs believe that the information is easy to use, tailored, cost-effective or integrated 17% of IT budgets for storage hardware and storage management software and people More than 60% of CEOs believe their business needs to access and understand information faster to make swift decisions 30–50% of design time is copy management 85% of information is unstructured 37% growth of disk storage in 2005 40% of IT budgets may be spent on integration Source: IBM [email_address] Documents Transactions Customers Partners Employees Organizations Financials Products E-mails Databases Media Web content Reports
  • 4.
    Corporate information insilos – a need for connecting layer 2011-05-11 [email_address] Partners R&D Production ICT Sales & Marketing HR
  • 5.
    What is ontology in IT system scope? Enterprise Architect Dmitry Bogachev – Topic Maps 2007 -metadata
  • 6.
    IM Layers (a’laHeimo) 2011-05-03 project technology service staff solution customer Business Ontology/Domain model : -Business entities -With associations and -Properties -Across domains Taxonomy : -Categorize -Hierarchical -Domain specific Information : -with metadata -identified -content and data Direct link Indirect link Value chain : -Processes document
  • 7.
    Phased modelling work2011-06-17 Preparation [email_address] iterative iterative
  • 8.
    Recursive ontology development[email_address] 2011-02-09 Visualize Formalize Populate Analyze & Refine Biz reqs Value chain User study Conceptual model  Cmap Logical model  UML class diagram Implementation model  RDF(S), OWL, Topic Maps, XML schema test Use case Prototype Pilot Use case Process walk
  • 9.
    About the designwork A kind of data modeling and knowledge engineering 80% of communication, 20% of design Top down: business needs the information Bottom up: inventorying where to get the data Middle out: start with a motivated gang, a small system Skills required: SMEs, often a guy who has been around long enough (find them) Communication and facilitation skills Data modeling (visualization) skills Team work: to create understanding on situation and common goal Deliver something tangible to review (PoC approach) Focus on the business not on the technology “ Keep stakeholders close and enemies closer” 2011-02-09 [email_address]
  • 10.
    System development withsemantic technology approach Understanding semantics of biz entities and taking good care of managing identities of entities and information related. Understand business (value proposition & process) Extract info objects needed in process: Entities, their relationship, properties and info resources (links to docs) Model the formal ontology in W3C: RDF/OWL or ISO: Topic Maps Analyze and build adapters to sources (mapping) Analyze biz app interface and provide methods and Create ontology queries and package results for methods 2011-03-04 [email_address]
  • 11.
    Finding critical informationassets and mapping them to business objects 1. Biz Process Walk through 2. Describe biz entities, properties and relations in common biz language 3. Identify sources. Understand local glossaries 4. Create mapping from local to global glossary Business ontology Heimo Hänninen [email_address] R&D Logistics Manufac- turing Marketing Sales Distri- bution Customer Support Product Design PDM, ERP CMS BI Sales & Marketing Partner’s data CRM
  • 12.
    Faced problems inthe real world Poor data quality in source systems missing metadata, no identifiers, ad hoc field naming etc. Security mgt. do you have to copy ACL or request it while user runs a query? Performance if running complex queries Governance difficult to define, who owns centralized knowledge service (it's not CMS, nor DW, nor CRM, nor PDM but can benefit from those all). Usage: everyone want to ride on a bus but nobody wants to run the bus company. Ontology is redundant metadata about data in other systems. Not good for real time critical systems. Virtual ontology engine  which reads data directly from source is not here yet. Difficult to calculate the business case. Not in ICT main stream. Educate and lobby managers, train developers to use sem tech [email_address] 2011-03-04

Editor's Notes

  • #3 Brain will not get bigger – amount of information will
  • #4 The data integration. It is where we spend most of our time,
  • #11 Development of Ontology driven system is like any other dev, except perhaps one thing: focus is on subject centric computing. = system development with sem tech approach
  • #12 Top down – bottom up
  • #13 Biz case requires a large understanding or business needs, application landscape, integration challenges, data sources and interfaces Lack of risk taking in times when you should innovate and generate new type of business?