MOD Ontology Ian Bailey, Model Futures y, (ian@modelfutures.com) Michael Warner, MOD ICAD -- 1 of 19 -- Introduction • MOD Ontology Team – Michael Warner Bose Lawanson MOD – Michael Warner, Bose Lawanson, MOD – Ian Bailey, Model Futures – Chris Partridge, 42 Objects Chris Partridge, 42 Objects – Fariba Hozhabrafkan, Serco • Background on MOD Ontology Background on MOD Ontology – Why we started it – What it’s founded on – IDEAS • The BORO Method • The MOD Ontology Demonstrator – Semantic integration, naming, de‐confliction g g -- 2 of 19 -- The MOD Ontology gy • Originated as a key enabler to MOD’s Enterprise Architecture Programme Architecture Programme • Intended to provide a common, reusable t i t t MOD enterprise structure across MOD • Based on the IDEAS Model • Not intended for inference and reasoning purposes • Goals are: – de‐confliction of terminology & data gy – accurate model of enterprise – processes, people, systems, etc. -- 3 of 19 -- About the MOD Ontology gy • For the ontologists amongst you, it is; – Extensional Extensional – Fully Four‐Dimensional (Sider, Hawley) – Higher‐Order • For the non‐ontologists; – It’s very precise about identification I h dl i i i – It handles time in a consistent manner – It embraces and reflects the complexity of the real world • Developed using the BORO method – Precise methodology for developing extensional ontologies gy p g g – Also the basis of the IDEAS model – Great for getting to the “true meaning” of legacy data and d fli ti l i d t de‐conflicting overlapping data sources -- 4 of 19 -- IDEAS • The IDEAS Model is a data integration model – Co‐developed by the Australian, Canadian, UK & US defence Co developed by the Australian, Canadian, UK & US defence departments – Sweden & NATO also involved as observers MOD has taken the lead on the modelling by suggesting the use of – MOD has taken the lead on the modelling, by suggesting the use of the BORO methodology to de‐conflict different data types from each country IDEAS is a formal ontology it is flexible extensible and based on – IDEAS is a formal ontology ‐ it is flexible, extensible and based on sound mathematical principles • MOD Ontology uses IDEAS Foundation – Extends the IDEAS concepts with MOD‐specific elements – …whilst still retaining international compatibility at the upper levels • http://www ideasgroup org • http://www.ideasgroup.org -- 5 of 19 -- IDEAS Structure• Provides a common semantic foundation IDEAS Foundation Defines a common language for Ontology and a standard XML encoding for exchange of for multiple uses • The common foundation enablesMOD U O t l IDEAS Taxonomy Defines a common language for Ontology and a standard XML encoding for exchange of Information, based on the W3C’s OWL and RDF specifications IDEAS Upper Ontology Defines common elements and patterns for re-use and commonality interoperability across domains and applications MOD Ontology Introduces MOD specific terminology to existing IDEAS concepts and adds any new concepts that MOD Upper Ontology Establishes any additional high-level elements and patterns required by MOD which are not in the core IDEAS set IDEAS Taxonomy Extends the upper ontology with elements that are common to all the IDEAS participating nations – increasing the level of semantic interoperability and commonality between the nations applications • All traces back up to IDEAS, so also offers b l f Introduces MOD specific terminology to existing IDEAS concepts, and adds any new concepts that are required for MOD business possibility of international interoperability • Data sources act as requirements on the ontology, feeding up the stack into the areas of stronger governance “standardisation by adoption” the areas of stronger governance – standardisation by adoption -- 6 of 19 -- Traditional Analysis Methods “process” “activity” “service” business l data d ll consultant analyst modeller “f ti ” “procedure” “function” policy maker technical architect IT Traditional analysis techniques rely on figuring out what each person or system calls something – in this example, a process – and it never, ever works. But yet we do it over d d h l f l b ’ h k IT and over again in grand IT projects that always fail. Maybe it’s time to think again ? This slide © Model Futures Ltd 2006‐2007 – Permission for MOD Re‐Use Granted -- 7 of 19 -- The BORO Method • Very simple method – hence it’s easy to get people to sign up to it – …hence it s easy to get people to sign up to it – Uses physical extent as the criterion for identity • Names mean nothing Names mean nothing – No, really – Confusion over naming is a big problem Confusion over naming is a big problem • In IT systems • In conversation i • In taxonomies • Basis of BORO is “Can I kick it ?” Id tifi ti f i di id l b h i l t t – Identification of individuals by physical extent – Identification of classes by their members – Identification of relationships by their ends – Identification of relationships by their ends -- 8 of 19 -- BORO Flowchart Select a Concept for Analysis Does it have spatial and temporal extent ? NO (not indi id al) temporal extent ? (not individual) YES (individual) Does it have members ? YES (type)Add to Ontology NO (tuple) Identify exemplar members, and feed back into BORO Identify things related by tuple BORO process related by tuple and feed back into BORO process -- 9 of 19 -- The Naming Pattern g • Although names are unimportant in BORO, they are quite important to users are quite important to users – Different people use different names for the same things – Different applications also use different names – Foolish to try and impose one naming policy on lti l t k h ld d li ti multiple stakeholders and applications • It’s costly, it never works in large organisations, and it annoys people who are trying to get on with their jobs • Also, different communities use different terminology for good reasons – So we have a naming pattern So…we have a naming pattern • Once the identity of a thing has been established using BORO (i.e. its extent), we can name it as many times as we like with context like, with context. -- 10 of 19 -- The Naming Pattern g • Things have names, and these are related by the namedBy relationship • Names are typed – i e they belong to a set of names Names are typed i.e. they belong to a set of names <<TupleType>> namedBy <<TupleType>> nameTypeInstance <<Type>> Thing <<Type>> Name <<Type>> NameType <<Name>> <<NameType>> <<nameTypeInstance>> <<Name>> “Chair” <<NameType>> Ians Names <<nameTypeInstance>> <<Name>> <<NameType>> <<nameTypeInstance>> “Seat” Michaels Names -- 11 of 19 -- The MOD Ontology Demonstrator gy • Centred around the naming pattern • Derived from a pre existing taxonomy of geo • Derived from a pre‐existing taxonomy of geo‐ political areas – Regions countries bodies of water Regions, countries, bodies of water – Whole‐Part structure – Multiple names added, with context p , – Border information also added • Intended to show power of ontologies for p g integration – Data from multiple sources – ISO 3166, STANAG1059, FIPS10, CIA World Factbook, etc. -- 12 of 19 -- BORO in Use ‐ Borders • Borders data came from CIA World Fact Book : • BORO only deals with the facts presented. In this case: y p 1. There are countries 2. They each have a border y 3. Some parts of their borders coincide -- 13 of 19 -- Analysing Borders y g Country A’s border Country A the borders overlap …in other words, there is a part of one Country B whole border that is also a part of the other y part The border is a part of the country Country B’s border and that’s all the information we have We could add the complications of neutral …and that s all the information we have. We could add the complications of neutral territory (e.g. no‐mans land), but we don’t know that from the data source. BORO is precise and forensic – we are not allowed to invent anything new (rational reconstruction) -- 14 of 19 -- Country has Border Individual tuple IntentionallyConstructed Individual couple GeoPolitical Area GeoPolitical Border wholePartArea Border Country part Country Border borderOf Country whole • A CountryBorder is the entire border surrounding a given Country Th C B d i f h C h • The CountryBorder is a proper part of the Country, hence we use a specialisation of wholePart to relate them ‐ borderOfCountry -- 15 of 19 -- One Country Borders Another Individual wholePart Proper Overlap overlapPart GeoPolitical Border l B d whole Shared Border overlapBorder part part Country Border overlapCountry Border whole p • Classic overlap pattern. Each country has a border, those borders overlap. • The overlapping bit of border is a part of each border F h C t B d S it l d h C t B d Th l • France has CountryBorder, Switzerland has a CountryBorder. They overlap, and the overlapping part is the SharedBorder -- 16 of 19 -- Applications for Ontology in MOD pp gy • Semantic Integration – Current point‐to‐point solutions expensive to maintain – Current point‐to‐point solutions expensive to maintain – Need accuracy of data conversion • Intelligence Analysis Intelligence Analysis – Ability of a 4D ontology to handle change over time lends it well to evidential analysis – BORO method is simple enough to allow subject matter experts to make first‐cut at “ontologising” their information their information • More Geographic Work Location rationalisation (lat long geode GPS) – Location rationalisation (lat‐long, geode, GPS) • Theatre Airspace (ATOs, ACOs) Pilot project underway – Pilot project underway -- 17 of 19 -- Applications for Ontology in MOD 2 pp gy • Contract Management C bi ti f 4D O t l d i tt – Combination of 4D Ontology and naming pattern allows for a temporal contact management system Remembers where people use to work their previous – Remembers where people use to work, their previous ranks, phone numbers, addresses, etc. • Enterprise Architecture • Enterprise Architecture – The discipline of EA covers a broad range of topics and information. information. – EA models are very “webby” – Ontology provides an ideal underpinning for EA Ontology provides an ideal underpinning for EA – IDEAS Model used to exchange DoD (DoDAF) architecture data with MOD (MODAF). ( ) -- 18 of 19 -- Closing g • BORO is proving useful across MOD – Decision makers like defensible information – Decision makers like defensible information. – BORO is defensible – you can prove where semantic overlaps and disconnects exist between systems – Theatre Airspace • IDEAS is a good foundation – Foundation works well and is easy to implement in RDBMS technology P tt t t d d h t k ( h l t – Patterns tested and shown to work (whole‐part, overlap, naming) – We could have chosen OWL & RDF but weaknesses in We could have chosen OWL & RDF, but weaknesses in these are well known – Could have chosen ISO15926, but doesn’t have a f l f d formal foundation -- 19 of 19 --