1 Universal Core Semantic Layer Barry Smith, Lowell Vizenor and James Schoening National Center for Ontological Research, University at Buffalo 701 Ellicott Street, Buffalo NY, 14214 Army Net-Centric Data Strategy Center of Excellence, Building 1209, Fort Monmouth, NJ 07703 phismith@buffalo.edu, lowell.vizenor@ctg.com, james.schoening@us.army.mil Abstract – The Universal Core (UCore) is a central element of message content, thereby enabling multiple different types of the National Information Sharing Strategy that is supported by information retrieval, reasoning and consistency checking. multiple U.S. Federal Government Departments, by the intelligence community, and by a number of other national and The UCore taxonomy consists of terms (such as ‘Person’ international institutions. The goal of the UCore initiative is to or ‘Organization’) which are universally understood in the foster information sharing by means of an XML schema sense that they require no domain-specific expertise for their providing consensus representations for four groups of understanding. The taxonomy can thereby be shared by many universally understood terms under the headings who, what, different types of users, and thus it provides the opportunity when, and where. We here describe a project to create an for interoperability over many different sorts of domain- ontology-based supporting layer for UCore, entitled ‘Universal specific exchanges. As M. Daconta expresses it: Core Semantic Layer’ (UCore SL), and describe how UCore SL if I have a UCore-wrapped National Information Exchange can be applied to further UCore’s information sharing goals. Model [NIEM] message from Immigration and Customs Index Terms – Ontology, Data Integration, Semantic Enforcement about illegal immigrants wounded during Technology, OWL DL, Universal Core criminal activity and I have a UCore-wrapped Health and Human Service Department message on visitors to emergency rooms, I have enabled immediate cross-domain search. … I. THE UNIVERSAL CORE UCore is a process of extracting cross-domain commonality The Universal Core (UCore) [1] is a US Federal from your message flows, thereby massively broadening the Government information sharing initiative that is supported possible adoption and use of your shared information. In information sharing, adoption by consumers is the key value by the US Departments of Defense, Energy, Justice, and metric. [2] Homeland Security, by the Intelligence Community, and by a large number of other national and international agencies. The UCore 2.0 taxonomy in its current form is well UCore supports the principles of the Department of Defense adapted to realizing this strategy of information sharing on (DoD) and Intelligence Community (IC) Data Strategies by the basis of universally understood terms. UCore 2.0 as a defining a small set of common data elements that are whole, however, still has a number of problems, including a implemented in a lightweight information exchange schema mismatch between this taxonomy and UCore’s larger XML that is shared across multiple agencies. schema. The latter includes a number of elements that are not The prime focus of the UCore initiative is messaging. represented in the taxonomy, including spatial and temporal UCore is designed to promote information sharing across terms: multiple message domains by means of a simple XML GeoLocation: A physical location with coordinates, or a simple message format built on a taxonomical structure comprising geospatial region; four groups of terms under the headings who, what, when, TimeInterval: An interval in time, defined by two instants in and where. Table 1, below, represents the taxonomy as time. released in UCore Version 2.0, which is the version upon which we focus in what follows. Table 2 represents the Since these elements do not have a corresponding relations contained within the UCore 2.0 xsd:schema. representation in the taxonomy, their intended semantics remain implicit, and no straightforward way exists to link The UCore strategy is to require message-creators to them to, say, spatio-temporal ontologies. construct for each message a digest, a summary built out of a restricted vocabulary of UCore terms, and to link elements II. UCORE AND THE ARMY NET-CENTRIC DATA from the message payload to this digest. Developers of STRATEGY information systems are encouraged to use these terms UCore is designed not only to support messaging and the wherever practical in order to realize the goal of facilitating retrieval and analysis of message content. It is also built in automated sharing of information within and across agencies. such a way as to support interoperability of information To reap maximal benefit from its messaging resources, systems of a variety of different types. The strategy is to have participants in the UCore initiative offer validation processes UCore serve as the consensus starting point for the and tools intended to promote machine understanding of construction of successive layers of more inclusive artifacts, 2 TABLE I: UCORE 2.0 TAXONOMY uc:Entity uc:Event uc:Cargo uc:LivingThing uc:AlertEvent uc:LawEnforcementEvent uc:CollectionOfThings uc:Animal uc:CommunicationEvent uc:MigrationEvent uc:CyberAgent uc:Person uc:CriminalEvent uc:MilitaryEvent uc:Document uc:MicroOrganism uc:CyberSpaceEvent uc:NaturalEvent uc:Environment uc:Plant uc:DisasterEvent uc:ObservationEvent uc:Equipment uc:Organization uc:EconomicEvent uc:PlannedEvent uc:Facility uc:PoliticalEntity uc:EmergencyEvent uc:PoliticalEvent uc:FinancialInstrument uc:Sensor uc:EnvironmentalEvent uc:PublicHealthEvent uc:GeographicFeature uc:Vehicle uc:EvacuationEvent uc:SecurityEvent uc:GroupOfOrganizations uc:Aircraft uc:ExerciseEvent uc:SocialEvent uc:GroupOfPersons uc:GroundVehicle uc:FinancialEvent uc:TerroristEvent uc:InformationSource uc:Spacecraft uc:HazardousEvent uc:TransportationEvent uc:Infrastructure uc:Watercraft uc:HumanitarianAssistanceEvent uc:WeatherEvent uc:InfrastructureEvent creating a growing terminology framework within which III. THE OBO FOUNDRY AND BASIC FORMAL ONTOLOGY there can be threaded interoperability corridors tailored to the needs of specialist groups of users. The idea of creating consistent extensions on the basis of a common core in order to serve interoperability has been Against this background, the Army Net-Centric Data thoroughly explored in the biomedical domain, where the Strategy Center of Excellence is supporting experiments to Gene Ontology (GO) was established already in 1998 [3] to use UCore as the basis for fostering the interoperability of provide a resource for the consistent description of biological information artifacts created by Communities of Interest functions and processes across a multiplicity of different (COIs) in the Command and Control (C2) and other domains. species, including humans. Although initially a logically The idea is that such COIs will create new vocabularies weakly structured set of terms and definitions, the GO was tailored to meet their unique requirements and thus go nonetheless extraordinarily successful in terms of both beyond the narrow set of UCore terms. By providing an numbers of users and of the variety of different types of use. evolving resource of common terms UCore will serve as a As the need began to make itself felt to extend the reach of central hub designed to maintain a joint community the GO through the construction of new ontologies designed perspective. The long-term goal is that these common terms to serve, for example, the description of clinical phenomena, will create a common reference platform allowing data from it was recognized by the GO community that a more diverse COIs to be understood by systems across the DoD systematic approach to logical structure was required in order and IC. This approach is also designed to allow a level of to ensure cross-domain consistency and thereby enable information sharing between unanticipated users and systems integration of data across species and disciplines [4]. and to reduce the time and cost to implement information sharing across the DoD and IC enterprise, while allowing In 2006, accordingly, the Open Biomedical Ontologies COIs to focus on their community specific needs. (OBO) Foundry [5] was established, comprising a suite of ontologies built and maintained in such a way as to be To achieve these ends, UCore will need to accommodate interoperable with the GO. Basic Formal Ontology (BFO) [6] new requirements from its partner agencies, while at the same plays the role of core for the extension ontologies within this time remaining faithful to its key principle of providing a framework [7], so that each ontology is required to employ small set of essential terms and relations. This set will BFO’s restricted set of logically defined ontological relations however need to be expanded in order to include those [8]. Some Foundry ontologies are being created ab initio to universally understandable terms (such as ‘weapon’) not so satisfy the Foundry principles. Legacy ontologies will be far included. UCore has accordingly established a subjected to an incremental process of logical reconstruction Configuration Control Board (CCB), whose role is to manage that designed to ensure that they, and the large quantities of change in such a way that successive UCore versions remain legacy data annotated in their terms, become progressively useable throughout the change lifecycle. linked together in a computable way. 3 IV. UNIVERSAL CORE SEMANTIC LAYER The UCore SL Taxonomy (version 1.0) consists of 144 We describe in what follows an initiative on the part of the terms organized into an is-a (subclass) hierarchy, of which, Army Net-Centric Data Strategy (ANCDS) [9] Center of following UCore 2.0, the top two terms are sl:Entity and Excellence to create an analogous logical infrastructure in sl:Event (see table below), corresponding roughly to the support of the UCore endeavor, focusing especially on the continuant and occurrent terms standardly used in upper-level application of UCore in the creation of domain and COI- ontologies such as BFO. The UCore SL taxonomy specific extensions. The role of logical core is played in this comprehends the entirety of the UCore 2.0 taxonomy in the case by the UCore Semantic Layer (UCore SL), version 1.0 sense that each one of the 55 terms in the UCore 2.0 of which was released on June 15, 2009. UCore SL is the taxonomy is mapped to a corresponding UCore SL term. As a product of work by researchers from the National Center for result, it is possible to translate UCore 2.0 into UCore SL in Ontological Research (NCOR) in Buffalo, with considerable order to take advantage of the latter’s enhanced logical input from the intelligence community under the sponsorship resources . As UCore itself is expanded, additional resources of the Office of the Director of National Intelligence (ODNI) will be added to UCore SL in order to ensure that this CIO. translatability is preserved. UCore SL is designed to work behind the scenes in UCore UCore SL contains 16 relations, with definitions relying on 2.0 application environments as a logical supplement to the those provided in BFO [8]. 12 UCore SL relations have UCore messaging standard. Where UCore 2.0 is based on the counterparts in UCore 2.0. In keeping with the W3C XML format, in which definitions are logically unarticulated recommended best practice for reuse of OWL resources, and thus logically based merging of content is not allowed, ucore:DistinctFrom and ucore:SameAs are not mapped to UCore SL employs the W3C’s OWL DL web ontology corresponding UCore SL relations but rather to language, which allows logically articulated definitions to be owl:differentFrom and owl:sameAs respectively. Four other formulated in such a way as to support such merging. UCore UCore SL relations taken over from BFO do not correspond SL offers the entirety of the content UCore 2.0, both to any UCore 2.0 relations but are included in order to ensure taxonomy and relations, in a form which satisfies the needs of logical decomposability of definitions. These are: inheres_in, users with a need for enhanced logical resources. It provides part_of, participates_in and agent_in. for logical decomposition of terms and definitions, the ability to reason logically on the basis of the content of these TABLE II definitions, and thereby also enhanced support for the creation of consistent extension modules. UCore SL is being UCORE 2.0 AND UCORE SL RELATIONS used as a tool for validation of UCore itself and for the UCore 2.0 UCore SL generation of proposals for changes and additions both to Relations Relations UCore 2.0 and to its extensions. It also provides accessibility of UCore message content to W3C-standard OWL-DL rdfs:subClassOf rdfs:subClassOf technology. ucore:AffiliatedWith slr:affiliated_with Where UCore 2.0 provides for syntactic interoperability ucore:CauseOf slr:cause_of through its XML framework and controlled vocabulary, ucore:Controls slr:controls UCore SL offers a logically organized vocabulary of terms, ucore:DistinctFrom owl:differentFrom relations and definitions which can serve the semantic ucore:EmployedBy slr:employed_by interoperability of UCore message content. ucore:HasDestinationOf slr:has_destination_of UCore SL is already helping to provide semantic ucore:HasFamilialRelationTo slr:has_familial_relation_to interoperability in the results of work sponsored by the ucore:HasOriginOf slr:has_origin_of ANCDS COE on Biometrics and C2 Ontologies carried out ucore:InvolvedIn slr:involved_in by NCOR researchers in Buffalo. We are currently evaluating ucore:LocatedAt slr:located_at the ability of UCore SL to provide more powerful reasoning ucore:OccursAt slr:occurs_at and message-checking capabilities as compared with UCore ucore:SameAs owl:sameAs 2.0 without the added logical support. We and others are also testing the capacities of UCore SL to provide facilities for ucore:SubordinateTo slr:subordinate_to enhanced data sharing by helping to ensure that extension ucore:WorksAt slr:works_at modules created by different domains or COIs, for example slr:agent_in within the C2 framework [10], are created in a logically slr:inheres_in consistent fashion on the basis of logically sound and easily slr:part_of understood definitions. At the same time C2 and Biometric slr:participates_in test extensions are themselves being used to test the adequacy and clarity of UCore SL terms and definitions. 4 V. DEFINITIONS IN UCORE 2.0 AND UCORE SL 3). In UCore SL it is possible to state not merely that The UCore 2.0 definitions are derived primarily from the sl:Organization and sl:GroupOfPersons are distinct, but also Concise Oxford English Dictionary (OED), which, while that they share no instances in common, since UCore SL helpful to human users, unfortunately only goes part of the includes explicit disjointness axioms. way to specifying the intended meaning of the terms in a VI. CURRENT PROJECTS AND FUTURE PLANS fashion useful to computers. A further problem with this C2 Core approach is that there are cases where the provided definition is not in agreement with UCore’s own is-a hierarchy. An C2 Core, a DoD-level initiative pursuing C2 data example is uc:Animal: interoperability, is exploring a combined top-down/bottom-up approach, which both extends semantics down from UCore A non-human organism which feeds on organic matter, 2.0 while also addressing the bottom-up requirements for has specialized sense organs and nervous system, and is information exchange brought by specific user groups. The able to move about and to respond rapidly to stimuli. NCOR team is achieving logical consistency through a top- (Derived from OED) down extension of UCore 2.0 terms, logically defined using Given that uc:Person is a subclass of uc:Animal, this the resources of UCore SL, and applying the result to create a definition entails that a uc:Person is a non-human organism. C2 conceptual data model called ‘C2 Core’. The latter This problem has now been corrected through UCore’s currently contains over 120 high-frequency terms that define change management process by removing ‘non-human’ from the C2 domain. These terms pertain to situational awareness, the definition of ‘animal’, but further problems remain. (For structuring a military organization, planning and assigning example Alert Event is treated by UCore 2.0 is a sibling, tasks, decision making, and assessing progress. rather than as a child, of Communication Event; Weather Examples of potential targets for extensions of the existing Event, similarly, is treated as sibling rather than as child of C2 Core include sub-domains such as Strike, Unit Readiness, Natural Event. Planning and Operations, and the Military Decision Making Other examples of UCore 2.0 definitions are: Process (MDMP). Experience in creating UCore SL has • uc:GroupOfPersons =def A number of people located, yielded a proven process for creating such extensions which gathered, or classed together. (Derived from OED) results in definitions which are optimized for use both by • uc:Organization =def An organized body of people with a humans (for teaching and doctrine writing) as well as use by particular purpose, e.g. a business or government department. computers (in validation and reasoning). (Verbatim from OED) • uc:PoliticalEntity =def An organized governing body with Using UCore SL to Support Reasoning with UCore Messages political responsibility in a given geographic region. (Derived As summarized in [11], we are developing a system which from OED) will allow software agents to better understand and reason The definition of ‘Organization’ does not make it clear with UCore-2.0 messaging content in an approach based once whether or not organizations are groups of persons. The again on the logical resources provided by UCore SL. The definition of ‘PoliticalEntity’ suggests that it should be a underlying idea is to treat the XML-labels used in UCore 2.0 subclass of ‘Organization’, but this is not reflected in the messages as annotations for particulars (for instance UCore 2.0 taxonomy. individual agents) about which these messages contain UCore SL, in contrast, rigorously utilizes the information. Some particulars are referred to in these structure of the taxonomy in the formulation of its definitions. messages directly (for instance the military unit that has been Every UCore SL term is defined in terms of necessary and given an order to move from place A to place B); others are sufficient conditions following the Aristotelian schema, particulars that must exist for the messages to be correctly which defines each child term ‘A’ in terms of its immediate interpretable by software agents and whose existence can parent ‘B’ together with the differentia ‘C’ which determines thereby be indirectly inferred. To make such inferences what it is about the B’s which makes them A’s (as in: a XML-labels are mapped to ontologies based on UCore SL. human =def. an animal that is rational). Examples from Depending on the quality of the mappings, and the quality of UCore SL are: the associated ontologies, more and better inferences can be made about the portion of reality described in the messages. • sl:Government =def. An Organization with political responsibility for governing in a specified GeospatialRegion. We are working on a method to quantify the quality of these mappings and the ontologies in such a way that we can • sl:Organization =def. An Agent that has (1) members which are Agents, (2) one or more Objectives, and (3) MemberRoles (and demonstrate that one ontology is to be preferred over another, other AffiliateRoles) which are realized in the pursuit of the or that one mapping to an ontology is to be preferred over Objective or Objectives another mapping. By using such quantified measures, we can • sl:GroupOfPersons =def. A Group that includes only Persons. engineer an evolutionary improvement of ontology resources, which can be used across the entire domain of messaging in The fact that sl:Government is a subclass of sl:Organization areas such as C2, where tight integration of messages is reflected in both the definition and the taxonomy (see table deriving from disparate sources is required. 5 UCore SL as Basis for a Cyberwarfare Operations Ontology pertaining to cyberwarfare. Doctrine and law can be written While standard military operations doctrine is thoroughly only if experts agree on the semantics of the domain [12]. documented in Joint Publications (JPs), Field Manuals (FMs), Drawing, again, on our experience with UCore SL, we and other reference materials, this is not the case for military propose to identify the semantic content pertaining to operations in cyberspace. Now, however, with the increasing cyberwarfare, defining and establishing relations between the importance of cyberwarfare, there is a need for standardized high-frequency terms that are common to the relevant subject terminological resources which can serve as the basis for matter experts. We will then use UCore SL as basis for a formulation of sound doctrine and also be applied to other Cyberwarfare Ontology, proposing to UCore 2.0 additional purposes such as the development of international law terms for inclusion as necessary. TABLE III: UCORE SL TAXONOMY sl:Entity sl:Event sl:InformationContentEntity sl:Infrastructure sl:Act sl:Analysis sl:Materiel sl:ActOfCommunication sl:Objective sl:Consumable sl:ActOfHumanitarianAssistance sl:ObjectiveSpecification sl:Organization sl:ActOfObservation sl:Opinion sl:Government sl:CriminalAct sl:Plan sl:PhysicalObject sl:ImmigrationEvent sl:TaskSpecification sl:LivingThing sl:LawEnforcementEvent sl:PhysicalEntity sl:Animal sl:TerroristAct sl:Agent sl:Person sl:CyberSpaceEvent sl:Artifact sl:InfectiousOrganism sl:Danger sl:ArtificialAgent sl:MicroOrganism sl:Disaster sl:Equipment sl:Plant sl:EconomicEvent sl:Facility sl:Vehicle sl:FinancialEvent sl:Sensor sl:SpaceRegion sl:EnvironmentalEvent sl:Environment sl:Property sl:Epidemic sl:GeographicFeature sl:Capability sl:EvacuationEvent sl:GeospatialBoundary sl:PhysicalProperty sl:HazardousEvent sl:GeospatialRegion sl:AtmosphericProperty sl:Incident sl:AdministrativeDivision sl:GeographicProperty sl:InfrastructureEvent sl:ControlFeature sl:OceanographicProperty sl:MigrationEvent sl:CoverageFeature sl:SpaceEnvironmentProperty sl:MilitaryEvent sl:GeopoliticalEntity sl:Role sl:MissileLaunchEvent sl:Route sl:AffiliationRole sl:NaturalEvent sl:Track sl:AgentRole sl:AtmosphericEvent sl:Group sl:CargoRole sl:GeographicEvent sl:GroupOfOrganizations sl:ControlFeatureRole sl:NaturalEvent (cont.) sl:GroupOfPersons sl:ControlledSubstanceRole sl:OceanographicEvent sl:InformationBearingEntity sl:InformationSourceRole sl:SpaceEnvironmentEvent sl:Database sl:MaterielRole sl:PlannedEvent sl:Datafile sl:WaypointRole sl:PoliticalEvent sl:Document sl:PublicHealthEvent sl:Program sl:SecurityEvent sl:Website sl:NationalSpecialSecurityEvent sl:SocialEvent sl:StructuralCollapse sl:Task sl:TransportationEvent 6 Crosswalks between UCore SL, DOLCE, and SUMO ACKNOWLEDGEMENTS As part of the design process, UCore SL has a built-in Thanks are due to Paul Birkel, Brian Haugh, William crosswalk between UCore SL and BFO. In an effort to make Mandrick, Randall Dipert, Leo Obrst, Ron Rudnicki, and UCore as widely applicable as possible, additional crosswalks especially to Deborah Nichols, for valuable contributions to to the other major upper-level ontologies DOLCE [13] and the development of UCore SL. SUMO [14] will be created in order to leverage the knowledge sources that utilize these artifacts. REFERENCES UCore 2.0 andJC3IEDM [1] http://ucore.gov/. The Joint Consultation, Command and Control Information [2] http://www.gcn.com/Articles/2009/06/15/Reality-Check- Exchange Data Model (JC3IEDM) is a model that aims to commentary-UCore-info-sharing.aspx. enable international interoperability of C2 information [3] The Gene Ontology Consortium (2000). “Gene Ontology: tool systems at all levels in order to support multinational for the unification of biology,” Nature Genetics, 25: 25–29. (including NATO) combined and joint operations and the [4] Amy Adams, “Your database is talking; is anybody listening? advancement of digitization in the international arena [15]. Biological linguists develop new ontologies for richer, and Recognizing that the integration of UCore 2.0 with relevant cross-database searches,” The Scientist 2005, 19(17):26. portions of JC3IEDM would represent a significant [5] Barry Smith, Michael Ashburner, Cornelius Rosse, et al., “The interoperability gain, and being aware also that a direct OBO Foundry: Coordinated evolution of ontologies to support biomedical data integration,” Nature Biotechnology, 25 (11), mapping between UCore 2.0 and JC3IEDM is likely to November 2007, 1251-1255. produce inexact results, NCOR is exploring the option of use [6] http://www.ifomis.uni-saarland.de/bfo. the logical resources of UCore SL to build semantic bridges [7] http://www.ifomis.org/bfo/users. between the two resources. [8] Barry Smith, Werner Ceusters, Bert Klagges, et al., “Relations VII. CONCLUSION in biomedical ontologies,” Genome Biology, vol. 6, 2006, R46, 2005. UCore SL, an ontology-based supporting layer for UCore, is designed to work behind the scenes in UCore 2.0 [9] Army Net-Centric Data Strategy (ANCDS). http://data.army.mil/. application environments as a logical supplement to the UCore messaging standard. UCore SL builds upon previous [10] Winters, Leslie and Andreas Tolk, “C2 Domain Ontology within Our Lifetime”, Proceedings of the 14th International work in the biomedical domain on creating consistent Command and Control Research and Technology Symposium extensions on the basis of a common core ontology in order (ICCRTS), June 15-17, 2009, Washington, DC. to serve interoperability. UCore SL provides the logical http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA503107& resources for the UCore initiative to do this work. Location=U2&doc=GetTRDoc.pdf. UCore SL is currently in the beta phase of development, [11] Shahid Manzoor, Werner Ceusters and Barry Smith, “Referent with several current and potential users who are testing it in Tracking for Command and Control Messaging Systems”, their application environments and providing valuable Proccedings of the 3rd Conference on Ontology for the feedback in order to help improve future versions of UCore- Intelligence Community, George Mason University, Fairfax, VA, October 21-23, 2009 (in this volume). SL. In order to demonstrate the true value of UCore SL it is necessary to develop a significant user community around [12] Randall Dipert, “Ethical and Ontological Aspects of Cyberwarfare”, to be presented at International Society for UCore SL, one where multiple extension ontologies are Military Ethics Conference, University of San Diego, January subjected to rigorous logical analysis and testing, linked 26-29, 2010. together in computable ways, and used to annotate large [13] http://www.loa-cnr.it/DOLCE.html. quantities of data. In this way it will be possible to show how [14] http://www.ontologyportal.org/. UCore SL’s added logical resources can meaningfully [15] http://www.mip-site.org/035_Public_Organ_DMWG.htm. advance UCore’s information sharing goals.