Formal framework for semantic interoperability in Supply Chain networks

750 views

Published on

Milan Zdravković, PhD Defense, 9.10.2012, Faculty of Mechanical Engineering in Niš, University of Niš

Published in: Technology
0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
750
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
21
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide
  • Outline of the presentation Research questions 1,2 – “Problemation (problem + motivation) puzzles” 3,4 – Methodology puzzles 5,6 – Implementation puzzles
  • Metaphor of multitasking
  • Complexity and volume of supply relationships , high frequency of transactions between parties. In Supplier Relationship Management, 80% is human effort and 20% information technology. There is a tendency to reduce number of suppliers because of possible relation cost reductions . Costs of SCM up to 8-10% of sales.
  • New organizational forms. Although significant innovation is made in this topic, the essence of the supplier-customer relationships remains the same as in what is considered as traditional supply chains. The economic phenomena, such as globalization, outsourcing, increased demand for customization and specialization do not change this essence. This is the reason why the title of this thesis still refers to the supply chains, and not to the new terms of Virtual Enterprise or Collaborative Networked Organization.
  • First, enterprises in a supply chain need to speak the same language.
  • Source connects to supplier Deliver connects to customer Not all companies have make We can model as far up or down the supply chain as we view important (not limited to two tiers) Customers and / or suppliers can be internal or external Plan (Processes that balance aggregate demand and supply to develop a course of action which best meets sourcing, production and delivery requirements). Balance resources with requirements, Establish/communicate plans for the whole supply chain Source (Processes that procure goods and services to meet planned or actual demand). Schedule deliveries (receive, verify, transfer) Make (Processes that transform product to a finished state to meet planned or actual demand). Schedule production Deliver (Processes that provide finished goods and services to meet planned or actual demand, typically including order management, transportation management, and distribution management). Warehouse management from receiving and picking product to load and ship product. Return (Processes associated with returning or receiving returned products). Manage Return business rules SCOR describes processes not functions. In other words, the Model focuses on the activity involved, not the person or organizational element that performs the activity.
  • Because SCOR spans boundaries of the enterprises.
  • Each of the systems speaks its own language. So, we need a common dictionary, which can be used to reconcile the languages of systems and SCOR. In other words, we need to make implicit SCOR – explicit.
  • English translation of Welsh: “I am not in the office at the moment. Please send any work to be translated”
  • Networking is defined as a simple information exchange for some benefit. Coordinated networking implies aligning activities of two parties . Cooperation also involves resource sharing for achievement of the compatible goals. Collaboration means that common goal is setup .
  • Outline of the presentation Research questions 1,2 – “Problemation (problem + motivation) puzzles” 3,4 – Methodology puzzles 5,6 – Implementation puzzles
  • Supply Chain Management becomes more transparent and decisions are made upon the real conditions. ZAMENI OVU SLIKU
  • Systems are insensitive to the not-so-obvious and non-functional contexts, such as communication culture, etc. You have to be explicit when communicating with a system. Teaser: What do we know about SCOR ?
  • Common misconception: differences between semantic interoperability an semantically facilitated interoperability.
  • A sender's system S is _semantically operable_ with a receiver's system R if and only if the follow condition holds for any data p that is transmitted from S to R: For every statement q that is implied by p on the system S, there is a statement q' on the system R that (1) is implied by p on the system R, and (2) is logically equivalent to q. the receiver must at least be able to derive a logically equivalent implication for every implication of the sender's system.
  • Adding contexts improves expressiveness of a framework if there exist systems S 1 and S 2 , driven by the ontologies O 1 and O 2 , and if there exist alignment between these ontologies O 1 ≡O 2 , the competence of O 1 will be improved and S 1 will be enabled to make more qualified conclusions about its domain of interest
  • Can you consider all this knowledge about SCOR explicit ? Even if it is explicit, is it represented in such a way so it can be computed by the systems.
  • This is why we developed SCOR-OWL and SCOR-Full models. First we represent the implicit knowledge. Now, it can be computed.
  • D escribes an executive role and entails all entities which perform individual or set of tasks within the supply network, classified with the concepts of equipment, organization, supply chain, supply chain network, facility and information system. A gents do not have explicit definition of functions. Functionality is defined as a property of a course, performed by an agent. Hence, agents are functional in a context of a course they execute. The basic formal consequence of the assumptions above is that agents do not exist if they do not perform some course of doable things.
  • C lassifies prescriptions or descriptions (independent of the time dimension) of ordered sets of tasks . C ourse generalizes “doable” or “done” things with common properties of environment (corresponding to the enabling and resulting states, constraints, requirements, etc.), quality (cost, duration, capacity, performance, etc.) and organization (agent and business function). The second necessary condition for a Course is that it has some impact to the environment (a goal, objective or state) and/or it receives some feedback from the environment or it considers some of its features (such as constraint, requirement, rule or assumption). In other words, the course must have its own setting. Subproperties of has-setting are has-postcondition and has-precondition.
  • Setting concept provides the description of environment of a course. It aggregates semantically defined features of the context in which course take place – its motivation, drivers and constraints. T hey must correspond to some quantifiable notions which describe the specific values or states. Otherwise, they would be only of abstract nature. So, the necessary condition for a setting is to be realized by some configured item (to be described later) .
  • Quality is the general attribute of a course, agent or function which can be perceived or measured . Like in the case of Setting concepts, those attributes are only semantically described abstract categories. Hence, they need to be mapped to the actual specific values or states. The necessary condition for the instances of the Quality concept is that they need to be associated to at least one instance of the “configured-item” concept .
  • Function concept entails elements of the horizontal business organization, such as stocking, shipping, control, sales, replenishment, return, delivery, disposition, maintenance, production, etc. Although it may have some qualities associated, the concept of function is an abstract concept, which basic purpose is to semantically define the context of the course.
  • Configured items model state semantics of the resource – physical or information item . Information items are the atomic concepts which can be semantically defined when mapped to other enterprise ontologies . For the expressive process model, it is crucial to define how resources are communicated among activities and their corresponding actors . This is why communicated item concept is introduced. It aggregates specific concepts of Notice (or its child concept - Signal), Request, Response and Receipt .
  • Configured items are characterized by one or multiple states of information or a physical item, assigned numerical (textual or date) value or realized by another configured item . Available states are identified in the analysis of SCOR model and include 25 possible attributes of the configured item, which can be associated to different information and physical items. Some of the examples of the states are: Adjusted, Approved, Authorized, Completed, Delivered, Installed, Loaded, Planned, Released, Returned, Updated, Validated . I nformation items become configured when at least one of their properties is defined or configured, whether this property can be described by numerical, textual or date information; or the state. Sometimes, it is not possible to “configure” the information item with a simple object, such as data type or state. Hence, information item can also be “realized” with a configured item, as a complex property.
  • SCOR-MAP is a central ontology. It imports (blue arrows) domain ontologies, implicit SCOR model represented in OWL (SCOR-KOS OWL), SCOR’s semantic enrichment (SCOR-FULL OWL) and all local ontologies. SCOR-MAP stores the SWRL rules which are used to represent correspondences between all these models. Focus of this paper is on what is inside purple boxes.
  • SCOR-MAP is a central ontology. It imports (blue arrows) domain ontologies, implicit SCOR model represented in OWL (SCOR-KOS OWL), SCOR’s semantic enrichment (SCOR-FULL OWL) and all local ontologies. SCOR-MAP stores the SWRL rules which are used to represent correspondences between all these models. Focus of this paper is on what is inside purple boxes.
  • Outline of the presentation Research questions 1,2 – “Problemation (problem + motivation) puzzles” 3,4 – Methodology puzzles 5,6 – Implementation puzzles
  • SCOR-MAP is a central ontology. It imports (blue arrows) domain ontologies, implicit SCOR model represented in OWL (SCOR-KOS OWL), SCOR’s semantic enrichment (SCOR-FULL OWL) and all local ontologies. SCOR-MAP stores the SWRL rules which are used to represent correspondences between all these models. Focus of this paper is on what is inside purple boxes.
  • Outline of the presentation Research questions 1,2 – “Problemation (problem + motivation) puzzles” 3,4 – Methodology puzzles 5,6 – Implementation puzzles
  • Typically, a photo like that can be associated to infinite pleasure and joy of flying, time is frozen to enjoy the perfect view that only you could enjoy, blood is quickly going through your vens. However, there is also a pessimist perspective: once he lands, no way that this guy will not suffer a serious and complicated bone fracture!
  • When all local languages are translated to universal domain knowledge, this domain knowledge is then used as a facilitator for the communication. The pre-condition for implementing the above scenario is to have all local languages and domain knowledge - formally described, by using the same formalism. When same formalism is used for all those formal descriptions, it is also possible to define correspondences between the notions of the local languages and domain knowledge. Now, domain knowledge can be considered as advanced dictionary, which is used to formally define meanings of all terms of the exchanged messages. The meaning is formally defined because it is intended to be computable or inferred by the different agents for the different purposes. This formal definition aims at bringing closer the symbols, used to formally describe a particular object, to its typical mental representation. With regard to this, the logical positivists strongly argued that the meaning is nothing more or less than the truth conditions it involves. Here, the meaning is explained by using the references to the actual existing (possibly also logically explained) things in the world. The process of the representation of such meanings is called intensional conceptualization. In linguistics, meaning is what the sender expresses, communicates or conveys in its message to the receiver (or observer) and what the receiver infers from the current context (Akmajian et al, 1995). The diversity of the contexts in which the same message is inferred may easily lead to different interpretations of the meaning of this message. The pragmatic meaning considers the contexts that affect the meaning and it distinguishes two of their primary forms: linguistic and situational. The linguistic context refers to how meaning is understood, without relying on intent and assumptions. The situational context refers to non-linguistic factors which affect the meaning of the message. The linguistic context of the meanings depends on the expressivity of the vocabulary used to describe those meanings and a level of abstraction applied in its development. Both factors significantly influence the capability of the receiver to understand the transmitted messages. The expressivity of vocabulary basically refers to the number and diversity of the concepts (and their properties) used to describe one domain of knowledge. The higher levels of expressivity are important for the cases of very specific communication about highly focused issues of the domain. In most cases, it is very likely that the outside listener will not understand the communication between two domain experts. The level of abstraction has more profound impact. The human reasoning of an unknown term is done by attempting to refer to the known related concepts (or truth conditions). When this is not enough to classify a term, humans reduce or eliminate some truth conditions in attempt to infer a more general, more abstract, known term, which may help in understanding the initial one. Sometimes, even more truth conditions are added so the unknown term is specialized to a known one. Hence, existence of the different levels of abstraction of similar terms or groups of terms may certainly help in understanding the domain knowledge. Typically, higher level of abstraction used in development of one vocabulary, implies lesser expressivity and vice-versa. However, the advantages of both factors can be combined by developing different vocabularies whose concepts are referenced to each other. Hence, highly abstract, less expressive knowledge may be related to a very specific one. If we consider the above-mentioned communication between two experts on the focused domain issues, it is clear that the references to the known generalizations of the specific terms would certainly help the outside listener to understand this communication.
  • When all local languages are translated to universal domain knowledge, this domain knowledge is then used as a facilitator for the communication. The pre-condition for implementing the above scenario is to have all local languages and domain knowledge - formally described, by using the same formalism. When same formalism is used for all those formal descriptions, it is also possible to define correspondences between the notions of the local languages and domain knowledge. Now, domain knowledge can be considered as advanced dictionary, which is used to formally define meanings of all terms of the exchanged messages. The meaning is formally defined because it is intended to be computable or inferred by the different agents for the different purposes. This formal definition aims at bringing closer the symbols, used to formally describe a particular object, to its typical mental representation. With regard to this, the logical positivists strongly argued that the meaning is nothing more or less than the truth conditions it involves. Here, the meaning is explained by using the references to the actual existing (possibly also logically explained) things in the world. The process of the representation of such meanings is called intensional conceptualization. In linguistics, meaning is what the sender expresses, communicates or conveys in its message to the receiver (or observer) and what the receiver infers from the current context (Akmajian et al, 1995). The diversity of the contexts in which the same message is inferred may easily lead to different interpretations of the meaning of this message. The pragmatic meaning considers the contexts that affect the meaning and it distinguishes two of their primary forms: linguistic and situational. The linguistic context refers to how meaning is understood, without relying on intent and assumptions. The situational context refers to non-linguistic factors which affect the meaning of the message. The linguistic context of the meanings depends on the expressivity of the vocabulary used to describe those meanings and a level of abstraction applied in its development. Both factors significantly influence the capability of the receiver to understand the transmitted messages. The expressivity of vocabulary basically refers to the number and diversity of the concepts (and their properties) used to describe one domain of knowledge. The higher levels of expressivity are important for the cases of very specific communication about highly focused issues of the domain. In most cases, it is very likely that the outside listener will not understand the communication between two domain experts. The level of abstraction has more profound impact. The human reasoning of an unknown term is done by attempting to refer to the known related concepts (or truth conditions). When this is not enough to classify a term, humans reduce or eliminate some truth conditions in attempt to infer a more general, more abstract, known term, which may help in understanding the initial one. Sometimes, even more truth conditions are added so the unknown term is specialized to a known one. Hence, existence of the different levels of abstraction of similar terms or groups of terms may certainly help in understanding the domain knowledge. Typically, higher level of abstraction used in development of one vocabulary, implies lesser expressivity and vice-versa. However, the advantages of both factors can be combined by developing different vocabularies whose concepts are referenced to each other. Hence, highly abstract, less expressive knowledge may be related to a very specific one. If we consider the above-mentioned communication between two experts on the focused domain issues, it is clear that the references to the known generalizations of the specific terms would certainly help the outside listener to understand this communication.
  • Formal framework for semantic interoperability in Supply Chain networks

    1. 1. Formal framework forsemantic interoperability in Supply Chain networks Milan Zdravković PhD Defense 9.10.2012 Faculty of Mechanical Engineering in Niš, University of Niš
    2. 2. Puzzle #1Why isinteroperabilityimportant fornetworkedenterprises?
    3. 3. Problems of “traditional” supply chains• High-speed, low-cost – Focal partner can’t respond effectively to structural changes in demand• Cost reduction is a key aspect of collaboration – Supplier Relationship Management becomes key aspect of SCM – Number of suppliers is reduced – Only dyadic relationships are managed – High level of integration • Both suppliers and focal partner are having high costs • Supplier suffers from reduced flexibility Why is SCM important for suppliers?
    4. 4. Why is Supply Chain Management important for suppliers What is expensive in SCM?
    5. 5. What is expensive in Supply Chain Management Virtual organizations
    6. 6. Virtual organizations – Supply chains of the future ? Opportunity 1 Opportunity n Configuration Configuration *Virtual Breeding Selection Selection **Virtual Enterprise 1 **Virtual Enterprise n Environment Ent21 Ent2 Ent1 Ent2n Ent11 Ent5n Ent61 Ent3 Ent4 Ent4nEnt41 Ent3n Ent31 Dissolution Dissolution Ent6 Ent5 **Temporary network * Pool of organizations and related of independent supporting institutions that have both enterprises, who join the potential and the will to cooperate together quickly to with each other through the exploit fast-changing establishment of a “base” long-term opportunities and then cooperation agreement and dissolve (Browne and interoperable infrastructure. Zhang, 1999) (Sánchez et al, 2005) Many new forms for the VOs
    7. 7. Collaborative organization forms How the costs of SCM are reduced?
    8. 8. How the costs of Supply Chain Management are reduced What is interoperability?
    9. 9. What is interoperability ?• ISO/IEC 2382 – 01.01.47 interoperability: The capability to communicate, execute programs, or transfer data among various functional units in a manner that requires the user to have little or no knowledge of the unique characteristics of those units.• The main prerequisite for achievement of interoperability of the loosely coupled systems is to maximize the amount of semantics which can be utilized and make it increasingly explicit (Obrst, 2003) SCOR basic management processes
    10. 10. Supply Chain Operations Reference Model (SCOR) : Basic Management Processes Plan-Source-Make-Deliver-Return Plan Deliver Source Make Deliver Source Make Deliver Source Make Deliver Source Return Return Return ReturnSupplier’s Return Return Customer’s Supplier Customer Supplier Customer (Internal or (Internal or Your Company External) External) ..plus
    11. 11. ..plus: Plan Deliver Source Make Deliver Source Make Deliver Source Make Deliver Source Return Return Return Return Return Return• Each of the processes has its own activities, metrics and best practices• Each of the activities has inputs&outputs, metrics and best practices• Each of the metrics has performance attributes• Each of the best practices is implemented by the system Why is interoperability important for SCM?
    12. 12. Why is interoperability important for Supply Chain Management? Plan Deliver Source Make Deliver Source Make Deliver Source Make Deliver Source Return Return Return ReturnSupplier’s Return Return Customer’s Supplier Customer Supplier Customer (Internal or (Internal or Your Company External) External) Interoperability issues Asset flows between two SCOR processes
    13. 13. Assets flows between process elementsfor engineered-to-order production type
    14. 14. Systems do not “speak” SCOR
    15. 15. Puzzle #2Why is ontologyimportant forinteroperability?
    16. 16. “Lost in translation”
    17. 17. Issues source: “Lost in translation”• There is NO lingua franca for enterprises, they all “speak” different languages• However, some are “less different” than the others: – Enterprise models (loose alphabets) – Reference models (strict alphabets) – Ontologies (formal alphabets) What is ontology?
    18. 18. So, what is ontology?• Concepts can be related to other concepts – e.g. with parent and child relations• Concepts can be combined into propositions• Propositions can be clustered into mental models• When all this is specified, what we get is.. – ONTOLOGY
    19. 19. This is ontology
    20. 20. This is also an ontology (more formal and explicit)Concepts ∃p (information(p)), ∃e (enterprise(e)), ∃t (task(t)), ∃g (goal(g)), ∃r (resource(r)),...Propositions (statements) ∃e ∃n (enterprise(e) ∧ enterprise(n) ∧ network-with(e,n)) ∃e ∃n (enterprise(e) ∧ enterprise(n) ∧ coordinate-with(e,n)) ∃e ∃n (enterprise(e) ∧ enterprise(n) ∧ cooperate-with(e,n))Mental models (rules) network-with(A,B) ⇒ ∃p(information(p) ∧ (send(A,p) ∧ receive(B,p)) ∨ (send(B,p) ∧ receive(A,p))) coordinate-with(A,B) ⇒ network-with(A,B) ∧ ∃m ∃n(task(m) ∧ task(n) ∧ responsible-for(A,m) ∧ responsible-for(B,n) ∧ has-precondition (n, status(m,’completed’))) cooperate-with(A,B) ⇒ coordinate-with(A,B) ∧ ∃m ∃n(task(m) ∧ task(n) ∧ responsible-for(A,m) ∧ responsible-for(B,n) ∧ ∃r(resource(r) ∧ consumed- by(r,m) ∧ consumed-by(r,n)) ∧ ∃g∃f(goal(g) ∧ goal(f) ∧ has-goal(A,g) ∧ has- goal(B,f) ∧ is-compatible-with(g,f)) collaborate-with(A,B) ⇒ cooperate-with(A,B) ∧ ∃m(task(m) ∧ responsible- for(A,m) ∧ responsible-for(B,m)) ∧∃g(goal(g) ∧ has-goal(A,g) ∧ has-goal(B,g)) Representational languages
    21. 21. Representation languages for ontology• Less formal – UML (Unified Modeling Language), – E/R (Entity/Relationship) Syntax• More formal – OWL, SWRL
    22. 22. Puzzle #3What is semanticinteroperability(of systems)?
    23. 23. Why systems are good in communication
    24. 24. Why systems are bad in communication Human communication as a raw model for interoperability
    25. 25. Human communication as a raw model for interoperability Providing meaning to Selection of Stimulus sensory energy various sensations sensations In contexts of expectations, experience, Perception Sensation culture, etc. Perception SensationGaining psknowledge and ys ps Cognition Articulation iolcomprehension Cognition Articulation yc og ho icafrom the log l icasensations l Articulating Storage, reasoning, response Receipients, problem solving, imagining, language, means conceptualizing
    26. 26. Requirements for semantic interoperability ∃S(system(S)) Semantic Query Reasoner Mappings matching processing Web Ontologies services Articulation Cognition Ontologies Perception Sensation Sensation Perception Cognition Articulation ∀p ( (transmitted-from(p,S) ∧ transmitted-to(p,R)) ∧ ∃R(system(R)) ∀q(statement-of(q,S) ∧ p⇒q) ∃q’(statement-of(q’,R) ∧ p⇒q’ ∧ q’⇔q)• Sensation • Cognition ) ⇒ semantically-interoperable(S,R) – “Ask” & “Tell” interface – Triple store – No need for selective sensation – Formalized business rules• Perception – Rules-enabled reasoning – Semantic matching and – Assertion of new reasoning knowledge – Explicit enterprise knowledge – Formalized interoperability (ontologies) protocols Implementation of semantically interoperable systems
    27. 27. Implementation of semantically interoperable systems C1 MO1Oi≡f(ML1D1 , MD1D2, MLiD2) Si S1 OL1 ML1D1 OLi MO1O2≡f(ML1D1 , ML2D1) MLiD2 OD1 OD2 ML2D1 OL2 MD1D2 S2 MLnD1 • S1-Sn – Enterprise InformationC2 MO1On≡f(ML1D1 , MLnD1) Systems • OL1-OL2 – Local ontologies OLn Sn • OD1,2 – Domain ontologies Cn • MLiDi – Mappings between local and domain ontologies Adding contexts
    28. 28. Adding contexts improves expressiveness of a framework• if there exist systems S1 and S2, driven by the ontologies O1 and O2,• and if there exist alignment between these ontologies O1≡O2,• the competence of O1 will be improved and S1 will be enabled to make more qualified conclusions about its domain of interest
    29. 29. Puzzle #4Which semantics forinteroperability?
    30. 30. Framework for semantic enrichment of reference models Domain Domain ontology 1 ontology 2 Mapping Mapping Mapping Application rules rules rules ontology 1 Unifying model Semantically Mapping Mapping Mapping Application enriched model rules rules rules ontology 2Reference models Impor Sync Reference models (formats) t tools OWL model tools (native formats) SCOR-KOS OWL model
    31. 31. SCOR-KOS OWL Model• 418 metrics elements,• 166 process elements,• 25 process categories,• 164 best practices,• 282 Input/Output elements and• 108 system elements
    32. 32. SCOR-KOS OWL Model Web app for browsing SCOR-KOS OWL model
    33. 33. Web application for browsing the SCOR model SCOR-Full ontology
    34. 34. SCOR-Full Ontology• Explication of SCOR-KOS OWL• Developed by semantic analysis of SCOR-Full Input/Output elements SCOR-Full concepts
    35. 35. Agent concept• ∀a (agent(a)) ∃c (course(c)∧ performs(a,c))• Not functional
    36. 36. Course concept• Generalizes “doable” or “done” things with common properties of environment, quality and organization• ∀c (course(c)) ∃f (function(f)∧ has- function(c,f))• ∀c (course(c)) ∃s (setting(s)∧ has- setting(c,s))
    37. 37. Setting concept• provides the description of circumstances of a course• ∀s (setting(s)) ∃ci (configured-item(ci)∧ has-realization(s,ci))
    38. 38. Quality concept• general attribute of a course, agent or function which can be perceived or measured• ∀q (quality(q)) ∃ci (configured-item(c)∧ has-attribute(q,ci))
    39. 39. Function concept • entails elements of the horizontal business organization
    40. 40. Resource itemconcepts • Inf-Item defines the semantics of the relevant resource (atomic concept) • Conf-Item describes its dynamics
    41. 41. Configured items• (Inf-Item(?x) ∧ (has-numerical-value(?x, decimal) ∨ has-text- value(?x, string) ∨ has-date-value(?x, dateTime) ∨ (Inf-Item(?i) ∧ has-realization(?x, ?i)))) ∨ ((Phy-Item(?x) ∨ Inf-Item(?x)) ∧ has-state(?x,state(?y))) ⇒ Conf-Item(?x)• Examples – customer-credit(?x) ∧ in-state(?x, Adjusted) ⇒ SameAs (?x, Adjust_Customer_Credit) – return-to-service(?x) ∧ in-state(?x, Authorized) ⇒ SameAs (?x, Authorization_to_Return_to_Service) – product(?x) ∧ in-state(?x, Consolidated) ⇒ SameAs (?x, Consolidated_Product) Logical correspondences
    42. 42. Logical correspondences between implicit and explicit modelbusiness-rule(?x) ∧ return-process(?y) ∧ has-rule(?y, ?x) ⇒ SameAs(?x,Business_Rules_For_Return_Processes)available-to-promise(?x) ∧ time-range(?y) ∧ has-quality(?x, ?y) ⇒ SameAs (?y,Available_to_Promise_Date)capability(?x) ∧ return-process(?y) ∧ has-quality(?y, ?x) ⇒ SameAs (?x,Capabilities_of_the_Return_Processes)production-schedule(?x) ⇒ SameAs (?x, Production_Schedule) SCOR-Full validated
    43. 43. SCOR-Full Validated• All 282 SCOR Input/Output elements (with implicit meaning) are mapped to SCOR-Full concepts – All implicit meanings are now explained (explicated) Adding new contexts: TOVE
    44. 44. Adding new contexts: Logicalcorrespondences between SCOR-Full and TOVE • Facilitates the improvement of the structural and behavioural competence of the SCOR-Full model. Competency: – Whose permission (if any) is needed in order to perform the specific task of selected process element (activity)? – Who has authority to verify the receipt of the sourced part? – Which communication link can be used to acquire specific information?, etc. Formal framework for SC operations
    45. 45. Formal framework for supply chain operations Implicit Explicit Semantic Formal models semantics semantics enrichment of design goals Domain Ontologies SCOR-KOS OWL SCOR-FULL OWL SCOR-CFG OWL SCOR- MAP SCOR-GOAL OWL PRODUCT OWL SCOR Native formats, Exchange formats Sem interoperability of systems in SC network
    46. 46. Semantic interoperability of systems in supply chain network Enterprise Implicit Explicit Semantic Formal models SemanticInformation semantics semantics enrichment of design goals applications Systems Domain SCOR-SYS OWL Ontologies SCOR-KOS OWL SCOR-FULL OWL SCOR-CFG OWL SCOR-based SCOR- MAP systems SCOR-GOAL OWL PRODUCT OWL SCOR Native formats, Exchange formats EIS LOCAL ONTOLOGY database EIS LOCAL ONTOLOGY database EIS LOCAL ONTOLOGY database
    47. 47. Puzzle #5How thissemantics can beused forinteroperability?
    48. 48. Interoperability Service Utilities (ISU)• available at low cost,• accessible in principle by all enterprises (universal or near-universal access),• guaranteed to a certain extent and at certain level in accordance with a set of common rules,• not controlled or owned by any single private entity. S-ISU
    49. 49. Semantic Interoperability Service Utilities (S-ISU)• Take into account the restrictions of the functional approach and it assumes that enterprises should take their own decision on which part of their semantics should be made interoperable;• This semantics is described by the local ontologies. The main objective of the framework is to make those ontologies interoperable;• Minimum technical pre-requirements are foreseen;• The formal framework is not associated with some storage facility;• The formal framework facilitates delivery of the information by combining their sources (namely, local ontologies). – Only meta-information (other than a formal framework - common ontologies) about the interoperable systems is kept centrally; S-ISU: Component view
    50. 50. Component view of S-ISU architectureONTOLOGY DomOnt1 Mapping ProbOnt1 } Local Local Local Ontology Ontology Ontology Ontology DomOntn ProbOntm SemApp 1 EIS Database Native formats Exchange formats } SemApp n SQS ReaS Listener Semantic Apps Main Services EIS RegSApp RegS SRS AuthAppUTILITY ReaS SRSApp TrS Supportive Apps VE formation Services LOCAL CENTRAL S-ISU for semantically interoperable systems
    51. 51. S-ISU for Semantically interoperable systems Enterprise SemanticInformation Implicit Explicit applications Systems semantics semantics and services DOMAIN ONT DOMAIN ONT DOMAIN ONT Reconciliation service PROB ONT MAPPING ONTOLOGY PROB ONT Registration service Reasoning service Native formats, Exchange Semantic formats Query service EIS LOCAL ONTOLOGY Listener database Transformation service EIS LOCAL ONTOLOGY Listener database
    52. 52. Puzzle #6How the systemsare explicated andqueried by usingthe semantics?
    53. 53. Database-to-ontology er.owl entitymapping hasAttribute hasConstraint attribute Database hasType constraint hasSourceAttributeer:entity(x) ∧ not (er:hasAttribute only hasDestinationAttribute type(er:attribute ∧ (er:isSourceAttributeOf hasSourceMultiplicitysome er:relation))) ⇒ s-er:concept(x) output Data import and relatioer:entity(x) ∧ er:entity(y) ∧ er:relation(r) ∧ classification of ER entities n multiplicityer:hasAttribute(x, a1) ∧ er:hasAttribute(y,a2) ∧ er:isDestinationAttributeOf(a2, r) ∧ hasDestinationMultiplicityer:isSourceAttributeOf(a1, r) ⇒s-er:hasObjectProperty(x, y) imports Classification (inference) of outputs-er:hasObjectProperty(x, y) ∧ OWL types and propertieser:hasConstraint(a1,not-null) ⇒ s-er.owl data-types-er:hasDefiningProperty(x, y) hasDataTypeer:attribute and not hasDataProperty Lexical data-concept(er:isSourceAttributeOf some er:relation) hasFunctionalProperty⇒ s-er:data-concept Refinement hasDefiningDataProperty conceper:type(x) ⇒ s-er:data-type(x) t hasObjectPropertys-er:concept(c) ∧ er:attribute(a) ∧ hasDefiningPropertyer:type(t) ∧ er:hasAttribute(c, a) ∧er:hasType(a, t) ⇒ Local ontologys-er:hasDataProperty(c, t) generations-er:hasDataProperty(c, t) ∧ outputer:hasConstraint(a,not-null) ∧er:hasConstraint(a,unique) ⇒s-er:hasDefiningDataProperty(c, t) Query-driven vs massive dump population
    54. 54. Query-driven vs. massive dump population• Massive dump population – Local ontology is pre-populated with database instances – Querying local ontology at a runtime – Performance and synchronization issues Query-driven population
    55. 55. Query-driven population• Querying database at a runtime, real-time access to information• Issues – Centralized inference – all ontologies need to be in the reasoner’s memory space (static imports) – Data security / access authorization Semantic query execution
    56. 56. Semantic query hasResCompany someexecution (hasResCurrency some (hasName value "EUR") Input Query ) subject predicate some|only|min n|max m|exactly o bNode Decomposition subject predicate value {type} X bNode1 bNode2 hasResCompany hasResCurrency hasName some bNode1 some bNode2 value "EUR" SQL construct and execute bNode2 nothing ? Yes No SQL construct Assert to and execute temporary mdl bNode1 nothing ? No Yes SQL construct Assert to and execute temporary mdl X nothing ? No Yes Assert to End result temporary mdl graph
    57. 57. Manufacturing of custom orthopedic implants• Using custom implants over standard ones – Duration of operation decreased – Reliability of operation increased – Period of patient’s recovery reduced – Overal cost of treatment reduced – Risk of complications reduced Case implementation
    58. 58. Case implementation• Proposed models, knowledge and systems infrastructure• Interoperability and semantic interoperability issues analyzed• Infrastructure for collaborative supply chain planning implemented – Supply chain processes configuration problem resolved – Semantic querying of the production schedules for a given part enabled Semantic interoperability framework for this case
    59. 59. Semantic interoperability framework revisited Enterprise Implicit Explicit Semantic Formal models SemanticInformation semantics semantics enrichment of design goals applications Systems SCOR-FULL OWL SCOR- MAP SCOR-CFG OWL OpenERP OpenERP database LOCAL ONTOLOGY Web application for SCOR process configuration
    60. 60. Web application for SCOR process configuration • Features – Development of complex thread diagrams (multiple tiers, additional participants) – Generation of process models and workflows (including PLAN activities) – Generation of implementation roadmap SCOR-CFG OWL ontology
    61. 61. SCOR – CFG OWL, Example of application ontology• Design goal – Generation of SCOR thread diagrams SCOR thread diagram for manufacturing of custom implants
    62. 62. SCOR thread diagram formanufacturing custom implants Interoperability requirements (inferred)
    63. 63. Interoperability requirements(inferred from SCOR-KOS OWL) OpenERP ontology
    64. 64. OpenERP ontology • OpenERP PostgreSQL database with 238 tables is transformed to a local ontology, with 193 concepts, 493 data concepts and 2779 properties Fragment of UML representation
    65. 65. Fragment of UMLrepresentation of OpenERP local ontology Querying OpenERP
    66. 66. Querying OpenERP local ontology• Production schedule for the product (part) with name "Custom fixture F12"• By using SCOR-Full – has-realization some (production-schedule-item and has- product-information some (has-name value "Custom inner fixture F12"))• By using the local ontology of OpenERP system: – mrp_production and hasProductProduct some (hasProductTemplate some (hasName value "Custom inner fixture F12")) Result of query execution
    67. 67. Result of query execution
    68. 68. Conclusions (1/5)• Enterprises will continue to have mixed ICT environments for the foreseeable future – increase of the data complexity – further ICT developments• rate of the heterogeneity in the systems architecture will increase• interoperability is expected to become more critical feature of the EISs Conditional vs. unconditional interoperability
    69. 69. Conditional vs. unconditional (and universal) interoperability• The main pre-determined asset, which is needed so two system can interoperate is a common semantics• Traditional approaches structures interoperability problem into levels – This is not convinient, because individual level cannot be semantically analyzed (by implementing a full ontological commitment) in isolation from the others• Enterprise systems should not be exposed to the interoperable environment by the levels or any other conceptual categories, but by ontologies Possible restrictions
    70. 70. Possible restrictions• incompleteness and lack of validity of logical correspondences between two ontologies• expressiveness of the implicit models, namely local ontologies• expressiveness of the languages, used to formalize those models• restricted access to some of the information, modelled by the parts of local ontology Formalizing domains and systems semantics
    71. 71. Formalizing domains and systems semantics• NOT from the scratch. Issues: – Time and effort – Misbalance of the needed ontological commitment and epistemological dimension – Detachment from the common language of the domain• Task of the EIS conceptualization is not really to conceptualize the EIS models, but: – to make the assumptions on the mental models of the information systems’ designers – to make those models fully or partially equivalent to the real world semantics (ontological commitment)• This task is NOT yet achieved ! – Example 1: lack of logical implications of the cardinality of relationships and existential constraints (mandatory elements) – Example 2: semantics of the populated data rows remain hidden Human communication by logical positivists
    72. 72. Why considering a human communication ? Logical positivists:• The meaning is formally defined because it is intended to be computable or inferred by the different agents for the different purposes – This formal definition aims at bringing closer the symbols, used to formally describe a particular object, to its typical mental representation• The meaning is nothing more or less than the truth conditions it involves. – Here, the meaning is explained by using the references to the actual existing (possibly also logically explained) things in the world. Human communication by linguists
    73. 73. Why considering a human communication ? Linguists:• The meaning is what the sender expresses, communicates or conveys in its message to the receiver (or observer) and what the receiver infers from the current context• The pragmatic meaning considers the contexts that affect the meaning and it distinguishes two of their primary forms – The linguistic context refers to how meaning is understood, without relying on intent and assumptions • Expressivity, levels of abstraction – The situational context refers to non-linguistic factors which affect the meaning of the message • Descriptions of problems - intent Key contributions
    74. 74. Key contributions• 1) Common vocabulary, layered in different levels of abstraction for supply chain relevant systems interoperation• 2) Method for systems explication (conceptualization) and associated method for semantic querying of those systems Further research directions
    75. 75. Further research directions 1/2• General Semantic interoperability – Implementing method for evaluating semantic interoperability of two systems; – Further development of theoretical background for semantic interoperability, by following the principles of human communication;• Formal model for supply chain operations – Further explication of the SCOR-Full domain model by mapping with relevant and/or complementary domain models, such as RosettaNet , UNSPSC , AIAG and STAR , EDI , etc; – Development of new application models and ontologies which directly exploits SCOR-Full domain model; – Top-down validation of SCOR-Full domain model by semantic analysis of the logical correspondences with relevant upper ontologies, such as DOLCE;
    76. 76. Further research directions 2/2• S-ISU Transformation and Semantic Querying Service – Analysis of data patterns with goal to discover the semantics of the ambiguous notions of the local ontologies (e.g. type or status); – Semi-automatic classification of the concepts of local ontologies by analysis of necessary conditions for different concepts; – Developing universal method for semantic query rewriting, where source and destination queries are using the concepts of two ontologies, logically interrelated by using SWRL rules; – Developing method and tools for execution of “Tell” semantic queries;• General Semantic web tools – Implementing distributed reasoning capabilities for modular ontologies with dynamic imports; – Implementing security and access control levels to the parts of ontologies in distributed ontological frameworks; – Advance in performance and quality of ontology matching tools.
    77. 77. Thank you for your attention Q&A Milan Zdravković PhD Defense

    ×