1 / 17

Representation of Context-Dependent Relevance Relations with Fuzzy Ontologies

Representation of Context-Dependent Relevance Relations with Fuzzy Ontologies. Juan Gómez-Romero , Fernando Bobillo, Miguel Delgado University of Granada Department of Computer Science and A.I. ESWC 2008. outline. Information Overload CDR Ontology Design Pattern

avery
Download Presentation

Representation of Context-Dependent Relevance Relations with Fuzzy Ontologies

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Representation of Context-Dependent Relevance Relations with Fuzzy Ontologies Juan Gómez-Romero, Fernando Bobillo, Miguel Delgado University of GranadaDepartment of Computer Science and A.I. ESWC 2008

  2. outline • Information Overload • CDR Ontology Design Pattern • Fuzzy extension of the CDR pattern • Reasoning with fCDR • Conclusions and future work

  3. information overloadbasics • In general: • People get more information than they can process • In Information Systems: • Users are overwhelmed by the provided information • More information = Less knowledge • In Mobile Systems: • Easier to be “overloaded” with information • Solution: • To provide only relevant information • What is relevant? • User environment, preferences, previous behavior, etc. Context!

  4. information overloadexample • A doctor is attending to a patient outside the hospital • The doctor… • …uses a portable device to consult the patient’s clinical history, which is stored in the HIS, in order to prescribe a treatment • …retrieves a bunch of EHRs about the patient • …filters the results manually and grasps interesting information (it may take too long) Overload! • Solution: Use information about the context of use: • If the patient is “unconscious” and has an “hemorrhagic laceration”… • …information about if he has been diagnosed of “bad reactions to procaine” should be taken into account

  5. outline • Information Overload • CDR Ontology Design Pattern • Fuzzy extension of the CDR pattern • Reasoning with fCDR • Future work

  6. CDR design patterndescription • A design pattern for OWL • Represents which information is relevant in a given context • “Context is any information (implicit or explicit) that can be used to characterize the situation of an entity” (Dey and Abowd, 2000) • Relevance CDR ontology • Imports • Context Ontology: vocabulary to describe context situations • Domain Ontology: ontology to represent domain-specific knowledge • Includes links between context descriptions and domain expressions • Profiles: new concepts that connect contexts and domains

  7. CDR design patternformulation Domain Context Reasoning: To obtain the domain information which is relevant in a given context

  8. outline • Information Overload • CDR Ontology Design Pattern • Fuzzy extension of the CDR pattern • Reasoning with fCDR • Conclusions and future work

  9. fuzzy extension of CDRdescription • With the fuzzy CDR ontology: • Imprecise knowledge can be represented • E.g.: A patient is slightly unconscious • Partial simmilarities between contexts can be represented • E.g.: Anaphylaxis is quite similar to sepsis • Relevance relations hold to a degree • E.g.: Blood-borne diseases are less relevant than drug intollerances • Fuzzy extension of the CDR pattern • The CDR ontology is a fuzzy ontologyfCDR • The rules to create the fCDR ontology are a fuzzy adaptation of crisp CDR rules • fCDR is represented with a fuzzy Description Logic

  10. fuzzy extension of CDRfuzzy DLs • Fuzzy DLs extend DLs to the fuzzy case • Concepts are fuzzy sets • Roles are fuzzy relations • Axioms hold to a degree • Interpretation has fuzzy semantics • New reasoners are required • Fuzzy ontologies can be reduced to crisp ontologies (DeLorean) • Fuzzy ALC: • TBox. Fuzzy GCIs: C ⊑≥ α D (↔ <C ⊑ D, ≥ a >) • ABox. Fuzzy assertions; e.g. <a : C, ≥ a> • Interpretation; e.g.(C ⊔ D)I = C(x)I⊕D(x)I • Gödelimplicationfortheinterpretation of GCIs • Zadehfamilyfortheinterpretation of theremainingconstructors

  11. fuzzy extension of CDRformulation

  12. outline • Information Overload • CDR Ontology Design Pattern • Fuzzy extension of the CDR pattern • Reasoning with the fCDR ontology • Conclusions and future work

  13. fuzzy extension of CDRinference • Ranked-restricted domain I of a scenario E • Obtains the domain information which is relevant in a given context and its degree aggregation: min t-norm a  b greatest lower bound: glb = sup{a : K <t ≥ a>}

  14. fuzzy extension of CDRproperties • Complete • Computational complexity is upper-limited bythe sum of: • The complexity of reducing the fuzzy CDR ontology to a crisp ontology • Quadratic in space for fALC (at most) • It can be reduced to linear if the number of degrees (n) is fixed • It can performed only once under certain conditions • The complexity of the subsumption tests performed (steps 2, 3, 6): • log(n) subsumption tests for each glb • One subsumption test for each concept of the crisp ontology (ExpTime in ALC)

  15. outline • Information Overload • CDR Ontology Design Pattern • Fuzzy extension of the CDR pattern • Reasoning with the fCDR ontology • Conclusions and future work

  16. conclusions and future work • Conclusions: • Fuzzy relevance ontology improves the crisp approach • Context descriptions can be fuzzy • Retrieved domain knowledge can be ordered by relevance • Only the top-k most interesting domains can be provided • Future work : • Currently, we are working on the mobile healthcare problem (other domains?) • Offer a visual-edition tool (Protégé plug-in) • Reduce complexity of the reasoning process (other semantics? optimizations?) • Compare the fuzzy pattern with related proposals

  17. end questions? comments? thank you! ¡muchas gracias!

More Related