• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
An Ontology-underpinned Decision-Support System for Wastewater management
 

An Ontology-underpinned Decision-Support System for Wastewater management

on

  • 292 views

 

Statistics

Views

Total Views
292
Views on SlideShare
291
Embed Views
1

Actions

Likes
0
Downloads
1
Comments
0

1 Embed 1

http://www.linkedin.com 1

Accessibility

Categories

Upload Details

Uploaded via as Microsoft PowerPoint

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    An Ontology-underpinned Decision-Support System for Wastewater management An Ontology-underpinned Decision-Support System for Wastewater management Presentation Transcript

    • OntoWEDSSAn Ontology-underpinnedDecision-Support Systemfor Wastewater managementby Luigi Ceccaroni, Ulises Cortés and Miquel Sànchez-Marrè
    • June 26-27, 2002 2Outline Motivating tasks Background information The OntoWEDSS decision-supportsystem with the WaWO ontology Results Conclusions and perspectives
    • June 26-27, 2002 3Motivating tasks Improvement of the modeling of theinformation about the wastewatertreatment process and of wastewatermanagement Solution of complex problems relatedto wastewater using ontologies Integration of ontologies in thereasoning of decision support systems
    • June 26-27, 2002 4Outline Motivating tasks Background information The OntoWEDSS decision-supportsystem with the WaWO ontology Results Conclusions and perspectives
    • June 26-27, 2002 5Ontologies: definition An ontology is a formal and explicitspecification of a shared conceptualization,which is readable by a computer. An ontology describes the shared model ofa domain. Everybody following a particularontology understands all the categories andthe relations comprised in that ontology andbehave accordingly.
    • June 26-27, 2002 6PLANNING / PREDICTION/SUPERVISIONAIMODELSSTATISTICALMODELSNUMERICALMODELSGIS(SPATIAL DATA)DATA BASE(TEMPORAL DATA)USER INTERFACEBackground/ SubjectiveKnowledgeECONOMICCOSTSUSERDecision/ ActuationENVIRONMENTAL/ HEALTHREGULATIONSSpatial /GeographicaldataOn-linedataOff-line dataDATA MININGKNOWLEDGE ACQUISITION/LEARNINGEXPLANATION ALTERNATIVESEVAL.REASONING / MODELS’ INTEGRATIONBIOLOGICAL/ CHEMICAL/ PHYSICALANALYSESSENSORSON-LINE /OFF-LINEACTUATORSFeedbackENVIRONMENTAL SYSTEM / PROCESSDECISIONSUPPORTDATAINTERPRETATIONDIAGNOSISEnvironmental decision-support systems
    • June 26-27, 2002 7Outline Motivating tasks Background information The OntoWEDSS decision-supportsystem with the WaWO ontology Results Conclusions and perspectives
    • June 26-27, 2002 8OntoWEDSS: profile (1) Use of ontologies in domain modeling andclarification of existing terminologicalconfusion in wastewater domain Automatic, reliable discovery andmanagement of problematic states in real-world domains Composition, interoperation and reuse ofdifferent reasoning systems (rule-based,case-based and ontology-based)
    • June 26-27, 2002 9 Environmental process supervision andmanagement distributed in 3 layers:perception, diagnosis and decision support Incorporation of wastewater microbiologicalknowledge into the reasoning process andrepresentation of cause-effect relations Resolution of existing reasoning-impassesOntoWEDSS: profile (2)
    • June 26-27, 2002 10
    • June 26-27, 2002 11WaWO- Frame-based representation- Hierarchy used for:QueriesLanguage analysisReasoning- Standard but specialized:Storm is anOperational-ProblemBacterium is aWastewater-Biological--Living–Object- Metazoan represented:NematodeRotifer
    • June 26-27, 2002 12Reasoningwithontologies Role orPhenomenoncategories Occurrents Relations
    • June 26-27, 2002 13SupervisionSupervisionmodulemoduleRBESDoesRBES’sdiagnosticsexist?CBRSCBRS’sinferenceRBES’sinferenceNoYesNoNoYesNoDoesCBRS’sdiagnosticsexist?RBES’sDiagnostics=CBRS’sDiagnostics?YesCBRS’s >constant β ?YesDoesCBRS’sdiagnosticsexist?NoCBRS’sDiagnosticsYesRBES’sDiagnosticsCBRS’s DiagnosticsRBES’s DiagnosticsCBRS’sDiagnosticsRBES’sDiagnosticsWaWO’sDiagnosticsWaWOReasoningintegration
    • June 26-27, 2002 14Functionalities Input (modeling and execution) List of descriptors to use Weight of descriptors (optional) New-problem’s descriptors values Output (execution) Diagnosis of the current state of the WWTP(with reliability factor) Trace of the reasoning List of actions to take according to the currentsituation
    • June 26-27, 2002 15Interface for data exchange
    • June 26-27, 2002 16Action suggestion Change Sludge-Recirculation-External to 120 Destruction of filaments via chlorine addition Addition of inorganic coagulant Check out Food-To-Micro-Organism-Ratio Remove aeration-tank and clarifier foam Reduce waste-activated-sludge flow rate(FlowRate-WAS)
    • June 26-27, 2002 17Outline Motivating tasks Background information The OntoWEDSS decision-supportsystem with the WaWO ontology Results Conclusions and perspectives
    • June 26-27, 2002 18Database description Initial set: 790 days with 21 quantitative andqualitative descriptors (out of 170) Filters: missing values, labels Final set for CBRS training: 186 days Bulking-Sludge labeled: 29 days (16%) Lack of benchmarks High number of descriptors Multiple labelsProblems
    • June 26-27, 2002 19Evaluation results: CBRS and RBES Focus on themostrepresentativeproblematicsituation:bulkingsludge
    • June 26-27, 2002 20OntoWEDSS evaluation Averagesuccessfuloutcomes:65% Averagesuccessfuloutcomes:88%
    • June 26-27, 2002 21Outline Motivating tasks Background information The OntoWEDSS decision-supportsystem with the WaWO ontology Results Conclusions and perspectives
    • June 26-27, 2002 22Conclusions Research tool to explore the possibilities andthe potential of introducing ontologies intodecision support systems, using anenvironmental domain as case study Creation of an ontology for the domain ofwastewater treatment process Ontological representation of two kinds ofcause-effect relations: micro-organisms ↔ problematic situations state of the plant ↔ suggested actions
    • June 26-27, 2002 23Perspectives Further refinement and update ofcurrent AI modules Simulation and prediction of theevolution of a treatment plant’s state Integration of the ontology with sometemporal reasoning Reasoning with variations/transitionsof descriptors’ values
    • June 26-27, 2002 25…
    • June 26-27, 2002 26…
    • June 26-27, 2002 27Axioms Example of causality axiom: Physical entities may causally affect otherphysical entities Different views of the same entity may bedescribed with different words, definitionsand axioms. Each category in the hierarchy inherits allthe properties and axioms of every categoryabove it.
    • June 26-27, 2002 28Ontologies: languages KIF: meta-format for knowledge interchange Ontolingua: KIF-based; object-oriented using aFrame Ontology; Web interface (on-line collaboration);translation to various languages; large repository RDFS: resources as Web addresses; primitives forclasses and properties OIL: RDFS-based; entirely Web-driven; combinationof frame-based modeling and description logic DAML+OIL: designed for Web-agents; richermodeling primitives (e.g., properties with cardinality)
    • June 26-27, 2002 29Decision-support systems User friendliness Assistance in problem formulation Framework for information capture Specific KBs Integration of different AI systems(RBES and CBRS, generally) Generation of different strategies
    • June 26-27, 2002 30Rule-based expert system These systems express regularities asrules. They typically follow a situation-action paradigm: the set of rules letthem directly suggest what action totake in a given situation. The domain is so complex that causesother than the given action may alsocontribute to a resulting situation.
    • June 26-27, 2002 31Case-based reasoning system These systems express regularitiesand singularities as cases, each ofwhich encodes some effects of anaction under a specific situation. Theyalso follow a situation-action paradigm:the adaptation of the actions taken inprevious similar situations let themsuggest about the current actions totake.
    • June 26-27, 2002 32The chicken-and-egg paradoxin modeling and diagnosis The situations (set of descriptors’ values)cannot be defined without first knowing whatdiagnostics they correspond to. And most diagnostics can be hard to defineas such, until the corresponding situationshave been identified. Expert often have to use trial-and-errormethods.Set ofdescriptor valuesDiagnosticsDIAGNOSISSituationmodeling
    • June 26-27, 2002 33Functional parameters Activation cycle 1 hour (5 min in case of detected emergency) Accuracy (based on focused evaluation) Cost Allegro Common LISPExperiment Numberof dataCorrectclassificationG-1G-2G-381011100%90%70%