Supporting Domain Experts Understanding of Process Executions by means of Problem Solving Methods


Published on

Provenance information can be seen as a pyramid with four main levels: Data, Organization, Process, and, on top of the pyramid, Knowledge. The first three levels are focused on how data is transformed across the execution of a process, what the roles of the actors involved are, and which tasks were comprised in it. However, the increasing complexity of distributed data-intensive applications that produce larger amounts of provenance information require more advanced analytical capabilities with a higher level of abstraction. In this regard, we approach knowledge provenance as the provenance perspective focused on providing users with meaningful interpretations of process executions, explaining provenance in a way closer to how domain experts reason on a given problem and facilitating their comprehension. Our approach towards knowledge provenance is based on Problem Solving Methods (PSM). PSMs have been traditionally used in application development as generic and reusable strategies to model, establish, and control the sequence of actions required to accomplish tasks in different application domains. In this work, we use PSMs for a different purpose: we exploit their analytical power as high-level, domain-independent, knowledge templates to support user-focused interpretation of the execution of past processes. Our approach has been implemented as the Knowledge-Oriented Provenance Environment (KOPE) and evaluated through its participation in the Provenance Challenge.

Published in: Technology, Education
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • In our context, tasks can be understood as processes
  • We use the PSM analytic approach to interpret Provenance at the knowledge level
  • Currently only applied to interaction p-assertions. Might be applied to the other types. Annotations constrained to the data exchanged only, and only at domain level.
  • Several types of matching: is-a, subclass-of, part-of, etc Several degrees of matching
  • Supporting Domain Experts Understanding of Process Executions by means of Problem Solving Methods

    1. 1. Monterey, July 26th 2007 KOPE: Knowledge-Oriented Provenance Environment Jose Manuel Gómez-Pérez , Francisco Javier García (iSOCO), Rafael González (UPM), Chris Van Aart (Y’All) FP6-511513 OntoGrid: Paving the way for Knowledgeable Grid Services and Systems
    2. 2. Motivation <ul><li>Goals </li></ul><ul><ul><li>Increase understanding of process execution </li></ul></ul><ul><ul><li>Explain provenance in a way closer to how domain experts reason on a given problem </li></ul></ul><ul><li>Problem Solving Methods [McDermott,1988] </li></ul>Provenance (Source: UoS micro-site) Provenance Pyramid (Source: myGrid)
    3. 3. Problem Solving Methods (PSM) <ul><li>PSM are knowledge templates that </li></ul><ul><ul><li>Establish and control the sequence of actions required to perform a task </li></ul></ul><ul><ul><li>Define the kind of knowledge necessary at each task step </li></ul></ul><ul><ul><li>Hierarchically specify how tasks decompose into subtasks down to the level of primitive actions </li></ul></ul><ul><ul><li>Describe tasks at several levels of refinement </li></ul></ul><ul><li>PSM are domain-independent </li></ul><ul><ul><li>PSM inputs and outputs modelled as generic roles </li></ul></ul><ul><ul><li>Reusable across different domains </li></ul></ul>
    4. 4. Problem Solving Methods Visualization Paradigm Decomposition view Interaction view Knowledge Flow view
    5. 5. Applications of Problem Solving Methods <ul><li>Knowledge Engineering </li></ul><ul><ul><li>Knowledge acquisition : Guidelines to acquire problem solving knowledge </li></ul></ul><ul><ul><li>Reasoning : Enable flexible reasoning by selecting methods during problem solving </li></ul></ul><ul><ul><li>Process analysis : Description of the main rationale of (reasoning) processes </li></ul></ul><ul><li>Provenance Interpretation </li></ul><ul><ul><li>Explain the results of queries on process documentation </li></ul></ul>
    6. 6. Who defines Problem Solving Methods? <ul><li>Ideally, collaboratively defined by a community of domain experts </li></ul><ul><ul><li>Canonical specifications of domain processes </li></ul></ul><ul><ul><li>Agreed throughout the community </li></ul></ul><ul><ul><li>Examples: </li></ul></ul><ul><ul><ul><li>Regulations for good medical praxis </li></ul></ul></ul><ul><ul><ul><li>Diagnosis </li></ul></ul></ul><ul><ul><ul><li>Reasoning ( CommonKADS ) </li></ul></ul></ul><ul><li>Also possible: knowledge engineer with a little domain knowledge, e.g. population-based brain atlas </li></ul>
    7. 7. Provenance Interpretation Workflow
    8. 8. Semantic Resources PSM meta-model Domain ontology Roles (catalogue task) <ul><li>Bridge : Explicit definition of mappings between domain and PSM entities </li></ul><ul><li>Refiner : Specification of task decomposition into subtasks </li></ul>
    9. 9. Annotation of Process Documentation PASOA interaction p-assertion Automatically annotated against the domain ontology during process execution <content> <items xmlns=&quot;&quot;> <item1> </item1> <item2>     </item2> <item3> </item3> </items> </content>
    10. 10. Twig Matching Algorithm <ul><li>twig_join (D, i(P), o(P)) is a boolean function which checks whether a twig exists that connects i(P) and o(P) in D , where: </li></ul><ul><ul><li>P is a problem solving method </li></ul></ul><ul><ul><li>i(P) is the set of input roles of P </li></ul></ul><ul><ul><li>o(P) is the set of output roles of P </li></ul></ul><ul><ul><li>D is the provenance DAG of the documented process, returned by a provenance query </li></ul></ul><ul><li>We consider P as an interpretation of a process if twig_join(D, i(P), o(P)) = true </li></ul><ul><li>Bridges allow detecting occurrences of PSM roles in D </li></ul><ul><li>Refiners allow applying the algorithm recursively across the PSM hierarchy </li></ul>
    11. 11. Twig Matching for Brain Atlas workflow and Catalogue Brain Atlas Provenance Data Flow Prime Catalogue Method Knowledge Flow Brain Atlas Workflow
    12. 12. Demo
    13. 13. Thanks for your attention! iSOCO Valencia +34 96 3467143 Oficina 107 C/ Prof. Beltrán Báguena 4, 46009 Valencia iSOCO Barcelona +34 93 5677200 Edifici Testa A C/ Alcalde Barnils 64-68 St. Cugat del Vallès 08190 Barcelona iSOCO Madrid +34 91 3349797 C/Pedro de Valdivia, 10 28006 Madrid iSOCO Jose Manuel Gómez-Pérez [email_address] #T +34 91 334 9778 #M +34 609 077 103