Decision Mtg 15 Agenda
From Decision XG
Decision Meeting #15 Goal & Agenda
The dial-in information for this week's meeting is here. The Decision Incubator Activity is intended to explore the use cases and design of a Decision Representation format (e.g. in XML, RDF, OWL…) available on the wiki along with its use by participants in testbed settings for experiments and exercises. Each meeting will be an opportunity to summarize progress & contributions to the wiki, review next steps and action items, and to review the current representation format.
Summary from the Last Meeting
At the last meeting #14 (see minutes), three major topics were on the agenda. First up on the agenda was topic (1) Paper Status and Remaining Fixes. We have a position paper summarizing our incubator work appearing as a poster at the Workshop on Ontology Patterns (WOP) at ISWC 2010. The reviewers had valuable comments and we have now made the appropriate edits. At the meeting, we discussed the additional opportunity to contribute modeling problems which might be selected for work to be accomplished at the workshop. Three problems were proposed, one for ordering lists, another for modeling questions, and a third for modeling metrics.
A second topic on the agenda was (2) Modeling a Question as a Statement with Variable, Referencing using Quads, and rdf:Sequence for Options. The latter involves the best practice for ordering lists of things, such as by using the rdf:Sequence. Eva posted a link to a Collections Ontology which addresses this problem also for consideration. The discussion progressed to considering Named Graphs as an efficient solution for handling a way to reference a triple (or in this case a quad) such as when in a decision one wants to reference a "question" not simply as a string but as a triple. The Named Graph provides a 4th element, a URI, which specifies the context (the subset of triples to which this triple belongs), and this context could be limited to a single triple reference if desired. The discussion then turned to techniques for modeling a question, perhaps as a statement with a variable. Marion suggested considering upper ontologies for support and Eva suggested looking at question/answer literature and the concept of ensuring that the answer format matches the desired question, such as a "Who" question should return a "Person". Marion noted the importance of allowing for counter factuals to support coming up with a reference to a machine understandable version of the question that actually allows one to try to inference a solution.
A third topic was (3) Update and Insights on "Core Decision" Model Issues for Decisions. Piotr noted the importance of considering boolean questions. Eva suggested the situation pattern might be useful for modeling criteria. Piotr noted that he would need a few more days to explore this and in fact, at today's meeting, we will explore the model for criteria that Piotr has created based on the situation pattern.
As a final topic on the agenda (4) Status of Web-based Tool for Decisions with Linked Open Data, Jeff briefly summarized the concept of using options with weighted metrics for both filtering to a reduced set of options and for ordering the remaining options, and how linked open data could be used to drive both the options and to use the properties of the given data set as potential metrics.
Goal for this Meeting
At this fifteenth meeting of the Decision Incubator, we will take the opportunity to review Piotr Nowara's model for criteria based on the situation pattern. The model is both simple and clean, yet raises interesting issues about what distinguishes levels of determinism in making choices, which might be characterized in English by words such as a "process" versus a "determination" versus a "choice" and how the word "decision" can be used or represented in these different contexts. We'll consider different use cases, such as "software module checking", "purchasing a computer", "selecting a city for a vacation", or "selecting an evacuation route in an emergency", and see the interesting considerations when applying this model.
The following agenda items will be discussed at the next meeting. The topic, time, summary and discussion leader are provided for each agenda item.
(0) Review of Status, Goals & Agenda Overview (5 min, Jeff)
We'll briefly consider the current status and agenda items for any suggested changes.
(1) Paper & Ontology Modeling Problems Submission Status (10 min Eva/Marion/Jeff)
We will briefly discuss the current status of our paper and ontology modeling problem submission status and note any needed follow-up or actions
(2) Modeling Criteria Utilizing Situation Pattern (15 min Piotr)
Piotr will review his ontology model for criteria based on the situation pattern. We will consider how the model is constructed and its meaning by looking at the ontology layout and the provided example in the domain of software module checking.
(3) Implications and Next Steps for Adding to the Criteria Model (15 min, all)
After Piotr's review of his model, we should consider how the model would support representing criteria from other use cases ("purchasing a computer", "selecting a city for a vacation", "selecting an evacuation route in an emergency", etc.) and the important implications of the model overall. For example, we should consider the types of competency questions which it addresses. The model also raises some interesting thought-provoking issues about determinism versus choice (free will). The terms "Process" versus "Determination" versus "boolean (yes/no) filters" versus "numerical weighting and ordering" versus "Choice" and the implications for defining and modeling these and how one can be morphed into the other by simply being more specific and restrictive about the criteria will be considered.
(4) Next Steps for Overall Decision Modeling (15 min, all)
Based on the progress and discussions, the participants will consider the next modules for building out the decision model. Recommendations and action items will be considered to help guide and define progress for our next meeting.