Ontology Summit 2014 Theme: Big Data and Semantic Web Meets Applied Ontology

Track A: Common, Reusable Semantic Content, Mission Statement

Community Input Solicited

Please add your input as one-liners or short paragraphs, in bullets below, and make sure your include your name and date at the end for attribution, tracking and following-up purposes. Thanks. -Track-A co-chairs

Page Contents

• Summary of community email discussions, Feb 6 2014
• Initial list of questions related to reusable semantic content

Summary of Email Discussions

The following bullets attempt to distill the main discussion items in the various email threads of Track A. Please review, supplement and comment on the bullets, following the general guidelines above. These bullets, with your inputs and updates, will become the Track A "content synthesis". Thank you!

• Separate reuse of classes/concepts, from properties, from individuals and from axioms
  • By separating these semantics (whether for linked data or ontologies) and allowing their specific reuse, it is easier to target what is possible to reuse and reduces the amount of transformation and cleaning that is necessary
• Define and discuss concept naming
  • Names can be surrogate or human-readable identifiers, both approaches have their advocates, and pros and cons
  • Labels as documentation (such as from SKOS) are valuable regardless of the identifier scheme that is chosen
  • Better tooling is needed to create and use labels for searching ontologies and linked data
    • Tooling requirements in development environments are different than in repository or ontology search
• Discuss reuse issues in general
  • There are parallels and differences with software reuse
  • It is important to capture and understand the explicit range of conditions, contexts and purposes for which an ontology or linked data can be "safely" reused
  • Some specific items for consideration (capture and retrieval) are:
    • Content is accessible and can be found
    • The re-user is motivated to find the content
    • The content is in a form conducive to re-use or can be converted/transformed to a usable form
    • The re-user knows how to do the conversion/transformation
    • The content is logically consistent with the micro-theories of the re-user and this can be established
    • The re-user trusts the content and its quality, and believes that this quality will be maintained
  • It is important to have repositories and supporting metadata and tooling for any reusable content
• Define necessary metadata and possible repositories for content
  • It is necessary to have topical ontologies and linked data schemes in repositories with good search capabilities
  • Search requires metadata whose definition should be started by the work of Track A
    • Some possible metadata includes context, use cases, labels, governance information, etc.
    • A possible metadata definition is the Ontology Metadata Vocabulary
  • One possible repository is the OpenOntologyRepository
  • Reuse is enhanced by feedback and user input - Possibly both a recommendation system and feedback mechanisms should be available in the repository
  • Governance of the ontologies or schemes is critical and needs a process
    • The process should include open consideration, comment, revision and acceptance
    • It is important that multiple domains be represented such that the ontologies and schemes represent "common needs"
    • It is important to resist a single domain focus
• Collect and document approaches to modularization, best practices and specific patterns
  • It appears that small, more modular ontologies and schemes have more possibilities for reuse due to greater focus and cohesiveness, and likely less dependency on the original context
  • Dimensions of variability should be understood and addressed to improve modularity
    • There is variability across the contexts (for example, a certain concept or property may be present or absent in different contexts and uses)
    • There is variability over time (which addresses the evolution of a module and the need to take current trends and future directions into account so that the module is not obsoleted)
  • Tooling for modularity, documentation, etc. are critical
  • Some examples of best practices are:
    • "Integrating" modules may be defined for an application or domain - They employ owl:equivalentClass and OWL axioms to map between the concepts, properties, etc. of the complete set of modular ontologies that address an application/domain
    • It is important that each module and its concepts, properties, axioms, ... are well-documented via well-established labels and predicates
    • Patterns of concepts should be separated from patterns of usage, analysis, traversal and diagnosis
    • Modularity should be viewed from the perspective of the user, not the creator
    • Plans for variability and change should be documented with the modules
    • One must distinguish and separate constraints or axioms that are definitive (that "define" the concepts and are necessary in the core module) versus ones that are pragmatic (related to the business uses or a particular domain)

Reusable Content - Potential Foci for Track A

-- The following list was generated by the Track Co-Champions and included in the intro talk by GaryBergCross

• How can we characterize or measure semantic content reuse, both between ontologies and by Big Data and Semantic Web communities?  

• What building blocks of common semantic content exists now to enable interoperability?
  • What additions are needed to move forward and how are these best achieved?

• What is involved in reuse of Linked Data versus reuse of ontologies?

• What is an example of a small set of semantic content that the community might propose for reuse? 
  • Is there agreement on these or things like ODPs as building blocks?

• What is an example of a large set that the community might propose for reuse?  

• Is it reasonable to expect reuse of an entire ontology like DOLCE and Semantic Sensor Network (SSN)? 
  • If so under what conditions might this be reasonable?
  • Is it better to expect alignment rather than exact content reuse?

• Is reuse about semantics alone or should it also address reasoning and data analytics?

-- Ref. thread started by MikeBennett on 2013.01.20

To kick off Track A, we would like to ask:

• Is there a good body of common, re-usable semantic content which may be put to work in almost any ontology application?

• If so, would the same material be appropriate as a common semantic reference for Linked Data? For Big Data? Or are the re-use questions different?

• Is this even a matter of re-use, or should we be thinking more in terms of ontology design patterns? Or is there some more appropriate approach to making use of the different ontology resources that are available out there?