- From: Ken Laskey <klaskey@mitre.org>
- Date: Fri, 04 Jul 2003 17:18:46 -0400
- To: Jim Hendler <hendler@cs.umd.edu>, public-webont-comments@w3.org
Jim, Thank you for your email of 9 June 2003 which responds to comments I submitted during the OWL Last Call and my apologies for taking so long in responding back. As you noted, we had extensive discussions during the WWW conference in Budapest and in consideration of those discussions, I would like to make some specific suggestions beyond the more general nature of my previous comments. These suggestions, which currently focus on the use cases, address what I feel are gaps to which I alluded more generally in my previous comments. My intent in making these suggestions is to establish a more complete context for the current OWL contributions and (as my Ph.D. advisor always demanded) the significance of those contributions in advancing the Semantic Web. My suggestions also represent an evolution in the use cases from a set of relevant challenges to a statement which includes how the current work provides an enabler to eventual solutions. I have found this useful in other work because it maintains a clear connection between the initial motivators and the current state of advancement. Also, I believe this is important because while the use cases are comprehensive in illustrating the challenges, it is a disservice to give the impression that OWL in and of itself will provide the solution. Rather, OWL is seen as a critical enabling technology and the distinct OWL contributions should be made clear. The following are specific edits and additions which I propose: <original section="2" paragraph="1"> Ontologies can be used to improve existing Web-based applications and may enable new uses of the Web. In this section we describe six representative use cases of Web ontologies. Note that this is not an exhaustive list, but instead a cross-section of interesting use cases. </original> <comment> Add to the end of the paragraph: Also note that the present OWL charter bounds the scope to that necessary to capture sufficient information to support reasoning applications but does not in itself include the complete semantics required to capture probablistic or conditional aspects which these use cases may imply. </comment> <original section="2.1" paragraph="3"> Of course, such a technique relies on content providers annotating their pages with the web ontology language, but if we assume that these owners will try to distribute their content as widely as possible, then we can expect that they would be willing to do this. </original> <comment> This is a questionable assumption because providers are notoriously bad at creating useful metadata and thus I believe this undercuts your overall argument. Suggested alternate text: Such a technique relies on content providers using the Web ontology language to capture high-quality ontology relationships, and an objective of OWL is to enable sufficiently rich and useful metadata content to motivate the necessary effort. It is a separate challenge to minimize this effort and an ontology language will likely have a greater impact if it can facilitate metadata capture as an nonintrusive part of any information creation process. </comment> <original section="2.2" paragraph="3"> An example of such knowledge would be that a "Late Georgian chest of drawers" is typically made of mahogany. This knowledge is crucial for real semantic queries, e.g. a user query for "antique mahogany storage furniture" could match with images of Late Georgian chests of drawers, even if nothing is said about wood type in the image annotation. </original> <comment> OWL supports equivalence relationships but not probablistic relationships such as "typically made of mahogany". The concept "typically"would likely be application-specific reasoning which might be supported by a value mapping ontology, but this logic goes beyond OWL capabilities. Suggest adding to the end of the paragraph: While OWL in its present form does not intrinsically support such probablistic or conditional associations useful in real semantic queries, application-specific semantics could be encoded in OWL to support such functionality. </comment> <original section="2.3" paragraph="1"> A single taxonomy is often limiting because many things can fall under multiple categories. </original> <comment> The limitation of a single ontology is not in the possibility of multiple categories (relevant multiple categories can coexist in a single ontology) but that the categories included in any single ontology represent the view which the ontology author needed to capture. As a more accurate statement of the problem, I suggest the following be substituted for the original: A single ontology is often limiting because the constituent categories are likely constrained to those representing one view and one granularity of a domain, and the ability to simultaneously work with multiple ontologies increases the richness of description. </comment> <original section="2.3" paragraph="3"> A typical problem for each of these types of users is that they may not share terminology with the authors of the desired content. The salesperson may not know the technical name for a desired feature or technical people in different fields might use different terms for the same concept. For such problems, it would be useful for each class of user to have different ontologies of terms, but have each ontology interrelated so translations can be performed automatically. </original> <comment> In many situations, the usage of terminology within an ontology may not be exact in existing data stores, and we need to leverage existing resources rather than implying that these must be repairedto enable conclusive translation. To encompass this need, I suggest adding the following to the end of the paragraph: OWL does not account for fuzzinessin the alignment of terms in different vocabularies but it can capture sufficient usage examples to support application-specific processing. </comment> <original section="2.5" paragraph="2"> This type of agent requires domain ontologies that represent the terms for restaurants, hotels, etc. and service ontologies to represent the terms used in the actual services. When building the actual services, the information may come from a number of sources, such as portals, service-specific sites, reservation sites and the general Web. </original> <comment> The discussion of agents and services switches the focus a bit from information to the applications. A noncritical suggestion is to add the following text at the end of the paragraph: For a service environment, an ontology language will enable information capture necessary for applications to discriminate and balance among user preferences. </comment> <original section="2.6" paragraph="4"> The interoperation scenarios are dynamic in nature (i.e., devices appear and disappear at any moment as their owners carry them from one room or building to another) and do not involve humans in the loop as far as (re-)configuration is concerned. Given that device functionality can be modeled as web services, the needs for this use case are somewhat similar to the needs for DAML-S (particularly the issues surrounding the expressiveness of the language). The tasks involved in the utilization of services involve discovery, contracting, and composition. The contracting of services may involve representing information about security, privacy and trust, as well as about compensation-related details (the provider of a service may have to be compensated for services rendered). In particular, it is a goal that corporate or organizational security policies be expressed in application-neutral form, thus enabling constraint representation across the diversity of enforcement mechanisms (e.g., firewalls, filters/scanners, traffic monitors, application-level routers and load-balancers). Given that RDF-based schemes for representing information about device characteristics have started to be adopted (namely, W3C's Composite Capability/Preference Profile (CC/PP) and WAP Forum's User Agent Profile or UAProf), an additional need is compatibility with RDF at some level. </original> <comment> In reading paragraphs 4 through 7, paragraphs 5 and 7 feel like disjointed inserts and do not provide significant context or motivation. I suggest combining paragraphs 4 and 6, and incorporating elements of paragraphs 5 and 7 to give the following: The interoperation scenarios are dynamic in nature (i.e., devices appear and disappear at any moment as their owners carry them from one room or building to another) and do not involve humans in the loop as far as (re-)configuration is concerned. The tasks involved in the utilization of services involve discovery, contracting, and composition. The contracting of services may involve representing information about security, privacy and trust, as well as about compensation-related details (the provider of a service may have to be compensated for services rendered). In particular, it is a goal that corporate or organizational security policies be expressed in application-neutral form, thus enabling constraint representation across the diversity of enforcement mechanisms (e.g., firewalls, filters/scanners, traffic monitors, application-level routers and load-balancers). Thus, an ontology language will be used to describe the characteristics of devices, the means of access to such devices, the policy established by the owner for use of a device, and other technical constraints and requirements that affect incorporating a device into a ubiquitous computing network. The needs established for DAML-S (particularly the issues surrounding the expressiveness of the language) and the RDF-based schemes for representing information about device characteristics (namely, W3C's Composite Capability/Preference Profile (CC/PP) and WAP Forum's User Agent Profile or UAProf) directly relate to this use case and the resource infrastructure which will support applications which will negotiate and dynamically configure ad hoc networks. </comment> <original section="3" paragraph="1"> Design goals describe general motivations for the language that do not necessarily result from any single use case. In this section, we describe eight design goals for the Web ontology language. For each goal, we describe the tasks it supports and explain the rationale for the goal. We also describe the degree to which RDF and RDF Schema supports the goal. </original> <comment> This comment generally relates to all subsections of Section 3. The design goals need to address what the Web language will contribute to a perceived challenge and not just what the high-level challenge is. Why is a particular design goal important enough to include and what is OWLs intended contribution to meeting the challenge? To this end, I suggest each design goal include a section for OWL approach and current contributionbefore the RDF(S) Support. This improves the description of where we would like to be (the design goal), how far we had come (the RDF(S) contribution so far), and how the present work (the OWL contribution) moves us forward. Note, in the interest of not delaying this response further, I have not provided suggested text for each section but would be willing to discuss details later. Additionally, the chosen design goals have an uneven connection between them. As a reader, I am looking for a common thread and the presentation does not provide this. I suggest the following replace the current paragraph: Design goals describe general motivations for the language that do not necessarily result from any single use case. In this section, we describe eight design goals for the Web ontology language, in particular those dealing with - using established ontologies - changing established ontologies - interacting established ontologies - detecting inconsistencies across ontologies and instances of use - providing a balance between expressivity and scalability when creating ontologies - avoiding unnecessary complexity which may discourage widespread adoption - maintaining compatibility with other standards - supporting internationalization For each goal, we describe the tasks it supports and explain the rationale for the goal. We also describe the OWL approach in advancing these goals and the degree to which RDF and RDF Schema currently provides support. </comment> <original section="3.1" paragraph="1"> Ontologies should be publicly available and different data sources should be able to commit to the same ontology for shared meaning. Also, ontologies should be able to extend other ontologies in order to provide additional definitions. </original> <comment> In my suggested wording for Section 3, I more generally describe the focus of this section as using established ontologies. The current wording only includes shared use of the same snapshot of an ontology or defining ontology extensions, but general examples of use and extension quickly require the sibling concepts of pruning/contraction, reorganization, and combinations. The need for versioning to track such changes is alluded to in Section 3.2, but acknowledging these capabilities seems an obvious omission here. I suggest the following wording replace the original: Ontologies should be publicly available and different data sources should be able to commit to the same ontology for shared meaning. To ensure the maximum degree of semantic capture, support is needed to identify ontology extensions to provide additional definitions, ontology reorganization to modify the view of a domain without reinventing existing concepts, combining multiple ontologies to allow higher level views to use more detailed descriptions, and locally pruning shared ontologies to eliminate overlap in combined ontologies or to reduce overhead in dealing with ontology details that are not needed for the application at hand. </comment> <original section="3.4" paragraph="3"> Justification: The Web is decentralized, allowing anyone to say anything. As a result, different viewpoints may be contradictory, or even false information may be provided. In order to prevent agents from combining incompatible data or from taking consistent data and evolving it into an inconsistent state, it is important that inconsistencies can be detected automatically. </original> <comment> Adding a section following this on the OWL contribution is especially important because the solution space is not obvious. The author examples (Bucky Beaver or Penelope Ashe) I previously used may be a good basis for discussion of how OWL would support detecting and responding to such inconsistency. It is an extremely important concept because it deals directly with using Web resources as these exist (now and in the future) and does not just put the onus on the resource owner to redo things "correctly". </comment> <original section="3.5" paragraph="3"> Justification: There are over one billion pages on the Web, and the potential application of the Semantic Web to embedded devices and agents poses even larger amounts of information that must be handled. The web ontology language should support reasoning systems that scale well. However, the language should also be as expressive as possible, so that users can state the kinds of knowledge that is important to their applications. </original> <comment> The first four design goals describe realproblems which this and the following, while certainly as important, are less concrete situations. An OWL contribution section following the justification would be useful to indicate how the OWL 1.0 spec is believed to have accomplished a reasonable balance. I understand OWL Lite, OWL DL, and OWL Full, but a few words here would provide context and continuity. </comment> <original section="3.6" paragraph="3"> Justification: Although ideally most users will be isolated from the language by front end tools, the basic philosophy of the language must be natural and easy to learn. Furthermore, early adopters, tool developers, and power users may work directly with the syntax, meaning human readable (and writable) syntax is desirable. </original> <comment> Much the same as I said for Section 3.5. Something more germane to OWL is needed rather than a general motherhood statement that ease of use is good. In addition, OWL will require a firmer grounding in the underlying concepts than say did XML because for OWL the encoding of tag syntax is a trivial problem compared with organizing ontology concepts in a domain. While tool support will be important, keeping the underlying concepts accessible will be much more critical. </comment> At this stage, I have not gone into detailed comments on OWL syntax because I see no need to create further delay that would be introduced by suggesting revisions of existing syntax and the associated implementations to incorporate noncritical changes. However, the one shortcoming that seems to be easily addressed is in the area of disjoint classes. In section 5.3 of the OWL Language Guide, it notes that "the number of disjointness assertions grows proportionally to n^2". The rationale that "n is typically small" is the sort that is often quickly proved wrong. I would suggest the addition of a mutallyDisjointWith syntax something like <owl:mutuallyDisjointWith> <owl:disjointClass rdf:resource=class1/> <owl:disjointClass rdf:resource=class2/> <owl:disjointClass rdf:resource=class3/> <...> </owl:mutallyDisjointWith> Again, I apologize for the delay in submitting this response but my organization change has proven to be time consuming and I believe the changes I propose here are consistent with the issues we have previously discussed. I would be open to discussing these further as would be useful. Ken ------------------------------------------------------------------------------------- / Ken Laskey \ | MITRE Corporation, M/S H305 phone: 703-883-7934 | | 7515 Colshire Drive fax: 703-883-1379 | \ McLean VA 22102-7508 / ------------------------------------------------------------------------------------
Received on Friday, 4 July 2003 17:22:17 UTC