November 1999 - tdwg-content

RDF Schema design pb: inheritance of properties
by Jean-Marc Vanel 30 Nov '99

30 Nov '99

Hello We are designing XML vocabularies for the description of biological species. See : http://jmvanel.free.fr and Taxonomic Databases Working Group - Structure of Descriptive Data <TDWG-SDD(a)USOBI.ORG> (mailing list) archive available at http://usobi.org/archives/tdwg-sdd.html This is a large project: 250 000 species of plants, 1 000 000 of insects, relations beetwen plants and insects, etc... There is a containment hierarchy (e. g. a flower contains petals) which raises no problems in XML. But there is an inheritance hierarchy that I don't know how to handle using W3C's specifications alone. A genus has properties values (e. g. flower/petals/color = 'white' ) that MUST be inherited by the species in this genus. The properties values of a genus CAN be overriden by some species in this genus. It seems that the "refines" property in XML Schema would bring the desired semantics, together with the "Schema Information Set Contribution: augmentations to instance information sets which follow as a consequence of schema-validation". But the uncomplete state of the W3C document makes me unshure. Although RDF and RDF Schema's data model are very promising, allowing to : - make links between plants and insects, - state properties about properties (some are morphological, others are chemical, etc ...) , I don't see anything clearly specified in RDF and RDF Schema for inheritance of properties. Would this be a solution: - each instance of genus is a rdfs:Class, e.g. all daisies belong to (are rdfs:subClassOf) the genus Chrysantemum; - a genus has Class properties, - a species, instance of species, and subClassOf a genus, redefines some of the genus' properties, overrides the other properties, and adds properties unspecified by the genus. Sample RDFS: <rdfs:Class ID="Chrysantemum"/> <rdfs:subClassOf rdf:resource="http:botany.org/taxonomy#genus"/> <flower><petals> <color>white</color> <number>10--40</number> </petals></flower> </rdfs:Class> <species> <rdfs:subClassOf rdf:resource="Chrysantemum" /> <name>maximum</name> <flower><petals><color>yellow</color>  </petals></flower> </species> Info. for computer men: biology has a hierarchic classification of species whose 4 lowest levels are: order family genus species

1 0

Re: Item description data in XML (XML)
by Leigh Dodds 30 Nov '99

30 Nov '99

> > General question: Is there any reason why attributes exist at all? Robin Cover covers some of the issues here: http://www.oasis-open.org/cover/elementsAndAttrs.html This debate is also currently raging on the XML-DEV mailing list. There are proponents for both sides of the debate, which can get quite heated! If you care to dip in you can look at the thread here: http://www.lists.ic.ac.uk/hypermail/xml-dev/xml-dev-Nov-1999/0998.html There are strong arguments on either side - even the 'attributes are required for linking' viewpoint isn't firm, as <leaf shape="obovate" /> <leaf> <shape>obovate</shape> </leaf> are semantically identical. It largely depends on which you prefer, and how your schema is defined. The id/idref linking mechanism only works within documents. Its unlikely that this will apply in this context as links will be likely *between* documents - and we'll therefore be using XPointers (which are based on XPath). Its up to the application to decide whether it looks for a shape attribute or element. Cheers, L.

1 0

Re: Markup of text descriptions Vs. structured data (GEN)
by Gregor Hagedorn 30 Nov '99

30 Nov '99

Jim Croft wrote: > Can you explain in more detail why they should be fundamentally different? > Aren't the differences just a matter a matter of degree, different points > on a continuum of data as it were? And aren't the basic principles > applicable across the whole? >>The issues overlap, and it would be beneficial to use as much common >>syntax as possible, but fundamentally I believe them to be quite >>different. I completely agree with you that the issues overlap and should be treated together. However, I still see them a distinct, so that they can be treated by separate groups of commands, that use a common system. But maybe that is only to try to protect my brain from blowing up in confusion :-) Could you perhaps give examples where you see the continuum? My differences (see also the example in the other post) are: o Language in free text that is marked up may have different language, while the underlying data should be language independent o Similarly, on a more subtle level, the concepts used by one authors may only approximately map unto the concepts developed in the character definition o Data may be more detailed, which unambigously may map unto a course natural language representation o Much additional knowledge management attributes, mutliple authoring/editing/revision layers, annotation and assessment, quality control, status regarding data source and passing information up and down taxonomic hierarchy, assumptions about possible misinterpretations, etc. Much information may be used for markup as well as for data, so there will be overlap. However, for example, for a given character definition multiple independent hierarchical classifications (by morphological structure, perhaps according to different schools, by methods, by function) may be defined; the natural language definition will however usually only use a single classification. This classification is superfluous in the item description part of the data, but it is desirable to add to the markup. Or isn't it? But then we could not use the idea to just use a coarse structure-only markup as a first step to limit the scope of XML queries, an idea forwarded by Don Kirkup and Bryan Heidorn at the TDWG meeting. Gregor ---------------------------------------------------------- Gregor Hagedorn G.Hagedorn(a)bba.de Institute for Plant Virology, Microbiology, and Biosafety Federal Research Center for Agriculture and Forestry (BBA) Koenigin-Luise-Str. 19 Tel: +49-30-8304-2220 14195 Berlin, Germany Fax: +49-30-8304-2203 Often wrong but never in doubt!

1 0

Re: Item description data in XML (XML)
by Gregor Hagedorn 30 Nov '99

30 Nov '99

> My XML based application may not render the data at all - it might > just use the data to perform some other functionality - lets not > get too fixed up on the visual representation. OK, you are probably correct. > XSL is based on XSLT which is a means of taking an XML document > and producing another document from it (not necessarily XML). > XSLT currently supports three output methods - XML, plain-text > and HTML. Thanks for the insight into XSL and XSLT. I think I understand the context much better now; I was really thinking in terms of stylesheets alone. The problem is really that I believe using XML is a good idea because it allows reuse of existing tools, but then I really don't have any experience with an existing tools other than IE5, which does not do much. That in principal it is always possible (and necessary) to write custom applications for custom functionality is quite understood. However, it would be nice to have generally accepted mechanisms, that allow to use certain feature right out of the box. When I was asking about how to manage free text with markup and data _efficiently_, I was thinking about how to use standard tools (which don't exist that much yet) to edit the different views of the data, much more than thinking about reporting/screen display with a stylesheet. --- Also, I think I did not do a good job in explaining where I see the problems (which may be due to my lack of understanding XML). Lets go back to examples. Markup of the text: leaves usually obovate (except in young leaves), 3 cm long could look like <leaf>leaves <shape> <frequency modifier>usually</frequency modifier> <categorical value>obovate</categorical value> <reportable note>(except in young leaves)</reportable note> </shape> , <length>3 cm<length> long </leaf> Thats already quite complicated. But how do I add the data? The data contain, e.g. an identifier for obovate (i.e. an alphanumeric code or some GUID) to identify the term obovate in relation to some character schema. Note that the character schema may use the term obovate as identifier, so it could be identical. However, German, Chinese, or Russion descriptions will have quite other text within the markup... Further the data would contain more detailed information what frequency "usually" really is, where the original information has been stored, it may contain in addition to the 3 cm information whether "3 cm" is a single measurement, a guess, or the average of 30 measurements, for each of these informations. Yet another level is the knowlegde management, with authoring, editing/revision, quality control, multiple annotations etc. Welding all this into the above structure is quite complicated, so I am looking for general guidelines, how this can be achieved efficiently. My feeling is, that attributes within elements are not the solution, but I may be wrong. In Don's example <leaf shape="obovate" length_mm="31"/> It is difficult to add information about obovate or length. Noel already noted that we may be running into this "properties of properties" problem. General question: Is there any reason why attributes exist at all? What is the advantage of <leaf shape="obovate" length_mm="31"/> over <leaf> <shape>obovate</shape> <length>31</length> <MeasurementUnit>mm</MeasurementUnit>=" </leaf> ? Gregor ---------------------------------------------------------- Inst. for Plant Virology, Microbiology, and Biosafety Federal Research Center for Agriculture and Forestry (BBA) Gregor Hagedorn Net: G.Hagedorn(a)bba.de Koenigin-Luise-Str. 19 Tel: +49-30-8304-2220 14195 Berlin, Germany Fax: +49-30-8304-2203 Often wrong but never in doubt!

1 0

Re: (XML)Pedagogy
by Leigh Dodds 30 Nov '99

30 Nov '99

>Just say for the moment that we'll have several vocabularies by large domains: >biological descriptions (generalities) >botany >zoology >Virology >Microbiology >geography >ecology Just to comment here, one approach being taken by the XHTML working group is to take modular approach with future version of the specification (XHTML 1.1. onwards). This allows defined portions of the spec to be replaced with vendor specific extensions, e.g. replace all the form elements with some extended/amended syntax, or the same with tables, etc. Such an approach might benefit this effort, in that a core set of descriptive markup for characters can be defined, and then allow additional modular extensions for the differing disciplines. This has the advantage of providing a small initial spec and data model which can then be extended. (an approach thats worth considering even without modularity included in the final design). L.

1 0

Re: Item description data in XML (XML)
by Robert A. (Bob) Morris 30 Nov '99

30 Nov '99

[Forgive me if this duplicates. My mailer claims I sent it, but I never got it and I don't see it in the archives] Gregor Hagedorn writes: > > It is difficult to add information about obovate or length. Noel > already noted that we may be running into this "properties of > properties" problem. > > General question: Is there any reason why attributes exist at all? > What is the advantage of > > <leaf shape="obovate" length_mm="31"/> > > over > > <leaf> > <shape>obovate</shape> > <length>31</length> > <MeasurementUnit>mm</MeasurementUnit>=" > </leaf> > Popular wisdom often asserts there is no difference, but there are important differences for database work which are less so (except 1 below) for documents (hence the popular wisdom). These things are true of attributes in XML but not necessarily of elements (the tags + stuff between them) depending on the grammar defining the vocabulary: 1. Attributes are order independent 2. There can only be one attribute,value pair for each attribute. 3. With DTD's (which are the wrong thing to consider here but that is another story), you can enumerate the permitted values of attributes. You can work around this for elements but it may be pointless, verbose, and fragile, especially because of the order dependence of elements. 4. (Subtle and technical)References to XML objects essentially must appear as attributes. This is the only way you can get non-tree graph structure in XML documents, and also is the way you would make a data value conveniently be another object. So if "obvate" were actually some complex object that held more information, you would probably find it named as the value of an "idref", not as text data. While specimen and species documents may prove to representable by trees, keys probably will not be.

1 0

Re: Item description data in XML (XML)
by Noel Cross 30 Nov '99

30 Nov '99

On Tue, 30 Nov 1999, Gregor Hagedorn wrote: > > General question: Is there any reason why attributes exist at all? > What is the advantage of > > <leaf shape="obovate" length_mm="31"/> > > over > > <leaf> > <shape>obovate</shape> > <length>31</length> > <MeasurementUnit>mm</MeasurementUnit>=" > </leaf> > In general a tag can provide the same function as an attribute, as in the case you mention, but there are some situations where attributes are useful. For example, in cases where text is being marked up and the text itself cannot be altered, new data can be stored in attributes. In this case country IDs can added without altering the text: "They can be found in <country id="EN">England</country> and in <country id="IT">Italy</country>." Also in cases where you don't want your tagged items to be rendered by a browser. For example, if you add some RDF metadata: <rdf:RDF> <rdf:Description about="http://www.veggies.org/carrot.jpg"> <s:Family>Apiaceae</s:Family> </rdf:Description> </rdf:RDF> The this case the word "Apiaceae" would be rendered in the page, while in this case: <rdf:RDF> <rdf:Description about="http://www.veggies.org/carrot.jpg" s:Family="Apiaceae" </rdf:RDF> the word "Apiaceae" would not appear in the page. There are probably more fundamental differences that Liegh, Bob, and others could come up with -- these are just two that come to mind. Best, -Noel

1 0

Re: Markup of text descriptions Vs. structured data (GEN)
by Jim Croft 30 Nov '99

30 Nov '99

>These issues should be seen quite separately, and that we need both > >1. A markup language for text documents. This includes: > - existing descriptions, captured by OCR or other means, where the > markup would be manually (or automated/data mining?) > - computer generated natural language descriptions that are > published as electronic documents. A new CSIRO package could have > a ToNat command that automatically adds the necessary, hidden > markup code. >2. A data language for new observations, including repeated > measurements or repeated observations of categorical data, e.g. > shape of multiple leaves in a single specimen. Further, many > structures for knowledge managements (data revision, annotation, > quality control and assessment) need to be implemented here. > >The issues overlap, and it would be beneficial to use as much common >syntax as possible, but fundamentally I believe them to be quite >different. Gregor Can you explain in more detail why they should be fundamentally different? Aren't the differences just a matter a matter of degree, different points on a continuum of data as it were? And aren't the basic principles applicable across the whole? Is not the issue we are dealing with here one of how to deal with biological descriptive information whether at the level of individual components of the specimen (e.g. each of the leaves), the specimen as a whole (e.g. the variation of leaves on a specimen) or the taxon (e.g. the variation of leaves across all specimens of the species, or the higher level taxon)? Or at any other level, or mixture of levels, we might chose? Is it not possible, and indeed preferrable, to have a single scheme that covers all of these situations? I am finding it a bit difficult to get a grip of this discussion and where it is going (or where it is coming from, for that matter). I thought I understood (and agree with) your previous post about the need for a definition component as distinct from the the description and data in the same manner as the DTDs of SGML, but I am now not so sure if that is what, or all of what, is being suggested. jim __________________________________________________________________________ Jim Croft ~ jrc(a)anbg.gov.au ~ http://www.anbg.gov.au/people/croft.jim.html ph 02-6246-5500 ~ fx 02-6246-5248 ~ GPO Box 1600 Canberra ACT 2601

1 0

Re: (XML)Pedagogy
by Jean-Marc Vanel 29 Nov '99

29 Nov '99

Hello Gregor Hagedorn a écrit : > o Are Vocabulary and Namespace synonymous terms? Here is quotation from http://www.w3.org/TR/PR-rdf-schema (W3C's RDF schema spec.) 4.1.1. Terminology The phrase 'RDF vocabulary' is used here to refer to those resources which evolve over time; 'RDF schema' is used to denote those resources which constitute the particular (unchanging) versions of an RDF vocabulary at any point in time. Thus we might talk about the evolution of the Dublin Core vocabulary. Each version of the Dublin Core vocabulary would be a different RDF schema, and would have a corresponding RDF model and concrete syntactic representation. > o What is XML-DATA? It's a W3C Note of 1998 by Microsoft and others, that is to a large extent implemented in IE5 (but under the name of XML Schema). W3C's "true" XML Schema, still a Working draft, is not very different from XML-DATA. > o Reading about RDF and Dublin Core as RDF application I was left > confused as to what acutally is a resource. It usually seems to be > assumed that it is a document, but this impression may only be due to > the examples. Here is a quotation from http://www.w3.org/TR/REC-rdf-syntax (W3C's RDF spec.) Resources All things being described by RDF expressions are called resources. A resource may be an entire Web page; such as the HTML document "http://www.w3.org/Overview.html" for example. A resource may be a part of a Web page; e.g. a specific HTML or XML element within the document source. A resource may also be a whole collection of pages; e.g. an entire Web site. A resource may also be an object that is not directly accessible via the Web; e.g. a printed book. Resources are always named by URIs plus optional anchor ids (see [URI]). Anything can have a URI; the extensibility of URIs allows the introduction of identifiers for any entity imaginable. So we can have an URI for a plant species. The Dublin Core is a particular vocabulary in that it has only properties; other vocabularies, like RDF schema mentioned above have properties and types. See also my recent post about asserting in RDF a butterfly being a plant's pollinizator. One last remark: Dublin Core is a much publicized vocabulary, but it's simple with only 15 properties. A botanical vocabulary will count hundreds of properties (from both organs and features). > Thus, in a file with several item descriptions could: > - Each item be a resource, for which a creator/editor etc. is > described using DublinCore? > - Could even the characters information within an item have different > authors? (And if yes: Is it wise to do this?) Yes and yes. An article can have an author, and have quotations from others authors in it. > - Is there a limit to a hierarchy of RDF objects in an XML file? No. > > You can copy-and-paste this into an example.html file, and see it in > > Internet Explorer 5, or Mozilla, or any XHTML+CSS1 compliant browser. > Note: I assume that Mozilla does not apply to Netscape 4.7, which did > not seem to be able to display the code you appended Sad but true; I think that someday Mozilla and Netscape will merge (in functionality because the code and design is completely different). > > * how and why mix several vocabularies, > I would be very interested in this one. OK, warten Sie mal... Ich muss das überlegen. Just say for the moment that we'll have several vocabularies by large domains: * biological descriptions (generalities) * botany * zoology * Virology * Microbiology * geography * ecology May the forgoten disciplins forgive me... They'll have their XML vocabularies too. Jean-Marc

1 0

Re: Character types (RQT)
by Gregor Hagedorn 29 Nov '99

29 Nov '99

> This means that anyone using the format 'offline' doesn't need to > access some resource storage (i.e. the net) to make a successful > identification. My use of "external resource" was rather unclear. I am thinking in terms of a database information system, so external resource means any URL or file based document. Thus external in my use could just as well be a local resource as a internet resource. If an information system is distributed on CD, all images, formatted text, and audio etc. would be on that CD. Any better terms than "external resource"? I don't like media resource/hypermedia etc., because I feel this is not very adequate for resources like a query/identification interface. > Yes, rendering a graphical representation of (say) a leaf onto the screen > would be a good application for vector graphics. However again some > additional > text describing the image/shape would be useful - for those applications > that don't support vector graphics, or indeed visually impaired users. A resource in my model has specific attributes (which are different from characters), esp. indication of scope (to which items/character/state it belongs + flags whether it is to be used as default resource for this scope), a caption, which contains the description, media types, etc. Gregor ---------------------------------------------------------- Inst. for Plant Virology, Microbiology, and Biosafety Federal Research Center for Agriculture and Forestry (BBA) Gregor Hagedorn Net: G.Hagedorn(a)bba.de Koenigin-Luise-Str. 19 Tel: +49-30-8304-2220 14195 Berlin, Germany Fax: +49-30-8304-2203 Often wrong but never in doubt!

1 0