Javascript MinXML / RDF testbed

This doc shows the use of a simple RDF strawman dump syntax, parsed in Javascript using a syntactic profile of XML (the minXML proposal). The Javascript MinXML parser is used here for convenience; no endorsement or otherwise of MinXML, SML etc should be assumed. Note also that the RDF triple syntax used here is in no way standard.

The parsed RDF in turn can be fed to the Javascript RDF query engine. This could make a nice testbed for looking at the tradeoffs involved in the various XML profile proposals (eg. how do we do namespaces, language-tagging, datatypes etc...?) It should also make for a nice type-and-test query environment for RDF.

TODO:

Test on wider range of JS implementations. Works in NS4, IE4, Win and Linux at least...
learn javascript
redesign user interface so people know what is going on!
hook this into RDFViz for display of parsed and inferred data.
integrate DAML inference example (see rules in default kb) with the toyXML example
round-tripping. Get data from triples back into the xml window
try out xml representation of some rules
separate rules and data in query UI

dan brickley (2000-04-15)

Getting started. (1) select 'append parsed data into RDF query scratchpad' to parse and load the data. (2) select 'go' on the default query below to query the loaded data.

Parser Options

The first textarea contains pseudo-XML triples in a simple (undocumented :-) syntax. Second contains output of parser. There are options for outputting the parsed data to the RDF query engine (second half of the demo...).

Load RDF query scratchpad with parsed data
Append RDF query scratchpad with parsed data
parse for RDF arcs
Parse
View Tree

Parser Input (pseudo-XML format for RDF triples)

<webdata>
<arc>
  <subject>http://www.opensource.ac.uk/</subject>
  <predicate>http://purl.org/dc/elements/1.1/creator</predicate>
  <webobject>mailto:daniel.brickley@bristol.ac.uk</webobject>
</arc>

<arc>
  <subject>mailto:daniel.brickley@bristol.ac.uk</subject>
  <predicate>http://xmlns.com/fiction/misc/employer</predicate>
  <object>ILRT</object>
</arc>

<arc> 
  <subject>person:uk-nx930366b</subject>
  <predicate>http://xmlns.com/fiction/persons/age</predicate>
  <object>28</object>
</arc>

<arc> 
  <subject>person:uk-nx930366b</subject>
 <predicate>http://www.w3.org/1999/02/22-rdf-syntax-ns#type</predicate>
  <webobject>http://xmlns.com/wordnet/1.6/Person</webobject>
</arc>

<arc> 
  <subject>http://xmlns.com/wordnet/1.6/Person</subject>
<predicate>http://www.w3.org/2000/01/rdf-schema#subClassOf</predicate>
  <webobject>http://xmlns.com/wordnet/1.6/Mammal</webobject>
</arc>

<arc> 
 <subject>http://xmlns.com/wordnet/1.6/Mammal</subject>
 <predicate>http://www.w3.org/2000/01/rdf-schema#subClassOf</predicate>
 <webobject>http://xmlns.com/wordnet/1.6/Living_thing</webobject>
</arc>

</webdata>

Parser Info (output)

Test code:

Evaluate Result:

RDF Query tester

The data scratchpad here contains rules and data. Triples are represented in a simple prolog-like syntax, with {curly braces} around URIs. Rules can also be represented -- simple subClassOf rule included here for illustration.

Examples

This query lists all triples in the RDF graph

Semantic Web Screenscraper: Scheduling demo

This is a quick RDF/XSLT semantic web screenscraping demonstration, based on discussions with Dan Connolly. See Dan Connolly's RDF Interest Group message for background, or the 'semantic web accessibility' followup.

We show here how to scrape RDF data from HTML web pages, aggregate using RDF, and query using a prolog-like system in Javascript.

Raw materials:

HTML Data: pre-k.html
RDF, generated with XSLT
tinyProlog version (suitable for pasting into demo dataset below)

To use: get the tinyProlog data, paste into the textarea below, and enter queries / rules... You might try variations on these too...

find out who is busy...
	busy(dan, Day, Hour, What)

or... list members of groups...

	member(Person,Group)

or... ask about schedule of busy XML experts...

	busyXML(M,D,H,W,PP,P)

RDF Database

##### Your data goes here:

####
# Some constructs for talkign about Events.
# This schema from irc chat with Dan Connolly
# eats RDF data acquired via XSLT
# see: http://snowball.ilrt.bris.ac.uk/cgi-bin/rdf-dot/xrdf2tinyprolog?URI=http://www.w3.org/XML/2000/04soccer-sched/team-2.rdf
# for some prologized facts to paste in here 
# Enter this query above for example:
#	flatEvent(Event,TimeContext,Location,Hour,Day)

flatEvent(Event,TimeContext,Location,Hour,Day) :- when(Event,TimeContext),where(Event,Location),hour(TimeContext,Hour),day(TimeContext,Day).
when(X,Y) :- triple({foo:not-assigned-yetwhen},X,Y).
where(X,Y) :- triple({foo:not-assigned-yetwhere},X,Y).
day(X,Y) :- triple({foo:not-assigned-yetday},X,Y).
hour(X,Y) :- triple({foo:not-assigned-yethour},X,Y).

busy(Person, Day, Hour, What) :- member(Person, Group), who(Event, Group), when(Event, Time), day(Time, Day), hour(Time, Hour), label(Event, What).
member(dan,{http://www.bvsoccer.org/schedules/pre-k.html#team-2}).
who(E, W) :- triple({foo:not-assigned-yetwho},E, W).
label(E, L) :- triple({foo:not-assigned-yetlabel},E, L).

##### These represent some information from the core RDF schema model
	
# implied subClassOf relations:
triple({http://www.w3.org/2000/01/rdf-schema#subClassOf},C1,C2) :-triple(subClassOf2,C1,C2), triple({http://www.w3.org/2000/01/rdf-schema#subClassOf},C2,C3).
triple(subClassOf2,C1,C2) :- triple({http://www.w3.org/2000/01/rdf-schema#},C1,C2).
# SubClass implies type membership:

triple({http://www.w3.org/1999/02/22-rdf-syntax-ns#type},S, O) :- triple({http://www.w3.org/2000/01/rdf-schema#subClassOf},O2,O), triple({http://www.w3.org/1999/02/22-rdf-syntax-ns#type},S,O2).
 
## Searching on the Semantic Web example from DAML article at 
## http://www.zdnet.com/pcweek/stories/jumps/0,4270,2432946,00.html

# Data from the course catalogue
triple({http://www.w3.org/1999/02/22-rdf-syntax-ns#type},{http://www.bris.ac.uk/fiction/course_Web202},{http://xmlns.com/wordnet/1.6/Course}).
triple({http://purl.org/dc/elements/1.1/title},{http://www.bris.ac.uk/fiction/course_Web202},"Advanced Web Development").
triple({http://purl.org/dc/elements/1.1/description},{http://www.bris.ac.uk/fiction/course_Web202},"A high level course for Web developers familiar with basics of HTML, ECMAscript etc. Covers XHTML, XML, RDF, SVG and semantics of mixed-namespace document

formats...").

# Data from student annotations
triple({http://xmlns.com/fiction/techCoverage},{http://www.bris.ac.uk/fiction/course_Web202},{http://www.w3.org/XML/}).
triple({http://xmlns.com/fiction/techCoverage},{http://www.bris.ac.uk/fiction/course_Web202},{http://www.w3.org/RDF/}).
triple({http://xmlns.com/fiction/techCoverage},{http://www.bris.ac.uk/fiction/course_Web202},{http://www.w3.org/Graphics/SVG/}).

# Data from Bristol LDAP server or Staff profiles webpage
triple({http://www.w3.org/1999/02/22-rdf-syntax-ns#type},{http://www.bris.ac.uk/fiction/staffids/cm-ls},{http://xmlns.com/wordnet/1.6/Professor}).
triple({http://xmlns.com/fiction/teaches},{http://www.bris.ac.uk/fiction/staffids/cm-ls},{http://www.bris.ac.uk/fiction/course_Web202}).

## Someone is an expert in a technology if they teach a course about it
## (Ugh, ugly rules in tripled representation)
triple({http://xmlns.com/fiction/expertise},PERSON,TECHNOLOGY) :- triple({http://xmlns.com/fiction/teaches},PERSON,COURSE), triple({http://xmlns.com/fiction/techCoverage},COURSE,TECHNOLOGY).

## OK, this bit joins the two examples together. We can ask RDF what XML
## experts do with their time ;-)
## Note the funny business to join the data despite different person ids
## - we can do this because personalEmail uniquely picks out persons.

# mailbox, day, hour, what + stand-in identifiers for XML experts upcoming commitments
busyXML(M,D,H,W,PP,P) :- personalEmail(P, M), busy(P, D, H, W), identicalReferents(P, PP), expertise(PP, {http://www.w3.org/XML/}).

# expertise (might also be implied by contributing to home page for 
# the technology. S is the home page of a technology...
expertise(X, S) :- triple({http://xmlns.com/fiction/expertise},X, S).

# We don't use this yet. This fact is what allows us to have the
# following rule, though currently this is hardcoded knowledge.
secondArgUnifies(personalEmail).

# a deeper notion of identity (ie. acknowledge that temporary IDs often 
# get used, and that mailboxes are a socially-grounded basis for figuring
# out a deeper 'sameness' w.r.t. persons.
identicalReferents(X, Y) :- personalEmail(X, Z), personalEmail(Y, Z).

# given here as a known fact; 'dan' is really a temporary,transient id
# (the one used in the scheduling data...)
personalEmail(dan, {mailto:connolly@w3.org}).

## Following facts might come eg. from embedded metadata in /XML/ page...

# Some person (transiently identified) has expertise in XML
triple({http://xmlns.com/fiction/expertise},{genid:person-222},{http://www.w3.org/XML/}).

# That person has a certain email address... 
triple({http://www.w3.org/has-email/},{genid:person-222},{mailto:connolly@w3.org}).
personalEmail(X, M) :- triple({http://www.w3.org/has-email/}, X, M).

Testing, please ignore: