Am Dienstag, 11. Dezember 2018 19:03:30 UTC+1 schrieb Joe Armstrong:
>
> Thinking out loud here ...
>
> I've been thinking more about tags. One problem is that tags are rather
> vague and are written in different human languages.
>
> One way out of this might be to adopt the wikidata word definitions. For
> example, I am, unambiguously
>
> https://www.wikidata.org/wiki/Q1691321
> <https://www.google.com/url?q=https%3A%2F%2Fwww.wikidata.org%2Fwiki%2FQ1691321&sa=D&sntz=1&usg=AFQjCNFRT6Euza35njDPie_N6Crb_9vJUw>
>
> There are actually several Joe Armstrong's (for example,
> https://www.wikidata.org/wiki/Q712592)
>
> These Q numbers uniquely define subjects and objects. Verbs (or
> predicates) are given by P numbers
> so https://www.wikidata.org/wiki/Property:P178 means "the organisation or
> person who developed the item.
>
>
I really like this idea. The natural-language tags of a TiddlyWiki could be
linked to global URIs (like the ones from Wikidata) by adding a context to
a TiddlyWiki in the same way JSON-LD adds a @context to JSON documents to
give globally understandable meaning to application-specific property
identifiers: https://en.wikipedia.org/wiki/JSON-LD
> in RDF speak the triple
>
> {https://www.wikidata.org/wiki/Q1144644,
> https://www.wikidata.org/wiki/Property:P178,
> https://www.wikidata.org/wiki/Q17031730}
>
> (BTW I recommend clicking on these links and playing around - there's lots
> of interesting
> data in RDF tuples and the above links are a good place to start looking)
>
> Means "TiddlyWiki developer Jeremy Rushton"
>
> These triples encode facts in a hopefully reasonably clear manner.
>
> So now the N$ question - can we automatically analyse a tiddler and turn
> it into a set
> of RDF tuples. If we could then we could add these to the huge databases
> of RDF tuples
> and possible find stuff in a clever way.
>
>
That would be a very ambitious endeavor as it would require solutions to
two very hard problems:
1. The long-standing research problem of entity recognition and linkage:
https://www.stardog.com/blog/entity-linking-in-the-knowledge-graph/
2. Connecting the recognized entities semantically properly. For example,
how would you detect from the sentence "Jeremy Ruston started the
TiddlyWiki project in 2004." to use the
https://www.wikidata.org/wiki/Property:P178 property?
Regarding the first problem, which would already solve your initial tag
ambiguity problem, research has led to quite some progress. This library,
for example, offers a solution in
JavaScript: https://github.com/spencermountain/compromise
The filter notation in the tiddlywiki reminds me very much of prolog, and I
> guess with a but of
> work SPARQL queries might be possible (SPARQL is an RDF query language)
>
Even without a solution to the mentioned second problem (allowing to
generate proper RDF triples like the one you mentioned), we could already
do interesting SPARQL queries with recognized Wikidata entities for content
tags. We could for example query for all tiddlers tagged with a computer
scientist and would get the tiddlers tagged with "Jeremy Ruston". A SPARQL
engine for in-memory data in JavaScript can be found here:
https://github.com/antoniogarrote/rdfstore-js (I can't resist to also
mention that a SPARQL engine also exists for the BEAM:
https://github.com/marcelotto/sparql-ex ;-))
Cheers
>
> /Joe
>
>
Cheers,
Marcel
>
>
>
> On Monday, 10 December 2018 17:43:01 UTC+1, @TiddlyTweeter wrote:
>>
>> One of the things that interests me a lot that the talk raised a bit--and
>> which no one seems to know how to answer is ... :-)
>>
>> - WHAT exactly is an SU (Semantic Unit) in TW writing (or computing
>> writing In General, for that matter)?
>>
>> There is a kind of rule of thumb "its maybe a paragraph"? But, of course
>> that won't quite work for the one-sentence brevity of a Nietzsche.
>>
>> Its obviously highly context dependent. And I doubt much of that context
>> lives on the computer itself.
>>
>> The idea in TW towards writing "the shortest semantic whole possible"
>> (the word "fragment" here that is thrown around has muddied waters; they
>> are not fragments so much as whole-parts-of-wholes) allows for later
>> re-combinations to form more complex semantics.
>>
>> However, I think its bit of an, ultimately, moot and mute point, in the
>> sense that human meaning is often an interaction with technologies of
>> expression themselves (though no where ever fully defined by them). So its
>> an area of intuited understanding, not formal logic? On the other hand,
>> who's offering the horse which water?
>>
>> Josiah
>>
>> On Monday, 10 December 2018 12:49:14 UTC+1, PMario wrote:
>>>
>>> Hi,
>>>
>>> Here's the video:
>>> https://www.youtube.com/watch?v=Uv1UfLPK7_Q&index=9&list=PLvL2NEhYV4ZtWFBNOrApXaIoCTtj-yk7Y
>>>
>>> have fun!
>>> mario
>>>
>>
--
You received this message because you are subscribed to the Google Groups
"TiddlyWiki" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/tiddlywiki.
To view this discussion on the web visit
https://groups.google.com/d/msgid/tiddlywiki/091b83d3-9a94-4010-bb2a-e4feca649709%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
