Unpredicatable - ness

Wasted a whole morning trying to work out why the Quicktime here (which worked fine yesterday) dropped the audio today when encoding from .flv to .ogg files. After hours of scouting about I then discovered that Archive.org accepts .flv files, its only  WikiMedia commons that do not allow upload of video files that use proprietary codecs. After that I prepared a .csv file with all the metadata, got the python script to upload it Archive.org from here's:

http://archive.org/help/abouts3.txt

https://github.com/kngenie/ias3upload

this worked quite well for one file, but trying to batch from terminal gave this error:

298 file(s) to upload in Landscape-Portrait
Sent 4096 bytes (0%)
500 Can't read entity body: Connection reset by peer
500 Can't read entity body: Connection reset by peer

hopefully Jeff @ Archive might be able to help out. 

Encode, rename, upload, tag

Spent all putting together a workflow so I can encode, rename, upload and tag the video files from Bournemouth. First up the files are archived as flv's and Archive.org, and the WikiMedia commons do not allow proprietary codecs, therefore they only allow .ogv or .ogg files. I managed to find a ogg binaries for Quicktime here, meaning I could batch export the flv's. Then needed to rename the files from the native upload names using this. After that each file needs a csv of all the metadata, which can then be linked to via a RDF dump. Usefully Archive.org makes this metadata available as xml for both the parent project and the video files, so w should be able to use this in the RDF ?

Archive Org

The plan this week then is to upload a set of video's to Archive.Org - using this URI format:

www.archive.org/download/Landscape-portrait2010/Kevin_Carter/v10heora.ogv

from this I can develop the RDF - it's really important that I get a working version of this before plunging back into the writing.

Then need to publicise the content by uploading the RDF file to CKAN's datahub and notifying the semantic search engines via this:

semantic search engine notification URL and Software here.

http://pingthesemanticweb.com/api.php#1

RDF schema

Spent the day trying to visualise the schema that would be required to describe the Landscape-Portrait project overall as well as some of its video elements, from this I can hopefully develop a template which can be produced by Paul as part of a script.


Here is the schema:


Red is the entity, green properties, yellow classes etc, from this I was able to map different vocabularies for most of the entities, properties and classes; see here:

Interestingly I couldn't find a entity of either a 'question' or an 'answer' - I thought these might be sub properties or classes of a entity but could
not find it.

Found some interesting art related ontology's here:

http://dlib.york.ac.uk/ontologies/Openart/index.html
http://linkedevents.org/ontology#sec-namespace

Which were taken from here:

http://www.asis.org/Bulletin/Feb-12/FebMar12_Allinson.html


What is interesting is the degree of sophistication that is permissible in some spaces and not others, think we'll need to create 'community art' as a ontological term as none that I can see come near to it - Fro instance when looking at the FOAF ontology I was interested to see this property

http://xmlns.com/foaf/spec/#term_myersBriggs

Which has 16 personality classifications which can be used to describe a person -

RDF/RDFa/DataDump/Ontologies

Working through the different options I am now thinking to make either a RDF file, or include a RDFa snippet into the web page. Problem is that the Landscape-Portrait site is dynamically generated and there is not page that explicitly contains the video elements. Therefore we might create RDF pages for each postcode, an include all the portraits that relate to the postcode. What is also of interest is the idea that I might create (how do you create) a ontology to describe community artworks, this might include all differences of method and methodology.

Alongside all of this will be a data-dump of the metadata of Landscape-Portrait, so other users can download and make use of it.

I am not a programmer

Ever since deciding that if the rhetoric of Open Data is meaningful then i should be able to publish the Landscape/Portrait data fairly easily I have been bashing my head against a wall trying to think of a way this would work easily. Basically I have an ASP database with all the metadata that I need to translate into a RDF formatted LOD. Been messing about with Kasabi allday and whilst I like its point and shoot interface i can't see how to load large datasets into it. I think that maybe RDFa is the way forward, so a semantic snippet might be included on each page, but I need to work this through via a tutorial.

Open Data Art.

Read some interesting, if slightly optimistic things about the relationship of Open Data and art. I am now into the process of transferring the dataset created during the making of Landscape-Portrait into the public realm, ideally as linked open data (LOD) or/and as a Open licensed data dump. The problem with making available as LOD is I need to services of a programmer and this is not cheap. Also part of me thinks that their should be freely available tools so that a fairly technically adept person, such as myself, should be able  to do this, otherwise the rhetoric of Open Data is, as I have stated before, is severely constrained around technical ability and economics. So the option at moment is to output the data from LP in a CSV file and load that to various free platforms, such as OKF's Datahub, Kasabi (a commercial platforms currently free as its in Beta), Talis (who are commercial but offer free hosting for not for profits) as well as data sharing sites such as Googles Fusion Tables. The data will also be uploaded to Archive.org as well, to allow it to be used and viewed in the future. Another idea is to give the dataset to a hackday and allow it to be processed in that way.
It would be interesting to find a site that specialises in publishing art datasets, not data about art as Rob Myers has described, but the dataset as art object. Not found that yet ?