W3C Multimedia Semantics Incubator Group face to face meeting, Amsterdam, Day 2

11 Jul 2006


See also: IRC log


Ioannias, Vasileios, VassilisThomas, Raphael, Patrizia, George, Lynda, Jacco, Giovanni, Christian, Jeff, Pasquale, Alessio, Massimo, Roberto, Chris
Thomas, anadiotis, jacco




Demos by Patrizia & Giovanni

<jacco> scribe: Thomas

<alessio> hi all

<JeffP> Hi alessio

<pasquale> hi all, from Naples, Italy

<JeffP> are the slides online?

<jacco> participants: Ioannias, Vasileios, VassilisThomas, Raphael, Patrizia, George, Lynda, Jacco, Giovanni, Christian, Jeff, Pasquale, Alessio, Massimo

<Massimo> hi all

<JeffP> Hi Massimo

<jacco> Patrizia presenting slides at http://www.w3.org/2005/Incubator/mmsem/meetings/f2f-amsterdam/mmsem-member-presentation-pat3.ppt

<jacco> Patrizia explaining links with MUSCLE, mm ontology requirements effort etc

Patrizia explains the 4M (MultiMedia Metadata Management) Infrastructure

<vasileios> massimo, which software do you use for the extraction of MPEG-7 descriptors?

<JeffP> Hi Massimo: Yes, the irc log is public, we can either edit the minutes later to ensure the context of your texts are properly specified, or we can send it to the mailing list so that we don't have to dig it out from the minutes later.

<Massimo> we have extended emir and mpeg-7 encoder

<jacco> rssagent, pointer

<jacco> massimo, see above for the log

<Massimo> Hi Jeff/Jacco, ok

<jacco> Other resources, slides, minutes etc at http://www.w3.org/2005/Incubator/mmsem/#Meetings

Raphael asking Patrizia: can you give an example of how to use the Algorithms Thesaurus?

Patrizia: Algorithms Thesaurus used to establish provenance information
... ... to define formal processes that include the algorithms used

<jacco> jeff asks about the expressive power wrt swrl and owl that patrizia needs

<JeffP> jeff: what expressive power do you need from owl and swrl

<jacco> Patrizia: not thought very deeply about this, but we need some formal way to express rules

there is not yet a schedule for when the work on the algorithms ontology will be finished

<IoannisPratikakis> A question to Massimo:: Can you give an example of the existing relations in the Algorithms ontology ?

<Massimo> we thought of a sort of definition on how algorithms can be combined together

<Massimo> on the base of the multimedia object, the problem and the possible solutions

<jacco> jeff: asks what kind of rules you need, for example, do you need disjunction in the head, or do you need the vocabulary of the ontology in the rules

<Massimo> an example (thought now): suppose we need to obtain the contours of objects in an image and locate salient features (more specifically, how it is possible to extract features and locate a fracture on a Xray image)

<jacco> Patrizia: not sure, but we assume the rules are built on top of the ontology

<jacco> jeff: what are the rules used for?

<jacco> Patrizia: I'm not sure yet: we've just started

Giovanni presents DBin, on the DBin website there is a screen demo to follow

<jacco> DBin: http://dbin.org

DBin provides brainlets: environments for specific data/tasks that customize the user interface

Patrizia: can people modify the ontology?

Giovanni: creators of brainlets define the ontologies, you can install/deinstall ontologies

Raphael: please show how to annotate a photo

Giovanni: you can use any property given by the ontologies of the brainlet to add annotations to resources

<JeffP> -meeting_room offline ...

<jacco> we're calling back in ...

<JeffP> ok

<VassilisT> nice

<jacco> Having short coffee break

<VassilisT> ok

Multimedia Annotation Exercise Session

<jacco> Raphael explains set up of the annotation session.

<Jeff> I was out for minute, can you send the URL again?

<jacco> Mostly personal photos of beaches, anything, any tool, any vocabulary is allowed

<jacco> jeff, no url has been sent yet

<raphael> World Cup photos: http://www.apa-images.at/cms/site/EN/breakingnews.html;jsessionid=agCHvRUiTwE7

<raphael> World cup videos: http://www.euronews.net/create_html.php?page=detail_info&article=368313&lng=2

<raphael> Video: http://today.reuters.com/tv/videoChannel.aspx?storyid=2ba9545293e136702fab402c94af125ea8560bc1

<raphael> Photos: http://photos.reuters.com/Pictures/default.aspx

<raphael> Video Clip: http://www.rap2k.com/clip-1218-mas-que-nada-feat-black-eyed-peas-sergio-mendes.html#open

<raphael> Lyrics: http://en.wikipedia.org/wiki/Mas_Que_Nada

<raphael> Official Web Site: http://www.blackeyedpeas.com and http://www.brasil66.com/

<raphael> Temporary URLs of media files: http://homepages.cwi.nl/~troncy/tmp/W3C/

<jacco> thanks massimo

<jacco> massimo, you can phone in, we will just mute you

<Massimo> ok, i'll retry!

<raphael> there is now a Beach ontology at: http://homepages.cwi.nl/~troncy/tmp/W3C/

<raphael> for annotating the personal photos

<raphael> Massimo, please, DCC me the file

<raphael> you're flooding

<jacco> lunch break

<Massimo> see you later (sorry for the previous flooding)

<anadiotis> Sample RDF output : http://www.flickr.com/services/feeds/photos_public.gne?id=11026577@N00&tags=http://www.flickr.com/photos/11026577@N00/187187881&format=rss2

<jacco> my annotations are at

<jacco> http://homepages.cwi.nl/~jrvosse/spool/try.rdf, http://homepages.cwi.nl/~jrvosse/spool/try2.rdf, http://homepages.cwi.nl/~jrvosse/spool/ct.rdf

<anadiotis> More sample RDF output : http://homepages.cwi.nl/~anadioti/RDF/rembrandt.rdf

<vasileios> massimo, which annotation tool have you used for the extraction of the massimo.mp7 file?

<Massimo> our 4M

<vasileios> massimo, is this tool publicly available?

<Massimo> it's developed under the MUSCLE NoE, so i don't know the copyright policy

<vasileios> so, I assume there is no currently available version for download and testing

<Massimo> the thing we are thinking to propose to MUSCLE is if we can give an URI where people can connect and upload multimedia object, when MUSCLE will finish we would like to open the code

<IoannisPratikakis> Massimo, does '4M' support other descriptors than MPEG-7 ones ?

<Massimo> not at the moment

<Massimo> we started the developement less than one year ago, we would like to add some more features

<raphael> ALL: go on http://homepages.cwi.nl/~troncy/tmp/W3C/multimedia/

<IoannisPratikakis> have you used the MPEG-7 XM software for '4M'?

<Massimo> no

<Massimo> we started with vizir (emir) and mpeg7encoder, it's all written in java

<Massimo> we compiled XM6.1 on a windows pc (terrible experience) and we've use it to compare the results

<vasileios> massimo, i can see from the massimo.mp7 file that the descriptors have been extracted for the whole image (i.e. not region-based). Is this right?

<Massimo> but 4M extracts more features than XM now: some less for images, a lot more for audio, some for video

<Massimo> yes you're right

<Massimo> that is the next module under developing

<Jeff> http://www.csd.abdn.ac.uk/~jpan/w3c/mm-annotation/mm-annotation.rdf

my "handmade" rdf file is at http://www.uni-koblenz.de/~franz/manual.rdf

<jacco> jeff, <rdf:Description rdf:about="photo:DSC03139.JPG"/> should be <rdf:Description rdf:about="photo:DSC03139.JPG">

<jacco> without the slash in the end

<raphael> Jeff, all RDF files are at: http://homepages.cwi.nl/~troncy/tmp/W3C/multimedia/annotations/

<raphael> yours too

<Jeff> thanks

<jacco> jeff, can you put a real url to an image in your rdf:about

<jacco> ?

<Jeff> ok

<Jeff> jacco, done

<Jeff> http://www.csd.abdn.ac.uk/~jpan/w3c/mm-annotation/mm-annotation.rdf

<anadiotis> Round table - details about what everyone did and how

<anadiotis> Vassilis and Ioannis used PhotoStuff

<Massimo> (i can't see the last rdf)

<jacco> scribe: anadiotis

Limitations: no low-level feature extraction, no automatic segmentation

<jacco> jeff, we get an access denied on your RDF page

<Jeff> it should be fine now

<Jeff> ?

Could have used M-OntomatAnotizer, that deals with issues, but not yet publicly available

Thomas: tried to use PhotoStuff + Protege

Issues: could not load ontology to Protege

Did manual annotations..

Massimo: obtained MPEG-7 using 4M

<Massimo> RDF: foafer, image annotator (notepad to add the coordinates, and google to check). MPEG-7: 4M. OWL: Swoogle

<lynda> george used 2 different methods: uploaded picture to flickr, added regional annotations.

<lynda> Also added URL of picture.

<lynda> Then created RSS of this.

<lynda> Template Newsml of picture, found news codes, sample file -> converter for RDF

Jacco: tried to upload WordNet - crashed Protege

Then used Raphael's beach ontology

Raphael: created beach ontology with Protege

Then used it with PhotoStuff to annotate the pictures

<Massimo> (opps SWOOP2.3 for owl not swoogle)

PhotoStuff does not support properties

(it does in the latest version - Raphael was using a previous one)

Lynda: could not produce an RDF

tried to find all the ontologies

also tried with NewsML, but could not find the NewsCodes

<jacco> jacco: I used http://www.mindswap.org/2003/PhotoStuff/webstarts/PhotoStuff-3.33/PhotoStuff-20060531.zip

syntax was difficult to use too

Giovanni + Christian : set up environment, fixed broken soccer ontology, loaded ontologies

<jacco> giovanni: set up an environment in dbin.org to annote images

Giovanni + Christian : Found a simplified MPGE-7 ontology

Inserted new images by navigating the ontology and creating new 'image' instance

<jacco> added line to say place was a geo:spatialthing

<Jeff> The owl-full bug is fixed, see

<Jeff> http://www.csd.abdn.ac.uk/~jpan/w3c/mm-annotation/mm-annotation.rdf

Giovanni: also used geographic ontology
... in the end everything is exported in a single RDF file

wrap-up by Raphael : everyone went for images

initially at least

Giovanni: why is precise region-drawing important? Aren't most use cases covered by rectangle selction?

Ioannis: Precise region marking helps achieving better feature extraction

Raphael: purpose of the hands-on exercise was to show issues with ontology-based annotation
... discussion on the roadmap follows

MMSem XG roadmap

<raphael> http://www.w3.org/2006/07/10-mmsem-minutes.html

<jacco> Music use case: id of track, diff perf. of the same song; different encodings

<jacco> id3 tags, CDDB data, musicbrainz

<jacco> classification of music on rythm,tempo,gentre, which ontologies to use

<jacco> lyrics: paper about finding music based on lyrics

<jacco> thierry: there is a more or less standard linguistic annotation scheme

<jacco> (by us/ISO)

<jacco> link text of lyrics to audio? manual/automatic

<jacco> or use mpeg-7 linguistic scheme

<jacco> thierry was talking

<jacco> now giovanni

<Jeff> thanks

<jacco> raphael: we first need to describe the problem of the use case, than the general solution. only highlight specific tools as example

<jacco> thomas: what is the intention of listing these problems, this is not in the context of the real use case?

<jacco> What is the added value of SW?

<jacco> Raphael: Yes, this will be key to all use cases

<jacco> Thomas: yes, so we need a real scenario in which these problems occur

<jacco> raphael: we can be creative and can reuse all relevant RDF out there on the web, as long as it fits in a real scenario

<jacco> giovanni: music browsing, with intelligent "next" button, or locating scenario

<jacco> we can show how the problems listed so far relate to the use case

<jacco> Thomas:music browsing is too generic, need a real specific scenario

<jacco> patrizia: is a tool from vienna that does similar stuff

<jacco> ACTION: patrizia to find pointer to this tool [recorded in http://www.w3.org/2006/07/11-mmsem-minutes.html#action01]

<jacco> Rapahel: structured annotation of the lyrics?

<jacco> Thierry: we could do that, but we are not so strong in classifaction

<jacco> looking at http://freedb.org

jacco: aim is to produce readable documents, so compact and specific use-cases are required

<Jeff> agreed

<jacco> Low level feature extraction use case:

<Jeff> which use case are we talking about?

<jacco> current tools represent results of extraction in XML/MPEG7, why do we need RDF/OWL?

<jacco> Jeff, this is use case http://www.w3.org/2006/07/10-mmsem-minutes.html#action04

<jacco> Christian, there is overlap with Massimo's use case

<jacco> Christian: there is overlap with Massimo's use case

<jacco> So we should provide a link

<Jeff> thanks

<jacco> Advantages of RDF: combine with background knowledge, bootstrap and infer more information

<jacco> Patrizia:we also have the semantics of the algorithm

<jacco> Vassilis: two issues: we must have use cases that cover a broad range of languages and vocabularies

<jacco> not all use cases should be based on mpeg-7, not all shou

<jacco> not all use cases should be based on mpeg-7, not all should only use RDF, but also different flavours of OWL

<jacco> otherwise we cannot show interoperability

<jacco> Vassilis: we should use many metadata standards, not just three

<jacco> all agree

<jacco> Ioannis: we need spatio-temporal relations to be supported

<jacco> Vassilis: spatio temporal relations are, from a language perspective, just relations, so there is not a issue

<jacco> the problem could be on the reasoning level

<chalaschek> yes the problem is on the reasoning level

<jacco> jeff: I agree with Vassilis, can do something even if spatio-temporal reasoning is not supported

<jacco> Unstructured tagging use case: http://www.w3.org/2006/07/10-mmsem-minutes.html#action03

<jacco> show limitation of tags, using tags across modalities or applications

<jacco> show first efforts to bring more structure to tagging, show advantages and common approaches

<jacco> show how it is possible to use SW technology to use, for example picasa tags in flickr, or from blogging app to another

<jacco> thomas: I agree

<jacco> other issues are navigation between tags, clustering etc

<jacco> jacco: maybe reuse existing flickr2rdf converters etc

<jacco> make use case that works cross modality

<jacco> Next use case: News (http://www.w3.org/2006/07/10-mmsem-minutes.html#action05)

<jacco> Raphael: I will work with people of the newsml/iptc community

<jacco> use case should show what you can do more if you convert all current vocabs. to RDF and link to other material already out there

<jacco> including: cross modality search on news images and text

<jacco> recognize different version of the same story

<jacco> make explicit links among different versions

<jacco> patrizia: should speach be added?

<jacco> reuse exisiting metdatda, convert to RDF

<jacco> Massimo (by SMS): audio is not only music

<jacco> all agree

<Jeff> who is here?

<jacco> Raphael puts Massimo's use case sent yesterday on the beamer

<jacco> raphael: is this a representation problem, massimo?

<Massimo> i was only thinking of use cases

<Massimo> but yes, you're right raphael, it could be

<Jeff> BTW, do we have any use cases on videos?

<jacco> Massimo, could you write this up in terms of a concrete use case? For example one on e-science, or on image transforms on personal photos

<jacco> news domain and unstructured tagging deal with video, maybe the music use case also

<Massimo> ok we'll think about it

<jacco> Raphael: lets talk about deadlines

<jacco> We have a teleconf on the 20th of July, so submit half a page description on the 19th of July, so we can disccus 5 use cases during the teleconf

<jacco> Thomas will be on vacation but will contact George before

<jacco> Skip 3, 16 august teleconf because of vacations

<jacco> proposal: next teleconf on 31 august?

<jacco> resolved

<Jeff> next teleconfs: 20 July and 31 August?

<Vassilis> have we decided who is goin to do be the editor of this deliverable?

<jacco> no, we will do the individual use cases first

<jacco> raphael: we do not yet have talked about the structure of the document

<jacco> jeff: agrees with Vassilis, can we use the time left to look ahead?

<jacco> raphael: we will all be contributors to the XG notes, but we also need editors

<jacco> Vassilis volunteers to be one of the editors

<jacco> obections to Vassilis being an editor?

<jacco> Resolved

<jacco> Any more voluteers?

<jacco> No, want to see what the documents are going to look like

next f2f

<jacco> proposal: co-locate with next ISWC conference in November, Athens, Georgia

<jacco> Vassilis: There are many other competing events like ruleml, rif, owl2

<jacco> Other posibility is to co-locate with SAMT in Athens, Greece

<jacco> But this will be the second f2f in Europe

<Massimo> could be thougth the possibility of a videocon too ?

<jacco> 5 november is tutorials only

<raphael> http://iswc2006.semanticweb.org/program/dates.htm

<rellero_> could be a good idea make next f2f in Venice

giovanni: having the f2f coincide with ISWC workshops/demos may not be a good idea, since people in the XG will attend the workshops/. Maybe before ISWC?

<jacco> ACTION: Raphael and Jeff to contact US partners and propose f2f date and location [recorded in http://www.w3.org/2006/07/11-mmsem-minutes.html#action02]

<jacco> Vassilis: Workshop in Athens on cultural domain, presence from this XG will be appreciated

<jacco> ACTION: Vassilis to write proposal on the agenda for 20 jult teleconf [recorded in http://www.w3.org/2006/07/11-mmsem-minutes.html#action03]

<jacco> any other business?

<jacco> Jeff: we skipt discussing future deliverables, if any has an idea or volunteers to be an editor, send mail to public mailing list

<jacco> Chairs close the meeting and thank all particpants

<Massimo> hi all and thank you!

<rellero_> see you soon

<pasquale> bye all

<rellero_> bye

<alessio> see you next time :)

<jacco> scribe: jacco

Summary of Action Items

[NEW] ACTION: patrizia to find pointer to this tool [recorded in http://www.w3.org/2006/07/11-mmsem-minutes.html#action01]
[NEW] ACTION: Raphael and Jeff to contact US partners and propose f2f date and location [recorded in http://www.w3.org/2006/07/11-mmsem-minutes.html#action02]
[NEW] ACTION: Vassilis to write proposal on the agenda for 20 jult teleconf [recorded in http://www.w3.org/2006/07/11-mmsem-minutes.html#action03]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2006/07/11 14:49:06 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.127  of Date: 2005/08/16 15:12:03  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/pus/puts/
Succeeded: s/Vasilis/Vassilis/
Found Scribe: Thomas
Inferring ScribeNick: Thomas
Found Scribe: anadiotis
Inferring ScribeNick: anadiotis
Found Scribe: jacco
Inferring ScribeNick: jacco
Scribes: Thomas, anadiotis, jacco
ScribeNicks: Thomas, anadiotis, jacco
Default Present: meeting_room, Vassilis, Jeff_Pan, massimo
Present: Ioannias Vasileios VassilisThomas Raphael Patrizia George Lynda Jacco Giovanni Christian Jeff Pasquale Alessio Massimo Roberto Chris

WARNING: Replacing previous Regrets list. (Old list: Susanne, Boll)
Use 'Regrets+ ... ' if you meant to add people without replacing the list,
such as: <dbooth> Regrets+ SusanneBoll

Regrets: SusanneBoll
Agenda: http://lists.w3.org/Archives/Public/public-xg-mmsem/2006Jul/0009.html
Got date from IRC log name: 11 Jul 2006
Guessing minutes URL: http://www.w3.org/2006/07/11-mmsem-minutes.html
People with action items: jeff patrizia raphael vassilis

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]