History of the MO

[Norbert von Xanten]

Hi,

i'm interested in how the way the MO was published and maintained through the years since 2006, finally to understand why it's - in relation to other ontologies - so popular.

Is it because of the use in BBC Music, the offering of various API's, the community of developers, the popular topic or the full documentation or something else?

What was the intended way the creators want to publish the MO and make it available for other developers? When launched the first Website and when was it called a "good ontology" by the w3c? Was it published in various ontology-libraries? What have been done better?

You see, a lot of questions. In fact, its interesting how the history of the MO looks so far - maybe as a guideline for other ontology developers.

best regards

norbert

[Frederick Giasson]

Hi Norbert,

> i'm interested in how the way the MO was published and maintained
> through the years since 2006, finally to understand why it's - in
> relation to other ontologies - so popular.

I started the Music Ontology in 2006 for a really simple reason: I
wanted to create a dataset of musical things (artists, albums, tracks,
etc) from the musicbrainz.org dataset. At the time, no such ontology was
existing, and so I created one that would be expressive enough to be
able to express all the entities and properties of the things that were
existing in MBZ.

Here is the first blog post I wrote about it [1]

Additional stuff can be found here [2]

Not long after I started the project and created the first version of
the ontology and setuped the website and mailing list Yves Raymond
started to work on the project with me. At that time, he was not at the
BBC but was doing his Master and/or Doc on the subject (don't remember
exactly). This project started in the context of a startup that doesn't
exist anymore: Zitgist.

It has been popular because it was the first real ontology to describe a
wide range of things related to music. Eventually many other
contributors appeared and created all kind of extensions to the ontology
to describe all kind of really specific things and workflows related to
music composition.

Eventually because of the direction some of my work took, and because of
Yves' (and other) involvement in the music domain, my contributions
faded in time and the project was living by itself, supported by an
array of contributors.

> Is it because of the use in BBC Music, the offering of various API's,
> the community of developers, the popular topic or the full
> documentation or something else?

This certainly helped quite a lot for the adoption. However Yves could
tell you much more in that regard since he was working for the BBC at
the time.

> What was the intended way the creators want to publish the MO and make
> it available for other developers? When launched the first Website and
> when was it called a "good ontology" by the w3c? Was it published in
> various ontology-libraries? What have been done better?

I think that MO is a really good example of an ontology project. I would
say so because the project is still quite alive nearly 10 years after
its creation. A community growth and sustained it over time. All kind of
people worked on it to enhance its expressiveness. A good sign that this
is a good ontology is the way people adopted it. It even influenced the
Music entities created in Schema.org.

The real strength of an ontology project I think is its community. If a
project can lives by itself, without needing any kind of organization to
finance it, if people use it in their pet projects or in their million
dollars projects, then I think we can consider that ontology a good
ontology. An ontology is always an evolving thing... an ontology evolves
to be good. I don't think that an ontology starts really good.

I think that other good ontologies (as big, or as simple as they could
be) are: FOAF, SIOC, BIBO (Bibliographic Ontology) and others.

> You see, a lot of questions. In fact, its interesting how the history
> of the MO looks so far - maybe as a guideline for other ontology
> developers.

[1]
http://fgiasson.com/blog/index.php/2006/12/21/the_music_ontology_a_new_ontology_based_/
[2] http://fgiasson.com/blog/index.php/category/musicontology/


Thanks

Take care,

Fred

[Samer Abdallah]

Dear Norbert,

I think it would be fair to say that the Music Ontology has several parents :)

I would like to describe a little about the Queen Mary based branch of the family.

I started work on a relational basis for representing information about music

related data and music related computations in about 2003, shortly

after I started learning Prolog. After considering approaches based on tagging

(MPEG-7 etc), relational databases, and Datalog, I settled on a knowledge

representation based on deductive logic (basically, Prolog).  The motivation was 

(1) to be able to express all the circumstantial data (metadata) around a corpus of audio files,

(2) to use this to design experimental workflows (i.e. what algorithms to run on which files,

(3) to record the results of all computations and the parameters that went into them,

all within the same (relational/logic based) representation framework.

In fact, a pure knowledge representation system was not enough to achieve all of this,

but I found that the Transaction Logic invented by Bonner and Kifer in 1993 [4] fit

the task very well, expressing the logic of database updates without resorting to

Prolog's somewhat muddy procedural semantics with side-effects, and fitting in with

the deductive logic framework. (The transaction logic part is not so relevant to the

Music Ontology proper, but found its way into later work, e.g.  [11].)

The timeline ontology has its roots in in work I did on a project to do with segmentation,

which lead me to learn about temporal logics, and can be found in an unpublished technical 

report that was eventually published, but without the temporal logic bits, in [1,2].  

Around this time, 2004 I think, Yves Raimond joined the group as an intern (Yves, you may 

have to correct me on the details) and we started working towards a system based on 

description logics, OWL, RDF etc.  Yves and I developed the event ontology, driven by a desire 

to be able to describe the potentially complex production processes behind an audio signal, 

and he did a huge amount of work developing the framework, especially on the semantic 

web side of things, and turning it into a usable resource, rather than a collection of weird Prolog libraries. This, along with some of the earlier work on the timeline ontology and more 

information about the implementation details from [1,2] was published in 2006 [3,5,6,7], but

I found some unpublished papers and reports from 2004 and 2005, (including

Yves' internship report), lurking on my hard drive. I was amused to find the slides for this talk

http://c4dm.eecs.qmul.ac.uk/seminars-historical.html (see 1st December, 2004)

https://www.dropbox.com/s/y1zcj7y299ida7h/talk2004.pdf?dl=0)

which starts off talking about big fleas and little fleas… 

In 2006 or 2007, our work joins up with Frederick's [7].

I should also mention Oscar Celma's work from 2006 [12], which also talks about semantic

web technologies for music recommendation, but I don't know the history behind that, and

I do not know if it had any influence on the Music Ontology in its present form.

Best wishes,

Samer

[1] S. Abdallah, K. Noland, M. Sandler, M. Casey, and C. Rhodes. Theory and evaluation of a Bayesian music structure extractor. In J. D. Reiss and G. A. Wiggins, editors, Proceedings of the Sixth International Conference on Music Information Retrieval, pages 420–425, 2005.

[2] S. Abdallah, C. Rhodes, M. Sandler, and M. Casey. Using duration models to reduce fragmentation in audio segmentation. Machine Learning, 65:485–515, 2006.

[3] S. Abdallah, Y. Raimond, and M. Sandler. An ontology-based approach to information management for music analysis systems. In Proc. 120th AES Convention, Paris, France, 2006.

[4] A.J. Bonner and M. Kifer (1993), Transaction Logic Programming, International Conference on Logic Programming (ICLP), 1993.

[5] S. Abdallah, Y. Raimond, and M. Sandler. Proposal for a common multimedia ontology framework: Information management for music analysis systems. AceMedia (http://www. acemedia. org/) call for proposals on a common multimedia ontology framework, 2006.

[6] Y. Raimond, S. A. Abdallah, M. Sandler, and M. Lalmas. A scalable framework for multimedia knowledge management. In Semantic Multimedia, pages 11–25. Springer Berlin Heidelberg, 2006.

[7] Y. Raimond and S. A. Abdallah. The timeline ontology. OWL-DL ontology, 2006.

[8] Y. Raimond, S. Abdallah, M. Sandler, and M. Lalmas. A framework for multimedia processing and knowledge management. 2007.

[9] Y. Raimond, S. Abdallah, M. Sandler, and F. Giasson. The music ontology. In International Conference on Music Information Retrieval, ISMIR 2007, 2007.

[10] Y. Raimond and S. Abdallah. The event ontology. Technical report, Technical report, 2007. http://motools. sourceforge. net/event, 2007.

[11] D. Pastor, Y. Raimond, S. Abdallah, and M. Sandler. A logic-based framework for digital signal processing. 2008

[12] O`. Celma. Foafing the music: Bridging the semantic gap in music recommendation. In The Semantic Web-ISWC 2006, pages 927–934. Springer, 2006.

--
You received this message because you are subscribed to the Google Groups "Music Ontology Specification Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to music-ontology-specific...@googlegroups.com.
To post to this group, send email to music-ontology-sp...@googlegroups.com.
Visit this group at http://groups.google.com/group/music-ontology-specification-group.
For more options, visit https://groups.google.com/d/optout.

[Yves Raimond]

Hello Norbert,

Well I think there's not much for me to add to to Frederick and Samer's account :) There was an early version of an OWL-DL ontology that we wrote with Samer prior to us joining forces with Frederick, but sadly it looks like the site hosting it is down. Then it looks like in Aug 2007 we started working together with Frederick in combining the two ontologies - our old repository's SVN history should give you a pretty detailed account of what happened when [2] :) 

I guess I would also briefly mention dbtune.org - which was one of the first Linked Data repository (it appeared on the first LOD diagram in 2007 [1]) and made extensive use of MO, so I guess that it might have helped with adoption. An even more important aspect is, as you say, the community that grew around it - I don't think we'd have been able to keep the project alive that long without it.

That said I still think there's a lot of room for improvements. In particular with the upcoming release of the schema.org music extension, we'll need to make sure MO is fully compatible with it. There are also still some confusing bits left over from old times, which are marked 'deprecated' since roughly 6 years :)

Best,

Yves

[1] http://www.ibm.com/developerworks/library/wa-realweb5/lod-datasets_2007-11-10.gif

[2] http://sourceforge.net/p/motools/code/HEAD/tree/

[Norbert von Xanten]

@all: Thank your for your quick and so detailed response! :) Thats very helpful.

Seems that the success of and commitment to an ontology primary depends on the community contributing the ontology (as you said), and (in this case) of course the different 'roots' of the developers and the communities they're connected to. Correct me if i'm wrong. Are there any popular online-communities for (general) ontology-developing? I only found the ontolog-forum, but, however, i can't get familiar with that. 

Besides that: Are there any tools on the market to measure the proliferation of an ontology? I ask this question referring to the conclusion of the latest publication to the MO [1] that indicates the measure of the popularity of the MO as a future work. I think its possible in the case that the proliferation is measurable through the number and ways the ontology is mapped to other ontologies (see graphics at [2]). But, in my thoughts, this dependency is not attestable at all.

Thanks to all of you,

Norbert

[1] Raimond, Sandler: Evaluation of the Music Ontology framework. 2012.

[2] http://lov.okfn.org/dataset/lov/details/vocabulary_mo.html

[Frederick Giasson]

Hi Guys,

Thanks Yves and Samer for wrapping up the story :)

Well I think there's not much for me to add to to Frederick and Samer's account :) There was an early version of an OWL-DL ontology that we wrote with Samer prior to us joining forces with Frederick, but sadly it looks like the site hosting it is down. Then it looks like in Aug 2007 we started working together with Frederick in combining the two ontologies - our old repository's SVN history should give you a pretty detailed account of what happened when [2] :) 

I guess I would also briefly mention dbtune.org - which was one of the first Linked Data repository (it appeared on the first LOD diagram in 2007 [1]) and made extensive use of MO, so I guess that it might have helped with adoption. An even more important aspect is, as you say, the community that grew around it - I don't think we'd have been able to keep the project alive that long without it.

That said I still think there's a lot of room for improvements. In particular with the upcoming release of the schema.org music extension, we'll need to make sure MO is fully compatible with it. There are also still some confusing bits left over from old times, which are marked 'deprecated' since roughly 6 years :)

Yeah, we will have to make sure we can align both ontologies.

Take care,

Fred

[Frederick Giasson]

Hi Norbert,

> Seems that the success of and commitment to an ontology primary
> depends on the community contributing the ontology (as you said), and
> (in this case) of course the different 'roots' of the developers and
> the communities they're connected to. Correct me if i'm wrong. Are
> there any popular online-communities for (general)
> ontology-developing? I only found the ontolog-forum, but, however, i
> can't get familiar with that.

To my experience, what really works are the dedicated mailing list for
domain specific ontologies like MO, BIBO and SIOC. There is
public...@w3.org but it is mostly just related to Schema.org
development these days.

> Besides that: Are there any tools on the market to measure the
> proliferation of an ontology? I ask this question referring to the
> conclusion of the latest publication to the MO [1] that indicates the
> measure of the popularity of the MO as a future work. I think its
> possible in the case that the proliferation is measurable through the
> number and ways the ontology is mapped to other ontologies (see
> graphics at [2]). But, in my thoughts, this dependency is not
> attestable at all.

There are generally two aspect of this:

(1) how much an ontology is integrated in other ontologies (how many
other ontologies re-uses/link-to/etc to its classes and properties)
(2) how many entities are described using the ontology

They are distinct but probably as important. There were a few projects
in the past, but none of them I can remember at the moment.

You have LOV: http://lov.okfn.org/dataset/lov/

That does some of this, but it is constrained to the OKFN data. Here is
the page for MO: http://lov.okfn.org/dataset/lov/details/vocabulary_mo.html

But I don't think it is representative.


Thanks

Fred