Semantic Web, Music Ontology, Web

Major revision (1.01) of the Music Ontology

 

This is a draft of the Music Ontology revision 1.01. I took all the propositions posted by people on the mailing list since I released the revision 1.00 and then I wrote that new revision.

The ontology took a major shift by its deep integration with the FRBR (Functional Requirements for Bibliographic Records) and FOAF (Friend Of A Friend) ontologies.

As you will see with the classes schemas bellow, all the MO classes related with music are sub classes of FRBR classes. The FRBR ontology is used as the basement of musical works. So as you will see, a mo:Album is a sub class of a mo:MusicalWork and this class is a sub class of the frbr:Work class. This means that an album is ultimately a work in the sense of the FRBR ontology. The FRBR ontology is better explained by reading the Functional Requirements for Bibliographic Records – Final Report.

In this revision of the MO ontology, you should read these terms as:

 

Work: An abstract notion of an artistic or intellectual creation.

Musical work: Distinct intellectual or artistic musical creation.

Expression: A realization of a single work usually in a physical form.

Musical expression: Intellectual or artistic realization of a musical work.

Manifestation: The physical embodiment of one or more expressions.

Musical manifestation: Physical embodiment of an expression of a musical work.

Item: An exemplar of a single manifestation.

Corporate body: Organization or group of individuals and/or other organizations involved in the music market.

 

Music Ontology Classes Hierarchy

There are the schemas of the hierarchy of the Music Ontology classes. The pink bubbles are super-classes of external ontologies, green bubbles are super-classes of the MO ontology and blue bubbles are subclasses of the MO ontology.

 


 

 

 

 

 

 


 

Change log

There are the changes I made to the revision 1.00 of the MO ontology. All the changes have been wrote into the RDFS document describing the ontology. Please read it to see what changed for each classes and properties. Also note that I cleaned and redefined all the definitions in relation with the FRBR and FOAF ontologies.

So I outlined the changed bellow:

 

  • Suppressed class mo:Type
  • Suppressed class mo:Other

These classes were useless and confusing.

 

  • Suppressed class mo:EP
  • Suppressed class mo:Longplay
  • Suppressed class mo:Single

I deleted these three classes because they are all albums with less or more tracks. Since it doesn’t add anything to the semantic of the ontology, I choose to remove them to focus the ontology. Now one should describe each of these type of release as a mo:Album.

 

  • Changed the name of mo:Status for mo:ReleaseStatus

The class has a better meaning that way since the subclasses of mo:Status were in fact release status.

 

  • Added classe mo:Genre
  • Added classe mo:Classical
  • Added classe mo:Rock
  • Added classe mo:Jazz
  • Added classe mo:World
  • Added classe mo:Hiphop
  • Added classe mo:Country
  • Added classe mo:Blues
  • Added classe mo:Electronica
  • Added classe mo:Gospel
  • Added classe mo:Funk
  • Added classe mo: Pop
  • Added classe mo:Melodic
  • Added classe mo:Reggae

These classes handle a first level musical genre. One can now extend the ontology by creating new genres and/or subgenres of the existing genres.

 

  • Added classe mo:Instrument
  • Added classe mo:String
  • Added classe mo:Woodwind
  • Added classe mo:Brass
  • Added classe mo: Percussion
  • Added classe mo:Keyboard
  • Added classe mo: Digital

These classes handle a first level of instrumentation. One can now extend the ontology by creating classes for instruments belonging to these instrument categories.

 

  • Added classe mo:MusicalManifestation
  • Added classe mo: Dat
  • Added classe mo: Dcc
  • Added classe mo:Cd
  • Added classe mo:Md
  • Added classe mo: Dvda
  • Added classe mo:Sacd
  • Added classe mo:Vinyl
  • Added classe mo:Megnetictape
  • Added classe mo:Steam

These classes handle a first set of possible mediums of musical recording and distribution technologies.

 

  • Added classe mo:MusicalWork
  • Added classe mo:MusicalExpression
  • Added classe mo:MusicalManifestation

Creating subclasses of the frbr:Work, frbr:Expression and frbr:Manifestation specifically to express musical works.

 

  • Added classe mo:SoloMusicArtist
  • Added classe mo:MusicGroup
  • Added classe mo:CorporateBody
  • Added classe mo:Label

These classes describe musical people, group of people or corporate body in relation with the FOAF ontology. That way, all the FOAF, BIO, etc. ontologies can now be use to define an artist, a group of artist a corporate body and their relationship.

 

  • Changed the name of mo:Artist for mo:MusicArtist

Specifying the semantic of the mo:Artist by renaming it to mo:MusicArtist to explicit the fact that the ontology talk about musicians (and not writers, etc.).

 

  • Added property mo: possess_item
  • Added property mo:want_item
  • Added property mo:sell_item
  • Added property mo:exchange_item

These new properties are used by people to express the fact they possess exemplar of a musical manifestation, and if they want to sell or exchange it, or if they want an exemplar of a musical manifestation, and that, directly in their FOAF profile.

These properties will be quite useful to let people trade CDs for example. Check bellow for an example of how this will be implemented.

 

  • Added property mo:instrument

Link a person, a group of person, a musical work, the expression of a musical work, the manifestation of a musical work or an exemplar of a manifestation to a musical instrument.

 

  • Added property mo:key
  • Added property mo:timber
  • Added property mo: pitch
  • Added property mo:lyric

Adding expressiveness to describe mo:Track.

 

  • Added property mo:encoding
  • Added property mo:stream_url

 

  • Suppressed property mo:miscellaneoused
  • Suppressed property mo: performance_name

Deleting useless properties.

 

Finally I changed the domain and the range of most of the existing properties to reflect the changes in the way classes now work (the FRBR and FOAF ontologies).

 

An example of how the new mo:has_item, mo:sell_item, mo:exchange_item and mo:want_item are working

These new properties are quite interesting since it will enable people to trade music using their FOAF profile. There is an example of how these properties should be use. You have my FOAF profile, the properties that explicit the fact that I have and want to sell or exchange some musical albums and you have the description of these albums.

 

RDF document demonstrating the integration of the MO and FOAF ontologies

 

The resulting graph can be queried with SPARQL queries like:

This query will return the name of the people selling the album “Kill ’em All”.

PREFIX mo:      <http://purl.org/ontology/mo/>
PREFIX foaf:    <http://xmlns.com/foaf/0.1/>        
SELECT ?name
WHERE 
{
    ?seller a foaf: Person;
            mo:sell_item <http://mm.Music.org/album/a89e1d92-5381-4dab-ba51-733137d0e431>;
            foaf:name ?name.
} 

But what if someone doesn’t know the resource that define an album, or if the album is defined by more than one resource?

This query will return the name of the people wanting the albums having the word “Love” in its title.

PREFIX mo:      <http://purl.org/ontology/mo/>
PREFIX rdfs:  <http://www.w3.org/2000/01/rdf-schema#>
PREFIX foaf:    <http://xmlns.com/foaf/0.1/>        
SELECT ?name ?title
WHERE 
{
    ?seller a foaf: Person;
            mo:sell_item ?album;
            foaf:name ?name.

    ?album a mo:Album;
           rdfs:label ?title.
    FILTER regex(?title, "love", "i")
}

Would you like to make people happy? Then you can easily get the name of people selling an album and get the name of people that want that album.

PREFIX mo:      <http://purl.org/ontology/mo/>
PREFIX foaf:    <http://xmlns.com/foaf/0.1/>        
SELECT ?name_seller ?name_buyer
WHERE 
{
    {
        ?seller a foaf: Person;
                foaf:name ?name_seller;
                mo:sell_item <http://mm.Music.org/album/a89e1d92-5381-4dab-ba51-733137d0e431>.
    }
    UNION
    {
        ?seller a foaf: Person;
                foaf:name ?name_seller;
                mo:exchange_item <http://mm.Music.org/album/a89e1d92-5381-4dab-ba51-733137d0e431>.
    }
    UNION
    {    
        ?buyer a foaf: Person;
               foaf:name ?name_buyer;
               mo:want_item <http://mm.Music.org/album/a89e1d92-5381-4dab-ba51-733137d0e431>.
    }
}

Finally… do what you want 🙂

 

Conclusion

Please take the time to revise this draft if you are interested in the ontology. I will wait some days before rewriting the documentation of the ontology to make sure that people agree with the draft and to make sure that there is no major error.

 

 

Technorati: | | | | | | | |

11 thoughts on “Major revision (1.01) of the Music Ontology

  1. Are you sure that last SPARQL query does what you intend? It looks to me like it’ll list every named person as seller, even if they’re neither trying to sell nor exchange that album. I suspect you actually want a UNION of people willing to sell and people willing to exchange.

  2. Cool.

    I had a small role in the FRBR ontology. Have you talked to Rich and Ian about your modelling? I think some of it isn’t quite right.

    For example, foaf:Group should probably be a subclass of frbr:CorporateBody (the “corporate” does not refe to corporations).

    Also, I’m not so sure I’d call an Album a subclass of a Work. FRBR is particularly powerful with stuff like clasical music, where a work would be something like a symphony, which is expressed in recorded performances (as well as musical scores), and embodied in recordings (including albums).

    I don’t have time to get into this in detail, but just suggesting you look for feedback from those guys, and also the FRBR world (they have a mailing list).

  3. Hi Bruce,

    No I didn’t contacted them.

    About the frbr:CorporateBody, I thought about it but I wanted to use the foaf ontology instead. Since the frbr:CorporateBody is equivalent to the union clas of foaf:Organization, foaf:Group, etc, I wanted to use foaf:Agent instead, so all the BIO, REL, etc. ontology could be reuse to describe the relationship between people, groups, organization, etc.

    About mo:Album, you could be right, we should get feedback from people. But in my opinion, an album is a Work. In fact, it is an artistic creation. But since it is an abstract class, it is why I added a layer called a mo:MusicalWork that is a musical work (so a specialization of a Work).

    However, I am open to any suggestion to change that.

    Take care,

    Fred

  4. Frederick-

    There are a couple of issues related to your misunderstanding of the FRBR entity Expression. You define it as: “A realization of a single work usually in a physical form.” In fact Expression has no physical characteristics (this is part of the Manifestation entity); a better definition is “A realization of a single work in a fixed notation or expressive form.” For music, possible expressive forms would include sound recording, sheet music, etc. Some of your MusicalExpression subclasses seem to incorporate publication-related distinctions (e.g. “Track”, “AudioBook”) that are only meaningful at the Manifestation level.

  5. Hi Mr. Leybovich,

    Thanks for these comments. For your first statement, you are completely right, naturally. Unfortunately I don’t see where I wrote that definition? Could you please tell me where you read that?

    After that, for the distinction between MusicalExpression and MusicalManifestation I will have to clarify things.

    In fact, I see an Album and a Track not has manifestation, but as Work or expression. In fact, the physical embodiment of a track will be via a CD, etc (so, MusicalManifestation). But a track, is an artistic creation. When an artist (musicien) compose a track for an album, he perform an artistic creation, no?

    Please tell me how you see the distinction between MusicalWork, MusicalExpression and MusicalWork.

    Okay, I will take a look at this when I’ll get a moment. I should take the time to revize the ontology in a couple of days with people comments.

    Take care!

    Salutations,

    Fred

  6. Hi Fred,

    Near the top of the 1.01 revision you define: “Expression: A realization of a single work usually in a physical form.”

    Tracks and Albums are interesting in that- though I think they are obviously manifestation-derived entities (album of course refers to the housing in which phonographic records were distributed), in time they did in a sense become expression-related entities as artists consciously adapted their compositions to the structure and limitations of the medium- for example, with the “concept album” of unified songs, some of them merely short segues between other songs. And now with the iPod and MP3, the album may have finally reached the limits of its relevance.

    Yet if we consider both entities more abstractly, I think we see that “Track” is merely a singular/atomic musical composition, and “Album” is an aggregate musical composition composed of one or more distinct “Tracks”. The original nomenclature you used is not inherently bad, but I think does not sound quite right when applied to classical music. Should we really call an opera an instance of “Album” and an aria within that opera an instance of “Track”?

    Regards,

    Jonathan

  7. Hi Jonathan,

    I’m new to this ontology, but I’d like to say ‘good job!’ I thought I would make a couple of observations:

    1.) One of the cool things about FRBR is its ability to relate different expressions of the same work together. In music, this typically could mean that I would have the Audio and the Score be the same WORK, but a different EXPRESSION. This may also hold true for things such as remixes, etc. In fact, Barbara Tillet has produced a chart for the difference between identical works, and the same work but a different expression. (consult page 4 of this document: http://www.loc.gov/cds/downloads/FRBR.PDF) Are you specifically trying to limit this to audio-based expressions?

    2.) Instead of, or complimentary to, the FOAF Agent object, you might want to have a look at the draft proposal of the Functional Requirements and Number of Name Authority Records. (http://www.ifla.org/VII/d4/wg-franar.htm). According to this document, and to the FRBR document, a ‘corporate body’ is any entity that is a collection of two or more people. (‘corporate’ being understood not in the business sense, but in the collective sense). In that sense, a ‘musical group’ is a corporate body. This, along with the attributes described in FRANAR, allows you to encapsulate corporate history. You could, I suppose, also describe relationships within that corporate body. (i.e. “Mick Jagger” is a person, but is a member of “The Rolling Stones,” a corporate body.)

    3.) Who do you envision as the user group for this ontology?

    If I’ve misunderstood something, let me know.

  8. Hi Andrew!

    First of all, thanks for your comment about the Music Ontology.

    My first question would be: did you take a look at the latest revision of the ontology? (1.11). Many things changed since then. Also, I would suggest you to subscribe to the mailing list of the ontology to share these thoughts with the other member of the community!

    [quote post=”761″]1.) One of the cool things about FRBR is its ability to relate different expressions of the same work together. In music, this typically could mean that I would have the Audio and the Score be the same WORK, but a different EXPRESSION. This may also hold true for things such as remixes, etc. In fact, Barbara Tillet has produced a chart for the difference between identical works, and the same work but a different expression. (consult page 4 of this document: http://www.loc.gov/cds/downloads/FRBR.PDF) Are you specifically trying to limit this to audio-based expressions?[/quote]

    No definitely not. Please take a look at the Musical Creations Workflow. I think that it will answer to these questions. At first, the FRBR ontology hasn’t been properly integrated, howerver I think that with the latest revision, many things have been fixed. Please tell me what you think about it.

    [quote post=”761″]2.) Instead of, or complimentary to, the FOAF Agent object, you might want to have a look at the draft proposal of the Functional Requirements and Number of Name Authority Records. (http://www.ifla.org/VII/d4/wg-franar.htm). According to this document, and to the FRBR document, a ‘corporate body’ is any entity that is a collection of two or more people. (’corporate’ being understood not in the business sense, but in the collective sense). In that sense, a ‘musical group’ is a corporate body. This, along with the attributes described in FRANAR, allows you to encapsulate corporate history. You could, I suppose, also describe relationships within that corporate body. (i.e. “Mick Jagger” is a person, but is a member of “The Rolling Stones,” a corporate body.)[/quote]

    Well yeah. Two things in that situation then, or we put mo:CorporateBody as a subclass of frbr:CorporateBody, or we check if foaf:Organization can be equivalent to frbr:CorporateBody. The goal here would be to link these ontologies together at the ontologies level.

    [quote post=”761″]3.) Who do you envision as the user group for this ontology?[/quote]

    What you mean? The people that could use it? Anybody; from anyone who has a foaf profile, the edition company, the producer, the artists, the sellers (amazon), the bbc, etc. Check out my latest post about MO to get some projects developed by other people that shows a little bit of what can be done with that ontology.

    Take care!

    Salutations,

    Fred

Leave a Reply