AW: 回复:AW: About metadata implementation

Dear Jianbin,

if I understand you are defining your own ontology, rather than what this WG is proposing. We'd be interested to know the reasons why our proposed ontology is not suitable for you.

Concerning the representation: if you look at the examples of our API document, JSON is a representation we consider relevant for the client-side API.

Best regards,

Von: Dai Jianbin 00901725 []
Gesendet: Sonntag, 16. Jänner 2011 22:57
An: Bailer, Werner
Betreff: 回复:AW: About metadata implementation

Thanks Werner!
Yes, we are working the ontology for media resource, very much like media annotation group's work. But we have some properties that could contain very flexible values, such as a set of weighted keywords which best describe the content. The problem is to find a right representation method that supports modification and search well. Options I can think of are, xml and supported databases, JSON and CouchDB/MongoDB, Or Key-value Pairs (not sure
 if it's feasible). I would love to hear any suggestions.


> Dear Jianbin,
> is your request specifically about representing the ontology for
> media resources? We do propose a set of properties, but we do not
> specify the persistent representation.
> Some of the formats for which we have defined mappings do indeed
> have binary representations, e.g. MPEG-7 BiM, or SMPTE KLV.
> Best regards.
> Werner
> ________________________________________
> Von: [public-media-
>] im Auftrag von Dai Jianbin 00901725
> []Gesendet: Sonntag, 16. J?nner 2011 21:33
> An:
> Betreff: About metadata implementation
> Hi,
> I am new in metadata area. I have some questions regarding
> metadata implementation that someone here may help me out. It
> seems to me that XML is the only computer language to represent
> and store metadata on a computer. But for large set of metadata,
> which requires lots of editing work like modification, deleting,
> appending, etc., XML file is very inefficient for any real
> applications. Is it possible to use some other methods, such as
> key-va
> lue pair, to represent the metadata scheme?
> Thank you very much if you would like to share any related
> knowledge on this topic.
> JB-

Received on Thursday, 27 January 2011 12:40:17 UTC