[Metadata] Metadata group meeting this Thursday

classic Classic list List threaded Threaded
1 message Options
Schulte  Olaf A. Schulte Olaf A.
Reply | Threaded
Open this post in threaded view
|

[Metadata] Metadata group meeting this Thursday

Dear All

Thursday, April 29th,
at 8:00 am PST / 9:00 am Saskatoon / 10:00 am CST / 11:00 am EST / 4:00 pm GMT / 5:00 pm CET

will see a meeting of the metadata group; we'll meet in the second (!) Adobe Connect room at

http://ado.uvigo.es/opencast2.

Agenda
Series metadata, cf. http://opencast.jira.com/wiki/display/MH/Matterhorn+metadata+scheme+%28work+in+progress%29
Series vs. item metadata, cf. http://opencast.jira.com/wiki/display/MH/Proposed+metadata+mapping+and+inheritance+in+the+Admin+UIs 
RSS: Continue the discussion initiated by Brandon (cf. attached mail)

Regards

Olaf A.


>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

Von: Schulte Olaf A.
Gesendet: Dienstag, 2. März 2010 14:41
An: Brandon Muramatsu; [hidden email]
Betreff: AW: [Metadata] Questions regarding Matterhorn Metadata

Dear Brandon

I think some of your questions touch on the issues we currently deal with and will be discussed in the meeting today, but let me comment upon some of the points you make.

Matterhorn 0.5 - Matterhorn 1.0
The metadata scheme at

http://opencast.jira.com/wiki/display/MH/Matterhorn+metadata+scheme

originates from ETH Zurich's REPLAY scheme and is supposed to become the Opencast Matterhorn metadata scheme for Matterhorn 1.0; as an alpha release, 0.5 only features very rudimental implementations of the metadata scheme, especially with respect to the UI design.
As other institutions have different needs compared to ETH, the metadata scheme is subject to ongoing discussions (cf. http://opencast.jira.com/browse/MH-1279 and my recent e-mail to this list) to satisfy the requirements from various institutions; so please consider it under review for now.

Abstract, keywords
More to your point: dcterms:abstract was used in the REPLAY scheme, but - subject to the discussion we are having on my recent e-mail - will not feature in the Opencast scheme; instead, we will use dcterms:description for both item and series. As for "keywords" (tags?) I would like to see some discussion around this as I could imagine this to be mapped to either dcterms:description (item?) or dcterms:subject (item?). Is there such a thing as dcterms:keywords?

Technical metadata
Finally, this scheme is being completed by a set of more technical metadata which Matterhorn can - in most cases - automatically extract from the tracks, such as duration, resolution, frame rate, ratio etc. Would these be the ones you expect?

See you later

Olaf A.



Von: [hidden email] [mailto:[hidden email]] Im Auftrag von Brandon Muramatsu
Gesendet: Dienstag, 2. März 2010 13:28
An: [hidden email]
Betreff: [Metadata] Questions regarding Matterhorn Metadata

Hi all,

A colleague and I are planning on participating in the metadata meeting in a few hours. I followed a bit of the discussions last spring and summer but was unfortunately not able to engage to any significant extent. Now, my project is grappling with the need to define the metadata we collect, and define how we expose that metadata to a couple services we are developing, and then express that metadata externally for a number of uses.

I have no desire to create yet another spec. (Though if you're interested the last metadata spec I wrote for implementing in a system is available online [1].)

I should mention I haven't had time to play directly with v0.5, but am working under the assumption that I should be able to determine what's going on by looking in the documentation on the wiki and Jira [1].

However, I can't figure out what Matterhorn is doing with metadata from the public Matterhorn documents in Jira and the wiki. I did not go digging through the service definitions.

Specific Questions:

- Is the specification expressed at [2], the one implemented in v0.5?

- If [2] is implemented, how is the metadata mapped to an RSS feed?
---  A quick look at [3] shows that there's a mismatch in labels for fields. For example, if abstract is different than description in [3], then what is the best practice for feed.description?
--- The metadata schema does not appear to allow for tags--they could be encoded in dc.subject, but there does not appear provision for dc.keywords (which is where I would put them) and hence how does one populate feed.rsstags?

- The metadata schema does not appear to include any of what might be described as the technical metadata of the resource, such as the video dimensions, duration, codec, etc. Is this part of the schema? Is this decision documented somewhere that I didn't locate? Or are users expected to extract that themselves from directly from the media (for example, are the Matterhorn services extracting this from the media if they need it, I think the basic example is duration for use in say in captioning)?

- What are Matterhorn's plans for exposing or exchanging metadata outside the system? Only through RSS? Or for example, is there an XML specification for encoding Matterhorn metadata? (If there is I would have thought my searches for metadata would have found it.) Or was this out of scope for the work so far?

- What metadata is being encoded in the video files themselves? For example, are title and creators encoded with the video file?


Thanks.

Brandon Muramatsu
MIT

[1] http://educommons.com/documentation/manual/metadata-and-import-export/metadata-specifications-for-ocw-repositories
[2] http://opencast.jira.com/secure/StudioSearch.jspa?suggest-tracking=search&quickSearch=true&query=metadata
[3] http://opencast.jira.com/wiki/display/MH/Matterhorn+metadata+scheme
[4] http://opencast.jira.com/wiki/display/MH/Atom+and+RSS+Feeds

Dear Brandon



I think some of your questions touch on the issues we currently deal with and will be discussed in the meeting today, but let me comment upon some of the points you make.



Matterhorn 0.5 - Matterhorn 1.0

The metadata scheme at



http://opencast.jira.com/wiki/display/MH/Matterhorn+metadata+scheme



originates from ETH Zurich's REPLAY scheme and is supposed to become the Opencast Matterhorn metadata scheme for Matterhorn 1.0; as an alpha release, 0.5 only features very rudimental implementations of the metadata scheme, especially with respect to the UI design.

As other institutions have different needs compared to ETH, the metadata scheme is subject to ongoing discussions (cf. http://opencast.jira.com/browse/MH-1279 and my recent e-mail to this list) to satisfy the requirements from various institutions; so please consider it under review for now.



Abstract, keywords

More to your point: dcterms:abstract was used in the REPLAY scheme, but - subject to the discussion we are having on my recent e-mail - will not feature in the Opencast scheme; instead, we will use dcterms:description for both item and series. As for "keywords" (tags?) I would like to see some discussion around this as I could imagine this to be mapped to either dcterms:description (item?) or dcterms:subject (item?). Is there such a thing as dcterms:keywords?



Technical metadata

Finally, this scheme is being completed by a set of more technical metadata which Matterhorn can - in most cases - automatically extract from the tracks, such as duration, resolution, frame rate, ratio etc. Would these be the ones you expect?



See you later



Olaf A.







Von: [hidden email] [mailto:[hidden email]] Im Auftrag von Brandon Muramatsu
Gesendet: Dienstag, 2. März 2010 13:28
An: [hidden email]
Betreff: [Metadata] Questions regarding Matterhorn Metadata



Hi all,

A colleague and I are planning on participating in the metadata meeting in a few hours. I followed a bit of the discussions last spring and summer but was unfortunately not able to engage to any significant extent. Now, my project is grappling with the need to define the metadata we collect, and define how we expose that metadata to a couple services we are developing, and then express that metadata externally for a number of uses.

I have no desire to create yet another spec. (Though if you're interested the last metadata spec I wrote for implementing in a system is available online [1].)

I should mention I haven't had time to play directly with v0.5, but am working under the assumption that I should be able to determine what's going on by looking in the documentation on the wiki and Jira [1].

However, I can't figure out what Matterhorn is doing with metadata from the public Matterhorn documents in Jira and the wiki. I did not go digging through the service definitions.

Specific Questions:

- Is the specification expressed at [2], the one implemented in v0.5?

- If [2] is implemented, how is the metadata mapped to an RSS feed?
---  A quick look at [3] shows that there's a mismatch in labels for fields. For example, if abstract is different than description in [3], then what is the best practice for feed.description?
--- The metadata schema does not appear to allow for tags--they could be encoded in dc.subject, but there does not appear provision for dc.keywords (which is where I would put them) and hence how does one populate feed.rsstags?

- The metadata schema does not appear to include any of what might be described as the technical metadata of the resource, such as the video dimensions, duration, codec, etc. Is this part of the schema? Is this decision documented somewhere that I didn't locate? Or are users expected to extract that themselves from directly from the media (for example, are the Matterhorn services extracting this from the media if they need it, I think the basic example is duration for use in say in captioning)?

- What are Matterhorn's plans for exposing or exchanging metadata outside the system? Only through RSS? Or for example, is there an XML specification for encoding Matterhorn metadata? (If there is I would have thought my searches for metadata would have found it.) Or was this out of scope for the work so far?

- What metadata is being encoded in the video files themselves? For example, are title and creators encoded with the video file?


Thanks.

Brandon Muramatsu
MIT

[1] http://educommons.com/documentation/manual/metadata-and-import-export/metadata-specifications-for-ocw-repositories
[2] http://opencast.jira.com/secure/StudioSearch.jspa?suggest-tracking=search&quickSearch=true&query=metadata
[3] http://opencast.jira.com/wiki/display/MH/Matterhorn+metadata+scheme
[4] http://opencast.jira.com/wiki/display/MH/Atom+and+RSS+Feeds


_______________________________________________
Metadata mailing list
[hidden email]
http://lists.opencastproject.org/mailman/listinfo/metadata