W3C home > Mailing lists > Public > public-media-annotation@w3.org > May 2010

A first DRAFT of my action item

From: StrassnerˇˇJohn Charles <johns@postech.ac.kr>
Date: Tue, 25 May 2010 17:15:40 +0900 (KST)
Message-ID: <15856176.1274775340565.JavaMail.root@mail1.postech.ac.kr>
To: <public-media-annotation@w3.org>

.Bold { font-weight: bold; }
.Title { font-weight: bold; font-size: 18px; color: #cc3300; }
.Code { border: #8b4513 1px solid; padding-right: 5px; padding-left: 5px;color: #000066; font-family: 'Courier New' , Monospace;background-color: #ff9933; }

Hello all,

attached please find a first DRAFT of my review of the latest ontology document from Wonsuk.

I have tried to accomplish three things:

&nbsp; 1) correct and standardize English usage and grammar

&nbsp; 2) add a conformance section

&nbsp; 3) insert conformance statements into the document where appropriate

In performing the above, I found&nbsp;a LOT of mis-spellings, which I have tried to fix. I'm sure I may have missed some. I found a LOT of inconsistencies (e.g., 16 bit, 16bit, and 16-Bit all occurred multiple times in the document), which I have tried to standardize. This is one example of many. I'm sure I missed some. There was mixed American and British English - I defaulted to American English EXCEPT where the standard itself used British English. There were many minor differences between &quot;syntax&quot; and &quot;semantics&quot; - I have made every effort to reconcile these.

For instances such as:

&nbsp; &quot;The unique portable identification of an asset is the combinations of its Provider_ID and its Asset_ID.&quot;

I have tried to make this into a real conformance statement, as follows:

&nbsp; &quot;The unique portable identification of an asset MUST be defined as the concatenation of its 

&nbsp;&nbsp;&nbsp; Provider_ID followed by its Asset_ID.&quot;

This eliminates &quot;combinations&quot;, and provides something to test conformance against. Note, however, that not all elements in the mapping table had strict definitions. For example, several just had &quot;String&quot; as a datatype, with no character length indicated. This is not sufficient to provide a conformance statement against.

However, in doing the above, I have&nbsp;encountered a number of problems when edting the specification. I wasn't sure how to fix these, so left them alone. Note that these SHOULD :-) have conformance statements associated with them.

In general, the Actors elements in the mapping table are all not specified very well. The Studio_Royalty_Flat_Rate is ill-specified (e.g., no currency symbol is given, no upper limit is given). 

There were several elements, such as Screen_Format, that **seem** to imply conformance level values (e.g., enumerating of Standard, Widescreen, Letterbox, OAR), but did not explicitly do so. In addition, they did not indicate if this was a completely defined enumeration or not.

I was not sure what was meant by a &quot;numeric string&quot; (e.g., in ma:frameSize or in ma:numTracks).

For the DIG mappings, the ma:identifier datatype is ill-specified. It says &quot;ComplexType: sequence of UID (
) and ID_TYPE (
)&quot;; the obvious problem is, what does &quot;sequence&quot; mean (one instance or many). Note also that this is the first usage of UID, and as such, it should be defined. (Earlier, UUID was introduced, also without a definition!). The ma:license DIG mapping is ill-specified (i.e., &quot;Complex types using different fields...&quot; needs a more precise definition).

For the ID3 mappings, what exactly is a &quot;numeric string&quot;? This needs to be defined.

For the LOM mappings, the datatype &quot;string(vCard)&quot; is not precise. What version of vCard are you using? In addition, the vCard format allows private extensions, and is best described as a complex data type, not as a string, imho. In addition, what is a Duration datatype? (It was mis-spelled as Duraction).

Why, in MediaRDF, do you suddenly switch from &quot;string&quot; to &quot;plain literal&quot;? Datatypes shoudl be consistent. Note that most of this table is NOT specified...

In MediaRSS, several properties are complex types, but you have their datatypes as &quot;string&quot;.

For the MPEG-7 mappings, there are many questions on the datatypes. For example, what is meant by &quot;string + qualifier...&quot;? What type of string, exactly. Is only one qualifier to be specified, or can a set be specified? What is the format of the qualifier? etc.

One important class of problems encountered occurrences of text that did not propose a final goal, as well as text that did not state a clear objective. An example of the original text illustrating this is:

&nbsp;&nbsp;&nbsp; &quot;Some more fine grained definition of the properties has still to be done: we need to define their&nbsp;

&nbsp;&nbsp;&nbsp;&nbsp; formal properties (if they are symmetric, etc) to enhance more efficient concrete mappings. The 

&nbsp;&nbsp;&nbsp;&nbsp; mappings have to be as precise as possible to be efficiently used in the related API.&quot;

If I was a&nbsp;reviewer and saw this statement, I would immediately recommend that the document be returned&nbsp;to the working group because it was not complete. While the intent is of course very good, the&nbsp; problem is that the above statement implies that the document is not finished. Therefore, why is it worthy of being declared a Recommendation?

Another example is: 

&nbsp;&nbsp; &quot;The group still looks at rationale for including and excluding formats to be considered in the final 

&nbsp;&nbsp;&nbsp; mapping table&quot;. 

This is an open-ended statement, and as such, is not appropriate for a standard in its current form. There is no way that conformance can be defined for such a statement.

Sentences such as: &quot;Audio PID(s) shall correspond with Languages. Two character language code from 639-1.&quot; are typically considered incomplete. For example, &quot;639-1&quot; is not completely specified; ideally, it should have a citation. Similarly, sentences such as &quot;String, having a maximum of two characters per language, 1024 characters total, one language per element&quot; are confusing:

&nbsp; 1) can a language code have 0 or 1 characters? (that is what &quot;a maximum of two implies!)

&nbsp; 2) &quot;element&quot; isn't precisely defined

More fun (not!) is the following (referring to &quot;advisories&quot;):

&nbsp; &quot;String, one advisory per element, max 1024 characters for all advisories...There are at most

&nbsp;&nbsp;&nbsp;six occurrences fo &quot;Advisories&quot;, with a combined maximum of at most 12 characters&quot;.

The obvious problem is that 12*6 &lt;&lt; 1024!




Received on Tuesday, 25 May 2010 08:16:20 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 25 May 2010 08:16:21 GMT