[MM] Review of Image Annotation Document

GENERAL REMARKS:

There is a lot of good content in this document, I like where it is
going.  There was a good discussion of the overall field of images, and
the relevant issues. Much thought has gone into use cases, the use of
vocabularies, etc.  

However, there is much to do.  The main issues of concern, or
suggestions are:
1.	the objectives of the document need to be more clear, and more
exciting
2.	more emphasis needs to be made on motivating the benefits of
semantic technologies, whey should anyone bother to read this document?
3.	the descriptions of the use cases need to be presented in a
consistent manner
4.	the solutions of the use cases need to go into the main
document, at least in detailed summary form.
5.	overall, the document seems somewhat bitty, disconnected, a nice
coherent story does not emerge with the pieces connected in a natural
way. It looks like the pieces were independently written and pasted
together with just a skeleton plan about how they should be integrated
into a smooth-flowing document.

NB: I did not read the details of the use case solutions, because they
weren't there, and I was off-line when I initially read the document.

SPECIFIC COMMENTS:

Opening sentences in abstract: Suggest indicating that the goal is not
only to use Semantic Web technologies, but motivate WHY one might care
to use them. I.e. get reader excited by starting out with claims that
there are some tangible benefits for using Semantic Web technologies
--

Wordsmithing quibble: don't say "use cases to exemplify the use of" say
ways to exploit, or something.
--

Wordsmithing quibbles: Don't need the word 'however' in: "however, is
addressed in a separate document", nor do you need to repeat
'interoperability', make the link on the word 'document'.
--

You use 'non-professional' and 'personal' apparently as synonyms. Use
just one term, for consistency.
--

You say: "At the time of writing, most work done in this area is not
based on Semantic Web technology often because it predates the Semantic
Web."
This is not quite true, or is misleading. Much semantic technology that
is now called  Semantic Web technology has been around for a long time,
and the work in the area could easily have been based on semantic
approaches.
--

You say: "Trade offs along several dimensions make the task difficult:"
I'm not sure which task you mean the personal or the professional (or
both).




OBJECTIVES:
These seem a bit weak, incomplete, not very exciting. Objectives should
include:
*	convince reader of benefits of using semantic technologies 
*	guidelines for applying semantic technologies 

Image Annotation Basics: 

I like the use case on generic vs. specific.
--


You say that human annotators is a problem because different people see
different things. This has nothing to do with the annotators being
human, it is to do with there being DIFFERENT agents doing the
annotating. A machine-based approach is not going to be more uniform if
many different annotation programs developed by different persons. One
human doing all the annotating will be more uniform.

It is true, of course that for large scale annotation it is impractical
for one human to do it all.
--

Can you give a quick example of hi vs. lo level annotation in the
manual/vs. automatic use case?
--

Replace: 
Reusing metadata developed for one set of tools in another        is
often hindered by a lack of interoperability. <with>

Reusing metadata developed for one set of tools in another tool is often
hindered by a lack of interoperability.
--

You say: "Solving this problem is much harder and can be done
automatically only when the semantics of the vocabulary used is
explicitly defined for both tools."

This is true, but potentially misleading; even if the semantics of the
vocabulary is defined explicitly for both tools, it will generally be
intractable to automatically determine what the semantic relationship is
between two terms.
-- 

SEMANTIC WEB BASICS:
I'm not sure exactly what this section is aiming to achieve; is it a
quick summary of what the Semantic Web is all about? The first point is
about possible benefits, the others are different again. Point 3 is
pretty abstract, not sure how useful it will be. Point 4 is actually a
disadvantage of semantic approaches, because although there is a growing
number of tools, the number is still tiny, compared to traditional
approaches. 

USE CASES: 

Don't start this paragraph negatively. Say instead: We list some
representative cases, and there are many more that are out of the scope
of this document.
--

You say that the use cases are arranged at top level by the topic of the
media to be annotated, but this is not really true. Scientific images
can be on virtually any topic, though not as wide as news more
generally, media can also be on  any topic. Pick a different way to
categorize them, or don't try to categories them in a consistent way at
all, if nothing seems to work.

But DO tease out categories that are different in important ways. I
can't see how the scientific images use case differs from some of the
others. If is it just different subject matter, why is in an
interestingly different use case?
--

Use cases would be improved if they were presented in a more uniform
way. There should be a set of criteria or aspects that are addressed for
each. E.g. the overall situation/need, the kinds of images, the
topic(s), the user community, what the specific needs/services are in
each user situation, scale, etc.  The media production services one is
the best, from this regard. All should  be written in a similar way, so
it is easy to compare them.

In the media production use case, you get sidetracked and talk about how
technology helps, that is independent of the user need. Separate out the
user and their need, from the technologies that may support them.

You say: "In order to facilitate the above process, the annotation of
image content should make use of Semantic Web technologies, also
following the multimedia standards in order to be interoperable with
other archives."
Avoid 'shoulds' say instead, that using Semantic Web technologies holds
promise for addressing the interoperability issue. And anyway, the
technology is probably better discussed separately from the use case?
Not sure, could work either way.
--

Number disagreement here: "people that archives the".
--

Replace: "from a specific domain" with "from specific domains"
--
Personal photo collections, I suggest the following rewrite:

Replace: 
"Advances in digital technologies (cameras, computers, storage,
communication etc) have caused a huge increase of digital multimedia
information captured stored and distributed by personal users over the
web. Digital formats now provide the most cheap, safe and easy way to
broadly capture, store and deliver multimedia content. Many personal
users have thousands of photos (from vacations, parties, traveling,
conferences, everyday life etc), usually stored in several resolutions
on the hard disk of their computers in a simple directory structure
without any metadata. Ideally, the user wants to easily access this
content, view it, create presentations, use it in his homepage, deliver
it over the Internet to other people, make part of it accessible for
other people or even sell part of it to image banks etc. Too often,
however, the only way for this content to be accessed is by browsing the
directories, the name of which usually provides the date and describes
with one or two words the original event captured by the specific
photos. Obviously, this access becomes more and more difficult since the
number of photos increases everyday and the content will quickly become
practically inaccessible. See also example solution
<file:///C:\Documents%20and%20Settings\uscholdm\My%20Documents\UscholdM\
aa-PROJECTS\2005%20Projects\SWBPD\MM\ToReview\image_annotation_galway.ht
ml> ."
<with>

Many personal users have thousands of digital photos from vacations,
parties, traveling, conferences, everyday life etc. Typically, the
photos are stored on personal computer hard drives in a simple directory
structure without any metadata. The user wants to easily access this
content, view it, create presentations, use it in his homepage, deliver
it over the Internet to other people, make part of it accessible for
other people or even sell part of it to image banks etc. Too often,
however, the only way for this content to be accessed is by browsing the
directories, the name of which usually provides the date and describes
with one or two words the original event captured by the specific
photos. Obviously, this access becomes more and more difficult since the
number of photos increases everyday and the content will quickly become
practically inaccessible. 

More sophisticated users leverage simple photo organizing tools allowing
them to provide keyword metadata, possibly along with a simple taxonomy
of categories. This is a step in the right direction, to a
semantically-enabled solution. See also example solution
<file:///C:\Documents%20and%20Settings\uscholdm\My%20Documents\UscholdM\
aa-PROJECTS\2005%20Projects\SWBPD\MM\ToReview\image_annotation_galway.ht
ml> .
--

Does it really matter for this note, that they are in multiple
resolutions?

PLEASE PUT THE USE CASES IN THE MAIN DOCUMENT, at least a decent summary
description. You can always have more details elsewhere. Also, I want to
see a general description or some pictures of the personal collection
ontology, raw RDFS is pretty useless for humans.

Cultural Heritage Use case:

This is a good, traditional style of use case description.

Also, there is a strong overlap with the personal collection use case.
Sunsets, regions, etc.
--

Media production services is not a topic. 

Also, please describe the need for annotations for media production
services. What is the actual use case?
--

This was surprising:
"A media production house requires several web services in order to
organize and implement its projects."
Do they really REQUIRE Web services? Or do they just happen to use Web
services in their architecture?
--




Section 3: Vocabularies

Say something about why you need vocabularies in the first place. Why
not just mark things on the fly (as one does for a personal photo
collection).
--

Give a quick example or two of what VR Core has, don't force me to click
the link (I'm not on-line anyway!).
--

Section: Tools

You might want to call this section "Available Tools".
--

You say: "Besides the hundreds of tools used for image archiving and
description there are also many tools that are used for semantic image
annotation. The aim of this section is to give a brief description of
the above tools"
The word 'above' is ambiguous, do you mean semantic or both semantic and
non-semantic annotation?
--

Define 'annotation level" it could mean many things, more/less specific,
more less abstract, the whole media, or part of the media, etc...
--

[terms quibble] The term 'operation level' means a lot more that whether
it is Web-based or not. If all you mean is the latter, then use call
this :Web based or not. The former term suggests a lot of other things
that you may mean.
--

You say: "Here, all the annotation tools found in the internet have been
categorized according to the characteristics described above." Where is
'here'?
--

You say: TO DISCUSS DURING F2F: Should the use case solutions be
included into this document or should they be kept separate so we can
update them after publication of this note.

I say: it would be good to have some well worked examples in this
document.
--

You say: "Furthermore the information of the image can be separated into
two categories, the general image characteristics and the image
content."

Above you called this granularity, use terminology consistently.
--

Section 5.1: Personal Photo Collections
I really wanted to see what you have to say here, but there is little
content. I can's see the link to the complete use case description since
I'm off-line.
--

CONCLUSION: 
It is absent, you need to say something here, review the goals of the
documenting, what interesting observations / conclusions can you report
on or summarize?

Too abrupt an end.

Received on Thursday, 3 November 2005 14:34:25 UTC