Comments to be shared with the authors


1. What area(s) of ... does the paper address?

Data management and retrieval, data modeling, geographic data
retrieval implementation. GIS interfaces (?)

2. Summarize in 2-3 sentences the main contribution of this paper:
The author proposed a geographic information retrieval model (YYY)
that uses two ways
for indexing geographical documents that are self-complementary: geographical model and the 
traditional (textual) vector space model. 
Retrieval and visualization are presented in terms of cognitive
concepts.
Most of the paper is taken by background material, and thus
there is very little space left to the actual model and its
implementation.

3. What are the major strengths of this paper?
- The paper is well written, and justifies the implementation
decisions taken
- Actual implementation of a prototype of the interface

4. What are weaknesses or deficiencies?
Some of the terminology used is not standard in information
retrieval, and thus the text loses value because information
retrieval experts will not understand several concepts.

The paper indicates it will present XXX, but the actual
presentation of the model occupies only 1/10 of the whole paper.
Similarity based-retrieval and measurements are extremely
complex, and must be accompanied by some description of their
effectiveness if they are to be validated. 


5. What could the authors do to improve the paper?
- Change terminology to more standard terms.
- Change order of boxes in figure 1.
- Section 3 presents the vector space model and the geographical model. However in 
  section 2 the author already comments on deficiencies of these
models. I suggest the 
  exchanging of the sections
- Still in section 2, a document example should be presented for showing the way of indexing
  in each model
- Correct some spelling mistakes and verify the reference Y]
- The main strength of the paper is the implementation of blah
and thus you could extend the section on the description of
this system, and decrease the amount of space dedicated to 
comparing vector space to geographic model.
- Provide measurements about effectiveness of your similarity
model (eg precision and recall) otherwise there is no point
in proposing it. (How is your model better than others in 
terms of retrieval power or cognitive interaction with the
interface???)

6. On a scale from 1 to 7, with 
1 = very low, 
7 = very high
how do you rate this paper for:

6.a     Relevance [1 - 7]: 6
6.b     Originality/Novelty [1 - 7]: 4
6.c     Scientific and/or Technical Quality   [1 - 7]: 4
6.d     Presentation (i.e., readability, structure, and English) [1 - 7]: 5
6.e     Overall recommendation    [1 - 7]: 5

7. Reviewer's familiarity with the subject:  [low, medium, high]: 
high
8. Additional comments to the author(s):
- In introduction (page 2) the author did not explain what is blabla.
Since this is one of the paper's motivations, and is not a well
known concept, it should be detailed.
- In section 4.2, you did not suggest any example of the function f(*)
- In section 7.1, you did not explain world metaphor
  - There are  some spelling mistakes, like 'focued' on page 2 last paragraph

In some places of the paper appear references to section 2 (introduction, last paragraph and
section 5, after [Principle I] "...ambiguity[4]. In section 2, we mentioned ...") but the 
subject seems to be of section 3. It seems that section 2 was included after the paper was finished.

Conceptualization of geo-space as cube - as opposed to vector space.
This is OK. However, any IR model nowadays treats all kinds
of attribute spaces as multidimensional vectors (hypercubes).
A geo-space is just another kind of attribute partition.
This is how data mining is supported.
Thus, any kind of attribute space (geographic or not) can
be described by hypercubes.

Terminology problems. You present the paper in terms of Information
Retrieval (IR) theory, but the terms you use are not compatible
to those employed in IR research, and furthermore use terms . This confounds the reader.

Figure 1 is also counterintuitive.
There is no such thing as conceptual and spatial entities.
I believe you mean that real world entities have spatial and
descriptive (non-spatial) properties.
Calling a descriptive/thematic property by the name "conceptual"
is confusing.
The term "conceptual view" has a specific connotation in IR and does not correspond to
data retrieval, but to a mental conceptualization of a problem.
A conceptual view is specified by means of a conceptual data model
(e g, ER, OMT).
You should choose another term.
By the same token, a spatial view is not "visible" in itself. 
Both views should be placed between entities and encoding.
Maybe you mean something completely different in figure 1.
If so, you must describe the information flow within the figure.
As I read it, the figure describes real world entities which have
spatial and descriptive/thematic properties (or, as your figure suggests,
real world entities may be either spatial or descriptive).
These either are encoded into some computer system and then can
be retreived by cartographical or textual documents.

Finally, the figure talks about visual and textual documents.
Anything you can see is visual.
I believe you mean some sort of graphical visualization 
as opposed to textual visualization.
If so, you cannot name visual documents by this name;
rather, call them cartographical or something similar.

Comments you want to share only with PC chairs:
kf;kajsdlkfa