Reviews for paper The Relational XQuery Puzzle: A Look-Back at the Pieces Found So Far, submitted to Computer Science—Research and Development.
Overall rating: accept
The paper provides a survey on XQuery processing using relational database technology. This has been an area of vivid research over the past decade and clearly deserves a Look-Back (and also a Look-Forward). From the perspective of a co-developer of a successful relational XQuery system, the author highlights some of the most important features in relational XQuery processing, namely
This is a nice survey which is of great interest both to readers who come from the “XML world” and for readers from the “relational DB world”. The paper is written on a sufficiently high level of abstraction that also the non-expert reader may profit from this work. The author admits that the paper is biased towards the Pathfinder XQuery compiler – a system in whose development he has been heavily involved. Nevertheless, the paper appears to be fair in the way it also treats other systems – in particular commercial RDBMs.
I only have two criticisms:
In summary, I definitely recommend to accept the paper with a minor revision along the lines mentioned above.
p.2, col.1, first par. of Section 2 and footnote 1: The footnote is confusing: First you mention that the work “fell short in providing a convincing implementation”. Then you write in the footnote that “with an appropriate choice of oid value”, problem (a) could be easily solved. I would not expect such an easy fix for a “not convincing implementation”. Please explain or reformulate.
p.2, col.2, equation (2) and Figure 1: Please add also the value of post(v) to the table. Ideally you would also extend the labels of the tree in (a). Why not write pairs (pre,post) on the left-hand side of each node. The equation (2) would then be easier to understand.
p.11/12: References: Please make sure that the references are presented uniformly and with full information, e.g.: the page number is missing in some references (e.g., [3], [7], [24], [25], …). Please check carefully.
Some typos:
p.1,col.2: “need to mapped” -> “need to be mapped”
p.2, col.1: “an convincing” -> “a convincing”
p.2,col.2: “using each nodes rank” -> “using each node’s rank”
p.4, col.1: “data data by ignoring…” -> “data by ignoring…”
p.9, col.2: “can efficiently handled” -> “can be efficiently handled”
The author surveys state-of-the art in the area of Relational XQuery processing, but focuses in most aspects on the results of the Pathfinder project. The paper is well written and organized and offers a good motivation for and overview of the area.
My only major criticism is that the paper is not self-contained. Some concepts are described too superficially to be of value. For example, from the text it is not apparent how a staircase join works (not even an example is provided). The same holds for PathStack and TwigStack. Also, there are not enough examples in the paper to illustrate the main concepts found. In a lot of ways, the authors assume that the reader has significant background knowledge of the subject. This raises the question about the audience of the paper. For a reader who is not familiar with XML query processing approaches, the paper doesn’t seem to be comprehensible without inquiring additional sources.
The paper focuses in most aspects on the Pathfinder system (this is more or less admitted by the author by stating that he provides a biased viewpoint), but at least from the title of the paper gives the impression that it intending to survey the state of the art with a much broader perspective. This should be rectified, either by making the paper more comprehensive, or by changing the title. For example, the title could already reflect the focus (“… the pieces WE found so far.”?).
The paper could be further improved by providing a more thorough comparison with or survey of other approaches in the field.
Minor comment:
The paper provides a survey on XQuery processing using relational database technology. This has been an area of vivid research over the past decade and clearly deserves a Look-Back (and also a Look-Forward). From the perspective of a co-developer of a successful relational XQuery system, the author highlights some of the most important features in relational XQuery processing, namely
This is a nice survey which is of great interest both to readers who come from the “XML world” and for readers from the “relational DB world”. The paper is written on a sufficiently high level of abstraction that also the non-expert reader may profit from this work. The author admits that the paper is biased towards the Pathfinder XQuery compiler – a system in whose development he has been heavily involved. Nevertheless, the paper appears to be fair in the way it also treats other systems – in particular commercial RDBMs.
I only have two criticisms:
In summary, I definitely recommend to accept the paper with a minor revision along the lines mentioned above.
p.2, col.1, first par. of Section 2 and footnote 1: The footnote is confusing: First you mention that the work “fell short in providing a convincing implementation”. Then you write in the footnote that “with an appropriate choice of oid value”, problem (a) could be easily solved. I would not expect such an easy fix for a “not convincing implementation”. Please explain or reformulate.
p.2, col.2, equation (2) and Figure 1: Please add also the value of post(v) to the table. Ideally you would also extend the labels of the tree in (a). Why not write pairs (pre,post) on the left-hand side of each node. The equation (2) would then be easier to understand.
p.11/12: References: Please make sure that the references are presented uniformly and with full information, e.g.: the page number is missing in some references (e.g., [3], [7], [24], [25], …). Please check carefully.
Some typos:
p.1,col.2: “need to mapped” -> “need to be mapped”
p.2, col.1: “an convincing” -> “a convincing”
p.2,col.2: “using each nodes rank” -> “using each node’s rank”
p.4, col.1: “data data by ignoring…” -> “data by ignoring…”
p.9, col.2: “can efficiently handled” -> “can be efficiently handled”
The author surveys state-of-the art in the area of Relational XQuery processing, but focuses in most aspects on the results of the Pathfinder project. The paper is well written and organized and offers a good motivation for and overview of the area.
My only major criticism is that the paper is not self-contained. Some concepts are described too superficially to be of value. For example, from the text it is not apparent how a staircase join works (not even an example is provided). The same holds for PathStack and TwigStack. Also, there are not enough examples in the paper to illustrate the main concepts found. In a lot of ways, the authors assume that the reader has significant background knowledge of the subject. This raises the question about the audience of the paper. For a reader who is not familiar with XML query processing approaches, the paper doesn’t seem to be comprehensible without inquiring additional sources.
The paper focuses in most aspects on the Pathfinder system (this is more or less admitted by the author by stating that he provides a biased viewpoint), but at least from the title of the paper gives the impression that it intending to survey the state of the art with a much broader perspective. This should be rectified, either by making the paper more comprehensive, or by changing the title. For example, the title could already reflect the focus (“… the pieces WE found so far.”?).
The paper could be further improved by providing a more thorough comparison with or survey of other approaches in the field.
Minor comment: