SIGMOD 2008
2008 ACM SIGMOD International Conference on Management of Data
June 9-12 2008
Vancouver
Canada
Paper ID: 388
Title: Keyword Reminiscence Search in Relational Databases
Track Name: Research
Review Count: 3

Review: 1
Reviewer: Narain Gehani
Email: gehani@adm.njit.edu
Organization: New Jersey Institute of Technology
Review:
 QuestionResponse
1Overall Rating Weak Accept
2Reject due to technical incorrectness No
3Novelty Low
4Technical Depth Medium
5Is the relevant and reasonably well known state of the art fairly treated (note this may come from outside the database literature and even outside computer science)? to a limited extent
6Experimental results should meet the following expectations: deal with materially relevant cases (e.g., updates as well as queries, different scales) allowing for space limitations; statistically meaningful results; and use standard datasets or benchmarks (as opposed to a cherry-picked one) when possible. How did the experiments rate? adequate
7If experiments were presented, would you want them checked for repeatability (as might be required in a future sigmod)? yes
8Presentation Adequate
9Reviewer Confidence Low
10Name of External Reviewer (if applicable)
11Summary of the paper's main contributions and impact (up to one paragraph) Database search technique that returns a ranked list of the terms and phrases that are suggestive ("reminiscent")of the search keywords. Presumably this information can be used to narrow or expand a query when retrieving data. They then give algorithms to produce this list efficiently - without doing any joins.
12Three strong points of the paper (please number them S1,S2,S3)
13Three weak points of the paper (please number them W1,W2,W3) W1. Presentation could be better - there are typos, wording can be improved, etc.
see in detailed comments
14Detailed comments (please number each point) In the abstract, the authors say that they are proposing a new type of search problem over relational databases. Probably, "problem" should be "technique". In any case, the "new type of search ..." is not explained.

Writing could be improved

Typos etc.

- Why are the words "Search Engine" (line 2, Intro) capitalized
- knowledge on --> knowledge of (line 4, Intro)
- Example 1: "operation system" should be "operating system"
15Comments for the Program Committee did not read the math.
16Is this paper a candidate for the Best Paper Award? No
17Would author feedback be useful for this Review? (if "Yes", please answer Q. 18) No
18List specific clarifications you seek from the Authors (if you have answered "Yes" to Q. 17)

Review: 2
Reviewer: Tomasz Imielinski
Email: timielinski@ask.com
Organization: Ask.com and Rutgers University
Review:
 QuestionResponse
1Overall Rating
2Reject due to technical incorrectness
3Novelty
4Technical Depth
5Is the relevant and reasonably well known state of the art fairly treated (note this may come from outside the database literature and even outside computer science)?
6Experimental results should meet the following expectations: deal with materially relevant cases (e.g., updates as well as queries, different scales) allowing for space limitations; statistically meaningful results; and use standard datasets or benchmarks (as opposed to a cherry-picked one) when possible. How did the experiments rate?
7If experiments were presented, would you want them checked for repeatability (as might be required in a future sigmod)?
8Presentation
9Reviewer Confidence
10Name of External Reviewer (if applicable)
11Summary of the paper's main contributions and impact (up to one paragraph)
12Three strong points of the paper (please number them S1,S2,S3)
13Three weak points of the paper (please number them W1,W2,W3)
14Detailed comments (please number each point)
15Comments for the Program Committee
16Is this paper a candidate for the Best Paper Award?
17Would author feedback be useful for this Review? (if "Yes", please answer Q. 18)
18List specific clarifications you seek from the Authors (if you have answered "Yes" to Q. 17)

Review: 3
Reviewer: Susan Dumais
Email: sdumais@microsoft.com
Organization: Microsoft Research
Review:
 QuestionResponse
1Overall Rating Weak Reject
2Reject due to technical incorrectness No
3Novelty Medium
4Technical Depth Low
5Is the relevant and reasonably well known state of the art fairly treated (note this may come from outside the database literature and even outside computer science)? yes, allowing for space limitations
6Experimental results should meet the following expectations: deal with materially relevant cases (e.g., updates as well as queries, different scales) allowing for space limitations; statistically meaningful results; and use standard datasets or benchmarks (as opposed to a cherry-picked one) when possible. How did the experiments rate? not good
7If experiments were presented, would you want them checked for repeatability (as might be required in a future sigmod)? no or not applicable
8Presentation Adequate
9Reviewer Confidence Medium
10Name of External Reviewer (if applicable)
11Summary of the paper's main contributions and impact (up to one paragraph) The paper describes a new idea, called Keyword Reminiscence Search”. The basic idea is that in some applications it might be desirable to retrieve related terms instead of joined tuple trees, in response to a keyword query. Three simple methods for identify related keywords are described and evaluated. The applications are not particularly well motivated for me. And the evaluation consists of assessing the overlap of the proposed method with conventional keyword searches over databases. If this is the gold standard, why not use standard methods directly?
12Three strong points of the paper (please number them S1,S2,S3) S1: Describes a new problem, namely returning words associated with tuple trees that match keyword queries, which the authors call Keyword Reminiscence Search.
S2: Three simple techniques for ranking associated terms are proposed and evaluated.
13Three weak points of the paper (please number them W1,W2,W3) W1: I’m not sure how important a problem this is. Are searchers looking for associated keywords or joined tuple trees?
W2: The datasets considered have simple schema, involving at most three relations. How realistic is this?
W3: The “gold standard” for evaluation is the top-20 terms returned by a conventional keyword search over the database. If this is what you want to retrieve, why not do so directly?
14Detailed comments (please number each point) The paper described a new functionality, Keyword Reminiscence Search, which amounts to returning related keywords rather tuples in response to a keyword database search. My main concern is that there is little motivation for the problem (who wants this and for what kinds of task). This kind of background should shape the approaches used. I also have concerns, detailed above, about the evaluation which involves datasets with simple schemas and a gold standard (top-20 terms returned by a conventional keyword search over the database) that is not well motivated.
15Comments for the Program Committee
16Is this paper a candidate for the Best Paper Award? No
17Would author feedback be useful for this Review? (if "Yes", please answer Q. 18) Yes
18List specific clarifications you seek from the Authors (if you have answered "Yes" to Q. 17)