A sample review for Pottinger and Levy, "A Scalable Algorithm for Answering Queries Using Views", VLDB 2000 available at http://ww.cs.ubc.ca/~rap/publications/pottinger-levy-vldb00.pdf if you want to see the original paper for comparison ********************************************************************** A sample review that would receive a 3: ********************************************************************** This paper describes MiniCon, a scalable algorithm for answering queries using views, specifically for the case of rewriting queries for data integration systems. The algorithm is a descendent of the Bucket Algorithm, but it prunes potential views in an earlier stage than the bucket algorithm did, so it's much faster in practice than the bucket algorithm would be, even if both have the same asymptotic running time. In particular, it realizes that in order for a view to be used to rewrite a particular subgoal of a given query, the view may also have to be able to rewrite other subgoals of that query. Aside from presenting the algorithm, the paper also presents the first large scale evaluation of algorithms for answering queries using views. That's quite nice because it shows that in a number of cases, the MiniCon algorithm really is practical. Before this, particularly after reading about the bucket algorithm, I had my doubts as to how useful LAV would be for a mechanism for describing the mediated schema's relation to the source schemas. The strengths of this paper are the motivations, the effort to make sure that there's an intuitive explanation for why it succeeds, and the experimental results. It's too bad that the experiments, while for a number of different query shapes, don't give an indication of queries that are used in practice. In addition, the queries that are studied are only conjunctive queries (with some results for queries with arithmetic comparisons), which is a very limited set of queries compared to those used in the real world. I wonder if the techniques used here could be generalized to more complicated queries, or other types of queries, such as queries over XML or queries in different applications. A question for discussion: this seems like an awfully restrictive class of queries to consider for the real world. What's the point of coming up with techniques that are tractable and yet don't allow us to actually solve the kinds of queries that real users need? ********************************************************************** A sample review that would receive a 2: ********************************************************************** This paper describes MiniCon, a scalable algorithm for answering queries using views, specifically for the case of rewriting queries for data integration systems. The algorithm prunes potential views in an earlier stage than previous algorithms. Aside from presenting the algorithm, the paper also presents the first large scale evaluation of algorithms for answering queries using views. The strengths of this paper are the motivations, the effort to make sure that there's an intuitive explanation for why it succeeds, and the experimental results. I wonder if the techniques used here could be generalized to more complicated queries, or other types of queries, such as queries over XML or queries in different applications. ********************************************************************** A sample review that would receive a 1: ********************************************************************** This paper describes MiniCon, a scalable paper for answering views. The algorithm is better than other algorithms because it is faster. It talks about how it's faster than the other algorithms, like the bucket algorithm. I wonder if this works well in the real world - it seems like it might not be that useful to real people. ********************************************************************** A sample review that would receive a 0: ********************************************************************** This paper describes the MiniCon Algorithm for Answer Queries Using Views. It shows that the MiniCon algorithm scales up well and significantly outperforms previous algorithms.