Title of Paper: Caching Schemes in Mobile Databases Authors: Rooma Rathore, Rohini Prinja Reviewer Team (Name, Student Ids): G10, Kuo-Wei Hsu, 3483317 Date Review Completed: 11/13/2006 SUMMARY: XML, extensible markup language, becomes more popular as the Internet develops rapidly. Besides, researchers, engineers and developers start using XML as a tool for exchanging data. As the business becomes more complicated, as the amount of data generated from business activities grows rapidly, and as the variety of database used by enterprises increases, the need for XML database becomes stronger. Basically, there are two major ways to implement an XML-enabled database: one is to provide a set of add-on XML functions, which are built over the existing database (e.g., relational or object-relational database), and the other one is to design and a native XML database, in which the XML data are stored and treated as other native data structures. The native XML database is more popular since generally it provides better performance. This project report focuses on the native XML database. Once we have an XML database, the following question will be: “how can we query data?” or “how can we use find part of XML data in which we are interested?” Researchers have done many outstanding works on XML query and published many papers talking over this topic. In the first section of this project report, authors give us a short yet good summary for research works. Basically, XML data is of tree structure. In order to do the query optimization, the query processing subsystem has to take into account the tree structure, and so does the indexing subsystem. Furthermore, XML is a kind of semi-structured format, which means, it is very flexible and tags are allowed to be absent so that these subsystems have difficulties in grouping related data. There are many other issues about XML query processing. In this project, the major issue or the topic is to find the relationship between the query performance and the XML structure or DTD (Document Type Definition). The concept behind this project is simple yet powerful, that is, we can make the XML query more efficient in the runtime if we can provide the better schema (XML structure) in the design time. This project report provides five hypotheses and use experiments to verify them: 1) searching an attribute of an element will be faster than searching an element in a nested element., 2) id/idref attributes will make query faster, 3) searching an attribute with key/keyref will be faster than searching an attribute without key/keyref, 4) highly nested structure will increase the query time, and 5) using verbose element name will slower the query. The first one is an interesting topic since some developers prefer using attributes but others prefer using elements to describe the data or data element. Therefore, it is useful to provide an experiment result to say which one is better. The second as well s third ones are from the nature of XML, while the fourth and fifth ones are straightforward but still worth experiments that can show how bad the performance could be if we increase the degree of nesting or the length of element names. FOCUS: Does the paper clearly identify the problem it is addressing ? YES. The problem is described clearly, or we can restate the problem as: “given a query, how can the XML structure or DTD influence the query performance?” Does the paper clearly explain related work and their limitations? YES and NO. Authors summarize 14 papers well, but they do not provide a detailed comparison between these papers. Does the paper identify its key contributions? YES. Authors clearly identify their contribution as that, there is no paper discussing the relationship between query performance and XML structure (or DTD). Does the paper present any evidence to support the contribution claim? YES and NO. Authors design several examples to address several design issues. They make experiments as clear as possible, which is pretty good. However, they do not provide the experiment results. (I believe they will make up the results required.) TECHNICAL EVALUATION: Is the literature survey complete? YES. They refer 14 (pretty new) papers and give a good summary in the first section. Is the work novel relative to the literature? Explain. YES. The references do not carefully talk over the relationship between query performance and XML structure (or DTD). So, this project is comparatively novel to them. As a reviewer do you agree with the contribution claims? Explain. YES and NO. I agree with their experiment design (and hypotheses), but I can not agree with the contribution claims since there are no experiment results. (Hope authors can make up the results as soon as possible). READABILITY AND ORGANIZATION: Is the paper easy to read and understand to students in this course (Csci 8715)? YES. This report is well-organized and well-written. It is pretty easy to find the key points of each section. Is the paper self-contained ? N/A Is the paper length reasonable ? YES. But, it will be more complete if authors make up the experiment results. Does it include sufficient number of figures and tables? NO. There are neither tables nor figures. However, there are several good examples used to describe the hypotheses. STRENGTHS: What are the strengths of this paper? Authors provide good experiments and examples used to explain experiments one by one. AREAS FOR IMPROVEMENT: How can this paper be improved? If you were to rewrite this paper, what revisions would you consider? The topic is quite interesting and practical. This will be a good technical report if authors can 1) provide the experiment results, 2) attach corresponding references/papers to each hypothesis, and 3) discuss more design issues/hypotheses or describe if there is any other one.