Alternative representations for database responses
Cass Shum · 1989
Two issues in retrieval from a database are addressed: effectiveness and efficiency. Effectiveness is concerned with the representation of answers; and efficiency is concerned with fast response to queries. Conventional responses to queries in database systems, usually consisting of lists of tuples, do not necessarily provide effective communication between a user and the system. The first part of this thesis suggests how, when responding to a user query, the system can improve effectiveness of information transfer by providing concise intensional answers, as a set of concepts and predicates, rather than just providing the set of facts satisfying the query. Abstracting responses before presenting them to the user is an effective means of information exchange. Aggregation is a commonly used abstraction process. The second part of this thesis investigates aggregate responses that are expressions of quantified concepts. The tradeoff between the conciseness and impreciseness of an aggregate response is studied. Conciseness is measured by the length of the expression and preciseness is measured by the entropy of the expression. The objective of conventional query optimization is usually to minimize the total time required for evaluating a query. The last part of this thesis argues that in some interactive environment, the initial response time of a query may also be considered important. For selection queries, a long initial response time can be shortened without compromising on the total time for completely evaluating a query. The same, however, is not always true for join queries. But the tradeoffs between the total response time and the initial response time of a query are quantifiable. Semantic query optimization techniques can also improve the initial response time of a query. Moreover, in addition to exact rules, inexact rules can be utilized to improve the initial response time.