Effective Databases For Text Document Management PDF Download

Abstract: "In this thesis, we study the problem of representing and manipulating a document to facilitate browsing, editing, string/content searches and document assembly. Two major data models in which documents are represented and stored are: 1. a relational data model, where all text contents in a document are represented in relations, each with several attributes, or 2. a text data model, where documents are represented as contiguous characters, typically interspersed with tags to capture their various logical, semantic, and presentational features and relationships [sic] Each approach has its own strengths and limitations. In our work, we study how a hybrid system based on a combined text/relational model can support document management. We describe database design trade-offs involving the appropriate placement of information in the text and relational database components. With an appropriate design, the advantages of both models can be exploited, while the shortcomings of using them individually are diminshed. We propose a set of primitive operations and a methodology for using them to evaluate the various alternatives for data placement. The methodology consists of simulating pre-defined, representative, document management tasks using the primitive operations and studying the numbers, types, and the time performance of the operations involved. Using some representative document management tasks as examples, we demonstrate the use of the methodology and the primitive operations to study and compare the processing of the tasks in the various data models metntioned above."