The Study of Generating Recommended Documents Based on Multiple Concepts and Document Structure
| Autor: | Yu Jen Lin, 林于荏 |
|---|---|
| Rok vydání: | 2002 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 90 In reality, a large portion of the available information appearing in textual and unstructured forms is valuable to people. Techniques specifically for analyzing textual data become necessary to extract information from such kind of textual datasets. The searching of similar documents also plays an important role in every aspect of text mining research. Similarity searching is an essential task for document management. Most work of the past researches focused on comparing different algorithms of classification by considering the contents of the documents or improving the performance of the algorithms. We propose a multiple-concept mechanism composed of different algorithms to solve the similarity searching problem. Furthermore, another factor-“distribution of structure” ignored by previous researches is also considered in this study. According to the empirical evaluation result, the proposed technique was more effective than the traditional approaches. Namely the effectiveness of “distribution of structure” had been proved. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|