Most of the existing methods for XML keyword search are based on the notion of Lowest Common Ancestor (LCA). However, as we explore the most important fundamental flaw inside those result models is that the search results are eternally determined and nonadjustable. In order to serve better results, we propose a novel and flexible result model which can avoid all these defects. Within our model, a scoring function is presented to judge the quality of each result. The considered metrics of evaluating results are weighted, and can be updated as needed. Based on the result model, three heuristic algorithms are proposed. Moreover, a mechanism is employed to select the most suitable one out of these algorithms to generate better results. Extensive experiments show that our approach outperforms any LCA-based ones with higher recall and precision.
CITATION STYLE
Yang, W., Zhu, H., Li, N., Zhu, G., Huang, J., Cao, L., & Srivastava, J. (2011). Advances in Knowledge Discovery Yang, W., Zhu, H., Li, N., Zhu, G., Huang, J., Cao, L., & Srivastava, J. (2011). Advances in Knowledge Discovery and DatYang, W., Zhu, H., Li, N., Zhu, G., Huang, J., Cao, L., & Srivastava, J. (2011). Advances in Knowledge, 6634, 423–434. Retrieved from http://www.springerlink.com/content/pjlv874811744p6m/
Mendeley helps you to discover research relevant for your work.