The following information was submitted:
Transactions: WSEAS TRANSACTIONS ON INFORMATION SCIENCE AND APPLICATIONS
Transactions ID Number: 32-903
Full Name: Hsukuang Chang
Position: Ph.D. Candidate
Age: ON
Sex: Male
Address: No.1, Sec. 1, Syuecheng Rd. ,Dashu Township, Kaohsiung,
Country: TAIWAN
Tel: 07-6577711-6533
Tel prefix:
Fax:
E-mail address: hkchang@isu.edu.tw
Other E-mails:
Title of the Paper: Incremental Singular Value Decomposition for Managing Dynamic Collections of Semantic Semi-structured Documents
Authors as they appear in the Paper: Hsu-Kuang Chang, I-Chang Jou, King-Chu Hung
Email addresses of all the authors:
Number of paper pages: 12
Abstract: The number of XML documents has been increasing as electronic document systems adopt XML as the standard format for document exchange. Like the weather, XML document databases are always changing. Information is constantly added or removed, meaning that catalogs and indexes become obsolete or incomplete, sometimes in a matter of seconds. With the tremendous increase in online information, the dynamic update of XML documents has become critical for efficient document organization and navigation and for the retrieval of large quantities of XML documents. XML documents can be modeled as a matrix, and a user¡¦s query and update of an XML database can be represented as a vector. Relevant XML documents in the XML database are then identified via vector operations. In the latent semantic indexing (LSI) model, LSI is used to recompute the singular value decomposition (SVD) of the new path-by-document matrix. This procedure is very costly in time and space for large XML do!
cument databases. This paper proposes a new method of folding-in combined with SVD updating. The folding¡Vupdating method offers a significant improvement in computation time compared with either recomputing or updating the thin SVD in practical operations. The folding¡Vupdating hybrid method applied to the thin SVD offers a significant reduction in CPU execution time and an increase in querying precision compared to either economic SVD or the original SVD. This is also true for the conventional methods of folding, updating, and recomputing the thin SVD.
Keywords: thin SVD, LSI, folding-in, SVD updating, SVD folding¡Vupdating, SVD recomputing
EXTENSION of the file: .pdf
Special (Invited) Session:
Organizer of the Session:
How Did you learn about congress:
IP ADDRESS: 140.127.194.25
