.png){kind=icon}
.png){kind=icon}
返回
Annotation alignment method on P2P-TV system
.png){kind=icon}
PRE
AI
DOI：10.1007/s12083-014-0264-4.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
P2P-TV is a TV system that receives content through a peer-to-peer network. Content is stored in the distributed manner then to be serviced to users, and thus expenses incurred by the content delivery are regarded as a near zero cost. This is why various studies are being conducted on the P2P-TV of late. Furthermore, the spread of smart televisions with combined functions of the Internet and PC enhanced the feasibility of P2P-TV greatly. Current discussions surrounding the P2P-TV, however, are mostly focused on studies that aim to improve the quality of content delivery. On just rare occasions are there studies that investigate the possibility of a significant profit source, P2P-TV based T-Commerce. As for the T-commerce to work, product information shall be extracted from content to be delivered to online shopping mall websites as a basic service. In this respect, this paper suggests methods and system to align content with annotation data so as to be used for T-Commerce through the P2P-TV system while evaluating the system performance. The alignment is divided into total matching process and partial matching process. When content is sent through the P2P network and stored afterwards, there can be unintentional content corruption depending on the network quality. There are five types of such content variations suggested as follows: (1) frame loss; (2) key frame corruption; (3) shot loss; (4) changes in the number of frames; and (5) changes in the resolution. These variations cause mismatching between content and annotation data. Against this backdrop, the paper proposes reasons for such mismatching as well as methods for mismatching-related decisions while conducting an experimentation and performance evaluation. The experimentation on these five different types of content corruption has generated relatively strong annotation results of 82.63 %. This shows that the proposed methods for alignment between content and annotation data are effective.
Keyword:
P2P-TV
T-Commerce
Annotation data
Alignment
Quality
期刊
IF:
2.6
论文数:
2.2K
被引数:
2.9K
