347 0

시맨틱 멀티미디어 검색을 위한 위키피디아 기반 태깅 및 개인화 정제 기법

Title
시맨틱 멀티미디어 검색을 위한 위키피디아 기반 태깅 및 개인화 정제 기법
Other Titles
Wikipedia-based Tagging and Personalized Refinement Techniques for Semantic Multimedia Retrieval
Author
홍현기
Advisor(s)
이동호
Issue Date
2017-08
Publisher
한양대학교
Degree
Doctor
Abstract
The volumes of multimedia content and users have increased on social multimedia sites due to the prevalence of smart mobile devices and digital cameras. User’s information needs for multimedia are diversifying and becoming increasingly complicated. In order to improve user’s satisfaction, in this thesis, we propose novel Wikipedia-based tagging and personalized refinement techniques for semantic multimedia retrieval. The proposed approach consists of 1) semantic multimedia tagging phase and 2) personalized refinement phase. In semantic multimedia tagging phase, we present novel tag recommendation method exploiting Wikipedia to conduct accurate multimedia tagging. On the basis of precise tagging, we conduct personalized refinement using domain ontology for multimedia search results in semantic multimedia retrieval phase. For semantic multimedia tagging, we propose a semantic tag recommendation technique exploiting associated words that are semantically similar or related to each other using the interwiki links of Wikipedia. First, we generate a word relationship graph after extracting meaningful words from each article in Wikipedia. The candidate words are then rearranged according to importance by applying a link-based ranking algorithm and then the top-k words are defined as the associated words for the article. When a user uploads multimedia content, we collect visually similar multimedia content from a social multimedia database. After propagating the proper tags from the collected multimedia content, we recommend associated words related to the candidate tags. Our experimental results show that the proposed method can improve the accuracy by up to 14% compared with other works and that exploiting associated words makes it possible to perform semantic tag recommendation. To support personalized multimedia content retrieval, we propose an efficient semantic analysis and personalized refinement technique for searching multimedia content. This system makes it possible to provide more personal-tailored content to users with minimal input operations in smart devices. For the search results from diverse multimedia content sharing sites, the proposed method semantically analyzes user information and multimedia content, and then measures the semantic relatedness among user’s query, multimedia content, and user preference exploiting domain ontology. In order to provide accurate personalized results of the multimedia content search, it carries out filtering, grouping, and ranking for the search results based on the semantic relatedness. The various experimental results show the effectiveness of the proposed approach.
URI
http://hdl.handle.net/20.500.11754/33704http://hanyang.dcollection.net/common/orgView/200000431441
Appears in Collections:
GRADUATE SCHOOL[S](대학원) > COMPUTER SCIENCE & ENGINEERING(컴퓨터공학과) > Theses (Master)
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE