標題: 合作式網路文件分級系統之研究
Collaborative Platform for Internet Content Selection
作者: 林耀聰
Yao-Tsung Lin
曾憲雄
Shian-Shyong Tseng
資訊科學與工程研究所
關鍵字: 合作式;文件分級;網際網路;機器學習;Collaborative;Internet Content Selection;Clustering;Machine Learning
公開日期: 1998
摘要: 隨著網際網路的普及化,生活上利用網際網路的現象也越來越多,而使用者的年齡層也越來越低,如何避免接觸到網路上無用或不良的資訊,成為一項相當重要的課題。根據研究指出,網際網路內容管理是目前在網際網路上最受到重視的問題之一。針對此一問題,國外提出了許多研究報告,而其中較具規模與影響力的是由W3C(World Wide Web Consortium)所推動的PICS協定(PICS,Platform for Internet Content Selection),內容包括完整的分級規格與方法。但目前網路文件分級在實施上仍有相當的問題,包括了蒐集與儲存數量龐大的網路分級資料之方法、散佈網路分級資料之型式。為了解決上述問題,本論文首先提出合作式網路分級系統之基本架構,以解決資料難以收集之問題;並在散佈網路分級資料的層面上,以原本Domain Name System的概念為基礎,提出一個分散式的架構來應付大量的查詢,藉以提昇分級服務的效能。此外根據所實施的問卷調查結果顯示,受訪者對於網路分級制度有很高的支持度,後續我們將根據本論文的架構,在國內推動進一步的研究計劃來驗證與推動分級制度。
Due to the rapid growth of the amount of web pages, web becomes an important media for information communication and much inappropriate information is transferred through Internet. PICS protocol provided a systematic and complete architecture for Internet content selection. However, based upon PICS protocol, that may still have some problems need to be solved. In this work, a collaborative platform for Internet content selection is proposed. After the rating information are extracted by the collecting and extracting phases of the platform, the characteristics of all rating group are learned and can be used to help users to determine which group is suitable for joining. In the distribution of rating information, a distributed environment similar to DNS structure is proposed to solve three critical problems, including problems on storage, performance, and integrity of rating information. We also design a questionnaire to understand the opinions concerning the Internet content selection of users. According to the statistics of 600 answers of our designed questionnaire, over 72% answers would support the collaborative platform for Internet content selection. Therefore, we propose a plan to construct a national project based upon the architecture in this work.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT870394010
http://hdl.handle.net/11536/64148
Appears in Collections:Thesis