Academic journal article Interdisciplinary Journal of e-Skills and Lifelong Learning

SOAF: Semantic Indexing System Based on Collaborative Tagging

Academic journal article Interdisciplinary Journal of e-Skills and Lifelong Learning

SOAF: Semantic Indexing System Based on Collaborative Tagging

Article excerpt

Introduction

There is a need for the design and development of information management systems based on semantics to guarantee the reusability of the Learning Objects and learning assets in different learning contexts. These systems have to be able to categorize extensive collections of Learning Objects and learning assets in the increasingly expanding repositories and to infer their real meaning, taking into account user's qualitative descriptions based on real learning experiences. However, existing technologies for concept based description of web resources are still not very advanced. At the same time, the multiple and distinct media types used in the design of Learning Objects require specific techniques for each of these types.

The analysis of textual information is usually more affordable as the use of keywords for syntactic searches makes it possible to break the text into separate paragraphs and words. However, in the case of multimedia assets, the minimum unit is much more difficult to divide and there is still a great distance between high-level descriptions (or concept based descriptions) and low-level visual features that can be automatically extracted from data, like color histograms, texture indexing, shape, or contours. The main challenge in this field has been quoted as automatically generated concept based description for multimedia information, and this problem is known as "bridging the semantic gap". Currently, there is a lack of tools that can automatically manipulate the high-level concepts of an image or video.

There have been many attempts to solve the problem of the semantic description of multimedia assets (Moenne-Loccoz, Janvier, Marchand-Maillet, & Bruno, 2005; Stamou & Kollias, 2005). One of the first--using the MPEG-7 format to store multimedia metadata--was made by Hunter in 2001, who first developed a specific ontology using DAML-OIL. As part of the aceMedia Project (Bloehdorn et al., 2005), an ontology was created for visual description of multimedia content that has been incorporated in the MPEG-7 visual metadata. The MPEG-7 format provides detailed formatting information and fine-grained descriptions of the structural and low-level audiovisual features of multimedia content as follows: the Description Definition Language (the basic building blocks of MPEG-7), Audio (the descriptive elements for audio), Visual (those for video) and Multimedia Description Schemes (the descriptors for capturing the semantic aspects of contents, e.g. places, objects, events.)

Some of the proposed solutions for automatic semantic description include dividing the image into regions with similar visual features and assigning them semantic labels using statistical methods (Fan, Gao, Luo, & Xu, 2004; Pan, Yang, Duygulu, & Faloutsos,, 2004), EM algorithms (Duygulu, Barnard, Freitas, & Forsyth, 2002), or recent probabilistic models like the Cross Relevance Model (Jeon, Lavrenko, & Manmatha, 2003).

Recently, more and more websites allow their users to use collaborative tagging of different types of resources in order to annotate and categorize Web content. The result is known as folksonomy, social classification or social indexing of web content. In contrast with the traditional method of adding semantics, metadata is not created by experts but it is spontaneously generated by consumers (Cattuto, 2006; Marlow, Naaman, Davis, & Boyd, 2006). Other sites, like Amazon, have implemented social filtering algorithms based on the similarity among users' profiles in order to offer better recommendations of their products (Linden, Smith, & York, 2003).

There are several websites that base their strategy on collaborative tagging, like del.icio.us, Tecnocrati, and Flickr (Marlow et al., 2006) which allow their users to annotate resources, like a web page, a blog post, or an image, normally using free sets of tags and enabling their sharing and reuse. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.