文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Symbolic Fuzzy Clustering for Mining User Profiles
侵权投诉

Symbolic Fuzzy Clustering for Mining User Profiles

ABSTRACT: The user modeling problem is generally considered a very important issue in the field of Adaptive Hypermedia Systems. In our work we investigate the particular context of Adaptive Educational Hypermedia Systems, where the users are learners. The

Symbolic Fuzzy Clustering for Mining User Profiles

G.Castellano, C.Castiello, A.M. Fanelli & M.Sidella

Computer Science Department, University of Bari Via Orabona, 4 – 70126 Bari, Italy

E-mail: [castellano, castiello, fanelli]@di.uniba.it, m_sidella@hotmail.com

ABSTRACT: The user modeling problem is generally considered a very important issue in the field of Adap-tive Hypermedia Systems. In our work we investigate the particular context of Adaptive Educational Hyper-media Systems, where the users are learners. The paper proposes a symbolic fuzzy c-means algorithm which can be used to prototype learners in form of profiles; the algorithm works on the basis of symbolic data repre-senting a set of answers to a questionnaire. The key feature of the proposed algorithm is the ability to handle data characterized by qualitative and/or quantitative features. The adopted strategy has been verified on a set of response data concerning an English grammar questionnaire and the obtained experimental results are re-ported.

base-knowledge and attitudes: information about us-ers must be collected before the learning process be-gins, in order to model distinct student profiles. On Hypermedia systems and user-model adaptive sys-tems have been generally considered incompatible the basis of this user modeling process, the system

approaches to information access. In the last few will be able to determine the didactic material and to years, however, the complexity and heterogeneity of tailor the teaching strategies according to the real in-current hypermedia systems, which are expected to terests and needs of the student (Roselli, 1995).

This paper proposes an approach to tackle the be used by a great variety of users, drove a growing

interest in the development of Adaptive Hypermedia user modeling problem when developing Adaptive (AH) Systems (Fink & Kobsa 2000, Brusilovsky Educational Hypermedia Systems. In particular, we

aim at deriving a set of learners’ profiles, gathering 1996).

AH systems arises from the need to make a single users with similar skills and attitudes on the basis of system available to several categories of users which empirical data obtained from questionnaires realized

on purpose. are characterized by different tasks, skills, compe-Cluster analysis represents one of the simplest tencies and background knowledge. A common fea-ture of all AH systems is the ability to build a model tool to generate learners profiles: users that produced

of the user and to generate profiles which encode similar data in their questionnaire answers are different information needs. In this way, the system grouped in the same cluster. In this way, the distinc-tive attributes can be automatically detected, instead could be able to exhibit its services (e.g. a hyperme-of intuitively searching the proper values for each dia page) adapting the contents to the specific de-mands. The AH systems and their benefits have been profile of student. Moreover, the clustering process

reviewed in many studies regarding user modeling implies that members closer to the cluster center are and evaluations (Fink & Kobsa 2000, Brusilovsky more representative for that profile than those at a

greater distance from the center. 1996, Specht, 1998, Brusilovsky, 2001).

In this paper, we adopt a symbolic fuzzy cluster-In this context, the user modeling problem plays a

ing algorithm to perform cluster analysis of data in key role in the design of AH systems and great at-tention is paid to the development of processes for the form of questionnaire answers. From each clus-ter a learner profile is directly derived. building and updating user profiles.

The paper is organized as follows. Section 2 in-Adaptive Educational Hypermedia Systems con-stitute a specific class of AH systems, which find troduces the proposed strategy for mining learner’s application in particular scenery where the users are profiles by categorization of questionnaire re-represented by students. Therefore the Educational sponses. In Section 3 the symbolic fuzzy clustering AH system deals with learners possessing different algorithm adopted to mine learner profiles from data 1 INTRODUCTION

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email