Dichotomic Node Network and Cognitive Trait Model

In the search of creating a representation, such as a cognitive trait model, of cognitive traits, such as working memory capacity or inductive reasoning ability, of a learner, it is hard to find a consensus model of the cognitive trait among different pers

2004 (August 30 - Sept 1, 2004, Joensuu, Finland), Los Alamitos, CA: IEEE Computer Society (ISBN 0-7695-2181-9), 702-704

Dichotomic Node Network and Cognitive Trait Model

Taiyu Lin and Kinshuk

Advanced Learning Technologies Research Centre

Massey University New Zealand

Abstract

In the search of creating a representation, such as a cognitive trait model, of cognitive traits, such as working memory capacity or inductive reasoning ability, of a learner, it is hard to find a consensus model of the cognitive trait among different perspectives of cognitive science. Dichotomic node network (DNN) is developed to provide a viable solution to this problem. DNN is a network representation of an entity of which the constituents are nodes that is consisted of a pair of dichotomic attributes. Through the contradiction detection mechanism and inclusion resolution mechanism, DNN is able to (1) represents of an entity contains multiple portrayals/perspectives, (2) select appropriate portrayals for any particular entity is very difficult or impossible, (3) handle nonlinear aggregation of portrayals in which combinations does not render result linearly, and therefore very suitable for cognitive trait model, and is potential for other applications.

losing the accuracy desired. With the regarded difficulty to understand and mysteriousness of the mind [2], not a single perspective could include a complete and all embracing description of how the mind works. It is indeed a difficult question on how to create a cognitive profile of a learner given all of the complexities.

In this paper, a mechanism is proposed to solve

this dilemma. The mechanism is called dichotomic node network. Dichotomic node network is employed in cognitive trait model [5][6], which profiles the learners using the cognitive traits such as working memory capacity, and inductive reasoning ability, to create the representation of the cognitive traits. Cognitive trait model will be briefly introduced along the discussion of the dichotomic node network as an exemplary application area.

Introduction

In the search of creating a representation of

cognitive traits, such as working memory capacity or inductive reasoning ability, of a learner, it is hard to find a consensus model of the cognitive trait among different perspectives of cognitive science. Taking working memory capacity as an example, some [8] viewed it structurally whereas others [1] viewed it functionally. The same disagreement applies to other cognitive trait such as inductive reasoning ability [3] [10]. Some of the perspectives came from theoretical frameworks [8], whereas others are derived from experimental results [4].

In close scrutiny, each of the different

perspectives proposed holds certain degree of truth about the characteristics of the cognitive trait. Taking any one of them and discard the rest means taking a great risk of

Dichotomic Node Network

Cognitive trait model [5][6] employs a mechanism that infers the learner’s cognitive traits from the navigational behaviours of the learner by finding certain navigational patterns (e.g. navigational linearity), called manifests, that give signs to the learner’s capacity of cognitive traits (e.g. working memory capacity). The basis of each manifest could be from different theories about the cognitive trait, and therefore dose not agree to each other. Therefore, dichotomic node network is developed to provide a viable solution to this problem.

Dichotomic node network (DNN) is a network

representation of an entity of which the constituents are nodes that is consisted of a pair of dichotomic attributes. Figure 1 depicts an exemplary network consisting of four nodes. Each node is a partial portrayal of the entity. The value of the entity is determined by its constituent nodes. Each node contains a numerical value called weight which determines the node’s influence over overall output value of the entity.

第1页下一页