11.11.2014 Views

itec - European Schoolnet

itec - European Schoolnet

itec - European Schoolnet

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

iTEC Project<br />

Title: ITEC-D10_2_V1-1 041102012.Docx<br />

implementation mechanisms according to this analysis. We describe below the results of applying<br />

the above-mentioned methodology to the design of the SDE, identifying the design decisions taken<br />

for each level:<br />

• Level 1. Definition of the Object of Decision. From the four decision targets identified in<br />

MCDM (choice, sorting, ranking and describing the items), the SDE addresses the<br />

relevance calculation and ranking of elements according to their estimated utility. It should<br />

be noted that we have both to identify candidate resources to perform an activity and to<br />

provide a sorted list of resources according to their relevance.<br />

• Level 2. Criteria modelling (definition of families of criteria). To identify the factors to<br />

take into account for relevance calculation, we thoroughly analysed all the properties of<br />

each resource type included in the semantic model to select those that may have an actual<br />

impact in resource selection. Once the collection of relevant properties was selected, we<br />

drew up a document to be presented to the Control Board to be discussed with the aim of<br />

providing an indication of the importance of each factor (cf. Appendix II). We then defined<br />

for each of the selected properties a formal criterion enabling the quantitative evaluation of<br />

the resource according to that property, as stated by the methodology. According to the<br />

formal definition of criterion in Section 1.2.1.1, the SDE will take as the general evaluation<br />

scale the interval . In this way, the criterion will take a value within this interval when<br />

considering users’ ratings or explicit numerical values (e.g. tool, people or event ratings,<br />

tool functionality, people’s expertise). For other criteria, we will adopt as a general rule the<br />

following solution: criterion will take value if the option analysed has the worst<br />

possible rating; if the criterion cannot be assessed because there is not enough<br />

information available; and if the option is totally relevant according to criterion .<br />

For example, the criterion of the cost factor applied to a tool ( ), will be assigned value<br />

<br />

when the tool is completely free of charge in the corresponding Technical<br />

Setting; <br />

when cost information is not available; and <br />

when tool usage<br />

is not free. Besides, when the object being assessed has some relevance according to a<br />

given criterion, this will always take a value in the range , to stress the difference<br />

between a clearly not recommended object ( ) and other objects. There are two<br />

basic strategies to compute these values:<br />

o Based on the number of properties (non-weighted properties): given a non-weighted<br />

property 13 on which we will establish a criterion, we measure how many relevant<br />

values are included in the resource.<br />

<br />

<br />

(4)<br />

This strategy is used, among others, for the criterion associated with the factor<br />

language applied to tools ( ). According to it, if a LARG Context is developed in<br />

English and Spanish, the assigned value will be <br />

for a tool<br />

<br />

13 A non-weighted property is a property defining a specific characteristic of the resource without taking into<br />

account specific numerical values. It supports relations such as resource X is available in language L or is<br />

targeted to audience A.<br />

Page 43/96

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!