E-LEARNING USABILITY: A LEARNER-ADAPTED APPROACH BASED ON THE EVALUATION OF LEANER S PREFERENCES. Valentina Terzieva, Yuri Pavlov, Rumen Andreev

Transcription

1 Titre du documet / Documet title E-learig usability : A learer-adapted approach based o the evaluatio of leaer's prefereces Auteur(s) / Author(s) TERZIEVA Valetia ; PAVLOV Yuri (1) ; ANDREEV Rume (2) ; Affiliatio(s) du ou des auteurs / Author(s) Affiliatio(s) (1) Cetre of Biomedical Egieerig, Bulgaria Academy of Scieces, Acad. G. Bochev Str., Bl. 105, 1113 Sofia, BULGARIE (2) Istitute of Computer ad Commuicatio Systems, Bulgaria Academy of Scieces, Acad. G. Bochev Str., Bl. 2, 1113 Sofia, BULGARIE Revue / Joural Title Commuicatio & cogitio. Moographies ISSN Source / Source 2007, vol. 40, o 1-2, pp [17 page(s) (article)] (3/4 p.) Lague / Laguage Aglais Editeur / Publisher Commuicatio ad cogitio, Ghet, BELGIQUE (1972) (Revue) Mots-clés aglais / Eglish Keywords E-LEARNING USABILITY: A LEARNER-ADAPTED APPROACH BASED ON THE EVALUATION OF LEANER S PREFERENCES Valetia Terzieva, Yuri Pavlov, Rume Adreev Abstract: E-learig usability is a goal of the user - cetred e-learig desig. Its achievemet depeds o the determiatio of the requiremets, characteristics ad prefereces of the users of a e-learig system or eviromet (learer, teacher ad admiistrator of a educatioal process). This paper presets differet aspects of e-learig usability. Oe of them requires a evaluatio of learer s prefereces that determies the idividual learig style of a learer. The results of this evaluatio are ecessary for learer modelig. The learer model has to ifluece the teacher activity i process of producig learig resources i harmoy with the learig style of the learer. It is obviously that the determiatio of learer prefereces is a elemet of the adaptatio techique that esures the developmet of a adaptive e-learig eviromet. The creatio of a adaptive e-learig eviromet depeds o the techique of learer s prefereces evaluatio. This paper presets a approach to evaluatio of huma s prefereces. The mathematical formulatio of this approach serves for basis of tool developmet. A usage of this tool i the evaluatio of learer s prefereces is illustrated. Key words: E-learig Usability, Adaptive E-learig Eviromet, Utility Fuctio, Evaluatio, User Cetred Desig INTRODUCTION E-learig is a implemetatio of a teachig-learig (educatio) process i a computer-based eviromet. There are two mai approaches to e-learig developmet, which result i differet usages of computers i educatio. The first approach uses a process-orieted desig model i the developmet of a e-learig system. I this case the computer techologies are used to support teacher activities i.e. carryig out the teachig process. This approach coicides with the teacher s viewpoit about e-learig. The secod approach bases o the service-orieted desig model. Accordig to it,

2 the computer techologies are ecessary for the realizatio of a e-learig eviromet that provides differet forms of learig iteractios. This aspect of e-learig developmet presets the viewpoit of a active learer with respect to the way of usig computer techologies. The evolutio of process-orieted e-learig systems requires a icreasig of their capacity. They have to support a richer set of teacher activities that have to be i harmoy with learer s prefereces. We distiguish three steps i the developmet of process-orieted e-learig systems: E-learig systems that esure a teacher activity kowledge delivery (bleded learig); Learig maagemet systems; Adaptive Web-based educatioal systems. The coceptio of bleded learig coceives of the idea that the teachig is a process of trasferrig kowledge (Sigh, H., Reed, Ch., 2001). This kid of learig cosiders the educatio as a process of kowledge delivery. The e-learig systems that realize this idea support oly oe teacher activity - kowledge deliverig. There are differet methods of kowledge deliverig that are i lie with the differet forms of learig (delivery optios): Offlie ad olie learig; Self-paced ad live learig; Formal ad iformal learig; Structured ad ustructured learig. All these forms of learig are implemeted by trasfer procedures (delivery formats). Some of them are sychroous physical formats, sychroous olie formats, self-paced asychroous formats ad etc. The e- learig systems that esure simultaeously differet delivery modes realize the coceptio of bleded learig. The learig maagemet systems (LMS) cover various teachig, learig ad admiistrative fuctios (Pasii, They support the followig teachig fuctios: authorig, classroom maagemet, kowledge maagemet, certificatio ad traiig, persoalizatio, metorig, video coferecig ad discussio boards. Geerally speakig, LMS support a school admiistrator ad teachers to achieve his educatioal objectives ad esure variety of learig resources. The Adaptive Web-Based Educatioal Systems (AWBES) are the third step i the evolutio of the process-orieted e-learig systems (Aroyo, L., Dicheva, D., 2004). They attempt to employ the Web techologies i order to achieve improved adaptatio ad flexibility for sigle or group users -

3 admiistrators, teachers, istructors ad learers. Furthermore, such educatioal systems apply ew methods ad types of courseware compliat with the Sematic Web visio. This promisig class AWBES forms the basis of the emergig Educatioal Sematic Web. Their realizatio guaratees the achievemet of iteroperability amog various educatio systems. I service-orieted e-learig systems, the teachers are a side of a educatio system that supplies the learers with learig resources. From the viewpoit of a learer this side is a learig eviromet, i which oe or more learers work. The learig eviromet is a factor of successful performace of a learig process. It esures learig objects ad services ecessary for the learig process. The service-orieted e-learig system supports the learig eviromet (e-learig eviromet) that is cosidered as a virtual computer system, with which the learers iteract (Kay, J., 2001). I this way, the e-learig eviromet is preseted by a sigle etity that is i bidirectioal coectio with learers. The secod sectio cosiders the usability aspect of e-learig desig. The third sectio describes elemets of our adaptatio model that esures a adaptive e-learig eviromet. Accordig to this model, the learig resources are adapted to learer s prefereces. The ext sectio presets a approach to evaluatio of learer s prefereces. The fifth sectio gives a example of learer modelig ad determiatio of the prefereces of a cocrete learer about the form of learer assessmet. E-LEARNING USABILITY E-learig desig as each oe desig process has two aspects that origiate a costructive desig approach ad a approach to harmoizatio with the user(s). The costructive aspect of e-learig desig is preseted by the process-orieted ad service-orieted desig models. E-learig desig is a costructive task that orgaizes the compoets of a computer-based eviromet i coformace with these models. I process-orieted e- learig desig the computer techology is used for the purposes of a teacher/ school admiistrator. The service-orieted e-learig systems guaratee the framework of a e-learig eviromet that supplies learers with learig resources (Fig. 1.). The developmet of a e-learig eviromet has to esure ot oly learig objects, but the realizatio of differet didactical methods: costructivist learig, collaborative learig, problem-based learig ad etc. The learig resources perform a dual role: they are products of istructor activities ad elemets of a learer-istructor iterface. This iterface meets the eeds of the learer(s) durig the educatioal process.

4 The learig resources have to esure all possible types of iteractio betwee a learer ad a teacher i a e-learig eviromet. These iteractios preset e-learig activities that iclude all learer ad teacher activities. The learig resources are a cosequece of the educatioal activities that support the implemetatio of differet istructioal strategies (Mehlebacher, B., 2002): the idetificatio, evaluatio ad itegratio of a variety of iformatio; collaboratio, discussio ad commuicatio of ideas; participatig i simulated experiece, appreticeships ad cogitive partership; the examiatio of learers. At the same time, the learig resources support the followig learer activities: Accessig iformatio learers idetify istructioal materials relevat to their educatioal objectives ad access them; Scaig iformatio learers search for particular headigs, iformatio items or istructios related to their problem represetatio; Uderstadig iformatio; Trasferrig iformatio durig a discussio, exercises, etc. Teacher/ Admiistrator E-learig Eviromet E-learig System Learig Resources Learer(s) Computer-based Eviromet Fig. 1 E-learig desig The harmoizatio of e-learig with its users is a objective of the user-cetered desig. The latter is closely liked with the cocept of usability, which comes from the field of Huma Factors (also kow as Ergoomics). Huma Factors is a form of egieerig, which puts the huma

5 at the cetre of desig rather tha machies ad equipmets. The usercetered desig takes as its basic premise the view that product developmet should be drive from user requiremets ad prefereces rather tha from techological capabilities. Hece, the costructive aspect of e-learig desig presets a system-orieted approach to desig. Accordig to the system theory e-learig ca be a system, if its parts iteract with oe aother to form a aggregate uified system. I this system the huma beig is cosidered as a user ad the computer implemets a computer-based applicatio system. The itegratio of these two heterogeeous parts requires the harmoizatio of their relatioships. A approach to achievemet of this goal is the creatio of user-adapted e-learig systems. Usability is about producig products ad systems that are easy to use ad suitable for cosumig by idividual users. Usability egieerig ad user-cetered desig help desigers to esure that their product ad system will meet the eeds of the users for whom it is iteded. This approach requires modelig ad idetificatio of the users that will use a service or product. The factors of user modelig are product characteristics. The usability characterizes the iteractio betwee a perso ad a computer-based system ad its products. For the clear determiatio, the cocept of usability is broke dow ito the followig measurable elemets: effectiveess (the ability to achieve specific goals ad perform certai tasks i the eviromet), efficiecy (i terms of used resources ad time for task support), usefuless (should do aythig that we expect to do ad serve a purpose), users satisfactio (icludig ecouragig users) ad learability (the ease ad speed with which users ca comprehed how to use a product i order to lear effectively ad retai the skills ad kowledge). We share the view that e-learig usability has two mai aspects: the usability of e-learig systems ad the efficiecy of computer-based teachig process that produces learer fit learig resources. Whe it is cosidered the e-learig system usability the focus is o the assessmet of the effectiveess ad efficiecy of the computer techologies that are implemeted for supportig of teacher/admiistrator tasks (Storey, M. A., et al., 2002). This assessmet usually bases o usability tests, field studies ad heuristic evaluatio (Miller, M. J. At the other had, the efficiecy of a teachig process determies the usability of the e-learig eviromet that is closely coected with the effectiveess learig process. This paper presets a approach to the productio of usable e-learig resources that have to esure effective learer activities, i.e. a usable leareristructor iterface i order to achieve their mutual educatioal goal. Sice

6 the usable iterface is guarateed by adaptive iteractios, we suggest achievemet of e-learig usability by the productio of learig resources that fit best the learer. It is a part of the developmet of a adaptive e- learig eviromet that depeds o the used adaptatio techique (Paramythis, A., S., http//: Adaptive_Learig_Eviromet_eLearig_Stadards) The latter is determied by the aswers of the followig questios: what would be adapted, how it is to be adapted ad what is the coditio it is to be adapted to. We propose techique to adapt learig resources, which are a compoet of the learig eviromet, to the mai characteristics ad prefereces of their users - learers. The realizatio of this approach is coected to a relatively ew directio of research o the crossroads of user modelig ad user-cetered desig of computer-based systems (Wiograd 1995, Brusilovski 2001). ELEMENTS OF THE USED ADAPTATION TECHNIQUE The usability of e-learig resources ca be viewed from both sides of educatio - teachers ad learers. They work together for achievemet of oe ad the same educatioal goal. From teacher s poit of view usability is grouded o desigig reusable learig materials, which could be reused i differet cotext, based o costructive learig theory ad applyig variety of pedagogical approaches ad istructioal scearios i order to achieve declared educatioal goal. From studet s poit of view, usability meas to be available such learig resources ad services allowig them to fid this oes that fit with learer s goal, kowledge ad prefereces. Therefore usable learer-teacher iterface has to be: Useful to help achievig predefied educatioal goal through providig appropriate learig materials ad coditios; Effective to allow easily performig of all learig activities i a quick ad proficiet maer; Adaptable to allow variety paths i the process of kowledge acquisitio ad checkig, i.e. some kid of persoalisatio; Ecouragig allowig users to perform their best. Usually usability of some product is evaluated a posterior. They test already created product if it satisfies some heuristics or tests (Miller, M. J., We assert that this approach is t appropriate i case of the learig process implemetatio. If oe discovers at the ed of some learig course that it is ot usable, i.e. learers could t perform their tasks ad assigmets successfully, it will be total waste of time ad will discourage learers from attedig courses. This situatio ca

7 be avoided if the teachers esure the achievemet of e-learig usability through a priory idetificatio of the learers. The ecessary iformatio about studet s characteristics could be derived from iquiry (Mehlebacher, B., 2002), user modellig, teacher s previous experiece, pedagogy theory ad practice. I our work we use elemets of user modellig which ivolves iferrig uobservable iformatio about the user from observable iformatio about him/her, e.g. his/her actios or utteraces. The user modelig techiques ca be classified by type of modelig method, by approach to model creatio ad by modelig objectives. There are two types of user modelig methods: empirical methods ad reasoig-based (theoretic) methods. The empirical user modelig techiques use both explicit ad implicit user model acquisitio to determie the user s characteristics or attributes (Zukerma ad Litma, 2001). The reasoigbased methods realize substatiated subjective user modelig that makes user models through cocepts. The classificatio of modelig techiques i correspodece with the modelig objectives depeds o the type of user models that the correspodig techique creates. The classificatio of the user models depeds o the followig aspects: Models of direct ad idirect users; Models of user s characteristics, backgroud, kowledge, prefereces, iterests ad idividual traits; Stereotypical user model or updatig user model. The direct user model represets a user who iteracts with a applicatio system ad determies its output (system products). This model cocers the creatio of a adaptive user-computer dialogue ad guaratees adaptive cotrol of the system. The model of a idirect user helps the creatio of usable system products. The stereotypical model is a result of a modellig approach that qualifies the behaviour or prefereces of a group of users i certai circumstaces as similar ad they are represeted by a fixed (costat) user model. The updatig user model bases o a prelimiary user model that is update i accordace with the actual user behaviour. I may cases, we are ot iterested i user modellig i ay geeral sese, but oly i correspodece with the purposes for which user models are formed. For this reaso, there are differet models for the represetatio of user s characteristics, prefereces (attributes), iterests or backgroud. Takig ito cosideratio the usability research, we ca summarize the mai learers characteristics as follows: Literacy computer, domai kowledge;

8 Cogitive abilities educatioal level, kowledge backgroud, testig ability; Persoal abilities ad prefereces: preferable learig style - deductive, iductive, sequetial, reflective; preferable cotet presetatio explaatory, expository, descriptive, argumetative; self passig pace ad course of learig activities - tasks, assessmets, tests, exams. Accordig to the proposed adaptatio techique the e-learig usability is achieved by producig adaptive learig resources (what would be adapted). They have to esure flexibility ad comfort to learers, cotet uderstadig, operatioal performaces as well as they ehace the learig process ad the users satisfactio. This goal is achieved by gradually adaptig pedagogical approaches, cotet graularity ad fuctioality (how it is to be adapted) to the level of competece ad iterests of the users i.e. the prefereces of a group of users (what is the coditio it is to be adapted to). Other beefit is that adapted learig resources are useful for a heterogeeous group of users. Matchig the learer s characteristics ad prefereces with pedagogical strategies ad applied educatioal approaches ad tools is at the core of the e-learig usability achievemet. MATHEMATICAL FORMULATIONS AND METHODS The estimatio of the prefereces of a group of learers bases o a mathematical approach that cocers the utility theory (Barbera, S., Hammod, P., Seidl, C., 2001). The group prefereces are reflected by the prefereces of a teacher (decisio-maker DM) for the way of carryig out a educatioal process. They have to be i harmoy with the prefereces of learers which the teacher observes durig his work. By meas of the mathematical approach preseted below we describe ad evaluate the prefereces of idirect users of a e-learig system - learers. The modellig techique that bases o stochastic algorithms for assessmet of expert utilities ad values o the basis of the expressed prefereces is a reasoig-based techique for modelig of learer s prefereces. It creates oe-dimesioal models. This techique is used for creatio of a stereotypical model of learer s prefereces for the form ad style of a examiatio. The learer (user) modelig uses the teacher s opiio about learers prefereces. It is a result of his experiece.

9 Stadard descriptio of the utility fuctio applicatio is preseted by Fig. 2. There are a variety of possible fial results that are cosequece of a learer activity determied by a educatioal objective, i.e. the activity cotext. A utility fuctio U(.) assesses each of these results. Cotext Actio Results P 1 U(X 1 ) U(X 2 ) P 2 P U(X ) Fig. 2 Utility fuctio applicatio The DM judgmet of these results is measured quatitatively by the followig formula: U ( p) = PiU ( xi), p is probability distributio Pi = 1 i We deote with P i (i =1 ) subjective or objective probabilities which reflect the ucertaity of the fial results. Strog mathematical formulatio of the utility fuctio is the ext: Let X be the set of alteratives ad P be a subset of the set of probability distributios over X. The DM s prefereces over P are expressed by ( ), icludig those over X. A utility fuctio is ay fuctio u(.) for which is fulfilled: 2 (p f q), (p, q) P ) ( u(.) dp > u(.) dq ). (1) The mathematical expectatio of the utility u (.) is a quatitative measure cocerig the teacher prefereces about the probability distributios over X. I practice the set P is a set of fiite probability distributio. We suppose that the sigleto distributios belog to P, (X P). A "lottery" is called every discrete probability distributio over X. We mark the lottery x with probability α ad y with probability (1-α) as <x,y,α>. There are differet systems of axioms that give satisfactio coditios of utility existece. The most famous of them is the system of Vo Neuma ad Morgester s axioms (Barbera, S., Hammod, P., Seidl, C., 2001). We start with the assumptio that ay covex combiatio of elemets of P belogs to P: (q, p) P 2 (αq+(1-α)p) P, for α [0,1] i

10 (Pavlov, Y., 2005, Pavlov, Y., Ljakova, K., 2003). This coditio ad (X P) determie the utility fuctio over X (whe this fuctio exists) with the accuracy of a affie trasformatio. The most used utility assessmet approach is comparisos of the kid: (z <x,y,α>), where (x z y), α [0,1], (x,y,z) X 3. It is well kow that the trasitivity of "~" is breached i practice because of the so cold certaity effect ad probability distortio idetified by Kahema ad Tversky (Prospect Theory) (Barbera, S., Hammod, P., Seidl, C., 2001). Here is proposed a procedure which resolves some of these difficulties (Pavlov, Y., 2005): <x,y,α> ( or or ( or o aswer ) z, α [0,1], (x,y,z) X 3 Every compariso of this kid defies a learig poit t=(x,y,z,α). With probability D 1 (x,y,z,α) the DM assigs the learig poit to the set A u or with D 2 (x,y,z,α) to B u : A u = {(x, y, z, α)/ (αu(x)+(1 α)u(y))>u(z)}, B u = {(x, y, z, α)/ (αu(x)+(1 α)u(y))<=u(z)}. The DM aswers ( 1; -1; 0) are with probability ad subjective ucertaity. The mai recurret stochastic procedure i the proposed approach has the form (Pavlov, Y., 2005, Pavlov, Y., Ljakova, K., 2003): + 1 ci = ci + D t + c t γ '( 1 ) (, Ψ ( 1 )) ξ 1 Ψ ( t + 1), i 2 γ = +, γ < +, γ 0. (2) Here ( c, Ψ ( t )) deotes scalar product ad D +ξ are the teacher aswers ( 1; -1; 0) were ξ is oise (ucertaity) i the teacher aswers with mathematical expectatio equal to zero. The scalar product has the form: ( c, Ψ( t)) = α( c, Φ( x)) + ( 1 α)( c, Φ( y)) ( c, Φ( z)) = αg (x)+(1-α)g (y)-g (z)=g (x,y,z,α). (3) The coefficiets c i take part i the decompositio of g (x) by a chose family of fuctios (Φ i (x)): g N ( x) = c i Φ ( x ). i = i 1 The lie above T= ( c, Ψ ( t)) meas that T = 1, if T >1, T = 1 if T <(-1) ad T = T if -1 < T < 1. It is kow that uder the procedure (2) coditios specified above the ext itegral coverges to the (mi):

11 if s ( t ) J ( G ( x, y, z, α )) = M( ( y D'( t)) dy) = ( ( y D'( t)) dy) df D' s ( t ) ( ( D '( t ) () t G D'( t) () t G D'( t) y D ' ( t )) dy ) df. (4) Here p.p. deotes almost sure ad s(t) deotes s(t)=αs(x)+(1-α)s(y)-s(z). After some calculatios the followig make the covergece clear: s( t) pp if ( y D'( t)) dydf lim( ( G ( t) D'( t)) df) 0 (5) s( t) D'( t) 2 Takig i to accout the covergece ad the structure of the fuctio G (x,y,z,α) (3) it is assumed that g (x) is approximatio of the empirical utility if () is sufficietly great. The learig poits are posed with the use of a pseudo-radom Sobol s sequeces. EVALUATION OF LEARNERS PREFERENCES WITH RESPECT TO EXAM FORM The estimatio of the prefereces of a group of learers is performed by a decisio support system for estimatio of idividual s utility fuctios developed o the basis of the preseted mathematical formulatios ad methods i a eviromet that cosists of Visual Studio, Visual Basic 6.0. The fial calculatios ad graphics are performed i the MATLAB eviromet. The objective of this example is the evaluatio of the learers prefereces for the form ad style of the exam. This estimatio bases o the teacher s opiio of these prefereces that is a result of his experiece. The examiatio (A) - form cocers the way of kowledge expressio by a learer: test or free expressio. The examiatio (B) - style regards learer s exam: oral ad writte. The possible criteria for the estimatio of the prefereces of learers which satisfactio is the objective of the teacher durig a examiatio are the followigs: (A) % test i relatio to the etire examiatio material (0% to 100%) illustrated i Fig. 3 ad Fig. 4; (B) % time for writte exam i relatio to the whole time that is ecessary for this exam (0% to 100%) - illustrated i Fig. 5 ad Fig. 6. The umber of teacher aswers is 64 for the Fig.3 ad Fig. 4. It is sufficiet oly for the first approximatio. The seesaw lies i Fig. 4 ad Fig. 6 recogize correctly more the 95% of the teacher aswers. Fig. 5 ad Fig. 6 are costructed by 1024 learig poits. The utility depedece o probability ca be assessed directly with the proposed procedure (2). For this purpose we search for a approximatio of the kid u(x, α), α [0,1], x X pp..

12 followig Kahema ad Tversky. The utility fuctios u(x, α) are show o Fig. 3 ad Fig. 5. The explicit formula of the utility u(.) has the form 1 u ( x) = u ( x, α ) dα. 0 Fig. 3 Utility f 1 (x,α) Fig. 4 Utility f 1 (x) Sice the teacher accepts that the factors (A) ad (B) are mutual idepedet i relatio to utility, the utility fuctio has the followig expressio: U(a,b)=K 1 *f 1 (a)+ K 2 *f 2 (b)+(1- K 1 - K 2 )*( f 1 (a)*f 2 (b)), (6) where a, b [0,100]%.

13 Fig. 5 Utility f 2 (x,α) Fig. 6 Utility f 2 (x) The determiatio of the coefficiets K 1 ad K 2 depeds o the determiatio of f 1 (.) ad f 2 (.). This utility fuctio is preseted by the Fig. 7 ad Fig. 8. The figure 8 illustrates the lies of idetical prefereces that show a way for partitio the group of learers i subgroups i accordace with their idetical prefereces. We ca determie to which subgroup belogs a learer through the costructio of his f 1 (.) ad f 2 (.). They are a source for the determiatio of а max ad b max ad U(а max,b max ) that shows the positio of the learer i the space preseted by Fig. 8, i.e. the subgroup to which the learer belogs.

14 Fig.7 Teacher Utility Fig.8 Partitio of E-Learer s group CONCLUSIONS AND FUTURE WORK The fudametal challege facig today's learig professioals is the satisfactio of the eeds ad prefereces of idividual learers or group of them. A oe-size-fits-all approach does ot work well eough. That is why teachers have to provide customized learig experieces for targeted groups. We preset a techology of the evaluatio of learers prefereces that helps the creatio of adaptive e-learig eviromet. The preseted mathematical method allows practical use of expert iformatio about complex system ad problems, which is difficult to use i aother way. This approach is used i our research for learer modellig with respect to the form of examiatio. It is also applicable to evaluatio of other learers prefereces, sice the preseted evaluatio techology is developed for

15 estimatio of huma s prefereces. The learer is a user determied i educatio settig. The user is a huma i the computig eviromet. REFERENCES Aroyo, L., Dicheva, D., 2004, The New Challeges for E-learig: the Educatioal Sematic Web, Educatioal Techology ad Society, (4), pp Barbera, S., Hammod, P., Seidl, C., 2001, Hadbook of Utility Theory, (1), Bosto, Priciples, Kluwer. Brusilovski, P., 2001, Adaptive Hypermedia, User Modellig ad User-Adapted Iteractio, (11), pp Kay, J., 2001, Learer Cotrol, User Modelig ad User-Adapted Iteractio, (11), pp Mehlebacher, B., 2002, Assessig the Usability of Olie Istructioal Materials, I Aderso, R. S., Bauer, J. F., ad Speck, B.W. (Eds), Assessmet Strategies for the O-lie Class: From Theory to Practice, New Directios for Teachig ad Learig Series, Sa Fracisco, CA, Jossey-Bass (91), pp Pavlov, Y., 2005, Subjective Prefereces, Values ad Decisios: Stochastic Approximatio Approach, Comptes redus de L Academie bulgare des Scieces, (58), pp Pavlov, Y., Ljakova, K., 2003, Machie Learig ad Expert Utility Assessmet, Proceedigs of the Iteratioal Coferece CompSysTech 03, Sofia, pp. III.A III.A 5-6. Sigh, H., Reed, Ch., 2001, A White paper: Achievig Success with Bledig Learig, ASTD State of the Idustry Report, America Society for Traiig ad Developmet. Storey, M. A., et al., 2002, Evaluatig the usability of web-based learig tools, Educatioal Techology & Society, (5), (3). Wiograd, T., 1995, From Programmig Eviromets to Eviromets for Desigig, Commuicatios of the ACM, (36) (6), pp Zukerma, I., D. Litma, 2001, Natural Laguage Processig ad User Modelig: Syergies ad Limitatios, User Modellig ad User- Adapted Iteractio, (11) (1-2), pp ABOUT THE AUTHORS Valetia Todorova Terzieva, Istitute of Computer ad Commuicatio Systems, Acad. G. Bochev Str., Bl. 2, 1113 Sofia, Bulgaria, valiabog@yahoo.com

16 Valetia Terzieva is a research associate i Istitute of Computer ad Commuicatio Systems. Her research iterests are i digital sigal processig, e-learig systems, reusable learig objects, usability, useradaptatio ad virtual laboratory. She is a member of the Bulgaria Uio of Automatio ad Iformatics. Yuri Pavlov Pavlov, Cetral Laboratory of Biomedical Egieerig, Bulgaria Academy of Scieces, Acad. G. Bochev Str., Bl. 105, 1113 Sofia, Bulgaria, Е- mail: yupavlov@clbme.bas.bg. Dr. Yuri Pavlov is a research associate i the Cetral Laboratory of Biomedical Egieerig. His research iterests are i systematic aalysis, utility theory, decisio makig, measuremet theory, optimizatio, stochastic approximatio, ad modelig. He is a member of the Bulgaria Uio of Automatio ad Iformatics. Rume Dimov Adreev, Istitute of Computer ad Commuicatio Systems, Bulgaria Academy of Scieces, Acad. G. Bochev Str., Bl. 2, 1113 Sofia, Bulgaria, rume@iccs.isdip.bas.bg Dr. Rume Adreev is a associate professor i the Istitute of Computer ad Commuicatio Systems. His research iterests are i computer graphics, user iterface, computer-aided desig, software egieerig, usability ad system desig. He is a member of the Huma Computer Iteractio Group i British Computer Society ad Bulgaria Uio of Automatio ad Iformatics. publicatio e-learig III ad the kowledge society