Towards Configuration of applied Web-based information system

T o w ards Configur atio n of appl i ed W eb-based inform atio n system Mark Sh. Levin ∗ In the pap er, combinatori al sy nthesis of structure for applied W eb-based systems is described. The problem is co nsidered as a combination of selected design alternatives for sy stem parts/comp onents in to a resultant com- p osite decisi on (i.e., system configuration design). The solving framew ork is based on Hierarc hical Morphological Multicriteria Des ign (HMMD) approac h: (i) multicriteria selection of alternatives for system parts, (ii) comp osing the selected alternatives into a resultant com bination (while taking into account ordinal q ualit y of the alterna- tives ab ov e and their compatibilit y). A lattice-based discrete space is used to ev aluate (to integrate) quality of the resultant combinations (i.e., composite system decisions or system confi gurations). I n addition, a simplified solving framew ork based on m ulticriteria multiple choice problem is considered. A m ultistage design pro cess to obtain a system tra jectory is describ ed as well. The basic app lied example is targeted to an applied W eb-based system for a communication service provider. Tw o other ap p lications are briefly d escribed (corp orate system and information system for academic app lication). Keywor ds: W eb-based system, System design, Communication p rovider, Configuration, Comp osition, S ynthe- sis, Com binatorial optimization 1. In tro duction W eb-based a pplied systems ar e increas ing in po pularity . Her e some bas ic tec hnologica l direc- tions can b e p ointed o ut as follows: (a) E-business and E-co mmer ce, for exa mple: (i) smart ma rketplaces ar e presented in [15], (ii) W eb services ar e studied in ([35] [61]), (iii) E- commerce bo o ks are pres ent ed in ([47] [52]); (b) W eb-ba sed informa tion systems are studied in ([11], [20], [56]); (c) E- gov ernment and E- demo cracy , for exam- ple: (i) e-g overnmen t systems (including their management, functionality , evolution, des igning, and innov atio n) are presented in ([1 7], [27], [42], [45], [50]); (ii) decisio n suppor t for participa- tory demo crac y is presen ted in [19], (iii) W eb- based public par ticipa tion geogr aphical informa- tion systems are descr ib e d in [24]; (d) W eb- based medicine systems: (i) W eb- based telemedicine systems for home-care are pre- sented in [4 ], (ii) developmen t o f W eb-bas e d c lin- ical information sys tems was intro duced in [9]); ∗ Mark Sh. Levin: Inst. for Infor mation T ransmis- sion Problems, Russian Academ y of Sciences, 19 Bol- sho j Karetn y lane, Moscow 127994, Russia. Email: mslevin@acm.org (e) W eb-based educational systems (e- learning, e-teaching, etc.), for ex ample: (i) W eb- based learning and tea ch ing t echnologies are studied in [1], (ii) dev elopment of ada ptive W eb-based courseware was int ro duced in [8], (iii) building a W eb-based educational system w as pr esented in [40], (iv) conce ptua l view o f web-based e-learning was sugg ested in [5 1]; (f ) W eb-based research suppo rt sys tems, for ex- ample: (i) special W eb-bas e d re search suppor t system was desig ned [57], (ii) fra mework for W eb- based resea r ch supp or t was sugges ted in [59]. Generally , it is p os sible to consider the follow- ing brief des cription o f an applied W eb-based sys- tem: (1) t here is a set o f user s and server(s), (2) each user ha s information and computing tasks (including W e b- based ta sks), (3) the server is a ba sis for information system (i.e., informa- tion pro c e ssing) and computing (for each user), (4) each us er has a p ers onal br owser, (5) use r s hav e their ac cess to the server(s) se pa rately , (6) there is a concurrent multiple user access, ( 7) there ar e limitations to the volume of infor ma- tion transmissio n, and (8) ther e ar e re quirements to p er formance, se curity , scalability , adaptabil- it y and upgradeability . As a r esult, there e x - 1 2 ists a need of W eb-ba sed s ystem life cy cle eng i- neering/mana gement (e.g., W eb engineer ing ) in- cluding req uirements engineering, de s ign, maint e- nance (e.g., [3], [5], [11], [12], [1 4], [16], [18], [2 1], [26], [35], [36], [38], [40], [44], [4 7], [60], [61]). Mainly , the desig n pr o cess o f W eb-ba sed ap- plied sys tems co ns ists in system co nfiguration design (i.e., selection or co mpo sition of design alternatives for s y stem co mpo nents/parts) (e.g., [2], [5], [6], [7], [10], [14], [1 8], [37], [4 1], [43], [55], [60]). Fig. 1 illustrates the design of system configuratio n as a selection of a lter- natives for s ystem parts. Here a comp osite (mo dula r) s y stem co nsists of m system par ts : { P (1) , ..., P ( i ) , ..., P ( m ) } . F or each system part (i.e., ∀ i, i = 1 , m ) there ar e cor r esp onding alter- natives { X i 1 , X i 2 , ..., X i q i } , where q i is the num b er of alterna tives for part i . The pro blem is: Sele ct an alternative for e ach system p art while taking into ac c ount some lo c al and/or glob al ob- je ctives/pr efer enc es and c onstr aints. Fig. 1. System configur ation pr oblem X 1 q 1 ... X 1 3 X 1 2 X 1 1 ☛ ✡ ✟ ✠ X i q i ... X i 3 X i 2 X i 1 ☛ ✡ ✟ ✠ X m q m ... X m 3 X m 2 X m 1 ☛ ✡ ✟ ✠ ❡ ❡ ❡ P (1) P ( i ) P ( m ) . . . . . . Comp osite system In Fig . 1 the following system configura tion example is depicted: S 1 = X 1 2 ⋆ ... ⋆ X i 3 ⋆ ... ⋆ X m 1 . T able 1 contains s ome approa ches to comp ositio n of applied W eb-based s y stems. No te a survey of combinatorial optimization mo dels, which can be used for system configur ation des ig n problems, is presented in [3 2]. The paper address es the problem of co mb in- ing some design a lternatives (D As) for W eb-ba sed system par ts/comp onents into a resultant com- po site decision (i.e., system configur ation). A sp ecial lattice-base d discrete s pace is used to ev al- uate qualit y of the resulta nt comp osite s ystem decisions or system co nfigurations. The ab ov e- men tioned discrete space integrates ordinal qual- it y of system elements and or dina l quality of com- patibilit y amo ng the system elements. T able 1 . Comp osition o f W eb-based sys tems Approach References 1. O b ject-oriented a ppr oach [14], [46] 2. Decla r ative approach [37] 3. Mo del- driven design [38] 4. AI tec hniques [43] 5. Self-s e rv environmen t [6] 6. Age nt-based approa ch [7] 7. O n tology- ba sed approa ch [2] 8. Q oS-aw are selection o f web services [36] 9. Dynamic selection [18] 10. QoS capable W eb service comp osition (multiple choice knapsack problem, ) shortest path pr oblem) [60] 11. Petri net a pproach to com- po sition of W eb s ervices [58] Hierarchical Morpholo gical Multicriter ia De- sign (HMMD) appr oach was suggested by Levin (e.g., [28], [30], [31]). This approach is used here as a general solving framework. I n addition, a simplified solving framew ork bas ed o n us age o f m ulticriteria multip le choice problem is co nsid- ered. This appr oach was suggested in ([34] [5 4]). Here the design problem do es no t inv olve elemen t compatibility . F urther a multistage design pro- cess to obtain a sys tem tr a jectory is descr ib e d. This design pro blem was presented in (e.g., [30], [31]). F or this pro blem the solving scheme is based on HMMD. HMMD approa ch inv olves the following phases: (i) design of a system tr ee-like mo del for the resultant comp osite decis ions, (ii) generation of (searching for) design alter natives for le af node s of the system model, (iii) ev a luation of the al- ternatives fo r s y stem parts, and (iv) comp osing the alterna tives (D As) into a resultant combina- tion as the s ystem decisio n(s) (while taking into account ordinal quality of the alter natives ab ov e and their compatibility o r in terconnection). Note HMMD gener alizes mo rphologica l analy sis that was crea ted by Zwicky (e.g., [62]). 3 HMMD implements modular mu lti-stage de- sign approa ch and provides the following: (1 ) hierarchical (Bottom-Up) design pro cess (m ulti- criteria ass essment, ev alua tion, selection, com- po sition of design alternativ es), (2) indepe n- dent ass e s sment and analysis of desig n alterna - tives f or ea ch system part/comp onent (includ- ing joint and/o r indep endent par ticipation of dif- ferent do main exp erts), (3) integration of an- alytical, co mputer -based, and exp ert-ba sed as- sessment of design alternatives and their inter- connection, (4) para llel (and concurrent) anal- ysis a nd design (ev a luation, selectio n, comp osi- tion) o f design alternatives for comp os ite system parts/co mp onents, (5) opp ortunity to use cogni- tive metho ds at each step and/o r part o f the de- sign pro cess. In the article, the basic applied example is tar- geted to a n applied informa tio n system for a com- m unication ser vice provider. Tw o o ther applica- tions are briefly descr ibed: corp ora te information system and information sy s tem for a n academic (scientific and/or educationa l) applicatio n. The same hierar chical design appr oach has b een used to W eb-ho sting sy stems [33]. Fig. 2 illustrates the intro duction par t. Fig. 2. Illustratio n for in tro duction part E-business ❄ E-gover- nmen t ❄ W eb-based medicine ❄ . . . Systems engineer ing o f W eb-based sys tems (e.g., W eb e ng ineering) ❄ ❄ . . . ❄ Mo dular design (sys- tem configura tion) ❄ ❄ ❄ ❄ Example 1: W eb-based system for provider Example 2: W eb-based system for corp ora te application Example 3: W eb-based system for academic application Example 4: W eb- hosting system [33] 2. Underlay ing Proble ms/Sc hemes 2.1. M ulticriteria Ranking Let H = { 1 , ..., i , ..., t } be a se t of items which are ev aluated up on criteria K = { 1 , ..., j, ..., d } and z i,j is an e s timate (quanti- tative, ordinal) o f item i on criterio n j . The matrix { z i,j } is a basis to build a partia l order on H , f or example thr o ugh the follo wing g en- eralized s cheme: (a) pairwis e elemen ts c ompari- son to get a pr eference (and/or incomparability , equiv alence) bina ry rela tion, (b) building a par - tial or der o n H . Here the following pa rtial o r- der (par tition) as linea r order ed subsets of H is searched for: H = ∪ m k =1 H ( k ), | H ( k 1 ) ∩ H ( k 2 ) | = 0 if k 1 6 = k 2 , i 2  i 1 ∀ i 1 ∈ H ( k 1 ), ∀ i 2 ∈ H ( k 2 ), k 1 ≤ k 2 . Set H ( k ) is ca lled layer k , and each item i ∈ H gets priority r i that equals the num ber of the cor resp onding lay er. This problem b elong s to class of ill-structured problems by classifica tion of Simo n and Newell [53]. The list of bas ic tech- niques for multicriteria selection is the following: (1) multi-attribute utilit y analysis [22]; (2) multi- criterion decisio n ma king [25]; (3) Analytic Hier- arch y Pr o cess (AHP) [49]; (4) outranking tech- niques [48]; e tc. 2.2. K napsac k Proble ms The description of knaps ack-lik e problems is presented in ([2 3], [39]). The basic (simplified) knapsack problem for m ulation is: max P m i =1 c i x i s.t. P m i =1 a i x i ≤ b , x i ∈ { 0 , 1 } , i = 1 , m , where x i = 1 if item i is selected, c i is a v alue (”utilit y”) for item i , and a i is a weigh t (or re- source required). Often nonnegative co efficients are assumed. The pr oblem is NP-hard and it is presented, for e xample in ([13], [39]). This pr ob- lem can b e solved by enumerative metho ds (e.g ., Branch-and-Bound, dy na mic progra mming), a p- proximate schemes with a limited relative erro r, for exa mple, the algor ithms are describ ed in ([23], [39]). In the ca se of a m ultiple choice pro blem, the items (e.g., actions) a re divided in to g r oups and we select elements from ea ch gro up while taking int o account a total res ource cons traint (or co n- straints): 4 max P m i =1 P q i j =1 c ij x ij s.t. P m i =1 P q i j =1 a ij x ij ≤ b, P q i j =1 x ij = 1, i = 1 , m , x ij ∈ { 0 , 1 } . In the ca se o f multicriteria descriptio n ea ch el- ement (i.e., ( i, j )) has a vector profit: c i,j = ( c 1 i,j , ..., c ξ i,j , ..., c r i,j ). A version of multicriteria multiple choice prob- lem was pr esented in [34]): max P m i =1 P q i j =1 c ξ ij x ij , ∀ ξ = 1 , r s.t. P m i =1 P q i j =1 a ij x ij ≤ b, P q i j =1 x ij = 1, i = 1 , m , x ij ∈ { 0 , 1 } . Evidently , in this case it is reasona ble to se a rch for Pareto-efficient solutions (b y the vector ob jec- tive function ab ov e). Here the following solving schemes can be used [34]: (i) dynamic pro gram- ming, (ii) heuristic based on preliminary mul- ticriteria ranking of elements to get their prio r i- ties and step-by-step pa cking the k na psack (i.e., greedy approa ch) , (iii) mult icriteria rank ing of elements to get their or dinal prio rities and usage of approximate solving scheme (as for kna ps ack problem) based o n dis crete space of sy stem excel- lence (i.e., lattice as in HMMD). In the ar ticle, greedy heuristic a bove is used la ter. 2.3. M orphological Des ign Hierarchical Mor phologica l Multicriteria De- sign (HMMD) approach, sugg ested by Levin (e.g., [28], [30], [3 1]), is based on the mo rphologica l clique problem. The comp osite (mo dular, de- comp osable) sys tem under exa mination consists of the comp onents and their int erconnections or compatibilities. B asic a ssumptions of HMMD a re the following: (a ) a tr e e -like structure of the sys- tem; (b) a co mpo s ite estimate fo r system quality that in tegrates co mpo nent s (subsys tems, pa r ts) qualities and qualities of interconnections (here- inafter referr e d as ’IC’) a cross subs ystems; (c) monotonic criteria for the system and its com- po nent s; and (d) quality of system comp onents and IC are ev aluated on the basis o f co ordinated ordinal sca les. The de s ignations are: (1) des ign alternatives (D As) fo r no des of the mo del; (2) priorities of DAs ( r = 1 , k ; 1 corres po nds to the bes t level); (3) ordina l compatibility es timates for each pair of DAs ( w = 0 , l ; l corres p onds to the b est level). The bas ic phases of HMMD are (Fig. 3): 1. design o f the tree-like system mo del (a preliminary phase); 2. g enerating D As for mo del’s le af no des; 3. hierarchical selection and comp osing of D As into compo site DAs for the cor- resp onding higher level o f the system hiera rch y (morphologic a l clique problem); a nd 4. analysis and improvemen t o f the resulta n t comp osite DA s (decisions). Fig. 3 . ’B ottom-Up’ scheme ✻ Generating DAs ✻ Assessment of D As . . . . . .   ✒ Ranking of DAs ✻ Generating DAs ✻ Assessment of D As ❅ ❅ ■ Ranking of DAs   ✒ Comp osition of DAs (morphologic a l clique problem)   ✒ ❅ ❅ ■ ✻ Comp osition of DAs (mor - phological clique problem . . . ✻ Analysis and improv ement of resultant comp osite DAs Let S b e a sys tem consis ting o f m parts (com- po nent s): P (1) , ..., P ( i ) , ..., P ( m ). A set o f de- sign alternatives is generated for each system part ab ov e. The pr oblem is: Find a c omp osite design alternative S = S (1) ⋆ ... ⋆ S ( i ) ⋆ ... ⋆ S ( m ) of D As ( one r epr esenta- tive design alternative S ( i ) for e ach system c om- p onent/p art P ( i ), i = 1 , m ) with non-zer o IC estimates b etwe en design altern atives. A discrete spa ce of the sys tem ex c ellence on the basis of the following vector is used: N ( S ) = ( w ( S ); n ( S )), where w ( S ) is the minimum of pairwise co mpatibilit y b etw een DAs which corre- sp ond to different system comp onents (i.e., ∀ P j 1 and P j 2 , 1 ≤ j 1 6 = j 2 ≤ m ) in S , n ( S ) = ( n 1 , ..., n r , ...n k ), where n r is the n um b er of DAs of the r th quality in S ( P k r =1 n r = m ). As a 5 result, we se arch for comp osite s y stem decisions which are nondominated by N ( S ) (Fig. 4 a nd Fig. 5 ). Here an en umera tive solving scheme (e.g., dynamic progra mming) is used (usually m ≤ 6) [28]. Fig. 4 . L a ttice of system quality Lattice: w = 2 < 3 , 0 , 0 > < 2 , 1 , 0 > N ( S 1 )    ✠ < 2 , 0 , 1 > < 1 , 1 , 1 > < 1 , 0 , 2 > ... < 0 , 1 , 2 > < 0 , 0 , 3 > < 1 , 2 , 0 > < 0 , 3 , 0 > < 0 , 2 , 1 > Lattice: w = 3 The ideal po int < 3 , 0 , 0 > < 2 , 1 , 0 > < 2 , 0 , 1 > < 1 , 1 , 1 > < 1 , 0 , 2 > < 0 , 1 , 2 > < 0 , 0 , 3 > < 1 , 2 , 0 > < 0 , 3 , 0 > < 0 , 2 , 1 > N ( S 2 ) ❅ ❅ ■ Fig. 5 . Illus tration of system q uality space ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✉ N ( S 2 ) ✉ N ( S 1 ) r ❤ The ideal po int w ( S ) = 1 w ( S ) = 2 w ( S ) = 3 Generally , the following layers of system excel- lence can b e c o nsidered: (i) ideal point; (ii) Pareto-efficient p oints; (iii) a neighbor ho o d of Pareto-efficient D As (e.g., a co mpo site decision of this set can b e tra nsformed into a Pareto-efficient po int o n the basis of a n impr ovemen t action(s)). Clearly , the compatibilit y co mpone nt of v ecto r N ( S ) can b e consider ed on the basis of a p oset- like scale to o (as n ( S )) ([29], [31]). In this case, the dis crete s pa ce of s ystem excelle nc e will b e a n analogica l lattice. Fig. 6 and Fig. 7 illustra te the co mpo sition problem (b y a numerical example fo r a system consisting of three parts S = X ⋆ Y ⋆Z ). P riorities of DA s a re shown in Fig. 6 in parentheses and are depicted in Fig. 7; compatibility estimates are p ointed o ut in Fig. 7). In the exa mple, the resultant comp osite decisions are (Fig. 4, Fig. 5, Fig. 6, Fig. 7): S 1 = X 2 ⋆ Y 1 ⋆ Z 2 , N ( S 1 ) = (2; 2 , 0 , 1); S 2 = X 3 ⋆ Y 1 ⋆ Z 3 , N ( S 2 ) = (3; 1 , 0 , 2). Fig. 6. Example of comp osition X 3 (1) X 2 (1) X 1 (2) ✞ ✝ ☎ ✆ Y 2 (2) Y 1 (3) ✞ ✝ ☎ ✆ Z 3 (3) Z 2 (1) Z 1 (1) ✞ ✝ ☎ ✆ ❡ ❡ ❡ ❡ r ❡ r ❡ r ❡ r ❡ r ❡ ❡ r r ❡ r ① X Y Z S = X ⋆ Y ⋆ Z S 1 = X 2 ⋆ Y 1 ⋆ Z 2 S 2 = X 3 ⋆ Y 1 ⋆ Z 3 Fig. 7 . C o ncentric presentation Z 3 Z 2 ☛ ✡ ✟ ✠ Z 1 Y 2 Y 1 ☛ ✡ ✟ ✠ X 3 X 2 ✞ ✝ ☎ ✆ X 1 2 2 3 3 3 3 3. Appli ed W eb-based System The str ucture (infrastructur e) of a n applied W eb ba sed system is examined as a combination of t wo main parts: so ftw are and hardware. T he basic example is ta r geted to a communication ser - vice provider (example 1). 3.1. H ierarc hical Mo del and Comp onen ts The tree-like mo del of the considered informa- tion system infras tructure is depicted in Fig . 8 . D As for sys tem comp onents ar e the following: (1) ser ver for DBs M : PC ( M 1 ), Sup ermicro ( M 2 ), and Sun ( M 3 ); 6 (2) ser ver for applica tions E : on server of DBs ( E 1 ), Sun ( E 2 ), Sup ermicro ( E 3 ), and PC ( E 4 ); (3) W eb-server W : Apache HTTP-server ( W 1 ), Microsoft I IS ( W 2 ), Bea W eblo g ic ( W 3 ), W eb Sphere ( W 4 ), and W eblogic cluster ( W 5 ); (4) DBMS D : Or acle ( D 1 ), Microsoft SQL ( D 2 ), and designed SQL ( D 3 ); and (5) o per ation system O : Windows 2000 server ( O 1 ), Windows 2003 ( O 2 ), Solaris ( O 3 ), F reeBSD ( O 4 ), and RHEL AS ( O 5 ). Fig. 8 . Str ucture of applied W eb-bas ed system ① S = A ⋆ B = ( M ⋆ E ) ⋆ ( W ⋆ D ⋆ O ) Hardware A = M ⋆ E Soft ware B = W ⋆ D ⋆ O ✉ ✉ r r r r r M E W D O Op eration system O 1 O 2 O 3 O 4 O 5 DBMS D 1 D 2 D 3 W eb server W 1 W 2 W 3 W 4 W 5 Server for applica- tions E 1 E 2 E 3 E 4 Server for DBs M 1 M 2 M 3 3.2. As sessment The following criteria are used for assessment of D As (’+’ corr esp onds to positive orientation of an or dinal sc ale a s [1 , 6] when the biggest esti- mate is the b est one, ’-’ co rresp onds to the neg- ative orientation of the scale when the smallest estimate is the b est one): ( a) cost C 1 (’-’), (b) per formance C 2 (’+’), (c) complexity of maint e- nance C 3 (’-’), and (d) scalability C 4 (’+’). The corres p o nding estimates for D A i a re as follows z i = ( z i 1 , z i 2 , z i 3 , z i 4 ). T ables 2 and 3 cont ain ordinal estimates of D As upo n the ab ov e-mentioned criteria (exp ert judg- men t). Cr iteria weights for three application ex- amples are co n tained in T able 4. E stimates of compatibility b etw een DAs ar e contained in T a- bles 5 and 6 (exp ert judgment). 3.3. C ommunica tion Service Provider The r esultant priorities o f DAs are o bta ined as res ult o f multicriteria ranking (Ele ctre-like metho d). The priorities of DA s for example 1 (communication service pr ovider) a re shown in Fig. 9 in par ent heses. T able 2. Estimates D As Criteria 1 2 3 4 M 1 M 2 M 3 E 1 E 2 E 3 E 4 W 1 W 2 W 3 W 4 W 5 2 2 3 2 5 4 4 3 6 5 5 5 1 2 3 1 6 5 5 5 5 4 4 3 2 2 3 2 1 5 2 2 4 3 3 5 5 3 4 4 4 4 5 3 6 3 4 5 T able 3. Estima tes D As Criteria 1 2 3 4 D 1 D 2 D 3 O 1 O 2 O 3 O 4 O 5 6 4 4 5 5 4 3 4 1 3 3 3 3 2 2 2 4 3 1 4 1 5 5 5 1 4 4 3 1 4 3 4 T able 4. Criteria weight s Application example Criteria 1 2 3 4 1.Provider 2.Corp or a te system 3.Academic system − 1 1 − 1 1 − 3 1 − 2 1 − 1 3 − 1 1 T able 5. Compatibility M 1 M 2 M 3 E 1 E 2 E 3 E 4 3 3 3 3 3 3 3 3 3 3 3 3 T able 6. Compatibility W 1 W 2 W 3 W 4 W 5 D 1 D 2 D 3 D 1 D 2 D 3 O 1 O 2 O 3 O 4 O 5 3 3 3 2 2 3 3 3 3 3 3 3 3 0 0 0 3 3 3 3 3 3 3 3 3 3 3 0 3 3 0 0 3 3 3 3 3 3 3 3 3 3 3 0 1 3 3 0 0 1 3 3 3 3 3 F or system part A , we get the following Pareto- efficient c o mpo site DA (sup erscr ipt fo r A , B , and S co rresp onds to the nu mber of applied appli- cation as 1, 2, 3): A 1 1 = M 2 ⋆ E 2 , N ( A 1 1 ) = 7 (3; 1 , 1 , 0). Fig. 10 illustra tes the space of q uality for N ( A 1 1 ). F or sys tem pa rt B , we g et the follow- ing Pareto-efficient comp osite D As: B 1 1 = W 1 ⋆ D 3 ⋆ O 3 , N ( B 1 1 ) = (3; 2 , 1 , 0); B 1 2 = W 2 ⋆ D 2 ⋆ O 2 , N ( B 1 2 ) = (3; 2 , 1 , 0); and B 1 3 = W 1 ⋆ D 2 ⋆ O 5 , N ( B 1 3 ) = (1; 3 , 0 , 0). Fig. 9. Communication provider ① S = A ⋆ B S 1 1 = A 1 1 ⋆ B 1 1 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 3 ) S 1 2 = A 1 1 ⋆ B 1 2 = ( M 2 ⋆ E 2 ) ⋆ ( W 2 ⋆ D 2 ⋆ O 2 ) S 1 3 = A 1 1 ⋆ B 1 3 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 2 ⋆ O 5 ) Hardware A = M ⋆ E Soft ware B = W ⋆ D ⋆ O ✉ B 1 1 = W 1 ⋆ D 3 ⋆ O 3 B 1 2 = W 2 ⋆ D 2 ⋆ O 2 B 1 3 = W 1 ⋆ D 2 ⋆ O 5 ✉ A 1 1 = M 2 ⋆ E 2 r r r r r M E W D O Op eration system O 1 (3) O 2 (2) O 3 (1) O 4 (3) O 5 (1) DBMS D 1 (3) D 2 (1) D 3 (2) W eb server W 1 (1) W 2 (1) W 3 (3) W 4 (3) W 5 (3) Server for applica- tions E 1 (3) E 2 (1) E 3 (2) E 4 (3) Server for DBs M 1 (3) M 2 (2) M 3 (3) Fig. 11 illus tr ates comp osite DAs for pa rt B . Clearly , the resultant comp osite DAs a re the fol- lowing: (1) S 1 1 = A 1 1 ⋆ B 1 1 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 3 ); (2) S 1 2 = A 1 1 ⋆ B 1 2 = ( M 2 ⋆ E 2 ) ⋆ ( W 2 ⋆ D 2 ⋆ O 2 ); (3) S 1 3 = A 1 1 ⋆ B 1 3 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 2 ⋆ O 5 ). Fig. 1 0 . Space of system qua lit y for A ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✉ N ( A 1 1 ) , N ( A 2 1 ) , N ( A 2 2 ) r ❤ The ideal po int N ( A 3 1 ) w = 1 w = 2 w = 3 Fig. 1 1 . Space o f system quality for B ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✓ ✓ ✓ ✓ ✓ ✓ ❙ ❙ ❙ ❙ ❙ ❙ ✉ N ( B 1 3 ) ✉ N ( B 1 1 ) , N ( B 1 2 ) N ( B 2 1 ) , N ( B 2 2 ) , N ( B 3 1 ) r ❤ The ideal po int w = 1 w = 2 w = 3 In addition, it is reaso nable to consider the fo l- lowing technological system pr oblems [31]: (a ) revelation o f “b ottlenecks” and (b) improv ement of so me obtained solution(s). F or example, let us examine comp osite DAs for B : B 1 3 = W 1 ⋆ D 2 ⋆ O 5 with N ( B 1 3 ) = (1; 3 , 0 , 0). Here compatibility ( D 2 , O 5 ) (that equals 1) is the “b o ttleneck”. As a result, a sp ecial activity for improving this com- patibilit y can b e conside r ed as an impr ov ement op eration. 3.4. C orp orate Application The priorities o f DAs for exa mple 2 (corp orate application) are shown in Fig. 1 2 in par ent heses. F or system part A , we get the following Pareto- efficient comp osite DAs: A 2 1 = M 1 ⋆ E 1 , N ( A 2 1 ) = (3; 1 , 1 , 0); a nd A 2 2 = M 2 ⋆ E 2 , N ( A 2 1 ) = (3; 1 , 1 , 0). Qualit y o f decisio ns A 2 1 and A 2 2 is depicted in Fig. 10. F or system par t B , we g et the following Pareto-efficient comp osite D As (the ideal solutio ns): B 2 1 = W 1 ⋆ D 3 ⋆ O 5 , N ( B 2 1 ) = (3; 3 , 0 , 0); and B 2 2 = W 2 ⋆ D 3 ⋆ O 2 , N ( B 2 2 ) = (3; 3 , 0 , 0). Quality of decisions B 2 1 and B 2 2 is depicted in Fig. 11. As a result, we get the following four final comp osite DAs: (1) S 2 1 = A 2 1 ⋆ B 2 1 = ( M 1 ⋆ E 1 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 5 ); (2) S 2 2 = A 2 1 ⋆ B 2 2 = ( M 1 ⋆ E 1 ) ⋆ ( W 2 ⋆ D 3 ⋆ O 2 ); (3) S 2 3 = A 2 2 ⋆ B 2 1 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 5 ); (4) S 2 4 = A 2 2 ⋆ B 2 2 = ( M 2 ⋆ E 2 ) ⋆ ( W 2 ⋆ D 3 ⋆ O 2 ). Fig. 1 2 depicts infor mation system a nd com- po site decisions fo r example 2. 8 Fig. 12. Corp ora te applica tion ① S = A ⋆ B S 2 1 = A 2 1 ⋆ B 2 1 = ( M 1 ⋆ E 1 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 5 ) S 2 2 = A 2 1 ⋆ B 2 2 = ( M 1 ⋆ E 1 ) ⋆ ( W 2 ⋆ D 3 ⋆ O 2 ) S 2 3 = A 2 1 ⋆ B 2 1 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 5 ) S 2 4 = A 2 1 ⋆ B 2 2 = ( M 2 ⋆ E 2 ) ⋆ ( W 2 ⋆ D 3 ⋆ O 2 ) Hardware A = M ⋆ E Soft ware B = W ⋆ D ⋆ O ✉ B 2 1 = W 1 ⋆ D 3 ⋆ O 5 B 2 2 = W 2 ⋆ D 3 ⋆ O 2 ✉ A 2 1 = M 1 ⋆ E 1 A 2 2 = M 2 ⋆ E 2 r r r r r M E W D O Op eration system O 1 (3) O 2 (1) O 3 (2) O 4 (2) O 5 (1) DBMS D 1 (3) D 2 (2) D 3 (1) W eb server W 1 (1) W 2 (2) W 3 (3) W 4 (3) W 5 (3) Server for applica- tions E 1 (1) E 2 (3) E 3 (3) E 4 (3) Server for DBs M 1 (2) M 2 (2) M 3 (3) 3.5. Academi c Appl ication The prior ities of D As for example 3 (aca demic application) are shown in Fig. 1 3 in parentheses. F or system part A , we g et the following Pareto- efficient comp osite D A: A 3 1 = M 3 ⋆ E 2 , N ( A 3 1 ) = (3; 2 , 0 , 0). Q uality of decision A 3 1 is depicted in Fig. 10. F or system part B , we get the following Pareto-efficient comp osite DA: B 3 1 = W 1 ⋆ D 2 ⋆ O 3 , N ( B 3 1 ) = (3; 3 , 0 , 0). Quality of decision B 3 1 is depicted in Fig. 11. The re sultant comp osite D A is the following: S 3 1 = A 3 1 ⋆ B 3 1 = ( M 3 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 2 ⋆ O 3 ). Fig. 13 depicts information system a nd co m- po site decisions fo r example 3. 3.6. T ow ards Analysis of Decisions T able 7 summar izes the r esultant comp osite decisions for three c o nsidered applied exa mples ab ov e and it is a ba sis to analyze a nd/ or com- pare the co r resp onding resultant decisio ns. Evidently , cer ta in requir e men ts and constra ints lead to sp ecific results. F or example, in the corp o- rate applicatio ns maintenance requirements can be imp ortant, in the a cademic applica tions p er- formance requir ement s can b e often cruc ia l ones . In the ar ticle, numerical r esults hav e o nly illus- trative character to explain the metho dological approach (i.e., steps o f so lving scheme). The de- sign and usage o f sp ecial a pproaches to analy- sis a nd compar ison of different r esultant applied decisions is a pr osp ective topic for future stud- ies (e.g., m ulticriteria compa r ison, stability anal- ysis). Fig. 13. Academic applica tion ① S = A ⋆ B S 3 1 = A 3 1 ⋆ B 3 1 = ( M 3 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 2 ⋆ O 3 ) Hardware A = M ⋆ E Soft ware B = W ⋆ D ⋆ O ✉ B 3 1 = W 1 ⋆ D 2 ⋆ O 3 ✉ A 3 1 = M 3 ⋆ E 2 r r r r r M E W D O Op eration system O 1 (3) O 2 (2) O 3 (1) O 4 (2) O 5 (1) DBMS D 1 (3) D 2 (1) D 3 (2) W eb server W 1 (1) W 2 (2) W 3 (3) W 4 (3) W 5 (3) Server for applica- tions E 1 (3) E 2 (1) E 3 (2) E 4 (3) Server for DBs M 1 (3) M 2 (2) M 3 (1) T able 7 . Resultant comp osite decisions # Comp osite DA s 1. 2. 3. S 1 1 = A 1 1 ⋆ B 1 1 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 3 ) S 1 2 = A 1 1 ⋆ B 1 2 = ( M 2 ⋆ E 2 ) ⋆ ( W 5 ⋆ D 2 ⋆ O 2 ) S 1 3 = A 1 1 ⋆ B 1 3 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 2 ⋆ O 5 ) S 2 1 = A 2 1 ⋆ B 2 1 = ( M 1 ⋆ E 1 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 5 ) S 2 2 = A 2 1 ⋆ B 2 2 = ( M 1 ⋆ E 1 ) ⋆ ( W 2 ⋆ D 3 ⋆ O 2 ) S 2 3 = A 2 1 ⋆ B 2 1 = ( M 2 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 3 ⋆ O 5 ) S 2 4 = A 2 1 ⋆ B 2 2 = ( M 2 ⋆ E 2 ) ⋆ ( W 2 ⋆ D 3 ⋆ O 2 ) S 3 1 = A 3 1 ⋆ B 3 1 = ( M 3 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 2 ⋆ O 3 ) 3.7. Usage of Multipl e Choice Problem In this case estimates of compatibility are not used and the mo del is mor e simple. Here we co n- sider the g r eedy heuristic for a pplied example 1 (communication serv ice provider). Let us com- pute for ea ch D A( µ ) a prior it y r ( µ ) by three cri- teria C 2 , C 3 , and C 4 . After that it is p os sible to get for each DA the v alue (as “rela tive utility”) λ ( µ ) = ( b r − r ( µ )) /z µ (where b r = max µ { r ( µ ) } and z µ is the estimate of cost for DA( µ ) by criterion C 1 ). As a re s ult, we ca n get a linea r or dering 9 of all DAs by λ ( µ ) to get the n umber of linear order π ( µ ). T a bles 8 and 9 contains es timates r ( µ ), λ ( µ ), a nd π ( µ ). T able 8. Ordering D As r ( µ ) λ ( µ ) π ( µ ) M 1 M 2 M 3 E 1 E 2 E 3 E 4 W 1 W 2 W 3 W 4 W 5 3 0 . 0 0 13 2 0 . 2 0 11 1 0 . 3 3 7 3 0 . 0 0 14 1 0 . 3 3 8 2 0 . 2 0 12 3 0 . 0 0 15 1 2 . 0 0 1 2 0 . 2 5 10 3 0 . 0 0 16 3 0 . 0 0 17 3 0 . 0 0 18 T able 9. Ordering D As r ( µ ) λ ( µ ) π ( µ ) D 1 D 2 D 3 O 1 O 2 O 3 O 4 O 5 1 0 . 3 3 9 1 0 . 4 0 6 3 0 . 0 0 19 3 0 . 0 0 20 3 0 . 5 0 5 1 1 . 0 0 2 2 1 . 0 0 3 2 1 . 0 0 4 As a result, the following s olutions ar e ob- tained: (1) total cost constr aint ≤ 15: e S 1 1 = M 1 ⋆ E 2 ⋆ W 1 ⋆ D 2 ⋆ O 3 ; (2) total cost constraint ≤ 18: e S 1 2 = M 2 ⋆ E 2 ⋆ W 1 ⋆ D 2 ⋆ O 3 ; and (3) total cost constra in t ≤ 19: e S 1 3 = M 3 ⋆ E 2 ⋆ W 1 ⋆ D 2 ⋆ O 3 . 3.8. Desig n of System T ra jectory The scheme of m ultistage design co nsists of tw o phases (Fig. 14): 1. design o f c o mpo site DAs for each time stage (HMMD); 2. design of a system tra jectory based on D As which were obtained at phase 1 (HMMD). Note a ch ange of elements into the tra jector y ca n require s ome efforts, and it is necessary to s o lve an additional top-level comp o- sition problem (phase 2) as follows: Combine a tr aje ctory (i.e., sele ction of a system solution at e ach stage ) while taking into ac c ount quality of c omp osite DAs at e ach stage and a c ost of the c omp onent changes. Fig. 14. Illustration of multistage design ✄ ✄ ✄ ❈ ❈ ❈ Phase 1 Stage 1 ✄ ✄ ✄ ❈ ❈ ❈ Phase 1 Stage 2 ✄ ✄ ✄ ❈ ❈ ❈ Phase 1 Stage 3 Comp osite D As ✟ ✟ ❍ ❍ ✁ ✁ ✁ ❆ ❆ ❆ Phase 2: Design of tra jector y T ra jectory ❍ ❍ ✞ ✝ ☎ ✆ ✘ ✘ ✘ ✘ ✿ ✞ ✝ ☎ ✆ ❍ ❍ ❍ ❍ ❥ ✞ ✝ ☎ ✆ Fig. 15. Communication provider (stage 2) ① S = A ⋆ B b S 1 1 = b A 1 1 ⋆ b B 1 1 = ( M 3 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 1 ⋆ O 3 ) b S 1 2 = b A 1 1 ⋆ b B 1 2 = ( M 3 ⋆ E 2 ) ⋆ ( W 5 ⋆ D 1 ⋆ O 3 ) Hardware A = M ⋆ E Soft ware B = W ⋆ D ⋆ O ✉ b B 1 1 = W 1 ⋆ D 1 ⋆ O 3 b B 1 2 = W 5 ⋆ D 1 ⋆ O 3 ✉ b A 1 1 = M 3 ⋆ E 2 r r r r r M E W D O Op eration system O 1 (3) O 2 (3) O 3 (1) O 4 (3) O 5 (2) DBMS D 1 (1) D 2 (2) D 3 (3) W eb server W 1 (2) W 2 (1) W 3 (3) W 4 (3) W 5 (2) Server for applica- tions E 1 (3) E 2 (1) E 3 (2) E 4 (3) Server for DBs M 1 (3) M 2 (2) M 3 (1) This pro blem (tra jector y desig n) is presented in ([30], [31]). Here example 1 (communication service provider) is considered for thr ee stages. Stage 1 corresp onds to Fig. 9 with s olutions S 1 1 , S 1 2 , and S 1 3 . F or stage 2 (near future) and stage 3 (future) other weigh ts of criteria a re used: stage 2: − 1, 3, − 1, a nd 3; stage 3: − 1, 5, − 3 , and 5 . Fig. 1 5 and 16 depict results for stag es 2 a nd 3. The comp osite DA s for stage 2 ar e the follow- ing: b A 1 1 = M 3 ⋆ E 2 , N ( b A 1 1 ) = (3; 2 , 0 , 0); b B 1 1 = W 1 ⋆ D 1 ⋆ O 3 , N ( b B 1 1 ) = (3; 2 , 1 , 0); b B 1 2 = W 5 ⋆ D 1 ⋆ O 3 , N ( b B 1 2 ) = (3; 2 , 1 , 0); b S 1 1 = ( b A 1 1 ⋆ b B 1 1 ) = ( M 3 ⋆ E 2 ) ⋆ ( W 1 ⋆ D 1 ⋆ O 3 ); 10 b S 1 2 = ( b A 1 1 ⋆ b B 1 2 ) = ( M 3 ⋆ E 2 ) ⋆ ( W 5 ⋆ D 1 ⋆ O 3 ). The co mpo site DAs for stag e 3 are the follow- ing: A 1 1 = M 3 ⋆ E 2 , N ( A 1 1 ) = (3; 2 , 0 , 0); B 1 1 = W 2 ⋆ D 2 ⋆ O 2 , N ( B 1 1 ) = (3; 2 , 1 , 0); S 1 1 = ( A 1 1 ⋆ B 1 1 ) = ( M 3 ⋆ E 2 ) ⋆ ( W 2 ⋆ D 2 ⋆ O 2 ). Fig. 16. Communication provider (stage 3) ① S = A ⋆ B S 1 1 = A 1 1 ⋆ B 1 1 = ( M 3 ⋆ E 2 ) ⋆ ( W 2 ⋆ D 2 ⋆ O 2 ) Hardware A = M ⋆ E Soft ware B = W ⋆ D ⋆ O ✉ B 1 1 = W 2 ⋆ D 2 ⋆ O 2 ✉ A 1 1 = M 3 ⋆ E 2 r r r r r M E W D O Op eration system O 1 (3) O 2 (2) O 3 (1) O 4 (2) O 5 (1) DBMS D 1 (2) D 2 (1) D 3 (3) W eb server W 1 (2) W 2 (1) W 3 (3) W 4 (3) W 5 (3) Server for applica- tions E 1 (3) E 2 (1) E 3 (2) E 4 (2) Server for DBs M 1 (3) M 2 (2) M 3 (1) Fig. 1 7 depicts systems solutions at thr ee stages and an exa mple of the res ultant system tra jectory: α = < S 1 2 , b S 1 2 , S 1 1 > . Fig. 17. Illustration for s ystem tra jectory ✲ T Stage 1 Stage 2 Stage 3 α : ✲ ✘ ✘ ✘ ✘ ✘ ✘ ✿ S 1 1 S 1 2 S 1 3 b S 1 1 b S 1 2 S 1 1 4. Conclus ion and F uture Rese arc h In the pap er, a new mo dular approa ch to c om- po se a co nfiguration of applied W eb-based sys- tems is sug gested. The appro ach is based on Hierarchical Morphological Multicr iteria Design (HMMD) of mo dular systems a nd is illustr a ted by three simplified applied examples. In HMMD a sp ecia l lattice-based discr ete space is used to ev aluate quality of the resultant comp osite sys- tem decisions or sy stem configuratio ns. The lat- tice ab ov e int egrates ordinal quality of system ele- men ts and o rdinal q uality of co mpa tibilit y among the system e le men ts. Note, the system structure in HMMD is con- sidered as a tree. This is useful from the fo llow- ing viewp oints: (i) it often allows to construct solving schemes and/o r s o lving alg o rithms with a po lynomial complexity; more generalized system structures lead to NP- hard or / and NP-complete problems; (ii) tree- like structures are more eas y and under standable for r eaders and end-user s , and it is very impo rtant to facilitate co mprehen- sion o f a new metho dolo gy at the 1st steps via simplified structures; and (iii) tree-like struc- tures can b e used as a basis for ex amination of more complica ted s ystem struc tur es (e.g., hierar- chies) and approximation of the complicated sys- tem str uctures by tre e-like str uctures is a n imp or- tant underlaying a pproach in so lving pr o cesses. In the future it may be reasonable to consider the following r esearch dir ections: 1. extensio n of the considere d system architec- ture (i.e., examina tio n of hierarchical structures instead of trees ); 2. analyzing some issues of system adapta bilit y and upgra deability; 3. examina tion of sp ecia l new appr o aches to analysis/ comparison of the resultant decisio ns; 4. usage of the desc r ib e d lattices o f integrated system quality for other combinatorial problems which lead to comp osite solutions (e.g., kna psack problem, multiple choice problem); 5. usa ge of a more c o mplicated la ttice- based discrete spac e of system q uality that in v olves po set-like scale for element compatibility as it was suggested in ([2 9], [31]); 6. usa ge of fuzzy set appr oaches and AI tech- niques; and 7. examination of other netw or k applica tions. REFERENC ES 1. A. Agg arwal, (Ed.), W eb-Bas ed Lear ning and T eaching T echnologies: Opp ortunities a nd Challanges. IGI Globa l, Her shey , P A, 2 000. 2. I.B. Arpinar , R. Zhang, B. Aleman-Meza, A. Maduko, Ontology-driven W e b services com- po sition platform, Informatio n Systems and E-Business Mana g ement 2(2) (2005 ) 1 75-19 9. 11 3. S. Battacharjee, R. Ramesh, S. Zio nt s, A de- sign framework fo r e-business infrastructur e int egratio n and resource manag emen t, IEEE T rans. on SMC, Part C 31(3) (2001) 304 – 319. 4. R. Bellazzi, S. Montani, A. Riv a, M. Ste- fanelli, W eb- based telemedicine sys tems for home-care: tec hnical issue s and ex per i- ences, Computer Metho ds and P rogra ms in Biomedicine 64(3) (20 01) 175- 1 87. 5. B. Benatalla h, M. Dumas, M.-C. F auvet, F.A. Rabhi, Q.Z. Sheng, Overview of some pat- terns for ar chit ecting and managing co mpo s - ite w eb services, ACM SIGecom Exchanges 3(3) (2002) 9 - 16. 6. B. B e natallah, Q.Z. Sheng, M. Dumas, The self-serv environmen t for web services compo - sition, IEEE Internet Computing 7(6) (2003) 40-48 . 7. M.B. Blake, H. Gomaa, Agent-based comp o- sitional approa ches to s ervices-ba sed cr oss- organiza tional workflow, Decision Suppo rt Systems 40 (1) (200 4) 31-50 . 8. P . Br us ilovsky , J. Eklund, E. Sc hw ar z, W eb- based education for all: A to o l for devel- opment ada ptiv e courseware, Computer Net- works and ISDN Systems 30(1-7) (199 8) 29 1- 300. 9. J.J . Cimino, S.A. So c r atous, P .D. Clayton, Int ernet as clinica l information system: ap- plication developmen t using the W orld Wide W eb, J . America n Medical Informatics Asso- ciation 2(5) (1 995) 273-2 8 4. 10. S. Dustdar, W. Schreiner, A survey on web services comp osition, Int. J. o f W eb and Grid Services 1(1) (20 05) 1 -30. 11. L. F eng, G. W ang, C. Zeng, R. Huang (E ds.), W eb Infor mation Systems - WISE 20 06 W ork - shops, LNCS 4256, Spring er, New Y o rk, NY, 2006. 12. P . F raterna li, T o ols a nd a pproaches for de- veloping data-intensiv e W eb applications: a survey , A CM Computing Surveys 31(3) (199 9 227-2 63. 13. M.R. Garey , D.S. J ohnson, Computers and Int ractability . The Guide to the Theory of NP-Completeness, W.H.F reeman and Com- pany , San F rancisco, 197 9. 14. H.-W. Geller sen, R. Wick e, M. Gaedke, W eb- Comp osition: an ob ject-oriented suppo rt sys- tem for the W eb eng inee ring lifecycle, Com- puter Netw orks and ISDN Systems 29(8-13 ) (1997) 1429 -1437 . 15. X. Geng, Y. Huang, A.B. Winston, Smart marketplaces: a step beyind W eb services, In- formation Systems and E-Business Manag e- men t 1(1) (200 3) 1 5 -34. 16. A. Ginige, S. Murug esan, W eb eng inee r- ing: An intro duction, IE EE Multimedia 8(1) (2001) 14-1 8 . 17. R. Heeks, Implementing and Managing e Gov- ernment. Sage Publications, Lo s Angeles, CA, 2005. 18. Y. Huang , X. Geng , A.B. Winston, Netw ork mapping serv ic es for dynamic selection o f W eb services: promises a nd challenges, Infor- mation Systems a nd E-Business Management 3(3) (2005) 2 8 1-297 . 19. D. R. Insua, G.E. Kersten, J. Rios, C. Crima, T ow a rds decision supp ort for par tic- ipatory demo cra cy , Information Systems a nd E-Business Mana g ement 6(2) (2008 ) 1 61-19 1. 20. T. Isakowitz, M. Bieb er, F. Vitali, W eb in- formation sys tems, Comm. o f the ACM 41(7) (1998) 78-8 0 . 21. G. Kapp el, B. Pr oll, S. Reich, W. Retschitzeg- ger, (Eds.), W eb Engineering - The Discipline of Systematic Dev elopment of W eb Applica- tions. J.Wiley & So ns, New Y ork, NY, 2006 . 22. R.L. Keeny , H. Raiffa, Decis ions with Multi- ple Ob jectives: Preferences and V a lue T rade- offs. J.Wiley & So ns , New Y ork, NY, 1976. 23. H. Keller er, U. Pfersch y , D. Pis inger, Kna p- sack Problems . Spr inger, Berlin, 2004 . 24. R. Kingsto n, S. Ca rver, A. Ev ans, I. T urton, W eb-based public participa tion geog raphical information systems: a n aid to lo ca l en viron- men t decision ma king, Co mputers, Environ- men t, and Urban Systems 24(2) (200 0) 109- 125. 25. P . K orhonen, J. W allenius, a nd S. Zionts, Solving the discre te multiple criteria pr ob- lems using co nv ex co nes, Ma nag. Sci. 30(11 ) (1984) 1336 –1345 . 26. M. Land, B. Fitzger ald, W e b- based sys- tems desig n: a study of co n tempo rary prac- tices and an explanato ry framework based on 12 “metho d-in-action” , Requirements Engineer - ing 12 (4 ) (200 7) 203-2 20. 27. K. Layne, J . Lee, Developing fully functional E-governmen t: A four sta ge mo del, Gov ern- men t Informa tion Quaterly 18(2) (200 1) 12 2- 136. 28. M.Sh. Levin, Combinatorial Engineer ing of Decomp osable Systems. Kluw er Academic Publishers, Dordr ech t, 1998. 29. M.Sh. Levin, System synthesis with morpho- logical clique problem: fusion of subsys tem ev aluation decisions, Information F us io n 2(3) (2001) 225- 2 37. 30. M.Sh. Levin, Mo dula r system synthesis: Ex- ample for compo site pack aged softw ar e, IEE E T rans. on SMC, Part C 35 (4) (2005 ) 544-5 53. 31. M.Sh. Levin, Compo site Systems Decisions, Springer, New Y o rk, 2 006. 32. M.Sh. Levin, Combin atoria l optimization in system configura tion design, Autom.&Rem. Control 70(3) (200 9) 519-5 61. 33. M.Sh. Levin, S.Y u. Shar ov, Hierar chical mor- phological comp osition of W e b- hosting sy s - tem, J . of Integrated Design and Pro c ess Sci- ence 13 (1 ) (200 9 ) 1-14. 34. M.Sh. Levin, A.V. Safonov, T ow ards mo d- ular r edesign o f netw orked s y stem. In: 2nd Int . Congr ess on Ultra Mo der n T elecomm. & Control Sy s t. a nd W or kshops ICUMT-20 10, Moscow, (2010 ) 1 09-11 4. 35. F. Leymann, D. Roller, M. Schmidt, W eb ser- vices and business pro cess management, IBM Systems J. 41 (2) (2002) 198– 211. 36. W.-L. Lin, C.-C. Lo , K.-M. Chao, M. Y o unas, Consumer-centeric QoS-aware selection o f web services, J. of Co mputer and Sy stem Sci- ences 74(2) (200 8) 21 1-231 . 37. T. Madhusudan, N. Uttamsingh, A decla r - ative appro ach to comp osing web se r vices in dynamic en vironments, Decision Suppor t Systems 41 (2) (200 6) 325–3 57. 38. I. Ma nolescu, M. Bra m billa, S. C e ri, S. Co - mai, P . F raterna li, Mo del-driven des ign and deploymen t of s ervice-enable d web applica- tions, ACM T rans. on Internet T e chnology 5(3) (2005 439 -479. 39. S. Mar tello, P . T oth, Knapsack Problem: Algorithms a nd Computer Implementation. J.Wiley & Sons , New Y ork, NY, 19 90. 40. C. McCormack, D. Jones, Building a W eb- Based E ducation System. J .Wiley & Sons, New Y ork, NY, 1 997. 41. N. Milanovic, M. Male k , Current solutions for web serv ice c o mpo sition, IEEE Internet Com- puting 8(6) (200 4) 51 -59. 42. M.J. Mo o n, The evolution of E-gov ernment among municipalities, Public Administration Review 62(4) (200 2) 424-4 33. 43. S.-C. Oh, D. Lee, S.R.T. K umara, A co m- parative illustra tion o f AI planning-ba s ed web s ervices comp os ition, ACM SIGecom Ex- changes 5(5) (200 5) 1-10. 44. M.P . Papazoglou, W.-J. v an den Heuvel, Service-or iented design a nd developmen t metho dology , Int. J. W eb Enginee r ing and T echnology , 2 (4 ) (2 006) 412 -442. 45. J.E.J. P rins, Designing E -Gov ernment . 2nd ed., Kluw er Academic Publis hers, Dor drech t, 2006. 46. I. Reinhartz- Berger, D. Dor i, S. Katz, OPM/W eb-o b ject-pro cess methodolo gy for developing W eb applications, Annals of Soft- ware Engineer ing 13 (1-4) (2 002) 141 -161. 47. J. Reyno lds , The Co mplete E-Commerc e Bo ok: Design, B uild and Maintain a Suc- cessful W eb-Based Busines s. 2nd ed., CMP Bo oks, Gilroy , CA, 2004. 48. B. Roy , Multicriteria Metho dology for De- cision Aiding. Kluw er Academic P ublishers, Dordrech t, 199 6. 49. T.L. Sa aty , The Analytic Hierar ch y P ro cess. MacGraw-Hill, New Y or k, NY, 1988 . 50. W.L. Sc herlis, J. Eisenberg, IT r esearch, in- nov ation, and e-g ov er nment , Commun. of the A CM 46(1 ) (200 3 ) 67-68. 51. K.-D. Schew e, B. Thalheim, A. Binerma nn- Zdanowitcz, R. Kaschek, T. Kuss, B. Tschiedel, A conceptual view of W eb-ba sed E-learning , Educa tion and Infor mation T ech- nologies 10(1- 2) (200 5) 8 3 -110. 52. M. Shaw, R. Bla nning, T. Stra der, A.B. Whinston, Handbo ok on Electr o nic Com- merce. Springer, New Y ork, NY, 20 00. 53. H.A. Simon, A. Newell, Heuristic problem solving: The next adv ance in op er a tions re- search, Op erations Research 6(1) (19 58) 1 -10. 13 54. J.P . Sousa, V. Poladian, D. Ga rlan, B. Schm erl, M. Shaw, T ask-ba sed adaptation for ubiquitous computing, IEEE T rans. on SMC - Part C 36 (3) (2006) 328– 340. 55. Y.-W. Sung, L. Zhou, The effect of allia nce information on W eb se rvice comp ositio n, In- formation Systems and E-Business Manag e- men t 6(4) (200 4) 403-4 17. 56. K. T ak a , E . Lia ng, Analy sis and design o f W eb-based information systems, Computer Net works and ISDN Systems 29 (8-13) (19 97) 1167- 1180. 57. H. T ang , Y. W u, J.T . Y ao, G. W a ng, CUP- TRSS: A web-based resea rch supp ort s ys- tem, In: J.T. Y a o, P . Lingr as (Eds.), Pr o c. of t he W orkshop on Applic ations, Pr o ducts ansd Servic es of Web-b ase d Su pp ort Systems WSS03 , Halifax, Cana da , pp. 21-2 8, 2003 . 58. P . Xiong, Y. F an, M. Zho u, A Petri net ap- proach to a nalysis and comp os iito n of W eb services, IEE E T rans. SMC - Part A 40(2) (2010) 376- 3 87. 59. Y.Y. Y a o, A framework for W eb-based re- search suppo rt sy s tems, in: Pro c. of 23rd I nt. Computer Softw ar e and Application Conf. COMPSAC 2 003, Da llas, TX, pp. 60 1-606 , 2003. 60. T. Y u, K.-J. Lin, Service selection alg orithms for W eb service s selection with end-to- end QoS constr aints, Information Systems and E- Business Manag emen t 3(2) (2005) 103- 126. 61. M. Y ounas, K.-M. Cha o, C. Laing, Co mpo si- tion o f web service in distributed se rvice o r i- ent ed design activities, Adv a nced E ngineering Informatics 19(2 ) (2 0 05) 143-15 3. 62. F. Zwicky , Discov ery Inv en tion, Resea rch Through the Morpholog ical Appr oach. McMillan, New Y o rk, 1969 .