Reputation Systems: A framework for attacks and frauds classification

Copy igh © 2023 by Au ho /s and Licensed by IADITI. This is an open access a icle dis ibu ed unde he C ea i e Commons A ibu ion License which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed. Jou nal o In o ma ion Sys ems Enginee ing and Managemen 2023, 8(1), 19218 e-ISSN: 2468-4376 h ps://www.jisem-jou nal.com/ Li e a u e Re iew Repu a ion Sys ems: A amewo k o a acks and auds classi ica ion Rui Humbe o Pe ei a1*, Ma ia José Angélico Gonçal es2, Ma a Alexand a Gue a Magalhães Coelho3 1 CEOS.PP/ Uni e si y o Maia, 4475-690 Maia, Po o, Po ugal 2 CEOS.PP, ISCAP, Poly echnic o Po o, 4465-004 Mamede In es a, Po o, Po ugal 3 ISCAP, Poly echnic o Po o, 4465-004 Mamede In es a, Po o, Po ugal * Co esponding Au ho : hpe ei [email protected] Ci a ion: Pe ei a, R. H., Gonçal es, M. J., & Magalhães, M. A. G. (2023). Repu a ion Sys ems: A amewo k o a acks and auds classi ica ion. Jou nal o In o ma ion Sys ems Enginee ing and Managemen , 8(1), 19218. h ps://doi.o g/10.55267/iad .07.12830 ARTICLE INFO ABSTRACT Recei ed: 05 Dec 2022 Accep ed: 13 Jan 2023 Repu a ion and ecommending sys ems ha e been widely used in e- comme ce, as well as online collabo a i e ne wo ks, P2P ne wo ks and many o he con ex s, in o de o p o ide us o he pa icipan s in ol ed in he online in e ac ion. Based on a epu a ion sco e, he e-comme ce use eels a sense o secu i y, leading he pe son o us o no when buying o selling. Howe e , hese sys ems may gi e he use a alse sense o secu i y due o hei gaps. This a icle discusses he limi a ions o he cu en epu a ion sys ems in e ms o models o de e mine he epu a ion sco e o he use s. We in end o con ibu e o he knowledge in his ield by p o iding a sys ema ic o e iew o he main ypes o a ack and aud ound in hose sys ems, p oposing a no el amewo k o classi ica ion based on a ma ix o a ibu es. We belie e such a amewo k could help analyse new ypes o a acks and aud. Ou wo k was based on a sys ema ic li e a u e e iew me hodology. Keywo ds: e-comme ce, us , epu a ion sys ems. INTRODUCTION Comme cial ansac ions equi e ha he pa icipan s us each o he . This us gi es pa icipan s a no ion abou he isk, hus, leading he pe son o conclude, o no , he ansac ion. In e-comme ce, his sense o isk/secu i y based on us is much mo e c i ical, pa icula ly when he e is no p io knowledge o he pe son on he o he side. In e-comme ce, he use s mus be awa e o some in e ela ed aspec s ega ding he o he pa icipan in he ansac ion: (1) he eal iden i y; (2) hones y; and (3) he quali y o he p oduc /se ice. Rega ding he eal iden i y o he pe son, his can be a complex p oblem in on- online ma ke places, due o he inco ec , o absence, o an e ec i e iden i y alida ion o he pe son associa ed wi h he use (p o iles o buye , selle , o bo h) in ha e-comme ce pla o m. On he o he hand, ega ding he hones y o he use and he quali y o he p oduc /se ice, hese issues ha e been add essed by means o epu a ion and ecommenda ion sys ems. In he case o B2C, when he e is a company o b and associa ed wi h he pla o m, he buye s’ us is mainly based on hei p io knowledge abou he c edibili y o ha company, b and, o p oduc /se ice quali y, which may be addi ionally complemen ed by a epu a ion and ecommenda ion sys em. Repu a ion and us a e dis inc and in e ela ed concep s. Jøsang e al. (2007) dis inguish “Reliabili y us ”, and “Decision us ”. In he o me concep , he au ho uses he de ini ion p oposed in (Gambe a, 1988). Howe e , he au ho s conside he concep o us o be mo e complex, e e ing o his as such: Decision us as: “T us is he ex en o which one pa y is willing o depend on some hing o somebody in a gi en si ua ion wi h a eeling o ela i e secu i y, e en hough nega i e consequences a e possible” (Jøsang e al., 2007, p. 620). Rega ding o he concep o epu a ion, he same au ho s de ine epu a ion acco ding o he Concise Ox o d Dic iona y, as: “Repu a ion is wha is gene ally said o belie ed abou a pe son's o hing's cha ac e o s anding” (Jøsang e al., 2007, p. 620). Thus, du ing an e-comme ce ansac ion, us and epu a ion a e wo subjec i e concep s on which he decision o conclude is based, accep ing a ce ain le el o isk. This obse a ion le s us iden i y he i s limi a ion o epu a ion sys ems. Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 2 / 10 As p e iously men ioned, he epu a ion o a p oduc o use can be de e mined by means o epu a ion sys ems. In he p esen wo k, we ocus on he epu a ion o he use s, since i is a dis inc p oblem o p oduc epu a ion/ ecommenda ion, despi e he ac ha hey sha e common p inciples. We should no ice ha he e a e o he ypes o h ea s, such as he ones in he ield o cybe secu i y, a in as uc u e and ne wo k le els, ha we do no conside in his wo k, because hey a e dis inc om he epu a ion sys ems as a mean o p o ide us o he use s. Repu a ion sys ems ha e been widely used in pee - o-pee (P2P) ne wo ks o es ablishing a use 's epu a ion sco e based on wha he gi es o he ne wo k and ge s, in e ms o he c i e ia o choosing iles (Damiani e al., 2002). C owdsou cing pla o ms also could apply simila p inciples, bu in his case, i is o es ablish a a ing sco e o he use ’s epu a ion in e ms o he alue o asks (Gong e al., 2021). In heo y, all collabo a i e ne wo k en i onmen s could bene i om epu a ion sys ems as a mean o p o ide us . In he con ex o ou wo k, e-comme ce ma ke places, such as Amazon and eBay, also apply epu a ion sys ems in o de o enable hei use s o a e o he use s. This is o say, ypically a buye a es a selle a e a inished ansac ion. Ou ocus will be on his la e case, in which he us o an e-comme ce use , a he momen o deciding abou a ansac ion, is based on he epu a ion o he o he pa icipan . We expec o con ibu e o he inc ease o knowledge in he ield, by answe ing he ollowing esea ch ques ion: Wha has been esea ched abou he mos common a acks and auds on e-comme ce pla o ms ha may a ec he use ’s us based on epu a ion sys ems? In o de o answe his ques ion, we choose a sys ema ic li e a u e e iew as me hodology using h ee ci a ions da abases: Web o Science, Scopus, and Google Schola . Nex , we p oposed a new amewo k o a acks and aud classi ica ion. We belie e ha such a amewo k can be a use ul ool o analysing new ypes o a acks and aud, hus, con ibu ing o he knowledge in he ield. In he nex sec ion, we p esen he concep s ela ed o epu a ion sys ems ocusing ou discussion on e-comme ce communi ies. In he sec ion Me hodology, we p esen he me hodology used, based on a li e a u e e iew. The discussion, abou he ypes o a acks, and auds and how hese secu i y issues a e classi ied in he li e a u e, is p esen ed in he sec ion Discussion. A e , we discuss ou p oposal o a new amewo k o classi ica ion and pe o m i s alida ion based on he scena ios ound in he li e a u e. We inish he pape p esen ing ou inal ema ks and he ocus o ou u u e wo k. BACKGROUND ON REPUTATION SYSTEMS In o de o con ex ualize he eade as o he weaknesses o epu a ion sys ems, we conside i impo an o cla i y some concep s, p inciples and s a egies ha a e usually adop ed in hese sys ems. The scope o he p esen pape is epu a ion sys ems o e- comme ce use s, howe e , he discussion o he ollowing aspec s and p inciples is gene alizable o o he ypes o collabo a i e ne wo ks, as well as o p oduc ecommenda ion sys ems in e-comme ce. Jøsang e al. (2007) dis inguish epu a ion sys ems om ecommenda ion sys ems e e ing o hem as collabo a i e sanc ioning and collabo a i e il e ing. In epu a ion sys ems (i.e., collabo a i e sanc ioning), he use is judged a e a ansac ion. This con as s wi h he ecommenda ion sys ems (i.e., collabo a i e il e ing), which a e based on di e en as es and he subjec i e opinions o he use s. Hend ikx e al. (2015) p opose a axonomy o epu a ion sys ems, which, a he i s , classi ies he epu a ion sys ems as implici and explici . Acco ding o he au ho s, implici epu a ion sys ems a e sys ems ha do no ha e a de ined epu a ion sys em, al hough epu a ion in o ma ion is used by i s membe s o assis in decision making. Examples o such epu a ion app oaches exis in social ne wo ks (e.g., Facebook and LinkedIn), in which we can ex ac some deg ee o us om he in o ma ion ga he ed h ough iends o iends. Ano he example is Google’s sea ch engine, in which he o de o he sea ch esul s ep esen s a anking o pages, based on he epu a ion o each page. The epu a ion is de e mined by he numbe o links ha poin o he page, and whe e he links o igina e (Hend ikx e al., 2015). On he o he hand, explici epu a ion sys ems ha e implemen ed a model ha enables he es ima e o a epu a ion using a sco e. The la e a e he ocus o he p esen pape . The epu a ion es ima ion model encompasses he h ee dimensions (1) sou ces and ypes, o da a, (2) he algo i hm based on ma hema ical calcula ions and (3) he ype o ou pu o he epu a ion sco e and how i is dissemina ed. In (Ho man e al., 2009), he au ho s e e o hese h ee dimensions as Fo mula ion, Calcula ion and Dissemina ion. The accu acy o he epu a ion sco e depends on he e ec i eness o he model, as well as he ypes o h ea s ha i is immune o. Ano he cha ac e is ic o he model is i s a chi ec u e, which is a cen al o dis ibu ed sys em. On he second le el o he axonomy p oposed by Hend ikx e al. (2015) he iden i ied aspec s a e sys ema ized, as well as discussed in o he wo ks (Ho man e al., 2009). In he ollowing sec ion, we will b ie ly discuss hese aspec s. Sou ces and ype o da a The sou ces o in o ma ion ha suppo he o mula ion o epu a ion p o ide he aw da a ha eeds he algo i hm implemen ed a he compu a ional le el. This da a is di e se and complemen a y o each o he . We can g oup i in o wo main sou ces: Manual and au oma ic. Ho man e al. (2009) sugges he ollowing classi ica ion o sou ces o in o ma ion:  Manual sou ces a e ob ained om human eedback, usually in he o m o use a ings o o he iden i ies based on he expe ience o a single ansac ion such as he eedback in a ma ke place, a speci ic ime pe iod o a bi a y eedback;  Au oma ic sou ces a e ob ained au oma ically ei he ia di ec o indi ec obse a ion. o Di ec obse a ions p o ide da a ega ding di ec ly obse ed e en s such as he success o ailu e o Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 3 / 10 in e ac ion, he di ec obse a ions o chea ing, o in he case o he P2P ne wo k, he measu emen o esou ce u iliza ion by neighbou s; o Indi ec obse a ions a e ob ained second-hand o a e in e ed om i s -hand in o ma ion. These sou ces o in o ma ion could in luence he epu a ion posi i ely, nega i ely, o neu ally, acco ding o i s le el o ele ance, calcula ed by he algo i hm implemen ed in he sys em. Rega ding o he axonomy o he da a ypes, his da a can be bina y, disc e e o con inuous. O he ypes a e possible bu could es ic he ype and accu acy o he esul s o he compu a ional algo i hm. In o de o achie e a quan i a i e me ic on use epu a ion, he quali a i e-inpu da a ypes could equi e he con e sion o a quan i a i e alue. Fo example, ee ex e iews can complemen a nume ical sco e bu equi e some manual analyses, which a e no iable, o p ocessed by means o a i icial in elligen mechanisms in o de o con e o a quan i a i e a iable. The compu a ional app oach applied o calcula e he epu a ion alue o he a ge The esul o he algo i hm ha compu es he da a, ob ained om manual o au oma ic da a sou ces, consis s o a me ic ega ding he epu a ion o a use , which in gene al is a quan i a i e one. Se e al algo i hm-based app oaches o epu a ion models can be ound in he li e a u e (C. Della ocas, 2000; Hend ikx e al., 2015; Panagopoulos e al., 2017). The main challenge placed on he designe o such algo i hms is choosing which a e he inpu a iables and hei espec i e weigh s o he ou pu me ic. The empo al a iable is ano he ac o ha some models conside in hei ma hema ical analyses, in which he impac (o ele ance) o he eedback, o di ec /indi ec obse a ion, dec eases wi h ime, which is da a ageing (Hend ikx e al., 2015). Rega ding new use s, who do no ha e his o ical eco ds o ansac ions, and o which an ini ial epu a ion sco e can be es ima ed, de aul neu al alue o epu a ion is ypically se o h. Panagopoulos e al. (2017) discuss some app oaches o dealing wi h newcome s, as well as o he economic and social issues such as inducing use pa icipa ion, using incen i es, and dealing wi h ecip oci y and e alia ion. O he app oaches such as he ones based on machine lea ning (Wang e al., 2020), o au oma ic de ec ion o alse o un ai a ings, o o he s blockchain-based (Zul iqa e al., 2021) app oaches, in which he inancial model is no iable o a dishones use . Fu he mo e, in dis ibu ed a chi ec u es epu a ion da a is sha ed among se e al e-comme ce pla o ms. Below, in he subsec ion 0, we de ail his discussion. The ou pu epu a ion sco e can be classi ied as ei he bina y, disc e e, o con inuous. A bina y one could ep esen i he use is epu able o no . The disc e e ou pu s, o example, one o i e s a s, de ine he le el o epu a ion, as well as he con inuous sco es, bu in his case, gi e a much ine-g ained classi ica ion. Accu acy and immuni y o he epu a ion model The accu acy o he model depends on he quali y o he inpu da a and he obus ness o he algo i hm and ma hema ical app oach. Addi ionally, he accu acy can be subjec o aud and manipula ion. These h ea s o he epu a ion sys ems ha e wo possible pu poses: o inc ease o dec ease he epu a ion o a use , based on a malicious s a egy (Kou ouli & Tsalga idou, 2012). In he Discussion Sec ion, we will examine hese h ea s in de ail, which is he ocus o he p esen pape . The incen i es o pa icipa ion in a ing he ansac ion a e one app oach o inc ease he olume o inpu da a, which is impo an o ge eliable ou pu s. Howe e , Panagopoulos e al. (2017) claim ha , al hough use pa icipa ion is necessa y o success ul eedback-based epu a ion sys ems, mos e-comme ce communi ies do no p o ide any kind o incen i es o encou age i . Tha is due o he ac ha e-comme ce pla o ms usually achie e good enough pa icipa ion h ough he mu ual exchange o a ings be ween he membe s in ol ed in he ansac ion, which akes place igh a e i s comple ion, by cou esy. The a o emen ioned app oaches, based on machine lea ning and public blockchain ne wo ks, could help o mi iga e hese p oblems. Cen alized s dis ibu ed a chi ec u e In epu a ion sys ems based on a cen alized a chi ec u e, he da a is managed only by one en i y. I a use has wo accoun s, each in a dis inc e-comme ce pla o m based on cen alized epu a ion sys ems, hen he has wo p o iles, each one wi h i s own epu a ion sco e, pe haps wo incohe en alues o epu a ion. Se e al p oposals o decen aliza ion can be ound in he li e a u e, bu o he simila p oblems eme ge (Panagopoulos e al., 2017). In ecen li e a u e (Ahn e al., 2018, 2019; Dennis & Owen, 2015; Dhakal e al., 2019; Ka ode e al., 2020; Mohe e al., 2009; Schaub e al., 2016; Zeynal and e al., 2021; Zul iqa e al., 2021), blockchain-based app oaches a e p oposed o enable a dis ibu ed a chi ec u e in epu a ion sys ems in e ms o sha ing da a. Acco ding o he au ho s, hese app oaches ensu e anspa ency and could help mi iga e some ypes o known aud. Zul iqa e al. (2021) s a e ha he cen al au ho i ies can po en ially il e , ampe , add, o ejec p oduc e iews based on hei p e e ence. Schaub e al. (2016) s a e ha , po en ially, a cen alized sys em can be abused by he cen al au ho i y. The managemen o use s’ epu a ion, based on paymen sys ems, a e also p one o manipula ion by malicious en i ies, which include he ad e ise s o owne s hemsel es, who may gi e ex emely high o low a ings on pu pose (Ahn e al., 2019; Dennis & Owen, 2015). Dhakal and Cui (2019) p esen he same a gumen s, s a ing ha he cu en cen alized sys ems a e silos and no anspa en in he e iew p ocess. Besides he lack o anspa ency, hese isola ed cen alized sys ems do no bene i om he epu a ion da a o each o he . Zeynal and e al. (2021) s a e ha i is ha d o de i e us models ha a e obus o a acks such as whi ewashing and Sybil a acks, i use s do no sha e in o ma ion. Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 4 / 10 Ka ode e al. (2020), in he con ex o a el e iew sys ems, s a e ha blockchain-based epu a ion sys ems enable consume s o be con iden ha he e iew sco e is no a ec ed by he pla o m p o ide s. Besides, he businesses can main ain he same a ing sco e ega dless o he pla o m hey ake pa in. Low-quali y e iew handling is a challenge o he global- scale e iew sys em; howe e , his p oblem can be add essed wi h au oma ic il a ion. METHODOLOGY Ou e iew can be ca ego ized as a sys ema ic e iew o he scien i ic li e a u e on secu i y p oblems in use epu a ion sys ems. Sys ema ic e iews a e a o m o me a-analysis designed o collec , in es iga e, and summa ise wha is known and wha is no known abou a “speci ic p ac ice- ela ed ques ion” (B ine e al., 2009). Sys ema ic e iews a e used ac oss a b oad ange o disciplines. Quali a i e s udies ha e es ablished a place o hemsel es wi hin he me hodologies, as e idenced by ini ia i es such as he Coch ane quali a i e me hods g oup (Dixon-Woods & Fi zpa ick, 2001) and ex books such as Sys ema ic Re iews in he Social Sciences (Pe ic ew & Robe s, 2005) and An In oduc ion o Sys ema ic Re iews (Thomas e al., 2017). In his s udy, besides conduc ing he li e a u e e iew ollowing i s p ima y objec i es acco ding o Mohe (2009) we also subs an ia e he esul s ob ained wi h a li e a u e e iew, p esen ing heo e ical pe spec i es and inno a ions om leading au ho s in he ield. Acco ding o he au ho s, he sys ema ic li e a u e e iew is ca ied ou in 3 s eps (Mohe e al., 2009). Fi s , he esea ch ques ion is de ined; his is ollowed by a esea ch p o ocol o e alua ing he selec ed scien i ic a icles. The las s ep in ol es answe ing he esea ch ques ions (in he i s s ep), based on he scien i ic a icles iden i ied as ele an (in he second s ep). Figu e 1 summa izes he s eps ollowed by he adop ed me hodology. The i s s ep o he adop ed me hodology is ela ed o he de ini ion o he esea ch ques ion o his s udy. The main esea ch ques ion in ends o iden i y he s a e o he a conce ning ou s udy cha ac e is ics. The e o e, ou esea ch ques ion can be o mula ed as ollows, as abo e men ioned in he in oduc ion sec ion: Wha has been esea ched abou he mos common a acks and auds on e-comme ce pla o ms ha may a ec he use ’s us based on epu a ion sys ems? A e he de ini ion o he esea ch ques ion, he second s ep was ela ed o he selec ion o he empi ical da a o be analysed. Da a collec ion ook place in Oc obe 2022. We did no apply any ch onological il e . In he i s phase, we ied a sepa a e sea ch o each keywo d. In Web o Science Co e Collec ion (WOS) we applied he ollowing s a egy: Sea ch: (TITLE-ABS- KEY(" epu a ion sys em" and axonomy and a ack) OR TITLE- ABS-KEY(" epu a ion sys em" and classi ica ion and a ack) OR TITLE-ABS-KEY(" epu a ion sys em" AND ype a acks)). In SCOPUS and Google Schola (GS) we ollowed he same c i e ia. This sea ch esul ed in 38 a icles selec ed om WoS, 75 a icles selec ed om Scopus and 29 selec ed om GS. The lis s we e expo ed o excel o u he analysis, and he ollowing ields we e chosen: au ho s, i le, yea , link, abs ac , and keywo ds. Figu e 1. Sys ema ic e iew s uc u e Then, we e alua ed he a icles based on he c i e ia o inclusion o de e mine hei ele ance o he s udy. An a icle had o include he sea ch e ms as he co e echnology unde analysis. This was ypically demons a ed by i s i le, abs ac , and keywo ds. We only selec ed academic pee - e iewed jou nal a icles and con e ence p oceedings and excluded o he s, namely: (a) a icles no ully a ailable, (b) a icles no a ailable in English, (c) duplica e a icles, and (d) a icles ha did no discuss secu i y issues in use epu a ion sys ems. Ou ini ial sea ch was ca ied ou in Oc obe 2022 and yielded 142 a icles. Once we elimina ed duplica es, we we e le wi h a popula ion o 96 a icles. A e , ou esea ch eam, including a p o esso and a mas e ’s s uden , e iewed his collec ion o a icles o ele ancy. In he i s ound, we assessed he a icles o ele ance based on i le, abs ac , and keywo ds. This p ocess led o he selec ion o 46 a icles. Any a icles we did no ag ee upon we e also excluded. In he second ound o e isions, we assessed he a icles based on he ull pape . We elimina ed 29 a icles ha did no p esen de ailed e e ence o a acks and auds on he epu a ion sys em. Thus, we iden i ied 17 ele an a icles o analysis, which a e lis ed in Table 1. Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 5 / 10 Table 1. Selec ed A icles Re e ence A icle Ti le (C. Della ocas, 2000) Immunizing online epu a ion epo ing sys ems agains un ai a ings and disc imina o y beha iou (C. Della ocas., 2000) Mechanisms o coping wi h un ai a ings and disc imina ion beha iou in online epu a ion epo ing Sys ems (Douceu , 2002) The Sybil A ack. In: D uschel, P., Kaashoek, F., Rows on, A. (eds) Pee - o-Pee Sys ems. (C. N. Della ocas, 2003) The Digi iza ion o Wo d - o - Mou h: P omise and Challenges o Online Feedback Mechanisms (Jøsang e al., 2007) A su ey o us and epu a ion sys ems o online se ice p o ision (Ho man e al., 2009) A su ey o a ack and de ense echniques o epu a ion sys ems (Swamyna han e al., 2010) The design o a eliable epu a ion sys em (F aga e al., 2012) A Taxonomy o T us and Repu a ion Sys em A acks (Kou ouli & Tsalga idou, 2012) Taxonomy o a acks and de ense mechanisms in P2P epu a ion sys ems— Lessons o epu a ion sys em designe s (Feng e al., 2012) Vulne abili ies and coun e measu es in con ex -awa e social a ing se ices (Yao e al., 2012) Add essing Common Vulne abili ies o Repu a ion Sys ems o Elec onic Comme ce (Sänge e al., 2015) Reusable componen s o online epu a ion sys ems ( Kou ouli & Tsalga idou, 2016) Repu a ion Sys ems E alua ion Su ey (Panagopoulos e al., 2017) Modeling and E alua ing a Robus Feedback-Based Repu a ion Sys em o E- Comme ce Pla o ms (Camilo e al., 2020) A Secu e Pe sonal - Da a T ading Sys em Based on Blockchain, T us , and Repu a ion (Zul iqa e al., 2021) E hRe iew: An E he eum - based P oduc Re iew Sys em o Mi iga ing Ra ing F auds (Zeynal and e al., 2021) A Blockchain - Enabled Quan i a i e App oach o T us and Repu a ion Managemen wi h Spa se E idence DISCUSSION Types o ulne abili ies and a acks Della ocas (2000, 2000, 2003) ocused his wo k on he audulen beha iou o use s when a ing o he s, in online ading communi ies. The au ho has iden i ied wo scena ios: (1) un ai buye a ings and (2) disc imina o y selle beha iou . In he case o un ai buye a ings, he e a e wo scena ios: (1.a) Un ai ly high a ings (“ballo s u ing”) and (1.b) Un ai ly low a ings (“bad-mou hing”). In bo h cases, a selle colludes wi h a g oup o buye s, bu in he i s case, “ballo s u ing” o inc ease his own epu a ion and, o “bad-mou hing” o damage he epu a ion o o he selle s, his compe i o s. This can, po en ially, inc ease his o de s and dec ease he o de s o his compe i o s. The second g oup o scena ios ela es o disc imina o y selle beha iou : (2. a) ha m ul disc imina ion - 1 h ps://en.wikipedia.o g/wiki/Sybil_(Sch eibe _book) a selle p o ides good se ice o he majo i y o buye s, excep a ew speci ic ones ha hey “don’ like”. This kind o ac ion does no ha e a g ea impac on he selle ’s epu a ion i he numbe o “ ic ims” is small; (2.b) posi i e disc imina ion – In his s a egic ac ion, he selle can po en ially inc ease his own epu a ion by p o iding an excep ional quali y se ice o a ew buye s and a e age quali y se ice o he es o he buye s. I he numbe o p i ileged cus ome s is su icien ly la ge, his ac ion is equi alen o Ballo s u ing, bu in an impe cep ible way, wi hou ha ing o conspi e wi h hi d pa ies. Douceu (2002), in he con ex o P2P ne wo ks, ollowing he sugges ion o B ian Zill, inspi ed by a book1 da ed 1973 wi h he same name, coined he e m Sybil a ack. In his ype o a ack on ne wo k se ices, an en i y o ges mul iple iden i ies in he sys em in o de o inc ease his in luence. In he con ex o e- comme ce, he ypical scena io is an en i y ha can in luence he epu a ion o a use , o p oduc by using mul iple iden i ies o inse un ai a ings in o he sys em. A simila ype o a ack, also based on a lack o e ec i e iden i y managemen , is Whi ewashing. In his ype o a ack, a dishones buye is able o whi ewash i s low us wo hiness by s a ing a new accoun wi h he ini ial us wo hiness alue o using some ulne abili y in he sys em. In (F aga e al., 2012), he au ho s dis inguish be ween Re-en y and Whi ewashing. Acco ding o he au ho s, in he o me case, i a acke s can c ea e new “iden i ies” eely, his p esen s he oppo uni y o emo e a bad epu a ion by c ea ing a new iden i y. In Whi ewashing, he a acke s can epai hei epu a ion comple ely by using some sys em ulne abili y. In ou opinion, Re-en y is a speci ic case o a Whi ewashing a ack, in which he malicious use uses ulne abili ies in iden i y managemen in o de o ge a new iden i y wi h a clean epu a ion. Despi e hei dis inc pu poses, Sybil, Whi ewashing o Re-en y a acks a e based on limi a ions in he iden i y managemen mechanism mainly due o low e o , o low cos , o ge a new iden i y. The cen alized na u e o he epu a ion sys ems, used in e-comme ce pla o ms ha don’ sha e he use iden i y da a, in la es he p oblem (Camilo e al., 2020; Zul iqa e al., 2021) In Swamyna han e al. (2010), he au ho s de ine he Chu n a ack in he con ex o P2P ne wo ks as high a es o pee u no e . In such a scena io, a signi ican numbe o pee s will ha e ela i ely sho - e m accumula ed epu a ion sco es as a esul o a small numbe o pas in e ac ions. In e-comme ce pla o ms, his undesi able scena io can be enabled by he low e o /cos o ge se e al iden i ies. The Sybil a acks a e equen ly coupled wi h Collusion a acks, in which he di e en iden i ies o a single en i y conduc coo dina ed ac ions in o de o in luence he epu a ion o a use o p oduc . Kou ouli and Tsalga idou (2012) iden i y h ee a ian s o Collusion: (1) collusi e badmou hing and (2) collusi e educing ecommenda ion epu a ion and (3) Collusi e decei . As we can obse e, all hese ypes o a acks on epu a ion sys ems esul om ulne abili ies in one componen o he unde lying epu a ion model o a chi ec u e o he sys em. The a o emen ioned p oblems o lack o anspa ency in he managemen o he epu a ion da a, discussed in he sec ion Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 6 / 10 Backg ound on epu a ion sys ems, subsec ion 0, he dis inc models and c i e ia o calcula e he use s’ epu a ion sco e, as well as he cen alized na u e o he cu en epu a ion sys ems make he mi iga ion o hose a acks a e y di icul ask. In he li e a u e, one can ind se e al axonomies and ca ego iza ions o hese ypes o a acks. In he nex sec ion, we discuss ou indings in e ms o axonomies and classi ica ions o a acks ha de i ed om ou sys ema ic li e a u e e iew. Taxonomies o a acks Feng e al. (2012) discussed h ee ypes o a acks: di ec , disguise, and misguidance a acks. A di ec a ack is common in mos basic a ing sys ems. The a acke s p o ide dishones a ings only on he i ems wi hin he a ack a ge se . Then all use a ings o an i em a e compu ed o ob ain i s agg ega e ecommenda ion sco e. Thus, he social ecommenda ion sco e will e lec use s’ mains eam opinions o . On he o he hand, disguise a acks and misguidance a acks a e ep esen a i e o us -enhanced social a ing sys ems. Acco ding o he au ho s’ de ini ion, in hese epu a ion sys ems, he use us can be de ined as a c edibili y measu emen o he a ing. Thus, in hese wo ypes o a acks sophis ica ed s a egies o non-di ec a ing a e applied in o de o educe he c edibili y o hones use s while inc easing he c edibili y o he malicious ones. Sel -p omo ing is an example o a disguise a ack. In he case o a misguidance a ack, he s a egy aims a s a egically making he sys em misjudge he hones a ing beha iou o be dishones and he dishones a ing beha iou o be hones . We classi ied he au ho s’ p oposal as simplis ic, because each ca ego y o a ack is e y comp ehensi e, in which many di e en ypes o a ack can be placed. F aga e al. (2012) p oposed an a ack axonomy based on he epu a ion sys em a chi ec u al model and a se o well- known secu i y opics. In one dimension, he au ho s p opose h ee basic p ocesses: us and epu a ion (T&R) in o ma ion ga he ing, T&R calcula ion, and T&R dissemina ion. In he second dimension, he au ho s p opose P ima y opics: Au hen ica ion, Au ho iza ion, A ailabili y and U ili y/p ocess; and De i ed opics: Iden i ica ion, Non- epudia ion, Con iden iali y, In eg i y and Time In eg i y. Fo ins ance, bad-mou hing and ballo -s u ing ega ds T&R in o ma ion ga he ing and U ili y/p ocess p ima y opic. Acco ding o he au ho s, U ili y means use ulness. I means ha he ac ion an agen wan s o pe o m o e a esou ce could be ca ied ou co ec ly. In his example o aud, he esou ce is he use s’ a ings. The au ho s aim o p opose a holis ic axonomy ha may include all known a acks and p o ides s uc u ed ools o iden i y new a acks. In ou opinion, he p oposed amewo k is e y complex, and no complian wi h he comp ehensibili y equi emen o a good axonomy, as s a ed by he au ho s. Yao e al. (2012) dis inguish he ulne abili ies o epu a ion sys ems in o wo ca ego ies: (1) sys em-based ulne abili ies – ha ela e o he ounda ion and en i onmen o epu a ion sys ems, and (2) me ic-based ulne abili ies ha ie o he selec ed epu a ion me ic and i s upda es. These wo ca ego ies encompass six secu i y equi emen s o he epu a ion sys em, a a lowe le el, and a a highe le el o compu a ion o epu a ion sco es and decision-making, i.e., he epu a ion me ic. In he i s ca ego y, he au ho s include ulne abili ies in he message exchange be ween sys em nodes, agili ies in he sys em nodes in e ms o da a in eg i y, when s o ing and p ocessing i , and he lack o e ec i e iden i y managemen ha a oids malicious ac ions igge ed by mul iple iden i ies o he same use . The second ca ego y ela es o he agili ies in he model in which he epu a ion sco es a e based on decision-making. Kou ouli and Tsalga idou (2012, 2016), in he con ex o P2P ne wo ks, and Panagopoulos e al. (2017), o e-comme ce, p opose he ollowing axonomy o ypes o a ack: (1) S a egy- Based A acks, (2) Iden i y-Based A acks, and (3) Un ai Ra ings. This p oposal s ic ly ocuses on he epu a ion model on which he me ics a e based. This axonomy p oposal includes he ypes o a acks and hei a ian s known in he li e a u e. Howe e , despi e i s co e age, we obse e ha o he possible combina ions o a ack could po en ially exis bu a e no included in his axonomy, such as ecip oci y and e alia ion, as discussed by he same au ho s in ano he pape (Panagopoulos e al., 2017). Panagopoulos e al. (2017) also e e o o he issues om he economic and/o social pe spec i es, which a e no a acks bu may a ec he e ec i eness o he epu a ion sys ems as a means o p o ide us in he e-comme ce communi y. The au ho s e e o (1) he impo ance o inducing use pa icipa ion using incen i es, as he a e age quan i y o eed-backs is insu icien o ge an accu a e sco e o he epu a ion o a pe son, (2) dealing wi h ecip oci y ( he au ho s men ion s udies whe e a s ong co ela ion be ween buye and selle a ings a e iden i ied), and e alia ion, and (3) how o deal wi h he epu a ion o newcome s, in e ms ini ial epu a ion o hese use s. Addi ionally, Panagopoulos e al. (2017) s ill e e o a numbe o issues ound in decen alized epu a ion sys ems, such as us p opaga ion (i.e., how o e ec i ely communica e us in o ma ion in la ge-scale ne wo ks o loosely connec ed en i ies (Zeynal and e al., 2021), and s o age o local and global epu a ion in o ma ion. Sänge e al. (2015) p opose a di e en app oach o axonomy o a acks. The au ho s p opose, a he highes le el, o dis inguish be ween selle a acks and ad iso a acks. In hese majo classes, he au ho s classi y e e y ype o a ack in o wo dimensions: a acke s and beha iou . A a lowe le el, he a acke ’s dimension e e s o he numbe and cha ac e is ics o he digi al iden i ies pa icipa ing in an a ack (one iden i y, mul iple iden i ies o mul iple en i ies), which dis inguishes Sybil and Collusion, and beha iou dimension (consis en o inconsis en ). The au ho s claim ha hey a e ocused on he gene al cha ac e is ics and symp oma ology o a acks, such as he con inui y and he numbe o a acke s. We conside his app oach o a axonomy e y compac and a he same ime comp ehensi e, which makes i e y in e es ing. Howe e , app oach can be limi ing in e ms o e ec i e iden i ica ion o he ype o a ack. Fo ins ance, we do no see how i is possible o classi y he whi ewashing a ack, and he au ho s, do no p o ide examples o addi ional in o ma ion. Ano he example is he scena io o a dis ibu ed epu a ion sys em in which he epu a ion da a is manipula ed. Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 7 / 10 Ho man e al (2009) classi y a acks agains epu a ion sys ems based on he goals o he epu a ion sys ems a ge ed by hese a acks. The au ho s p opose i e ca ego ies o a ack: (1) Sel -P omo ing, (2) Whi ewashing, (3) Slande ing, (4) O ches a ed and (5) Denial o Se ice (DoS). We conside ha his classi ica ion includes ulne abili ies a wo le els: a a highe le el, in he epu a ion model, which includes he i s ou ca ego ies, and a a lowe le el, he i h ca ego y ha ela es o he DoS a acks. We conside his classi ica ion incomple e, namely a he lowe le el, because DoS a acks a e jus one o he se e al possible ypes a his le el, in his case, a he ne wo k le el. We no e ha he e a e many o he ypes o ulne abili ies a a low le el. Below, in he p esen sec ion, we will discuss ou p oposal o a classi ica ion in which we e e o hose o he possible low-le el a acks. Ano he aspec o his p oposal is i s ocus on he a ack’s goal. Fo example, ega ding he Whi ewashing a ack, we ask i he inal goal is o escape om he consequences o a low epu a ion o o manipula e someone’s epu a ion sco e. In Table 2 we summa ize ou indings in e ms o classi ica ions, p esen ing a b ie desc ip ion o he s uc u e and i s ocus. Table 2. Summa y o p oposed classi ica ions Re e ence S uc u e Focus (Feng e al., 2012) Th ee ypes o a ack e-comme ce and collabo a i e ne wo ks (F aga e al., 2012) Bi-dimensional amewo k: basic p ocesses and opics Repu a ion sys ems in gene al (Kou ouli & Tsalga idou, 2012, 2016; Panagopoulos e al., 2017) Hie a chical classi ica ion o ype and a ian s P2P and e- comme ce (Sänge e al., 2015) A highes le el: selle a acks and ad iso a acks. In hese majo classes: wo dimensions: a acke s and beha iou . Elec onic ma ke places (Ho man e al., 2009) Fi e ypes o a ack Repu a ion sys ems in gene al Limi a ions o he cu en app oaches o ca ego iza ion In he p esen wo k, an a ack is an in en ional ac ion wi h aud as i s objec i e, which a ec he epu a ion sys ems. Addi ionally, o he issues can comp omise he accu acy o epu a ion sco es; howe e , hey a e no a acks. In (Jøsang e al., 2007) he "Bias owa ds posi i e a ing" is explained as posi i e a ings simply ep esen ing an exchange o cou esies; ei he he posi i e a ing is gi en in he hope o ge ing a posi i e a ing in e u n, o he nega i e a ing is a oided due o ea o e alia ion om he o he pa y. We conside ha he e is a gap o ex ensibili y in he p oposed classi ica ions ound in he li e a u e, as i is di icul o classi y all a ian s o a ype o a ack because a sligh a ia ion can change he gi en classi ica ion. In he nex sec ion, we will discuss hese limi a ions compa ed o ou p oposal. PROPOSAL FOR A NEW FRAMEWORK OF CLASSIFICATION Types o aud and a acks in epu a ion models Analysing he ypes o aud and a acks o he models used in epu a ion sys ems, ounded on he li e a u e, we obse ed ha he e a e wo undamen al le els o a ack, o dis inc na u e (o igin and echnique), as well as he ield o esea ch. The o me ca ego y ega ds all ulne abili ies a he ne wo k, in as uc u e and applica ion le els. These may esul om bad de ini ions in he sys ems, secu i y b eaches in he so wa e, w ong choices in e ms o ne wo k a chi ec u e, missing de ence ools, such as i ewalls, WAFs, IDS, lack o c yp og aphy in he s o ed da a o messaging exchanging, among o he s. On he o he hand, he second ca ego y encompasses he agili ies o he model ha de ines he algo i hm, which ga he s and calcula es all me ics in o de o es ablish a epu a ion sco e, as well as a chi ec u al issues, such as cen alized s dis ibu ed o iden i y managemen . Mo eo e , in gene al, hese a acks a he model le el a e accomplished by membe s o he e-comme ce communi y, e.g., a use which has been egis e ed o a long ime in he Amazon o eBay ma ke place, con as ing wi h he a acks a he le els o he ne wo k, in as uc u e o applica ion, which, in gene al, a e pe pe a ed by ou side s. Due o he na u e o he a acks, i s i s ca ego y is ou o he scope o ou wo k. In he p esen wo k, we ocused on he ulne abili ies o he model and he a chi ec u al issues on which he epu a ion o mula ion is based. Thus, we will con inue ou discussion ocused on he ulne abili ies o he epu a ion model. We s a ou discussion by obse ing ha any malicious ac ion alls in o one o wo aud cases: o inc ease o dec ease he epu a ion o an en i y o p oduc /se ice. Fo ins ance, ballo s u ing and bad-mou hing a e he same ulne abili y, bu wi h a di e en ype o aud as a goal. In ano he obse a ion, we no ice ha many a acks a e combina ions o p imi i e ypes o a ack leading o se e al a ia ions o he same ype o a ack. Fo ins ance, a g oup o en i ies can collude by means o un ai a ings (collusion + un ai a ings), o he collusion o se e al iden i ies, o he same en i y, combined in o de o gi e un ai a ions, i.e., a Sybil a ack. These obse a ions led us o p opose a no el app oach based on a classi ica ion o mul idimensional a ibu es. P oposal o a ma ix o a ibu es Ou p oposal is based on a classi ica ion o mul idimensional a ibu es. Each ype o a ack has he ollowing i e a ibu es, each wi h se e al possible alues: 1) Type o aud: (a) Inc ease i s own epu a ion, (b) dec ease he epu a ion o o he s, (c) inc ease he epu a ion o o he s o (d) dec ease ecommenda ions epu a ion o a use . Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 8 / 10 2) Le el: (a) Iden i y managemen , (b) model o mula ion/calcula ion (Ho man e al., 2009), (c) o mula ion/da a o (d) a chi ec u al – This a ibu e iden i ies an elemen in he model whe e he ulne abili y is.  Sybil and whi ewashing a e examples o an iden i y managemen -le el a ack.  The lack o an ageing mechanism, in he model o mula ion, can be exploi ed. A he model o mula ion le el, we also conside he lack o alida ion (policing) o he a ings by means o a manual (endo se s) o au oma ic mechanisms (e.g. machine lea ning);  In e ms o scena ios o a chi ec u al issues, in a cen alized sys em, he en i y ha manages he pla o m can manipula e he epu a ion da a. In he case o decen alized sys ems, he nodes can po en ially manipula e he da a sha ed in he ne wo k, e en i he da a is enc yp ed o signed. 3) Ca dinali y – (a) One en i y, (b) mul iple en i ies; (c) mul iple iden i ies o (d) Many en i ies o many en i ies.  In a Sybil a ack an en i y wi h mul iple iden i ies pa icipa es;  In he case o Collusion, he a ack is pe o med by mul iple en i ies. 4) Beha iou – (a) One ime, (b) cons an o (c) a iable  The scena io o un ai a ings gi en by one, o mo e buye s o selle s in a cons an o a iable beha iou ;  Following a a iable pa e n, e.g. Oscilla o y (Panagopoulos e al., 2017) o T ai o s (Panagopoulos e al., 2017) a acks. 5) Ac ion – (a) Un ai a ing, (b) disc imina o y a ing, (c) c ea ing a new iden i y o (d) da a manipula ion These i e dimensions p oposed o classi y he possible a acks on he epu a ion sys ems a e ocused on e-comme ce communi ies. Howe e , his ma ix could po en ially be applied o o he ne wo k se ices, such as c owdsou cing o P2P. The p oposed app oach also has he ad an age o ex ensibili y. Tha is o say, new a ibu es can be added o he amewo k, as well as new alues o hese a ibu es. Addi ionally, ou app oach can handle mul iple a ian s o he same ype o a ack, a oiding long and complex hie a chical axonomies. This is a subs an ial ad an age ega ding he axonomies ound in he li e a u e. Fo ins ance, one can pe o m bad-mou hing aud applying collusion by means o Sybil, o no . Thus, we conside bad moun ing as a ype o aud o he a ack, lack o e ec i e iden i y managemen a he le el in he case o a Sybil a ack and ca dinali y as mul iple iden i ies, in he case o a collusion-based a ack. E alua ion o he p oposed ma ix In his sec ion, we will es ou p oposal on he ypes o a acks on epu a ion sys ems ound in he li e a u e. In Table 3, o each a ack ( i s column) we p esen all i e a ibu es. When all a ibu es a e possible, we use “any”. In such cases, i means ha he same a ack (o a ian ) has a ian s, as many as he numbe o possible combina ions o a ibu es. Table 3. A acks mul idimensional analyse A ack Desc ip ion F aud Le el Ca dinali y Beha iou Ac ion Ballo s u ing (C. Della ocas., 2000; C. Della ocas, 2000; C. N. Della ocas, 2003) (a) (b) Any Any (a), (b) o (d) Bad-mou hing (b) (b) Any Any (a), (b) o (d) On-o (Alshamma i e al., 2021) Any (b) Any (c) (a) o (b) Oscilla o y beha iou (Panagopoulos e al., 2017) Any (b) Any (c) (a) o (b) Quali y a ia ions o e ime (Jøsang e al., 2007) Any (b) Any (c) (a) o (b) Sybil (Douceu , 2002) Any (a) (c) Any (c) and ((a) o (b)) Whi ewashing (F aga e al., 2012) (a) (a), (b) o (c) (a) (a) Any Re-en y (F aga e al., 2012) (a) (a) (a) (a) (c) Chu n (Swamyna han e al., 2010) (a) (a) (a) (a) (c) Collusion (Kou ouli & Tsalga idou, 2012, 2016) Any (b) (b) o (c) (b) o (c) (a) and/o (b) Collusi e decei (b) and (c) (b) (d) (b) o (c) (a) Collusi e badmou hing (b) (b) (b) o (c) (b) o (c) (a) Collusi e educing ecommenda ion epu a ion (d) (b) (b) o (c) (b) o (c) (a) Da a manipula ion by a cen al au ho i y (Schaub e al., 2016; Zul iqa e al., 2021) Any (c) (a) Any (d) Repu a ion T ap (Feng e al., 2012) (b) and (c) (b) (d) (b) o (c) (a) and (b) Bias owa d posi i e a ing (Jøsang e al., 2007) (c) (b) (a) (a) (a) Pe ei a R. H. e al./ J INFORM SYSTEMS ENG, 8(1), 19218 9 / 10 In Table 3, we can obse e ha , only by he a ack name, one canno know all he de ails o he malicious ac ion. Fo ins ance, a malicious use may apply a Sybil a ack o inc ease his own epu a ion o damage someone’s epu a ion. Thus, ou p oposal enables secu i y analys s o classi y new ypes o a ack, as well as iden i y wo ypes o a ack as e ec i ely being he same, o ins ance: On-o and con lic ing-beha iou a acks. E en i wo ypes o a ack ha e dis inc p oposals, i hei a ibu es a e he same, hen, po en ially, he same app oach o dealing wi h hem could be applied in bo h cases. CONCLUSION In he p esen wo k, we conduc ed a sys ema ic li e a u e e iew in o de o sys ema ize he se e al ypes o a acks and aud o epu a ion sys ems in he con ex o use epu a ion in e-comme ce. In ou discussion, we p esen some obse a ions ha lead us o conclude ha he ype o a ack/ aud does no in o m us abou all he necessa y de ails o unde s anding he malicious ac ion. In ac , each ulne abili y may be combined wi h o he s. Thus, he same ype o a ack could ha e dis inc names, o many a ian s, making i e y di icul o inbox i in a hie a chical o g oup-based classi ica ion, as he ones ound in he li e a u e. In o de o o e come his gap, we p opose a no el amewo k o classi ica ion. We a e con inced ha ou app oach has ad an ages o e o he p oposals based on axonomies, ca ego ies o hie a chical classi ica ions, which a e complex and edundan when ying o co e all ypes/ a ian s. We expec o con ibu e o he knowledge in his esea ch ield by means o ou p oposal o an inno a i e amewo k. We belie e ha ou amewo k can be use ul o epu a ion sys em de elope s in o de o p e iew and analyse new o ms o a ack, as well as o help o de elop e ec i e de ence mechanisms. Ou p oposal is s ill in i s i s e sion. Due o i s ex ensibili y, new a ibu es and alues can be added o he amewo k. The epu a ion sys ems s ill ha e open issues, mo i a ing us o con inue ou wo k. Thus, we expec o p esen new e sions o his classi ica ion amewo k in he nea u u e. REFERENCES Ahn, J., Pa k, M., & Paek, J. (2018). Rep o : A Model o De i ing T us and Repu a ion on Blockchain-based Elec onic Paymen Sys em. 2018 In e na ional Con e ence on In o ma ion and Communica ion Technology Con e gence (ICTC), 1431–1436. h ps://doi.o g/10.1109/ICTC.2018.8539641 Ahn, J., Pa k, M., Shin, H., & Paek, J. (2019). A Model o De i ing T us and Repu a ion on Blockchain-Based e- Paymen Sys em. Applied Sciences, 9(24), 5362. h ps://doi.o g/10.3390/app9245362 Alshamma i, S. T., Albesh i, A., & Alsubhi, K. (2021). Building a us model sys em o a oid cloud se ices epu a ion a acks. Egyp ian In o ma ics Jou nal, 22(4), 493–503. h ps://doi.o g/10.1016/j.eij.2021.04.001 B ine , R. B., Denye , D., & Rousseau, D. M. (2009). E idence- Based Managemen : Concep Cleanup Time? Academy o Managemen Pe spec i es, 23(4), 19–32. h ps://doi.o g/10.5465/AMP.2009.45590138 Camilo, G. F., Rebello, G. A. F., de Souza, L. A. C., & Dua e, O. C. M. B. (2020). A Secu e Pe sonal-Da a T ading Sys em Based on Blockchain, T us , and Repu a ion. 2020 IEEE In e na ional Con e ence on Blockchain (Blockchain), 379–384. h ps://doi.o g/10.1109/Blockchain50366.2020.00055 Damiani, E., di Vime ca i, D. C., Pa aboschi, S., Sama a i, P., & Violan e, F. (2002). A epu a ion-based app oach o choosing eliable esou ces in pee - o-pee ne wo ks. P oceedings o he 9 h ACM Con e ence on Compu e and Communica ions Secu i y - CCS ’02, 207. h ps://doi.o g/10.1145/586110.586138 Della ocas, C. (2000). Immunizing online epu a ion epo ing sys ems agains un ai a ings and disc imina o y beha io . P oceedings o he 2nd ACM Con e ence on Elec onic Comme ce - EC ’00, 150–157. h ps://doi.o g/10.1145/352871.352889 Della ocas., C. (2000). Mechanisms o coping wi h un ai a ings and disc imina o y beha io in online epu a ion epo ing sys ems. P oceedings o he Twen y Fi s In e na ional Con e ence on In o ma ion Sys ems (ICIS ’00), 520–525. Della ocas, C. N. (2003). The Digi iza ion o Wo d-o -Mou h: P omise and Challenges o Online Feedback Mechanisms. SSRN Elec onic Jou nal. h ps://doi.o g/10.2139/ss n.393042 Dennis, R., & Owen, G. (2015). Rep on he block: A nex gene a ion epu a ion sys em based on he blockchain. 2015 10 h In e na ional Con e ence o In e ne Technology and Secu ed T ansac ions (ICITST), 131–138. h ps://doi.o g/10.1109/ICITST.2015.7412073 Dhakal, Anup, & Cui, Xiaohui. (2019). DT us : A Decen alized Repu a ion Sys em o E-comme ce Ma ke places. Dixon-Woods, M., & Fi zpa ick, R. (2001). Quali a i e esea ch in sys ema ic e iews: Has es ablished a place o i sel . B i ish Medical Jou nal, 323, 765–766. Douceu , J. R. (2002). The Sybil A ack. In: D uschel, P., Kaashoek, F., Rows on, A. (eds) Pee - o-Pee Sys ems. Lec u e No es in Compu e Science, Vol 2429. Sp inge , Be lin, Heidelbe g, 2429. Feng, Q., Liu, L., & Dai, Y. (2012). Vulne abili ies and coun e measu es in con ex -awa e social a ing se ices. ACM T ansac ions on In e ne Technology, 11(3), 1–27. h ps://doi.o g/10.1145/2078316.2078319 F aga, D., Banko ic, Z., & Moya, J. M. (2012). A Taxonomy o T us and Repu a ion Sys em A acks. 2012 IEEE 11 h In e na ional Con e ence on T us , Secu i y and P i acy in Compu ing and Communica ions, 41–50. h ps://doi.o g/10.1109/T us Com.2012.58