A multiagent-based tool for the simulation of social production and management processes of urban ecosystems using the JaCaMo framework: a case study of San Jerónimo Vegetable Garden, Seville, Spain

A Mul iagen -Based Tool o he Simula ion o Social P oduc ion and Managemen P ocesses o U ban Ecosys ems: A Case S udy o San Je ónimo Vege able Ga - den – Se ille, Spain Flá ia Pe ei a dos San os1, Diana Adama i2, Hen ique Rod igues1, Glenda Dimu o3, Es eban De Manuel Je ez3, G açaliz Pe ei a Dimu o1 1Uni e sidade Fede al do Rio G ande, A . I alia s/n km 08, Rio G ande / RS, B azil 2Cen o de Ciência Compu acionais, Uni e sidade Fede al do Rio G ande, A . I alia Km 08 Ca ei os, Rio G ande / RS, 96201-090, B azil 3Uni e sidad de Se illa, A da. Reina Me cedes 4 A, 41012, Se illa, Spain Co espondence should be add essed o la iasan [email p o ec ed]; [email p o ec ed] Jou nal o A i icial Socie ies and Social Simula ion 19(3) 12, 2016 Doi: 10.18564/jasss.3128 U l: h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Recei ed: 29-02-2016 Accep ed: 17-05-2016 Published: 30-06-2016 Abs ac : The concep o social p oduc ion and managemen o u ban ecosys ems may be unde s ood as he gene a ion o newphysicalo ela ional si ua ions, by cons uc ing, ans o mingo elimina ing physicaland/o ela ional objec s o ensu ing he ul illmen o hei social and en i onmen al unc ions. This includes he ci izen pa icipa ion in he p ocess o u ban planning and ans o ma ion, o ming a ne wo k s uc u ed and suppo ed by ools allowing he equal dis ibu ion o powe in he decision making. The SJVG-MAS P ojec add esses, in an in e disciplina y app oach, he de elopmen o compu a ional ools based on Mul iagen Sys- ems (MAS) o he simula ion o he social p oduc ion and managemen p ocesses ha occu in u ban ecosys- ems, in pa icula , he San Je ónimo Vege able Ga den p ojec (Se ille, Spain). In his pape , we p esen a MAS-based simula ion ool de eloped in he JaCaMo amewo k. We concei ed a 5-dimensional BDI-like agen social sys em composed o he agen s’ popula ion, he social o ganiza ion, he en i onmen , he in e ac ional / communica ion and he egula o y s uc u es. Keywo ds: u ban ecosys em, social o ganiza ion simula ion, simula ion o social p oduc ion and managemen p ocesses, egula o y policy simula ion, mul iagen -based simula ions, JaCaMo amewo k In oduc ion 1.1 The concep o social p oduc ion and managemen o u ban ecosys ems may be unde s ood as he gene a ion o new physical o ela ional si ua ions, by cons uc ing, ans o ming o elimina ing physical objec s and/o ela ional objec s wi h he objec i e o ensu ing, in he new p oduced si ua ions, he ul illmen o hei so- cial and en i onmen al unc ions (O iz 2010; Pelli 2007, 2010). This includes he ci izen pa icipa ion in he p ocess o u ban planning and ans o ma ion, a icula ing he di e en in ol ed agen s (go e nmen , ins i u- ions, echnicians, ci izens), o ming a ne wo k s uc u ed and suppo ed by mechanisms and ools ha allow he equal dis ibu ion o powe in he decision making, so ha all agen s can pa icipa e and dialogue ac i ely in he whole p ocess o a ce ain p ojec , om i s planning o i s managemen , as discussed by (Dimu o 2010; Dimu o & Je ez 2010, 2011). 1.2 The social p oduc ion and managemen o u ban ecosys ems con ibu e o he s eng hening o communi y p ac ices, o he inc easing o esponsibili y o a collec i e p ojec , o he exe cise o democ acy, o he de el- opmen o mo e suppo i e ac ions, including bo h p oduc i e and economic issues, as well as en i onmen al issues. See also he discussions p esen ed by (Dimu o e al. 2015a,b). JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 1.3 The San Je ónimo Vege able Ga den (SJVG) p ojec , headed by he con ede a ion “ecologis as en acción” (EA)1, is an example o an u ban ecosys em loca ed in Se ille, Spain. The SJVG is main ained by i s own use s, unde he supe ision and coo dina ion o he EA. The ha es is en i ely ecological, he p oduc ion is only o sel - consump ion, bu people can exchange p oduc s and se ices. Then, he SJVG’s social o ganiza ion is cha ac- e ized o allowing and p omo ing a lo o in e ac ions and social exchanges be ween he pa icipan s. Ne e - heless, he beha io s, in e ac ions and communica ions a e egula ed by no ms es ablished by he communi y in assembly, unde he supe ision o he EA. 1.4 The SJVG-MAS P ojec 2, whose ini ial esul s we e p esen ed in (Fa ias e al. 2013; Rod igues e al. 2013; San- os e al. 2012, 2014a,b; Sil a e al. 2013), add esses, in an in e disciplina y app oach, The SJVG-MAS P ojec (Dimu o2010;Dimu o&Je ez2010,2011). TheSJVG-MASP ojec 3, whose ini ial esul s we e p esen ed in(Fa ias e al. 2013; Rod igues e al. 2013; San os e al. 2012, 2014a,b; Sil a e al. 2013), add esses, in an in e disciplina y app oach, The SJVG-MAS P ojec (Dimu o 2010; Dimu o & Je ez 2010, 2011). The P ojec is a join e o o in e ela ing knowledge, seeking collec i e in e p e a ions, adop ing as case s udy he cu en endency o ( e)app oaching he coun yside o he ci y h ough u ban ege able ga dens, in pa icula , he SJVG. We aim o con ibu e o he analysis o he ac ual eali y o he SJVG expe imen , p o iding esou ces o omen he discussions on he adop ed me hodology and o help he in es iga ion o new possible ideas ha may be ap- plied in he con ex o he SJVG’s o ganiza ion, o example, how possible changes in he social o ganiza ion (e.g., oles assumed by he agen s in he o ganiza ion, ac ions, beha io s, (in) o mal in e ac ion / communica- ion p o ocols, egula ion no ms), especially om he poin o iew o he agen ’s pa icipa ion in he decision making p ocesses, may ans o m his eali y, om he social, en i onmen al and economic poin o iew, hen con ibu ing o he sus ainabili y o he p ojec . 1.5 The objec i e o his pape is o p esen he mas-based ools de eloped in he SJVG-MAS con ex , discussing he adop ed solu ions and in oducing he esul s ob ained in simula ions. In o de o ake in o accoun all he sui able cha ac e is ics o he SJVG social o ganiza ion, we concei ed ou MAS as a mul i-dimensional BDI-like agen social sys em4, using he JaCaMo amewo k ((Boissie e al. 2013)), in ol ing he de elopmen o i e dimensions: 1. he agen s’ popula ion: he agen es ha may assume oles in he SJVG social o ganiza ion; 2. he social o ganiza ion: he o ganiza ional oles o ga dene , ins i u ion, echnician, e c. and hei hie a - chy; 3. he (physical) en i onmen ; 4. he in e ac ional/communica ion s uc u e be ween oles; 5. he egula o y s uc u e: cons i u i e and egula i e in e nal no ms es ablished by he SJVG communi y. 1.6 We adop ed he JaCaMo amewo k mainly because i o e s high-le el and modula acili ies o de elop he i s h ee dimensions men ioned abo e. I is composed o h ee sepa a e echnologies: 1. he Jason ((Bo dini e al. 2007)) in e p e e o an ex ended e sion o Agen Speak-L language, o he implemen a ion o he agen s’ popula ion (dimension (i)); 2. he CA AgO amewo k ((Ricci e al. 2011)), o modeling he en i onmen (dimension (iii)) using he con- cep o a i ac s (Ricci e al. 2007); 3. he MOISE+ model ((Hübne e al. 2007, 2010)), o modeling o he social o ganiza ion (dimension (ii)). 1.7 We ound ha he modula i y p o ided by he JaCaMo amewo k helps he modeling o eal wo ld o ganiza- ions, especially when an in e disciplina y wo king g oup is in ol ed, as in he SJVG-MAS p ojec . Mo eo e , his modula de elopmen acili a ed modi ica ions in he o ganiza ion, which is eally impo an o he analy- sis o he impac o hose changes in he social p oduc ion and managemen p ocesses and in he en i onmen . See also he discussions in (Hübne e al. 2010; San os e al. 2014a). 1.8 Obse e ha , since he implemen a ion o he dimension (i), namely, he agen popula ion, was done using Jason, he adop ed agen a chi ec u e was he bdi (Belie s, Desi es, In en ions) model. Ou choice o an agen model o in en ional na u e, whose beha io s can be explained by a ibu ing ce ain men al a i udes o he agen s, such as knowledge, belie s, desi es, in en ions, obliga ions, commi men s, is jus i ied by a la geamoun o wo k discussing he ole o such models in agen -based simula ion o human beha io , emo ions, and alue- based e alua ion, such as us and epu a ion. See, e.g., he discussions p esen ed by (An 2012), (Fila o a e al. 2013), (Subagdja e al. 2009) and (Adama i e al. 2014). JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 1.9 Fo heo he dimensions(i )and( ),whicha epa icula impo an , sincemodi ica ionsin hein e ac ion/communica ion/ egula o y s uc u es may di ec ly a ec he social p ocesses unde analysis in his p ojec , we in oduced a communica- ion and a egula o y s uc u es, based on he CA AgO amewo k and an adap a ion o he MSPP amewo k (Modeling and Simula ion Public Policies, in oduced in (San os & Cos a 2012)), espec i ely. Some addi ional commen s on ela ed wo k 1.10 This pape p esen ed he use o in eg a ed a i ac s o model an u ban ecosys em, based in a mul idimensional mul iagen app oach. These a i ac s ac in di e en dimensions o he social sys em: o ganiza ional, in e ac- ional/communica ion, egula o y and physical aspec s. In he li e a u e, we do no ind wo ks whe e a i ac s a e used in eal sys ems, in such gene al app oach, as in ou p oposal. Some wo ks p esen ed how JaCaMo in as uc u e could be applied, using “ oy examples”, as p esen ed by (Hübne e al. 2009), o “hypo he ical examples”, as in oduced in (Baldoni e al. 2010), in o de o show how a i ac s can be used and he ad ances o such app oach. Fo example, (Baldoni e al. 2010) p esen ed an applica ion o a i ac s ha explains how hey could be lexible and eusable. (Mokom 2015) p esen ed a p oposal o inse a i ac s in mul i-agen -based simula ions, whe e heses a i ac s a e dynamic using a i icial in elligence echniques, as gene ic algo i hms and cul u al algo i hms. in a simila line o wo k, (Von Lae e al. 2015) in oduced an e olu iona y agen socie y based on cul u al a i ac s, used o p omo e he sel - egula ion o exchange p ocesses, in he same di ec ion o he wo ks by (Dimu o e al. 2007, 2011; Dimu o & da Rocha Cos a 2015; Pe ei a e al. 2008). 1.11 Rela ed o he modeling and simula ion o (u ban o no ) ecosys ems and/o socio-ecological sys ems using agen -based o mul iagen sys ems, he wo ks ound in he li e a u e do no apply a i ac s, al hough agen echnology has been adop ed in se e al wo ks o pe o ming simula ions (see, e.g., he discussion on he chal- lengesand p ospec so agen -basedmodeling andsimula ionin he social-ecologicalsys emsp esen edby(Fi- la o a e al. 2013)). 1.12 In ac , he numbe o agen -based modelingapplica ions wi hin hesocio-en i onmen al con ex hasexploded o e he las decade, al hough he majo i y uses simple agen models (e.g., eac i e agen s). Di e en ly, we ha e adop ed a e y special cogni i e agen model, sui able o he kind o social o ganiza ion ound in he s udied u ban ecosys em. 1.13 Ne e heless, in he wo k by (Albe i & Waddell 2000), agen s and geo e e enced da a a e used o p opose a sus ainable ecosys em, ying o explain how he me opoli an a eas e ol e. (Manson 2003) p esen ed a new app oach o alida e and e i y mul i-agen sys ems applied o en i onmen al domains, acco ding o an spe- ci ic c i e ia ela ed o ecosys em managemen (see also (Janssen 2003), o se e al wo ks discussing complex- i y and ecosys em managemen , based on mul iagen app oaches). in (Adama i e al. 2005, 2009) and (Kou i a & Mak opoulos 2012), he mul iagen app oach is used o model a u ban wa e managemen . (Chen e al. 2012) used mul iagen sys ems o modeling he e ec s o social no ms on en ollmen in paymen s o ecosys em se ices. (Rai & Robinson 2015) discussed he empi ical in eg a ion o social, beha io al, economic, and en i- onmen al ac o s in hei agen -based modeling o ene gy echnology adop ion. (Sun & Mülle 2013) p oposed a amewo k o modeling paymen s o ecosys em se ices wi h agen -based models. 1.14 One in e es ing wo k is by (Sah bache e al. 2014), who p esen ed agen -based ools o he modeling and simula ion spa ial ela ionships be ween ecosys em se ices and ag icul u al p oduc ion, o he analysis o ag icul u al s uc u al changes, due o he collec i e impac s o a me s’ land managemen decisions on abo e g ound ecosys em se ices and hei implica ions o ag icul u e. (Iwamu a e al. 2014) used agen -based mod- eling o analysing in e ac ions be ween social and ecological sys ems in he con ex on indigenous lands. 1.15 In a mo e gene al app oach (Magliocca e al. 2014) in oduced he agen -based i ual labo a o y (ABVL) ap- p oach, which equi es mo e gene alized agen -based models (ABMs) o c oss-si e expe imen a ion, compa i- son, and syn hesis. B oadly, he ABVL app oach ha nesses hep ocess-based explana o y powe o ABMs wi hin a modeling sys em a chi ec u e explici ly designed o lexible, i e a i e expe imen a ion and c oss-si e com- pa ison. Howe e , ABVL is s ill limi ed o using simple eac i e agen s, so i is no possible o e alua e in which ex ension i can ake in o accoun he modeling o complex in e ac ions and beha iou s. 1.16 The es o he pape is o ganized as ollows. Subsec ion Some addi ional commen s on ela ed wo k p esen s a discussion on ela ed wo ks. In Sec ion 2, he JaCaMo amewo k and i s ela ed ools a e b ie ly explained.The MSPP amewo k is p esen ed in Sec ion 3. Sec ion 4 SJVG 5-dimension Social O ganiza ion, discussing he adop ed solu ions and he in eg a ion o he se e al kinds o a i ac s. Sec ion 5 p esen s some examples o simula ions. Sec ion 6 is he Conclusions. Some auxilia y in o ma ion abou he pe iodic ou ines o he se - e al oles iden i ied in he SJVG Social O ganiza ion is p esen ed in Appendix 1. Appendix 2 p esen s auxilia y JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 examples o he in eg a ion o a i ac s in he daily ou ines o some oles, in o m o ac i i y diag ams. The sou ce code o his wo k is a ailable as a supplemen a y ma e ial o his pape . The JaCaMo F amewo k 2.1 In his sec ion, we explain he echnologies ha a e encompassed by JaCaMo amewo k ((Boissie e al. 2013)), namely: Jason, Ca ago and MOISE+, p esen ing he main ea u es o hese echnologies, which co e some o he le els o abs ac ions ha a e equi ed o he de elopmen o sophis ica ed mas, which, in ou case, a e he agen s’ popula ion, he en i onmen and he social o ganiza ion. Jason and he agen s’ popula ion 2.2 Fo he implemen a ion o he agen s’ popula ion, JaCaMo p o ides Jason ((Bo dini e al. 2007)), which is an Agen Speak-L in e p e e ha p o ides a pla o m o de elop MAS, based on BDI agen model. Obse e ha he e a e many BDI ad hoc implemen a ions sys ems, howe e , an impo an cha ac e is ic o he Agen Speak-L language is i s heo e ic base. The Agen Speak-L p og amming language is an elegan ex ension o logic p o- g amming o BDI a chi ec u e o agen s. An Agen Speak-L agen co esponds o he speci ica ion o a se o belie s and plans ha o ms he ini ial knowledge base.5 2.3 Agen Speak-L dis inguishes wo ypes o goals: achie emen goals and es goals. Achie emen and es goals a e p edica es, such as belie s, bu hey ha e ixed ope a o s ’!’ and ’?’, espec i ely. Achie emen goals exp ess wha he agen wan s o achie e in an en i onmen s a e, whe e he p edica e associa ed wi h he goal is ue. In ac , hese objec i es s a he execu ion o subplans. A es goal e u ns he uni ica ion o a p edica e es wi h an agen belie , o ailu e i he uni ica ion is no possible wi h he agen belie s. 2.4 A igge ing e en de ines which e en s may ini ia e he execu ion o a plan. An e en can be in e nal, when gene a ed by he execu ion o a plan i a subgoal needs o be achie ed, o ex e nal, when gene a ed by he pe cep ion o he en i onmen . Ac i a ing e en s a e ela ed o he addi ion and emo al o men al a i udes (belie s o goals). Add and emo e men al a i udes a e ep esen ed by ixed ope a o s (’+’) and (’-’). The CA AgO amewo k and he MAS en i onmen 2.5 Fo he MAS en i onmen implemen a ion (and also o he acili ies, which a e discussed in he ollowing sec- ions), JaCaMo p o ides he CA AgO (Common A i ac In as uc u e o Agen s Open En i onmen s) ame- wo k ((Hübne e al. 2010; Ricci e al. 2011)), which is a MAS i ual en i onmen de elopmen and simula ion amewo k. CA AgO allows he implemen a ion o a i ual en i onmen as a compu a ional laye encapsula - ing he acili ies and non-au onomous se ices exploi ed by agen s du ing un ime.6 2.6 CA AgO is based on he Agen s and A i ac s (A & A) ((Ricci e al. 2007)) me a-model o model mas. This model in oduces a high-le el me apho , based in he idea ha human wo ke s ac in a coope a i e way wi h hei en i onmen : agen s a e compu a ional en i ies ha do some ype o goal-o ien ed ask (analogous o human wo ke s), and a i ac s a e he esou ces and ools dynamically c ea ed, handled and sha ed by agen s o sup- po hei ac i i ies, bo h indi idual and collec i e (as in he human con ex ). So, i is possible o de elop a i- ac s ha a e ins an ia edin he en i onmen , p o iding se ices o agen s, and able o do communica ion wi h ex e nal se ices (e.g., web-se ices). MOISE+ and he MAS social o ganiza ion 2.7 Fo modeling he MAS o ganiza ion, JaCaMo o e s he MOISE+ o ganiza ional model ((Hübne e al. 2007)), which encompasses he speci ica ion o h ee dimensions: he s uc u al, whe e oles, inhe i ance links and g oups a e de ined; he unc ional, whe e a se o global plans a e de ined, wi h missions o achie e hose goals; and he no ma i e dimension ha speci ies which ole has o commi o which mission. 2.8 In aS uc u al Speci ica ion (SS), indi idual, social andcollec i e le elscanbe de ined based on h ee concep s: oles (indi idual le el - se o beha io al cons ain s ha an agen accep s when joining a g oup), ela ionships JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 1: No ma i e Speci ica ion: linking SE and FS ((San os e al. 2014a)) be ween oles (social le el - ela ions allowed be ween he oles) and g oups (collec i e le el - a se o agen s wi h simila a ini ies and objec i es). The SS is de ined by a uple ss = ( g, ss , c), whe e g is he se o speci ica ion o oo s g oups o ss, ss is he se o all oles in he SS and cis he inhe i ance ela ionship on oles o ss . 2.9 The Func ional Speci ica ion (FS) is composed by a collec ion o Social Schemes (SS), which a e se s o goals s uc u ed by plans. The global a ge s ep esen he s a e o he wo ld ha is desi ed by he o ganiza ion, di e en om a local goal since he la e is o a single agen . he se o all se is deno ed by sch and a scheme sch is de ined by he uple sch = (g, p, m, mo, nm), whe e: 1. gis he se o goals in he scheme, 2. pis he se o plans ha builds he goals decomposi ion ee, 3. mis he se o missions, ha is, a se o global goals ha can be bound o a ole, 4. mo :m→pis a unc ion ha de e mines he se o goals in each mission, 5. nm :m→n×nde e mines he maximum and minimum numbe no agen s ha mus commi o each mission. 2.10 A scheme is a global goal decomposi ion ee, whose oo is he goal o he en i e scheme. The decomposi ion is made by plans (deno ed by he ope a o =), which poin a way o achie ing a goal. o ins ance, conside ing he plan g0=g1, g2, g3 The goal g0is decomposed in h ee plans, indica ing ha i will be achie ed only i plans g1,g2and g3a e also achie ed. 2.11 The No ma i e Speci ica ion (NS) is whe e ole missions a e speci ied wi h pe mission o obliga ion ype. This speci ica ion de ines a pe mission (pe ) o an obliga ion (obl) ela ed o a mission (m) ha an agen wi h a ole in he o ganiza ion is commi ed wi h (see igu e 1). They a e de ined as he uples pe (p, m, c)and obl(p, m, c), whe e whe e pde e mines ha an agen wi h he ole pcan be commi ed wi h he mission m, c de e mines ha empo al cons ain s a e es ablished, ha is, he e is a pe iod whe e he pe mission is alid, o example, “e e y day” o “e e y hou ”. The MSPP F amewo k o Modeling Public Policies 3.1 The MSPP (Modeling and Simula ion o Public Policies) amewo k ((San os & Cos a 2012)) aims o suppo agen -based models o he a ious ypes o sequen ial and non-sequen ial models o public policy p ocesses, as classi ied in (Hill 2009). I consis s o a se o p og amming schemes, classes and an API de eloped o he Jason-CA AgO pla o m. I s pu pose is o help he de elopmen o agen -based simula ions o public policy p ocesses ope a ing on agen -based simula ions o social, economic and en i onmen al con ex s. 3.2 In gene al, a policy is concei ed as a se o p inciples ha o ien and/o condi ion decisions and ac ions o he agen s ha ope a e in a gi en con ex , especially in wha conce ns he uses o he a ailable esou ces ((Eas on 1965)). A public policy in a gi en socie y, hus, is a policy conce ning he uses o esou ces ha a e conside ed o JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 be public in ha socie y, usually being issued by he go e nmen o ha socie y, as discussed in (Hill 2009). In he con ex o MSPP amewo k, a public policy is a se o no ms and ac ion plans, o be adop ed and ollowed by bo h he go e nmen agen s and he socie al agen s ha ope a e in he social con ex o conce n. 3.3 The MSPP oolki con ains he ollowing se o agen ypes: •go e nmen : an agen able o issue public policies ( o simplici y, he go e nmen o he socie y can be modeled as a single agen ); •socie al o social agen s: hose agen s o which he public policy is gene ally add essed, p esumably o sol e a public issued iden i ied in hei social con ex ; •go e nmen agen s: agen s ha ope a e as de ec o s and e ec o s o he go e nmen , as –no m en o ce s: de ec o and e ec o agen s ha pa icipa e in he p ocess o en o cemen o he no ms speci ied by he public policy: –no m de ec o s, which cap u e in o ma ion conce ning he agen s’ compliances o he policy no ms; –no m e ec o s, which apply he sanc ions p esc ibed by he no ms o he agen s ha do no comply o hem; –en i onmen al ope a o s: agen s ha pe o m plans speci ied by he public policy, aiming a he di- ec con olo aspec s o he physicalo social en i onmen o he socie y, in hesense o pe o ming ac ions ha ope a ionally in e e e wi h he s uc u e and/o he elemen s o hose en i onmen s (e.g.: ac ions on physical objec s, in e e ences on social ela ionships, damage o he na u al en i- onmen , e c.): –en i onmen al de ec o s, which cap u e in o ma ion conce ning he s a e o he en i onmen e- sou ces; –en i onmen al e ec o s, which ac on he en i onmen esou ces, changing hei ea u es, allowing o blocking he o he agen s accesses o hem, c ea ing o emo ing esou ces, e c. 3.4 The essen ial concep in he MSPP amewo k is ha o policy a i ac s, ha is, CA AgO a i ac s ha ei y he public policies ha a e add essed o he go e nmen agen s and socie al agen s o he socie y, so ha he com- ponen s o public policies a e conc e ely ep esen ed as a i ac s in he en i onmen . The ei ica ion o pub- lic policies as policy a i ac s amoun s o he ei ica ion o no ms and plans, so ha no m a i ac s and plan a i ac s should be de ined and ins an ia ed in he CA AgO amewo k, oge he wi h Agen Speak-L p og am schemes ha allow he agen s o he socie y o handle hem adequa ely. Modeling he SJVG Social O ganiza ion in JaCaMo 4.1 The u ban ege able ga den o he San Je ónimo Pa k (SJVG) (Se ille/Spain) (Figu e 2) is an ini ia i e o he con ede a ion EA in o de o p omo e social pa icipa ion in o ganic a ming p ac ices h ough he use o u ban ege able ga dens o ec ea ion, and conduc ing ac i i ies ela ed o en i onmen al educa ion. The main ea- u es o his p ojec is ha his u ban ecosys em is cen e ed on a nonp o i , social u ban ege able ga den ( ha is, he p oduc ion is dedica ed o i s own pa icipan s), he p oduc ion is all based on na u al and ecological p inciples, p omo ing he in eg a ion be ween human and na u al esou ces. 4.2 The ege able ga dens a e loca ed in he San Je ónimo Pa k, occupying abou 1.5 hec a es, di ided in o 42 indi idual plo s (o size a ound 75 m2), assigned o ga dene s o di e en ages, especially e i ees. Al hough he “owne ship‘’ o each plo is indi idual, he wo k in he ga den is some imes sha ed among o he amily membe s o e en iends, called he auxilia y ga dene s. A pe son who in end o en e in he p ojec is called an aspi ing ga dene . The EA con ede a ion has a collec i e plo , alloca ed o i s pa ne s, and ano he plo ha se es as a kind o “school plo ”, whe e classes on o ganic c ops a e e en ually augh . Figu e 3 shows he schema o a SJVG plo . 4.3 The ole o EA con ede a ion is o o e see he wo k o he ga dene s, p o iding echnical suppo , con olling he use o chemical pes icides, which is s ic ly o bidden, p omo ing o ien a ion and mo i a ional alks, and also ec ea ion ac i i ies. The ga dene s, in u n, o ensu e hei pe manence in he p ojec , mus comply wi h a se o de e mina ions es ablished in he SJVG’s In e nal Regula ion No ms, including, e.g., he equi emen o o ganic a ming and he o biddance o selling o ading he p oduc s, bu also o he ules such as o keep JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 2: Localiza ion o he San Je ónimo Pa k and he San Je ónimo Vege able Ga den he po ion clean, o ake ca e o he common a eas, o a end he assemblies, o i iga e by d ipping wa e , o collabo a e wi h he ope a ion o he acili ies and in as uc u es, o pay a mon hly ee, among o he s. 4.4 A key ea u e o he p ojec is he ho izon ali y ime o make decisions ha a e always aken in he SJVG Assem- blies and es ablished in he o m o consensus among he communi y o ga dene s and echnicians o he EA con ede a ion. Besides he cul i a ion o indi idual plo s, he SJVG p ojec includes he ca e o a g eenhouse o g owing seedlings and a chicken coop. EA also pe o ms some ag eemen s wi h Se illa Uni e si y and/o o he academic ins i u ions and suppo s s uden s in in e nships. Finally, depending on he annual budge , EA also ca ies ou wo k wi h neighbo hood schools h ough school ege able ga dens. 4.5 In his sec ion we p esen he solu ion we in oduce o modeling, in JaCaMo, he i e dimensions iden i ied in he SJVGsocial o ganiza ion, namely(i) he agen s’ popula ion, (ii) he social o ganiza ion, (iii) he en i onmen , (i ) he in e ac ional/communica ion s uc u e, and ( ) he egula o y s uc u e. This was done by he in eg a- ion o O ganiza ional, Regula o y, Communica ion and Physical A i ac s, as ini ially p oposed by (San os e al. 2014a). The O ganiza ional modeling and he o ganiza ional a i ac 4.6 The social o ganiza ion was i s ly modeled using MOISE+. Figu e 4 shows he s uc u al model o he SJVG, whe e oles, g oups and sub-g oups, ole ela ionships a e speci ied. in his S uc u al Speci ica ion (SS), we speci ied he oo g oup hsj_ ege able_ga den (SJVG p ojec ), andi ssub-g oupsea con ede a ion and pa cel (plo o cul i a ion). he oles ha canbeassumed in hese sub-g oups a e: ga dene and auxilia y ga dene ,aspi ing ga dene ,adminis a ion,sec e a y and ea echni- cian. The ela ionships be ween hese oles can be: au ho i y (which is he case o he EA adminis a ion in ela ion o he sec e a y, echnician and ga dene ), communica ions and compa ibili y (be ween auxilia y ga - dene and aspi ing ga dene ). Acco ding o MOISE+ model, agen s in di e en g oups canno assume di e en oles in di e en g oups. In ou model (Figu e 4), jus heauxilia y ga dene and aspi ing ga dene can assume he wo oles simul aneously. The o ganiza ional elemen s a e modeled as a i ac s, using o a4mas, which is an a i ac based in as uc- u e (based in CA AgO), whe e i s -class en i ies o he sys em a e also modeled. We conside he wo de aul CA AgO a i ac s, g oupboa d and schemeboa d, bo h belonging o he o a4mas.nopl package o CA AgO amewo k. JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 3: The San Je ónimo Vege able Ga den and a Schema o a Plo Figu e 4: S uc u al Speci ica ion o SJVG social o ganiza ion ((San os e al. 2014a)) JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Physical a i ac s and he en i onmen 4.7 The en i onmen is a compu a ional o physical space in which agen s a e si ua ed, and he no ions o pe cep- ions, ac ions and in e ac ions a e de ined and de eloped, and so, he agen can pe cei e and ac ((Ricci e al. 2011)). 4.8 Physical a i ac s o he SJVG we e de eloped using he CA AgO amewo k. They a e abs ac ions abou he en i onmen , ep esen ing esou ces o ools ha agen s can ins an ia e dynamically, sha e and use as suppo in hei daily ac i i ies in SJVG, and hese ac i i ies can be indi idual o in g oup. These a i ac s in he SJVG en i onmen ep esen , o example, g ubbe , sho el, ake, plo , wa e ing can, plan e , seeds, close , clock, all hem implemen ed inCA AgO. InFigu e 5, wep esen hesequenceo agen ac ions o using physicala i ac s. Whene e he a i ac is c ea ed by EA, he agen s ecei e a message (which is ans o med in o a belie ) and hen hey use he “lookupa i ac ” ope a ion (ac ion p o ided by CA AgO) and seek he physical a i ac . Figu e 5: The En i onmen and Physical A i ac s ((San os e al. 2014a)) 4.9 Agen s also ecei e signals om he “calenda ” a i ac o execu e a ailable ac ions in he i ual en i onmen . Figu e 6 shows he Jason code o c ea e he a i ac by he EA agen using he “makea i ac ” ope a ion, sea ch- ing he agen ha will use his a i ac by he “lookupa i ac ” ope a ion and, inally, he agen ecei ing he signal by he calenda a i ac . Some P oblems wi h JaCaMo 4.10 The social o ganiza ion o SJVG is based on he pe o mance o pe iodic ou ines by he o ganiza ional oles, and also on pe iodic no ms ha egula es hei beha io s. An example can be seen in he pe iodic ou ines o an agen playing he ole o a Ga dene shown in Figu e 7 in he o m o Venn Diag am, whe e one can obse e ha a “Ga dene , o join he SJVG p ojec , has he obliga ion o pay a ee mon hly”. Fo mo e de ails on he modelling o o he pe iodic ou ines o di e en oles in he SJVG social o ganiza ion, see he Appendix. 4.11 Howe e , in he JaCaMo amewo k, as discussed in (San os e al. 2014b), he modeling o such ole ou ines canno be easily done, since he e a e no na i e ools in he pla o m ha allow his kind o speci ica ion. In he ac ual de elopmen o JaCaMo in as uc u e, he allowed p ocesses in he MAS o ganiza ion, in e ms o he goals ha mus be achie ed, ha e o be desc ibed h ough he MOISE+ model. This ool p esen s a good abs ac ion le el o speci y hese objec i es, as well as he de ini ion o a hie a chy be ween hem. Howe e , a pe iodic ou ine in ol es he achie emen o pe iodic goals (e.g., in pe iods o one mon h, one week, one day), and MOISE+ model does no ha e s uc u es o do ep esen such pe iodici y. JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 11: In eg a ion o Physical, No ma i e and Communica ion A i ac s: he mon hly paymen by a Ga dene In Figu e 14, he speci ic “ga dene ” agen called “CICERO” (see Example 2), h ough he ope a ion “lookupa - i ac ”, sea ch he a i ac g oupboa d and hen i adop s he ole o “ga dene ” in he plo subg oup in SJVG (lines 12 and 13). 5.7 Obse e ha , in lines 15, 23, 27 and 32, he “social agen s” add belie s in hei belie bases, as he con i ma ion o he c ea ion o a i ac s (“makea i ac ” ope a ion), which is sen by he “EA” agen , h ough a “.b oadcas ” command. 5.8 A e , he agen s keep seeking communica ion a i ac s (line 16 and 17), he a i ac calenda (which sends sig- nals o agen s abou a ailable ac ions in he i ual en i onmen ) (line 24), physical a i ac s (line 28) and no - ma i e a i ac s (line 33) by he “lookupa i ac ” ope a ion, which is pe o med o sea ching o an a i ac by i s name and iden i ie (wi h “lookupa i ac (a name, ida )”). Finally, hey can execu e hei ac ions wi h hese esou ces in he SJVG i ual en i onmen . 5.9 Du ing his execu ion, he ope a ion ocus (lines 13, 20, 21, 25, 29 and 34) is pe o med by he agen s, so ha hey con inue obse ing changes which may occu in hose a i ac s o e en he exclusion o any o hem in he en i onmen . 5.10 An impo an concep o he o ganiza ion is he es ablished ules se by he con ede a ion EA, in o de o help agen s o comply wi h he ules o he SJVG p ojec . The ollowing example shows some no ma i e a i ac s (obliga ion, pe mission, p ohibi ion and igh ) ha we e de eloped in o de o simula e he SJVG’s egula ion no ms. Example 4: No ma i e a i ac s 5.11 The no ms a e c ea ed h ough plans by he agen “EA” a he beginning o he simula ion. Figu e 15 shows he implemen a ion o hese plans by he agen , he ac ions con ained in hem and he c ea ion o he no ms h ough he no ma i e a i ac s. a he bo om o igu e 15, we p esen a sample simula ion. 5.12 In Figu e 15 (line 63 o he implemen a ion, in he op o he igu e, and line 4 o he simula ion, in he bo om o he igu e), he no m “paga mensalidade” (which means: “ o pay he mon hly ee”) is manda o y and he non- compliance cons i u es a se e e and cumula i e misconduc (subjec o a punishmen ). This ac ion is e i ied by hee ec o /de ec o agen (go e nmen agen “admin”), which is esponsible o moni o ing he compliance wi h he no ms and check he no ma i e a i ac s o analyze i he pe o med ac ion is in ac a iola ion. 5.13 A e his e i ica ion, asshown inFigu e16, he agen “admin” sea chin i sbelie base, o headequa e penal y and no i y he o ende hei o ense (lines 3 and 4), egis e ing i in he a i ac “penal y egis a ion (RP)” (line JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 12: Agen s and he Plo Subg oup oles in SJVG o ganiza ion Figu e 13: Speci ica ion o he Plo Subg oup in SJVG o ganiza ion 23), in o de o ha e he sanc ion applied o he agen who pe o med he p ohibi ed ac ion. By checking he numbe o cumula i e penal ies eco ded in he RP a i ac , he go e nmen agen “admin” may con ene a mee ing (assembly) o he agen s pa icipa ing in SJVG p ojec ( igu e 16, line 25), so hey can o e wi h espec o he expulsion (o pe manency) o he o ending agen om he p ojec . In he simula ed example o Figu e 16, he agen “CICERO” ecei ed 5 o es, which, in his case, is su icien o ha e i expelled om SJVG p ojec (line 31). Obse e ha i i was he case o ie, he decision is done by he go e nmen al agen . 5.14 Communica ion A i ac s mus ul ill a unc ion o media ing communica ion, ha is, hey o wa d messages o hei ecipien s, acco ding o p o ocols, o e seeing he execu ion o de o sending hese messages. In he ollowing, we p esen an example o he use o communica ion a i ac s in he SJVG p ojec . JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 14: “lookupa i ac ” and “adop e ole” ope a ions ((San os e al. 2014a)) Figu e 15: C ea ing no ms: No ma i e A i ac s ((San os e al. 2014a)) Example 5: Communica ion a i ac s 5.15 Figu e 17 shows an agen called “LUCAS”, playing he ole o an auxilia y ga dene , asking pe mission ( h ough a eques message o he go e nmen agen “admin”) o cul i a ing ees in he ga den (line 6). The agen “admin”, in eply (using a message o ype in o m), in o ms ha his ac ion is no allowed (line 9). Ano he possible communica ion is he agen “CAIO” (line 1) ha sends an in o m message o agen “admin” abou i s JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 16: a simula ion: no ms punishmen and assembly ((San os e al. 2014a)) insc ip ion in he p ojec , which does no equi e a esponse om he ecipien . Figu e 17: Communica ion A i ac s ((San os e al. 2014a)) Conclusion 6.1 This pape p esen ed some MAS-based ools de eloped in he SJVG-MAS con ex , discussing he adop ed solu- ions and in oducing some examples o simula ions. 6.2 We ound ha o beable oconside all hesui able cha ac e is icso heSJVG socialo ganiza ion (e.g., hepe i- odici y o ou ines and no ms, he in e ac ional cha ac e o he social ela ionships and se ice exchanges), we had o concei e ou mas as a mul i-dimensional BDI-like agen social sys em, composed o i e dimensions: (i) he agen s’ popula ion, (ii) he social o ganiza ion, (iii) he en i onmen , (i ) he in e ac ional/communica ion s uc u e, and ( ) he egula o y s uc u e. so, we adop ed he JaCaMo amewo k, de ining and de eloping JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 o he a i ac -based in as uc u es o deal wi h he pe iodici y modeling, communica ion ea u es and he eg- ula o y policy. 6.3 The ools discussed in his pape a e o be used, in an in e disciplina y app oach o he simula ion o he social p oduc ion and managemen p ocesses ha occu inu ban ecosys ems, inpa icula , heSJVG, con ibu ing o he analysis o he ac ual eali y o he SJVG expe imen . Acco ding o he discussions on he adop ed me hod- ology, he in es iga ion o new possible ideas ha may be applied in he con ex o he SJVG’s o ganiza ion became possible. 6.4 Fu u e wo k is conce ned wi h he de elopmen o a simula ion in e ace, so o acili a e he s udy/analysis o he possible changes in he social o ganiza ion (e.g., oles assumed by he agen s in he o ganiza ion, ac ions, beha io s, (in) o mal in e ac ion/communica ion p o ocols, egula ion no ms) ha may in e e e he social p oduc ion and managemen p ocesses. Acknowledgmen s This wo k was pa ially suppo ed by he B azilian unding agency CNPQ (Conselho Nacional de Desen ol i- men o Cien í ico e Tecnológico), unde he p oc. no. 481283/2013-7, 306970/2013-9 and 232827/2014-1. Appendix Appendix A: Pe iodic Rou ines o he Roles in he SJVG Social O ganiza ion In o de o o ganize and es ablish he beha io s o he di e en oles and oles’ ou ines o SJVG’s o ganiza- ion, as well as he equency o hese ou ines, we used he so-called ellipses, a kind o Venn diag am o se heo y. The use o ellipses helps us o analyze he pe iodici y o he oles’ ou ines, helping he unde s anding o he agen s’ beha io , as well as he iden i ica ion o in e ac ions be ween hem and he en i onmen . As an example, Figu e 18 shows he ellipses o he ou ines o he EA’s sec e a y, desc ibed as: •daily ou ines: o ecei e candida es’ documen a ions desi ing o pa icipa e in he p ojec , called he aspi ing ege able ga dene , egis e ing hem in he wai ing lis ; o ecei e ans e eques o plo pos- session. •mon hly ou ines: o ecei e mon hly ees paid by he ege able ga dene o co e cos s wi h wa e (d ip), pes con ol ma e ial, use o common ools, e c.; o in o m he mee ings; o egis e auxilia eg- e able ga dene (in o med by ege able ga dene ). •biennial ou ine: o ecei e a eques om a ga dene o con inue in he p ojec . •seasonal ou ine: o sen he ecei ed documen a ion o he ea adminis a ion. Figu es 19-22 shows he pe iodic ou ines o he se e al oles iden i ied in he SJVG social o ganiza ion. JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 18: pe iodic ou ines o he sec e a y Figu e 19: pe iodic ou ines o he adminis a ion Figu e 20: pe iodic ou ines o an aspi ing ga dene JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 21: pe iodic ou ines o an auxilia y ga dene Figu e 22: pe iodic ou ines o a echnician Appendix B: Diag ams o some ou ines in SJVG To allow a clea isualiza ion o he in e ac ions be ween he class ins ances, we use UML ac i i y diag ams. An Ac i i y Diag am is a diag am de ined by he Uni ied Modeling Language (UML), ep esen ing he lows d i en by p ocesses. I is essen ially a low cha ha shows he low o con ol om one ac i i y o ano he . Usually his in ol es he modeling o sequen ial s eps in a compu a ional p ocess. In ou wo k, we use hese diag ams o isualize he in e ac ions be ween oles o he SJVG o ganiza ion. Figu e 23 is an ac i i y diag am showing in e ac ions be ween he oles o Auxilia y Vege able Ga dene , Veg- e able Ga dene and Technician. In he ollowing we explain i b ie ly. JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 23: Ac i i y Diag am - Auxilia y Vege able Ga dene - Fi s Day Ini ially, he agen ha assumes he Auxilia y Vege able Ga dene ole a i es a he EA’s building, and i lis ens o a lec u e. This lec u e is gi en by ano he agen , playing he echnician ole, and eaches some ules on how o ha es adequa ely. This ac i i y is execu ed as soon as he echnician agen pe cei es ha e e yone has a i ed a he hall. The in e ac ion be ween each ole is accomplished h ough o al communica ion, in which he la e agen alks o e e yone. Following his in e ac ion, ano he one is pe o med be ween he Auxilia y Vege able Ga dene and he Veg- e able Ga dene agen s. The o me one eques s au ho iza ion o use a cabine , using o al communica ion. JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Finally, he la e answe s he eques gi ing he o he agen pe mission o use he eques ed cabine , gi ing i he key o access ha objec . Figu es 24 and 25 show he diag ams o he insc ip ion in he SJVG p ojec (o an aspi ing ga dene ) and some daily ou ines o a ga dene , espec i ely. Figu e 24: In eg a ion o Physical, No ma i e and Communica ion A i ac s: he aspi ing ga dene insc ip ion in he SJVG P ojec JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128 Figu e 25: In eg a ion o Physical, No ma i e and Communica ion A i ac s: he ga dene daily ou ine No es 1h p://www.ecologis asenaccion.o g/ 2“SJVG-MAS P ojec : a mas o he simula ion o he social p oduc ion and managemen p ocesses in u ban ecosys ems, he case o he San Je ónimo U ban Vege able Ga den o Se ille” (FURG, B azil; Uni e sidad o Se illa, Spain) has been de eloped unde he con ex o he social simula ion ne o Rio G ande do Sul s a e, B azil (UFRGS, FURG, UFPEL, UFSM,UNISINOS). 3“SJVG-MAS P ojec : a mas o he simula ion o he social p oduc ion and managemen p ocesses in u ban ecosys ems, he case o he San Je ónimo U ban Vege able Ga den o Se ille” (FURG, B azil; Uni e sidad o Se illa, Spain) has been de eloped unde he con ex o he social simula ion ne o Rio G ande do Sul s a e, B azil (UFRGS, FURG, UFPEL, UFSM,UNISINOS). 4The BDI (belie s, desi es, in en ions) agen a chi ec u e is a pa icula cogni i e agen model in oduced in (Rao & Geo ge 1991). JASSS, 19(3) 12, 2016 h p://jasss.soc.su ey.ac.uk/19/3/12.h ml Doi: 10.18564/jasss.3128