CASE-BASED REASONING MODEL OF THE FISH DISEASE DIAGNOSIS Zetian Fu *, Guoyong Hong 1, Jian Sun 1 1 Depatment o Mechanical Electical Engineeing, China JiLiang Univesity,HangZhou, ZheJiang Povince, P R China 3118 * Coesponding autho, Addess: China Agicultue Univesity Key Laboatoy o Moden Pecison Agicultue System Integation Peking 183 Abstact: Keywods: The system o Case Based Reason is eseached in this pape, a kind o genetic algoithm is poposed to solve the poblem o eatue vecto space, and poposes an appoach o integation o CBR AND RBR, an example about ish disease diagnose ineing integation model is given method, CBR, diagnose, easoning 1 INTRODUCTION In a geneal the knowledge o expet system have a diect impact on the peomance o the taditional expet systems, because o the incomplete knowledge acquisition, to complete the constuction and use o the easoning ules o the system model is vey diicult, constuction expet knowledge acquisition system is the "bottleneck" poblem Since ish disease diagnosis system has the easoning complexity and uncetainty, using case-based easoning (CBR) appoach can educe the diagnosis in the pocess o manmade actos, the appoach to solve the poblem is vey close to the human, the key is the method can memoies the case, which is elated to the most simila case when the system encounteed a new poblem, to a lage extent the system is dependent on the quality o the case whethe the easoning ability o ish diseases can play its due ole in this pocess, in ode to get the bette guaantee o objectivity and accuacy o the esults, how to seach
1434 Zetian Fu, Guoyong Hong, Jian Sun the expected esult in a becoming huge case is a key pat o easoning in etieval CBR system 2 THE BASIC CONCEPT OF CASE-BASED REASONING 21 The concept o CBR CBR to develop atiicial intelligence in the ule-based easoning is dieent om a mode o easoning, it ees to the use o the old cases o expeience to solve the poblem and to evaluate solutions to explain the abnomal situation o undestand the new situation Case is contextual inomation with the knowledge, the knowledge o the easoning machine can achieve its objectives in the pocess, which can play a key ole in the expeience In CBR, the cuent situation o the poblems acing the case is known as the goal, but the memoy o the case is known as the souce o the poblem o case 22 The geneal pocess o CBR CBR to deal with all the main elements ae: identiication o new issues, and inding a souce o a simila case, the case is to use this new souce o the poblem a solution, the evaluation o this pogamme, and though case study and evise Libay CBR is actually a solution to the poblem; lean om expeience, to esolve the issue o a new cycle and integation pocess Geneal CBR cycle pocess includes the ollowing ou: case etieval, euse case, case modiication and case studies Case etieval is etieved om the case with most new issues simila to the case; euse is eusable case to case etieval o inomation and knowledge to solve new poblems; case o change is modiication pogamme; case study is to lean new Expeience and put it into the existing case libay Figue 1 can be used to descibe a cycle o the CBR New case Case seach Case seached Case euse Case leaned Case leaning Old case Geneal knowledge Case modiied Fig1: CBR Cycle Case modiy Case solved
Case-Based Reasoning Model o the Fish Disease Diagnosis 1435 Accoding to the above desciption, CBR can be ound on the coe issue includes ive aspects: that case method, case etieval method, case euse methods, case modiication method and case study methods 3 THE CONTENT OF FISH DISEASE DIAGNOSIS Fish Disease Diagnosis is though a vaiety o ways and means o obseving and monitoing the ish's living conditions and in accodance with the known symptoms, wee diagnosed to identiy possible tagets o ish o to be a disease, and led to the diect cause o these diseases, to ensue that Was diagnosed taget ish in cetain seasons and the living envionment o healthy gowth Fish Disease ineence system ees to the completion o a diagnostic unction o the ish's compute system, and its main task is unde obsevation by the use to show the ish and its elated symptoms, such as the quality o inomation, identiying the ish living in wate may exist disease and its elated popeties, and to detemine the coesponding contol measues o stategies Fish Disease Detection The scene o investigation Eye detection Glass detection Wate quality o the envionment Symptoms extact State ecognition Isolation and identiication o ish diseases Fish disease type, location, in season Fish Disease Diagnosis Fish Disease Diagnosis knowledge o ules Fish disease contol stategies o suggestions Fig2: Fish Disease Diagnosis pocess
1436 Zetian Fu, Guoyong Hong, Jian Sun Fish Disease Diagnosis has thee basic attibutes, that is, symptoms, diseases and causes One is the extenal maniestations o the disease o the esults o the symptoms o the disease occued beoe o duing the couse o ish in cetain aeas may show the abnomal phenomenon, which is the cause o diseases leading to the undamental attibutes collection Theeoe, a ish disease diagnosis system in the speciic diagnosis o the pocess is shown in Figue 2 4 CASE EXPRESSION OF FISH DISEASE DIAGNOSIS REASONING Nomally, a typical case should at least include a desciption o the poblem and a desciption o the solution Fo ish disease diagnosis system, the poblem is demonstated by the ish's symptoms, such as wate quality paametes; solution is to identiy the disease, causes, and gives the coesponding teatment In the ish disease diagnostic easoning system, accoding to the chaacteistics o ish diseases and content to eigenvecto as a case o that om, in ode to acilitate the naative, we head to the seizue pocess as an example, omitted the speciic case had little to do with the case-based easoning Phenomenon and disease contol methods, such as content, only concened with ish diseases elated to the case o attibutes We use muscle, skin, abdomen, scales, head, in, gill Depatment, intestinal and othe abnomal symptoms to ceate a disease case Listed in Table 1 is the basis o these chaacteistics established by the Lei Yue Mun in the ive disease cases, in each case on behal o its state o the disease demonstated by the dieent chaacteistics o the potolio In these cases did not conside the weight chaacteistics o the impact o the case, that is the assumption that all cases o the ole and inluence in the same Table1 Cap disease case ID Disease Muscle Suace Abdominal Scales Head Fin Gill Intestinal Case1 Baekdu-mouth A1 B4 Case2 Enteitis D1 H3 Case3 Red skin A4 E1 F1 C3 Case4 Black gill B1 C1 Case5 Scabies A7 F2 I each case as a ow vecto, om the above ive cases o the case though the matix that can be set as ollows:
Case-Based Reasoning Model o the Fish Disease Diagnosis 1437 A1 B 4 D 1 H 3 C A4 E 1 F 1 C 3 (1) B 1 C 1 A7 F 2 In pactice, the dieent chaacteistics o the ole and impact o cases thee is a cetain degee o dieence, but this dieence in peomance unde dieent cicumstances to dieent degees and othes Table 1 listed in the case cited, i the consideation o the chaacteistics o eight cases o the impact and ole o the dieences, they give dieent weights, and you can get a new set o cases, as shown in table 2 Table 2 Cap disease case ID Disease Muscle Suace Abdominal Scales Head Fin Gill Intestinal Case1 Baekdu-mouth A1 B4 Weight 5 5 Case2 Enteitis D1 H3 Weight 5 Case3 Red skin A4 E1 F1 C3 Weight 25 25 25 Case4 Black gill B1 C1 Weight 5 5 Case5 Scabies A7 F2 Weight 5 5 As in the case o dieent eatues and chaacteistics o the weight may change, as in the example o Black gill disease symptoms and weight in the peomance o the head and gills, and enteitis disease symptoms and weight ae elected in the abdomen and bowel, It is an extaodinay collection o coniguation case, we use this case the lagest collection o eatue set to solve the unusual coniguation poblems We can oganize the collection eatues the lagest case chaacteistic matix: 1,1 2,1 F n,1 1, 2 2, 2 n, 2 1, m 2, m n, m In chaacteistic matix, each line epesents a case, each column epesents a case o dieent chaacteistics in case the chaacteistics o the matix o evey element o the value (1 i n,1 j ), and we do the ollowing: m i, j (2)
1438 Zetian Fu, Guoyong Hong, Jian Sun the value o chaactei stic j in case i, when chaactei stic j F i, j no deinition, when chaactei stic j F When the eatue is a eatue o the case, it said in the case o the chaacteistics o value, o no pactical signiicance, can be used abitaily to ill the appopiate value o the element in the matix in the position Similaly, we have the chaacteistics o the case in the weight also simila manne in accodance with the above oganizations to become a matix om, saying the matix o the chaacteistics o the ight matix R 1,1 2,1 n,1 1, 2 2, 2 n, 2 1, m 2, m n, m Whee: R is the chaacteistics o the ight matix; n is the collection C eatues the lagest numbe o eatues, the case o the numbe o cases in the collection Chaacteistics o the matix R, each ow epesents a case o dieent chaacteistics in the case o the chaacteistics o the ight values, the matix R o evey element o the value is (1 i n,1 j ), we do the i m, j ollowing: the value o chaactei stic j in case i, when chaactei stic j F i, j no deinition, when chaactei stic j F When the eatue j is a eatue o the case i, said, is the ight value o chaacteistics, othewise i, j In the above deinition on the basis o a case that could set the case matix collection C eatues and chaacteistics o the matix ight to buy the poduct matix: C i j (3) T F R (4) Fo example, we can use matix shown in table 2 om the collection o ish diseases o the case said: C1 C 2 C 3 C 4 C 5 A1 A4 A7 D 1 E 1 B 4 B 1 F 1 F 2 C 3 C 1 H 3 5 25 5 25 5 5 5 5 5 5 (5)
Case-Based Reasoning Model o the Fish Disease Diagnosis 1439 5 FISH DISEASE DIAGNOSIS REASONING OF THE CASE RETRIEVAL SYSTEM In geneal, CBR system is suitable o expeienced, but the lack o knowledge o the ield, while RBR system can handle ich and knowledge o the system on the ield Theeoe, the CBR and RBR will be integated into one o the two systems will have the stengths to ovecome the shotage o both, so that the entie system to achieve a highe level o intelligence, is o geat signiicance It will not only geatly enhance the lexibility and compehensive easoning ability, but also signiicantly educe the case etieval and the buden o the case We will go into two ties: the bottom o all cases; uppe deck is a typical case, each case epesents the typical one o seveal months and it is vey simila to the case, typical case o the amended ules is essential, it should be the guidance o expets in the ield To be good to amend the ules, o a simila case semantic dieence between the measues is essential, semantic wod table can be deined by the index o keywods to descibe, indexing vocabulay o keywods can be used in case the same unction and the desciption o semantic netwok nodes Application o amended ules also equies case assessment that the subsystem is the evaluation o the answe, because once the amendment is not likely to be the esult o the establishment, but also uthe changes Sometimes, the poblem can not be with a cetain ules o the ist match, now need to sepaately image to the ist o seveal ules, the inal ules will be seveal intepetations o the esults integated into the whole question to answe 51 Case index The index case mechanism is an impotant issue, the objective is to povide a case o the seach mechanism, to make the etieval in the utue in line with the need to quickly identiy the cases o cases set The index should be speciic, clea and easy identiication Case o the index can have multiple, espectively, o dieent puposes Seach algoithms should be able to use the index case in time to meet the constaints to identiy the pemise o the case This means that the case should be indexed; making the case in any need at all times can be ound In this pape, C-means clusteing algoithm to a polymeization cente on behal o a disease, accoding to the type o disease the numbe is divided into seveal polymeizations cente, the cente o polymeization index C means clusteing model, established the objective unction: P m ( ) 2 J min X C (6) 1 i 1 Whee: Cluste Cente j
144 Zetian Fu, Guoyong Hong, Jian Sun C 1 m m i 1 X ( ) i ( i 1,2,, 1,2, P ), m N ( ) m X whee: is the numbe o sample; i X expess sample i belong to ;N is the numbe o sample;p is the numbe o cluste cente(2pn-1) We use genetic algoithms to solve the poblem dynamic clusteing Step 1 chomosome stuctue Genetic algoithms set up the elevant paametes, max_gen: the lagest numbe o iteative; N: population size; length: the length o chomosomes; Pc: coss-pobability;; Pm: mutation pobability;; P: initial cluste centes;; α: the itness unction o Paametes; Step 2: goups initialization o i = 1 to N do o j = 1 to length do Xi the ist-chomosome gene = andom (, P); Endo Endo STEP 3: Calculation o itness unction o i = 1 to N do Xi chomosome calculation o the objective unction Ji; endo Accoding to Xi chomosome taget unction o the value o Ji sot; o i = 1 to N do Individual itness calculation endo o j = 1 to N do An Intoduction to calculate the cumulative endo Step 4: The Executive select Options Step 5: the implementation o coss-opeation Step 6: The Executive mutation Step 7: etain the elite stategy i (Minimum o NewGeneation <Minimum o OldGeneation) then Fathe and with the best o the individual osping to eplace the wost o the individual; endo Step 8: The dynamic changes in the numbe o cluste I (the best goup o individuals in the M-geneation, has not changed) then P = Popt + n, and to Step 2; else Running; endi P 1 (7)
Case-Based Reasoning Model o the Fish Disease Diagnosis 1441 Step 9: temination o the conditions tested i (gen <max_gen) then Ode gen = gen +1, and tun to Step 3; else Stop, the output esults endi 52 Rule-based and case etieval algoithms Rule-based and case seach algoithm is as ollows: 1: input new case index geneated in accodance with the elevant ules, ceate a new case desciption o the ules; 2: judge ules coding RULE_CODE the degee o similaity, i the same case wee diectly tageted I not identical, the value o simila popeties in the aea theshold i the theshold value, then etieves the coesponding case set; i the theshold, based on the similaity case match and then etieves the coesponding The case I the seach to etieve cases o uneasonable o no case, to 4; 3: In Seach o the pocess, the pesevation o elevant case study o key knowledge and inteest in hobbies, cedited to the inspiing knowledge base 4: selected cases involving the taget system (object to sot though), in accodance with ou neighbos to impove the method o calculating tagets unde the similaity measue; 5: Does also involves othe object, i so, the system o cases involving the next taget, o else to 5; 6: calculate the case o the similaity system; 7: The similaity o the cases is the highest esult; 8: The Case uses o expets satisied with I satisied, such cases will be listed as a candidate case set, the end o easoning 6 CONCLUSION The system based on the chaacteistics o ish diseases and content o case-based easoning is studied in this pape, we can eectively solve the poblems in the diagnosis o ish diseases, and impove the opeating eiciency o the system CBR has a cucial impact on the accuacy and pactical application, but it is diicult to detemine the ight value, based on andom genetic algoithm global seach capability, a genetic algoithm to complete the space chaacteistics o the ight to diect the seach algoithm is pesented, Fo people to solve pactical poblems is not simply elying on expeience, not simply elying on theoetical knowledge, is oten a
1442 Zetian Fu, Guoyong Hong, Jian Sun combination o both the acts, it is a combination o methods, and pactice shows the method is eective ACKNOWLEDGEMENTS Funding o this eseach was povided by China Agicultue Univesity Key Laboatoy o Moden Pecision Agicultue System Integation (P R China) The ist autho is gateul to China Agicultue Univesity o poviding he with pusuing a PhD degee REFERENCES C R Sukumaan, B P N Singh Compession o bed o apeseeds: the oil-point, Jounal o Agicultual Engineeing Reseach, 1989,42:77-84 Cheng Wang Yinghong LI Hengxi Zhang The impovement o CBR technology in the diagnosis o ailue Compute Engineeing and Applications 23(15):44~46 E Davion, A G Meieing, FJ Middendo A theoetical stess model o apeseed, Canadian Agicultual Engineeing, 1979, 21(1): 45-46 J B Noh A case-based easoning appoach to cognitive map-diven tacit knowledge management Expet Systems with Application 2(19):249~259 J M Gaell I Guiu Automatic diagnosis with genetic algoithms and case-based easoning Atiicial Intelligence in Engineeing 13(1999)367~372 Pi Sheng Deng Using case-based easoning appoach to the suppot o ill-stuctued decisions Euopean Jounal o Opeational Reseach 1996 (93)511~521 S LA Salzbug, Neaest Hype ectangle Leaning Method Machine Leaning 1991,6:251-276 S MA Byant Case-based Reasoning Appoach to Bankuptcy Pediction Modeling T W Liao A case-based easoning system o identiying ailue mechanisms Engineeing Applications o Atiicial Intelligence 2 (13):199~213