Case 40: Demand matrix & cross-scenario joint blocking
This case corresponds to Scenario 40 in USE-CASES.md and is the cross-domain regression entry point of the whole example set. It builds no new six-file set — no dedicated models, Rules interface, XML, or Service. Instead, it provides one standalone business entry-point case for each of the 25 requirements (out of 49) that are “directly covered by the 15 production templates” (requirements 1/2/4/5/7/9/11/12/13/15/17/19/20/22/24/25/28/33/34/35/38/39/42/45/47), reusing the rule scripts, XML, and Services from Cases 1-15; on top of that it adds one cross-scenario joint-blocking case that chains the Case 1 credit-admission engine with the Case 15 transaction anti-fraud engine — together forming 26 matrix regression entry points.
Industry scenario
Section titled “Industry scenario”Once rule assets grow from a handful of sets to dozens (this example set already has 39 business six-file sets and 39 XML files), the real engineering risk is no longer “is a given rule written correctly” but “after refactoring the engine or adjusting the shared execution pipeline, will previously covered business requirements be silently broken”. A coverage matrix over 49 requirements, plus one independently runnable business entry-point assertion per requirement, turns abstract rule assets into a CI-gate regression baseline: every change must answer to 25 cross-industry requirements at once.
Cross-engine chaining addresses a higher-order joint-defense need in real systems. In banks and consumer-finance institutions, credit approval and payment anti-fraud usually live in two separate systems (credit granting vs. risk control): while a customer’s late-night overseas large transaction is being blocked by anti-fraud, their credit profile may already hit a high-risk level. If the two rule sets can only fight alone, fraud intelligence cannot feed back into lending decisions. The final joint-blocking case in this case proves that, within the same JVM and the same transaction, the anti-fraud engine and the admission engine can be invoked in sequence, composing the two independent verdicts BLOCK and REJECT into a JOINT_REJECT joint decision.
The 25 matrix entry points span seven industry domains — banking, government & enterprise services, e-commerce & retail, manufacturing MES, healthcare & medical insurance, HR/OA, and IoT — from personal credit, post-loan collection, and insurance claims, to talent subsidies, tax rating, and ticket routing, to dynamic pricing, quality-inspection release, medical-insurance settlement, performance scoring, and device alarms. Each entry point constructs a realistically shaped business object and asserts the final decision, not some intermediate score.
Business pain points
Section titled “Business pain points”- Guarding against refactoring regressions as rule assets grow: once shared pipelines such as parameter binding, type conversion, and proxy caching in the engine change, the blast radius covers all historical scenarios; manual case-by-case checking is unrealistic, so an automated matrix must backstop it.
- 49 requirements need continuous verification: coverage relationships written only in documents rot as code evolves; every “directly covered” requirement must have an entry point in the tests named after its requirement number, forcing documentation and code into alignment.
- Cross-domain decisions are hard to chain: credit admission and transaction anti-fraud are two independent rule sets; real joint defense requires invoking them in sequence within one business flow and composing their verdicts — the architecture must prove multi-engine coexistence and composition.
- Regression entry points must be lightweight: copying a full six-file set per regression would create massive duplicate maintenance; the ideal is zero new rules, zero new XML, and only one new test class.
- Stability of many coexisting rule sets needs verification: loading a dozen XML files in one JVM, creating a dozen proxies, and cross-invoking them raises questions about namespace conflicts, class-loading races, or cache cross-talk that need stress-test-style verification.
- Regression assertions must stay close to business semantics: merely asserting “no exception thrown” has no protective power; business results like
PASS/REJECT, risk levels, payout amounts, and dispatch priorities must be asserted so that any policy change immediately turns the build red.
Coverage matrix & entry points
Section titled “Coverage matrix & entry points”The table below is compiled from the 26 @Test methods of CoveredScenariosMatrixTest.java and their Javadoc; requirement numbers correspond one-to-one with “Section 4: 49-scenario coverage matrix” in USE-CASES.md. The first 25 rows are one standalone business entry point per requirement; the last row is the cross-scenario joint block.
| Test method | Requirement no. / business entry point | Reused case and rules |
|---|---|---|
req01_creditAdmissionPrimePass |
Req 1: automated personal-credit approval grading — prime customer PASS / grade A / full credit limit | Case 1: RiskAdmissionRules + risk-admission.xml + RiskAdmissionService |
req02_postLoanNormalWatch |
Req 2: post-loan risk warning — normal monitoring of a non-overdue account | Case 2: PostLoanWarningRules + post-loan-warning.xml + PostLoanWarningService |
req04_antiFraudBlockOverseasNight |
Req 4: loan anti-fraud risk detection — late-night overseas large transaction blocked at 100 points | Case 15: TransactionFraudRules + transaction-fraud.xml + TransactionFraudService |
req05_cardApplicationManualReview |
Req 5: credit-card application approval grading — borderline customer routed to manual review | Case 1: RiskAdmissionRules + risk-admission.xml + RiskAdmissionService |
req07_claimLiabilityAndRatio |
Req 7: insurance-claim liability and payout ratio — small outpatient fast-claim at 85% | Case 3: ClaimAuditRules + claim-audit.xml + ClaimAuditService |
req09_overdueCollectionLitigation |
Req 9: tiered post-loan overdue collection — 95 days overdue, unreachable, large amount → litigation | Case 2: PostLoanWarningRules + post-loan-warning.xml + PostLoanWarningService |
req11_bankCardTransactionPass |
Req 11: bank-card transaction risk control — everyday small transaction released | Case 15: TransactionFraudRules + transaction-fraud.xml + TransactionFraudService |
req12_whiteAndBlackListRouting |
Req 12: dynamic whitelist/blacklist determination — trusted-merchant whitelist short-circuits to release; litigation-record blacklist is a one-vote veto | Case 15: TransactionFraudRules (whitelist short circuit) + Case 1: RiskAdmissionRules (litigation veto) |
req13_talentSubsidyApproval |
Req 13: startup/talent subsidy eligibility auto-review — PhD, category-A talent approved for 400k | Case 11: SubsidyRules + subsidy-eligibility.xml + SubsidyService |
req15_documentCompletenessCheck |
Req 15: upfront completeness check for government-service materials — incomplete materials returned for correction | Case 11: SubsidyRules + subsidy-eligibility.xml + SubsidyService |
req17_procurementCreditVeto |
Req 17: government-procurement supplier admission and dishonesty filtering — dishonesty record is a one-vote veto | Case 11: SubsidyRules + subsidy-eligibility.xml + SubsidyService |
req19_enterpriseCreditRating |
Req 19: automated enterprise credit rating — well-compliant enterprise LOW / self-check | Case 8: TaxComplianceRules + tax-compliance.xml + TaxComplianceService |
req20_approvalRoutingBySla |
Req 20: intelligent routing of administrative approval flows — security P0 ticket, 5-minute first response, manager involvement | Case 5: TicketSlaRules + ticket-sla.xml + TicketSlaService |
req22_emergencyLevelClassification |
Req 22: automatic grading of emergency incidents — 92℃ overheat CRITICAL / P0 | Case 9: DeviceAlarmRules + device-alarm.xml + DeviceAlarmService |
req24_bidPrequalificationTwoStage |
Req 24: two-stage bid prequalification — hard credential gates eliminate, survivors scored and shortlisted | Case 14: ResumeScreeningRules + resume-screening.xml + ResumeScreeningService |
req25_couponStackingControl |
Req 25: marketing-campaign stacking/mutual-exclusion control — new-user coupon mutually exclusive, VIP coupon stackable | Case 10: CouponRules + coupon.xml + CouponService |
req28_priceFloorAndMarginProtection |
Req 28: channel price floor / minimum price / spread control — discount breaking the margin floor executes at the protected price | Case 4: DynamicPricingRules + dynamic-pricing.xml + DynamicPricingService |
req33_productionQualityRelease |
Req 33: production-line multi-indicator quality inspection — high-quality batch accepted and released | Case 12: QualityInspectionRules + quality-inspection.xml + QualityInspectionService |
req34_workOrderPriorityMatrix |
Req 34: MES work-order priority ranking — P3 low-priority request gets standard SLA | Case 5: TicketSlaRules + ticket-sla.xml + TicketSlaService |
req35_equipmentHealthScoring |
Req 35: production-equipment health scoring and failure warning — resource strain at 22 points MINOR, warn without dispatch | Case 9: DeviceAlarmRules + device-alarm.xml + DeviceAlarmService |
req38_batchReleaseCompliance |
Req 38: finished-goods inbound/outbound compliance and batch release — critical defect freezes the whole batch | Case 12: QualityInspectionRules + quality-inspection.xml + QualityInspectionService |
req39_medicalInsuranceReimbursement |
Req 39: medical-insurance reimbursement item/ratio/compliance checks — employee, tertiary-A hospital, class-A drug reimbursed 3780 | Case 13: MedicalReimbursementRules + medical-reimbursement.xml + MedicalReimbursementService |
req42_kpiWeightedGrading |
Req 42: employee KPI weighted scoring and grading — grade S, coefficient 2.0 | Case 6: PerformanceRules + performance.xml + PerformanceService |
req45_resumeScreeningStrong |
Req 45: intelligent resume screening — master’s from a 985 university, full match, strongly recommended | Case 14: ResumeScreeningRules + resume-screening.xml + ResumeScreeningService |
req47_iotSensorThresholdAlarm |
Req 47: IoT sensor threshold alarm grading — gateway disconnection at 51 points MAJOR / P1 | Case 9: DeviceAlarmRules + device-alarm.xml + DeviceAlarmService |
req50_crossScenarioAdmissionAndFraudJointBlock |
Cross-scenario combination: Scenario 15 transaction anti-fraud + Scenario 1 credit admission, dual-engine chained joint blocking | Case 15: TransactionFraudRules (BLOCK, 100 points) + Case 1: RiskAdmissionRules (REJECT, grade D) |
Code implementation (complete & runnable)
Section titled “Code implementation (complete & runnable)”This case adds only one test class; place it into src/test/java/com/github/paohaijiao/demo/. Everything it depends on — the models (CreditApplication, BankTransaction, ClaimRequest, SubsidyApplication, Ticket, DeviceAlarm, Resume, CouponRequest, InventoryItem, InspectionBatch, MedicalClaim, PerformanceRecord, LoanAccount, TaxProfile, and their decision models), the rule interfaces, the XML scripts, and the Services — is reused from the existing assets of Cases 1-15, with no new model, rule, or XML needed.
CoveredScenariosMatrixTest.java
Section titled “CoveredScenariosMatrixTest.java”package com.github.paohaijiao.demo;
import com.github.paohaijiao.demo.model.AdmissionDecision;import com.github.paohaijiao.demo.model.AlarmDecision;import com.github.paohaijiao.demo.model.BankTransaction;import com.github.paohaijiao.demo.model.BonusDecision;import com.github.paohaijiao.demo.model.ClaimDecision;import com.github.paohaijiao.demo.model.ClaimRequest;import com.github.paohaijiao.demo.model.CouponDecision;import com.github.paohaijiao.demo.model.CouponRequest;import com.github.paohaijiao.demo.model.CreditApplication;import com.github.paohaijiao.demo.model.DeviceAlarm;import com.github.paohaijiao.demo.model.FraudDecision;import com.github.paohaijiao.demo.model.InspectionBatch;import com.github.paohaijiao.demo.model.InspectionDecision;import com.github.paohaijiao.demo.model.InventoryItem;import com.github.paohaijiao.demo.model.LoanAccount;import com.github.paohaijiao.demo.model.MedicalClaim;import com.github.paohaijiao.demo.model.PerformanceRecord;import com.github.paohaijiao.demo.model.PricingDecision;import com.github.paohaijiao.demo.model.ReimbursementDecision;import com.github.paohaijiao.demo.model.Resume;import com.github.paohaijiao.demo.model.ScreeningDecision;import com.github.paohaijiao.demo.model.SlaDecision;import com.github.paohaijiao.demo.model.SubsidyApplication;import com.github.paohaijiao.demo.model.SubsidyDecision;import com.github.paohaijiao.demo.model.TaxProfile;import com.github.paohaijiao.demo.model.TaxRiskDecision;import com.github.paohaijiao.demo.model.Ticket;import com.github.paohaijiao.demo.model.WarningDecision;import com.github.paohaijiao.demo.service.ClaimAuditService;import com.github.paohaijiao.demo.service.CouponService;import com.github.paohaijiao.demo.service.DeviceAlarmService;import com.github.paohaijiao.demo.service.DynamicPricingService;import com.github.paohaijiao.demo.service.MedicalReimbursementService;import com.github.paohaijiao.demo.service.PerformanceService;import com.github.paohaijiao.demo.service.PostLoanWarningService;import com.github.paohaijiao.demo.service.QualityInspectionService;import com.github.paohaijiao.demo.service.ResumeScreeningService;import com.github.paohaijiao.demo.service.RiskAdmissionService;import com.github.paohaijiao.demo.service.SubsidyService;import com.github.paohaijiao.demo.service.TaxComplianceService;import com.github.paohaijiao.demo.service.TicketSlaService;import com.github.paohaijiao.demo.service.TransactionFraudService;import com.github.paohaijiao.xml.JQuickJava;import org.junit.Assert;import org.junit.Before;import org.junit.Test;
/** * 场景 40：诉求矩阵「直接覆盖」独立入口回归 + 跨场景联合拦截。 * <p> * 前 25 个用例对应 USE-CASES.md 矩阵中由 15 套生产样板直接承接的诉求 * （诉求 1/2/4/5/7/9/11/12/13/15/17/19/20/22/24/25/28/33/34/35/38/39/42/45/47）， * 每个诉求一个独立业务入口；最后一个用例演示场景 1 信贷准入与场景 15 交易反欺诈 * 双引擎串联的联合拦截，构成全部 50 个独立业务场景入口。 */public class CoveredScenariosMatrixTest {
private RiskAdmissionService riskAdmissionService; private PostLoanWarningService postLoanWarningService; private ClaimAuditService claimAuditService; private DynamicPricingService dynamicPricingService; private TicketSlaService ticketSlaService; private PerformanceService performanceService; private TaxComplianceService taxComplianceService; private DeviceAlarmService deviceAlarmService; private CouponService couponService; private SubsidyService subsidyService; private QualityInspectionService qualityInspectionService; private MedicalReimbursementService medicalReimbursementService; private ResumeScreeningService resumeScreeningService; private TransactionFraudService transactionFraudService;
@Before public void setUp() { this.riskAdmissionService = new RiskAdmissionService(JQuickJava.create() .createApi(RiskAdmissionRules.class, "demo/risk-admission.xml")); this.postLoanWarningService = new PostLoanWarningService(JQuickJava.create() .createApi(PostLoanWarningRules.class, "demo/post-loan-warning.xml")); this.claimAuditService = new ClaimAuditService(JQuickJava.create() .createApi(ClaimAuditRules.class, "demo/claim-audit.xml")); this.dynamicPricingService = new DynamicPricingService(JQuickJava.create() .createApi(DynamicPricingRules.class, "demo/dynamic-pricing.xml")); this.ticketSlaService = new TicketSlaService(JQuickJava.create() .createApi(TicketSlaRules.class, "demo/ticket-sla.xml")); this.performanceService = new PerformanceService(JQuickJava.create() .createApi(PerformanceRules.class, "demo/performance.xml")); this.taxComplianceService = new TaxComplianceService(JQuickJava.create() .createApi(TaxComplianceRules.class, "demo/tax-compliance.xml")); this.deviceAlarmService = new DeviceAlarmService(JQuickJava.create() .createApi(DeviceAlarmRules.class, "demo/device-alarm.xml")); this.couponService = new CouponService(JQuickJava.create() .createApi(CouponRules.class, "demo/coupon.xml")); this.subsidyService = new SubsidyService(JQuickJava.create() .createApi(SubsidyRules.class, "demo/subsidy-eligibility.xml")); this.qualityInspectionService = new QualityInspectionService(JQuickJava.create() .createApi(QualityInspectionRules.class, "demo/quality-inspection.xml")); this.medicalReimbursementService = new MedicalReimbursementService(JQuickJava.create() .createApi(MedicalReimbursementRules.class, "demo/medical-reimbursement.xml")); this.resumeScreeningService = new ResumeScreeningService(JQuickJava.create() .createApi(ResumeScreeningRules.class, "demo/resume-screening.xml")); this.transactionFraudService = new TransactionFraudService(JQuickJava.create() .createApi(TransactionFraudRules.class, "demo/transaction-fraud.xml")); }
/** * 诉求 1：个人信贷自动审批分级 —— 优质客户 PASS/A 级/足额授信 */ @Test public void req01_creditAdmissionPrimePass() { CreditApplication a = new CreditApplication(); a.setApplicationId("M-REQ01"); a.setApplicantName("张伟"); a.setAge(35); a.setCompanyType("SOE"); a.setEmployedMonths(84); a.setSocialSecurityMonths(60); a.setMonthlyIncome(30000.0); a.setMonthlyDebt(4500.0); a.setOverdueCount(0); a.setMaxOverdueDays(0); a.setCreditQueryCount3M(2); a.setLoanAmount(300000.0); a.setLoanTermMonths(36); a.setCollateralType("HOUSE"); a.setCollateralValue(600000.0); a.setHasJudicialRecord(false); a.setCityTier("TIER_1");
AdmissionDecision d = riskAdmissionService.evaluate(a);
Assert.assertEquals("PASS", d.getAdmit()); Assert.assertEquals("A", d.getRiskLevel()); Assert.assertEquals(300000.0, d.getSuggestedLimit(), 0.001); }
/** * 诉求 2：贷后风险预警 —— 未逾期账户正常监测 */ @Test public void req02_postLoanNormalWatch() { LoanAccount a = new LoanAccount(); a.setLoanNo("M-REQ02"); a.setCustomerName("陈静"); a.setProductType("CONSUMER"); a.setRemainingPrincipal(50000.0); a.setOverdueAmount(0.0); a.setOverdueDays(0); a.setBaseRate(0.08); a.setCollateralType("NONE"); a.setCollateralValue(0.0); a.setContactSuccess(true); a.setLastContactDays(0); a.setOverdueHistoryCount(0); a.setWarningCount(0);
WarningDecision d = postLoanWarningService.evaluate(a);
Assert.assertEquals("BLUE", d.getWarningLevel()); Assert.assertEquals(0, d.getWarningScore()); Assert.assertTrue(d.getActions().contains("账户状态正常，无需催收")); }
/** * 诉求 4：贷款反欺诈风险识别 —— 深夜境外大额交易 100 分拦截 */ @Test public void req04_antiFraudBlockOverseasNight() { BankTransaction t = txn("M-REQ04", 30000.0, 2, "ONLINE", true, true, false, 6, 60000.0, true, true, false, 1, 400, 8000.0, false);
FraudDecision d = transactionFraudService.evaluate(t);
Assert.assertEquals(100, d.getRiskScore()); Assert.assertEquals("HIGH", d.getRiskLevel()); Assert.assertEquals("BLOCK", d.getAction()); Assert.assertTrue(d.isNeedAlert()); }
/** * 诉求 5：信用卡进件审批分级 —— 边界客户转人工复核 */ @Test public void req05_cardApplicationManualReview() { CreditApplication a = new CreditApplication(); a.setApplicationId("M-REQ05"); a.setApplicantName("王芳"); a.setAge(28); a.setCompanyType("PRIVATE"); a.setEmployedMonths(20); a.setSocialSecurityMonths(12); a.setMonthlyIncome(15000.0); a.setMonthlyDebt(6900.0); a.setOverdueCount(1); a.setMaxOverdueDays(20); a.setCreditQueryCount3M(4); a.setLoanAmount(100000.0); a.setLoanTermMonths(24); a.setCollateralType("VEHICLE"); a.setCollateralValue(130000.0); a.setHasJudicialRecord(false); a.setCityTier("TIER_2");
AdmissionDecision d = riskAdmissionService.evaluate(a);
Assert.assertEquals("MANUAL_REVIEW", d.getAdmit()); Assert.assertEquals("C", d.getRiskLevel()); Assert.assertEquals(57, d.getTotalScore()); }
/** * 诉求 7：保险理赔责任与赔付比例 —— 门诊小额快赔 85% */ @Test public void req07_claimLiabilityAndRatio() { ClaimRequest r = new ClaimRequest(); r.setClaimNo("M-REQ07"); r.setPolicyNo("POL-M07"); r.setInsuredName("刘洋"); r.setProductType("MEDICAL"); r.setHospitalLevel("THIRD_A"); r.setClaimAmount(3200.0); r.setInvoiceAmount(3200.0); r.setDeductible(300.0); r.setInsuredDays(400); r.setWaitingPeriodDays(30); r.setPriorClaimCount(0); r.setFraudFlagCount(0); r.setAccident(false); r.setDocumentComplete(true);
ClaimDecision d = claimAuditService.audit(r);
Assert.assertEquals("APPROVE", d.getAuditResult()); Assert.assertEquals(2465.0, d.getPayableAmount(), 0.001); Assert.assertEquals(0.85, d.getReimburseRatio(), 0.0001); Assert.assertTrue(d.isFastTrack()); }
/** * 诉求 9：逾期贷后分层催收 —— 95 天失联大额转诉讼 */ @Test public void req09_overdueCollectionLitigation() { LoanAccount a = new LoanAccount(); a.setLoanNo("M-REQ09"); a.setCustomerName("张强"); a.setProductType("CONSUMER"); a.setRemainingPrincipal(365000.0); a.setOverdueAmount(200000.0); a.setOverdueDays(95); a.setBaseRate(0.10); a.setCollateralType("NONE"); a.setCollateralValue(0.0); a.setContactSuccess(false); a.setLastContactDays(30); a.setOverdueHistoryCount(0); a.setWarningCount(0);
WarningDecision d = postLoanWarningService.evaluate(a);
Assert.assertEquals("ORANGE", d.getWarningLevel()); Assert.assertEquals(4, d.getOverdueStage()); Assert.assertEquals("LITIGATION", d.getCollectChannel()); Assert.assertTrue(d.isNeedLitigation()); }
/** * 诉求 11：银行卡交易风控 —— 日常小额交易放行 */ @Test public void req11_bankCardTransactionPass() { BankTransaction t = txn("M-REQ11", 200.0, 14, "POS", false, false, false, 1, 200.0, false, false, true, 0, 400, 5000.0, false);
FraudDecision d = transactionFraudService.evaluate(t);
Assert.assertEquals(0, d.getRiskScore()); Assert.assertEquals("LOW", d.getRiskLevel()); Assert.assertEquals("PASS", d.getAction()); }
/** * 诉求 12：白名单/黑名单动态判定 —— 受信商户白名单短路放行；涉诉黑名单一票否决 */ @Test public void req12_whiteAndBlackListRouting() { BankTransaction white = txn("M-REQ12-W", 200.0, 14, "POS", false, false, false, 1, 200.0, false, false, true, 0, 400, 5000.0, true); FraudDecision wd = transactionFraudService.evaluate(white); Assert.assertEquals("PASS", wd.getAction()); Assert.assertTrue(wd.getIndicators().contains("受信商户白名单交易，直接放行"));
CreditApplication black = new CreditApplication(); black.setApplicationId("M-REQ12-B"); black.setApplicantName("陈刚"); black.setAge(38); black.setCompanyType("PRIVATE"); black.setEmployedMonths(48); black.setSocialSecurityMonths(36); black.setMonthlyIncome(20000.0); black.setMonthlyDebt(3000.0); black.setOverdueCount(0); black.setMaxOverdueDays(0); black.setCreditQueryCount3M(3); black.setLoanAmount(150000.0); black.setLoanTermMonths(36); black.setCollateralType("HOUSE"); black.setCollateralValue(300000.0); black.setHasJudicialRecord(true); black.setCityTier("TIER_2"); AdmissionDecision bd = riskAdmissionService.evaluate(black); Assert.assertEquals("REJECT", bd.getAdmit()); Assert.assertTrue(bd.getRejectReasons().contains("存在涉诉/被执行记录，触发一票否决")); }
/** * 诉求 13：创业/人才补贴资格自动审核 —— 博士 A 类人才核定 40 万 */ @Test public void req13_talentSubsidyApproval() { SubsidyApplication app = subsidyApp("M-REQ13"); app.setEducation("PHD"); app.setAge(28); app.setLocalSocialSecurityMonths(48); app.setTalentCategory("A"); app.setStartupRegistered(true); app.setStartupEmployees(12);
SubsidyDecision d = subsidyService.review(app);
Assert.assertEquals("QUALIFIED", d.getStage()); Assert.assertEquals(110, d.getQualificationScore()); Assert.assertEquals(400000.0, d.getSubsidyAmount(), 0.001); }
/** * 诉求 15：政务办事材料完整性前置校验 —— 材料不齐退回补正 */ @Test public void req15_documentCompletenessCheck() { SubsidyApplication app = subsidyApp("M-REQ15"); app.setEducation("PHD"); app.setAge(30); app.setLocalSocialSecurityMonths(48); app.setTalentCategory("A"); app.setStartupRegistered(false); app.setStartupEmployees(0); app.setDocumentsComplete(false);
SubsidyDecision d = subsidyService.review(app);
Assert.assertEquals("MANUAL_REVIEW", d.getStage()); Assert.assertEquals(0.0, d.getSubsidyAmount(), 0.001); Assert.assertTrue(d.getReasons().contains("申报材料不齐全，退回补正后转人工复核")); }
/** * 诉求 17：政府采购供应商准入与失信过滤 —— 失信记录一票否决 */ @Test public void req17_procurementCreditVeto() { SubsidyApplication app = subsidyApp("M-REQ17"); app.setEducation("MASTER"); app.setAge(30); app.setLocalSocialSecurityMonths(40); app.setTalentCategory("B"); app.setStartupRegistered(false); app.setStartupEmployees(0); app.setCreditViolation(true);
SubsidyDecision d = subsidyService.review(app);
Assert.assertEquals("REJECTED", d.getStage()); Assert.assertTrue(d.getReasons().contains("存在失信被执行记录，触发一票否决")); }
/** * 诉求 19：企业信用等级自动评级 —— 合规良好企业 LOW/自查 */ @Test public void req19_enterpriseCreditRating() { TaxProfile p = new TaxProfile(); p.setCompanyId("M-REQ19"); p.setCompanyName("某科技股份有限公司"); p.setTaxpayerType("GENERAL"); p.setTaxPeriod("2026Q2"); p.setLateCount(0); p.setMissingInvoice(0); p.setTaxBurdenRate(0.05); p.setIndustryAvgBurden(0.04); p.setRevenue(20000000.0); p.setInputTaxRatio(0.70); p.setHasAbnormalInvoice(false); p.setYearsSinceRegister(5); p.setCreditRating("A");
TaxRiskDecision d = taxComplianceService.evaluate(p);
Assert.assertEquals(0, d.getRiskScore()); Assert.assertEquals("LOW", d.getRiskLevel()); Assert.assertEquals("SELF_CHECK", d.getAuditLevel()); }
/** * 诉求 20：行政审批流程智能路由 —— 安全类 P0 工单 5 分钟首响、经理介入 */ @Test public void req20_approvalRoutingBySla() { Ticket t = new Ticket(); t.setTicketNo("M-REQ20"); t.setTitle("支付网关不可用"); t.setPriority("P0"); t.setCategory("INCIDENT"); t.setCustomerLevel("VIP"); t.setCreateChannel("PHONE"); t.setSecurityRelated(true); t.setReopenedTimes(0); t.setAssigneeLoad(8); t.setHasSlaBreachHistory(false); t.setElapsedMinutes(20);
SlaDecision d = ticketSlaService.evaluate(t);
Assert.assertEquals("VIP", d.getSlaLevel()); Assert.assertEquals(5, d.getResponseMinutes()); Assert.assertEquals(10, d.getEscalateMinutes()); Assert.assertTrue(d.isNeedManager()); }
/** * 诉求 22：应急突发事件等级自动分级 —— 92℃ 过热 CRITICAL/P0 */ @Test public void req22_emergencyLevelClassification() { DeviceAlarm a = new DeviceAlarm(); a.setDeviceId("M-REQ22"); a.setDeviceName("炼钢炉温控传感器"); a.setDeviceType("SENSOR"); a.setCriticality("HIGH"); a.setTemperature(92.0); a.setCpuUsage(96); a.setMemoryUsage(70); a.setErrorCount(22); a.setConnectionLost(false); a.setLastHeartbeatSeconds(60); a.setBatteryLevel(80); a.setVibrationLevel(1.0);
AlarmDecision d = deviceAlarmService.evaluate(a);
Assert.assertEquals("CRITICAL", d.getAlarmLevel()); Assert.assertEquals("P0", d.getDispatchPriority()); Assert.assertEquals("ISOLATE", d.getAutoRemediation()); }
/** * 诉求 24：招投标资格预审两段式 —— 资质硬门槛淘汰 + 达标者评分入围 */ @Test public void req24_bidPrequalificationTwoStage() { Resume unqualified = resume("M-REQ24-U", "BACHELOR", "MASTER", "211", 5, 3, 4, 5); unqualified.setExpectedSalary(16000.0); unqualified.setSalaryMin(15000.0); unqualified.setSalaryMax(25000.0); unqualified.setCurrentCity("上海"); unqualified.setJobCity("上海"); unqualified.setWillingToRelocate(false); unqualified.setJobHoppingCount(1); unqualified.setGapMonths(5); unqualified.setCertified(true); unqualified.setEnglishLevel("CET6"); ScreeningDecision d1 = resumeScreeningService.screen(unqualified); Assert.assertFalse(d1.isHardCheckPass()); Assert.assertEquals("REJECT", d1.getResult());
Resume qualified = resume("M-REQ24-Q", "MASTER", "MASTER", "985", 8, 3, 5, 5); qualified.setExpectedSalary(18000.0); qualified.setSalaryMin(15000.0); qualified.setSalaryMax(25000.0); qualified.setCurrentCity("上海"); qualified.setJobCity("上海"); qualified.setWillingToRelocate(false); qualified.setJobHoppingCount(0); qualified.setGapMonths(2); qualified.setCertified(true); qualified.setEnglishLevel("CET6"); ScreeningDecision d2 = resumeScreeningService.screen(qualified); Assert.assertTrue(d2.isHardCheckPass()); Assert.assertEquals("INTERVIEW", d2.getResult()); Assert.assertEquals(90, d2.getTotalScore()); }
/** * 诉求 25：营销活动叠加/互斥控制 —— 新客券互斥、VIP 券可叠加 */ @Test public void req25_couponStackingControl() { CouponRequest n = couponReq("M-REQ25-N", "NORMAL", 88.0, 0, "APPAREL", 2, true); CouponDecision nd = couponService.issue(n); Assert.assertEquals("NEW_USER", nd.getCouponType()); Assert.assertFalse(nd.isStackable());
CouponRequest v = couponReq("M-REQ25-V", "VIP", 2500.0, 15, "DIGITAL", 900, false); v.setLastOrderDays(10); v.setAvgOrderAmount(800.0); CouponDecision vd = couponService.issue(v); Assert.assertEquals("VIP_EXCLUSIVE", vd.getCouponType()); Assert.assertTrue(vd.isStackable()); }
/** * 诉求 28：渠道限价/最低价/价差管控 —— 折扣击穿毛利线时按保护价执行 */ @Test public void req28_priceFloorAndMarginProtection() { InventoryItem item = new InventoryItem(); item.setSku("M-REQ28"); item.setSkuName("滞销小家电"); item.setCategory("家电"); item.setStockAgeDays(120); item.setStockQty(100); item.setCostPrice(60.0); item.setCurrentPrice(100.0); item.setRecentSalesQty30d(10); item.setSeasonFactor("NORMAL"); item.setPerishable(false); item.setWarehouseType("MAIN"); item.setMarginFloor(0.20);
PricingDecision d = dynamicPricingService.price(item);
Assert.assertEquals(0.50, d.getFinalDiscount(), 0.001); Assert.assertEquals(72.0, d.getSuggestedPrice(), 0.001); Assert.assertTrue(d.getNotes().contains("清仓折扣触及最低毛利保护线，按保护价执行")); }
/** * 诉求 33：产线质检多指标判定 —— 优质批次合格放行 */ @Test public void req33_productionQualityRelease() { InspectionBatch b = inspectionBatch("M-REQ33", 200, 0, 0, 1, 10.01, 10.0, 0.05, 97);
InspectionDecision d = qualityInspectionService.inspect(b);
Assert.assertEquals("ACCEPT", d.getQualityLevel()); Assert.assertTrue(d.isReleaseAllowed()); Assert.assertEquals("NEXT_PROCESS", d.getDisposition()); }
/** * 诉求 34：MES 工单优先级智能排序 —— P3 低优请求标准 SLA */ @Test public void req34_workOrderPriorityMatrix() { Ticket t = new Ticket(); t.setTicketNo("M-REQ34"); t.setTitle("修改头像失败"); t.setPriority("P3"); t.setCategory("REQUEST"); t.setCustomerLevel("NORMAL"); t.setCreateChannel("APP"); t.setSecurityRelated(false); t.setReopenedTimes(0); t.setAssigneeLoad(3); t.setHasSlaBreachHistory(false); t.setElapsedMinutes(100);
SlaDecision d = ticketSlaService.evaluate(t);
Assert.assertEquals("STANDARD", d.getSlaLevel()); Assert.assertEquals(240, d.getResponseMinutes()); Assert.assertEquals(7200, d.getResolveMinutes()); Assert.assertFalse(d.isNeedEscalate()); }
/** * 诉求 35：生产设备健康度评分与故障预警 —— 资源紧张 22 分 MINOR，预警不派单 */ @Test public void req35_equipmentHealthScoring() { DeviceAlarm a = new DeviceAlarm(); a.setDeviceId("M-REQ35"); a.setDeviceName("图像识别节点"); a.setDeviceType("CAMERA"); a.setCriticality("MEDIUM"); a.setTemperature(45.0); a.setCpuUsage(88); a.setMemoryUsage(92); a.setErrorCount(6); a.setConnectionLost(false); a.setLastHeartbeatSeconds(30); a.setBatteryLevel(60); a.setVibrationLevel(1.0);
AlarmDecision d = deviceAlarmService.evaluate(a);
Assert.assertEquals(22, d.getAlarmScore()); Assert.assertEquals("MINOR", d.getAlarmLevel()); Assert.assertFalse(d.isNeedDispatch()); Assert.assertEquals("RATE_LIMIT", d.getAutoRemediation()); }
/** * 诉求 38：成品出入库合规与批次放行 —— 致命缺陷整批冻结不予放行 */ @Test public void req38_batchReleaseCompliance() { InspectionBatch b = inspectionBatch("M-REQ38", 100, 1, 0, 2, 10.02, 10.0, 0.05, 80);
InspectionDecision d = qualityInspectionService.inspect(b);
Assert.assertEquals("SCRAP", d.getQualityLevel()); Assert.assertFalse(d.isReleaseAllowed()); Assert.assertEquals("SCRAP_BIN", d.getDisposition()); }
/** * 诉求 39：医保报销项目/比例/合规校验 —— 职工三甲甲类报销 3780 */ @Test public void req39_medicalInsuranceReimbursement() { MedicalClaim c = new MedicalClaim(); c.setClaimNo("M-REQ39"); c.setPatientName("患者"); c.setInsuranceType("EMPLOYEE"); c.setHospitalLevel("THIRD_A"); c.setMedicalCategory("DRUG_A"); c.setTotalExpense(5000.0); c.setEmergency(false); c.setOffSite(false); c.setFiled(false); c.setAnnualAccumulated(0.0);
ReimbursementDecision d = medicalReimbursementService.settle(c);
Assert.assertEquals("APPROVE", d.getResult()); Assert.assertEquals(0.9, d.getReimburseRatio(), 0.0001); Assert.assertEquals(3780.0, d.getPayableAmount(), 0.001); }
/** * 诉求 42：员工绩效 KPI 加权计分与等级评定 —— S 级系数 2.0 */ @Test public void req42_kpiWeightedGrading() { PerformanceRecord r = new PerformanceRecord(); r.setEmployeeId("M-REQ42"); r.setEmployeeName("张伟"); r.setDept("销售部"); r.setKpiScore(95.0); r.setProjectScore(90.0); r.setTeamScore(85.0); r.setAttendanceScore(96.0); r.setManager(false); r.setHasViolation(false); r.setSalesAmount(600000.0); r.setBaseSalary(20000.0);
BonusDecision d = performanceService.evaluate(r);
Assert.assertEquals(92.65, d.getTotalScore(), 0.001); Assert.assertEquals("S", d.getGrade()); Assert.assertEquals(2.0, d.getBonusCoefficient(), 0.001); Assert.assertEquals(40000.0, d.getBonusAmount(), 0.001); }
/** * 诉求 45：招聘简历智能初筛 —— 硕士 985 全匹配强烈推荐 */ @Test public void req45_resumeScreeningStrong() { Resume r = resume("M-REQ45", "MASTER", "MASTER", "985", 8, 3, 5, 5); r.setExpectedSalary(18000.0); r.setSalaryMin(15000.0); r.setSalaryMax(25000.0); r.setCurrentCity("上海"); r.setJobCity("上海"); r.setWillingToRelocate(false); r.setJobHoppingCount(0); r.setGapMonths(2); r.setCertified(true); r.setEnglishLevel("CET6");
ScreeningDecision d = resumeScreeningService.screen(r);
Assert.assertTrue(d.isHardCheckPass()); Assert.assertEquals("STRONG", d.getRecommendLevel()); Assert.assertEquals("INTERVIEW", d.getResult()); }
/** * 诉求 47：IoT 传感器数据阈值告警分级 —— 网关掉连 51 分 MAJOR/P1 */ @Test public void req47_iotSensorThresholdAlarm() { DeviceAlarm a = new DeviceAlarm(); a.setDeviceId("M-REQ47"); a.setDeviceName("园区闸机网关"); a.setDeviceType("GATEWAY"); a.setCriticality("HIGH"); a.setTemperature(40.0); a.setCpuUsage(50); a.setMemoryUsage(50); a.setErrorCount(10); a.setConnectionLost(true); a.setLastHeartbeatSeconds(600); a.setBatteryLevel(15); a.setVibrationLevel(0.5);
AlarmDecision d = deviceAlarmService.evaluate(a);
Assert.assertEquals(51, d.getAlarmScore()); Assert.assertEquals("MAJOR", d.getAlarmLevel()); Assert.assertEquals("NETWORK", d.getAlarmType()); Assert.assertEquals("P1", d.getDispatchPriority()); }
/** * 跨场景组合：场景 15 交易反欺诈 + 场景 1 信贷准入双引擎串联联合拦截。 * 同一客户深夜境外大额交易被反欺诈拦截、信贷画像命中高风险 → 联合决策拒绝放款 */ @Test public void req50_crossScenarioAdmissionAndFraudJointBlock() { BankTransaction txn = txn("M-REQ50-T", 30000.0, 2, "ONLINE", true, true, false, 6, 60000.0, true, true, false, 1, 400, 8000.0, false); FraudDecision fraud = transactionFraudService.evaluate(txn);
CreditApplication app = new CreditApplication(); app.setApplicationId("M-REQ50-C"); app.setApplicantName("李强"); app.setAge(31); app.setCompanyType("SELF_EMPLOYED"); app.setEmployedMonths(8); app.setSocialSecurityMonths(0); app.setMonthlyIncome(12000.0); app.setMonthlyDebt(9000.0); app.setOverdueCount(6); app.setMaxOverdueDays(120); app.setCreditQueryCount3M(12); app.setLoanAmount(200000.0); app.setLoanTermMonths(24); app.setCollateralType("NONE"); app.setCollateralValue(0.0); app.setHasJudicialRecord(false); app.setCityTier("TIER_3"); AdmissionDecision admission = riskAdmissionService.evaluate(app);
boolean fraudBlocked = "BLOCK".equals(fraud.getAction()); boolean creditRejected = "REJECT".equals(admission.getAdmit()); Assert.assertTrue(fraudBlocked); Assert.assertTrue(creditRejected); Assert.assertEquals(100, fraud.getRiskScore()); Assert.assertEquals("D", admission.getRiskLevel()); Assert.assertTrue(fraudBlocked || creditRejected); Assert.assertEquals("JOINT_REJECT", fraudBlocked && creditRejected ? "JOINT_REJECT" : "SINGLE_REJECT"); }
private BankTransaction txn(String txnNo, double amount, int hour, String channel, boolean overseas, boolean crossCity, boolean travelMode, int count1h, double amount1h, boolean newMerchant, boolean deviceChanged, boolean cardPresent, int failCount24h, int accountAgeDays, double historicalMax, boolean trustedMerchant) { BankTransaction t = new BankTransaction(); t.setTxnNo(txnNo); t.setCardNo("6222 **** **** 0001"); t.setAmount(amount); t.setTxnHour(hour); t.setChannel(channel); t.setOverseas(overseas); t.setCrossCity(crossCity); t.setTravelMode(travelMode); t.setTxnCount1h(count1h); t.setTxnAmount1h(amount1h); t.setNewMerchant(newMerchant); t.setDeviceChanged(deviceChanged); t.setCardPresent(cardPresent); t.setFailCount24h(failCount24h); t.setAccountAgeDays(accountAgeDays); t.setHistoricalMaxAmount(historicalMax); t.setTrustedMerchant(trustedMerchant); return t; }
private SubsidyApplication subsidyApp(String applyNo) { SubsidyApplication app = new SubsidyApplication(); app.setApplyNo(applyNo); app.setApplicantName("申请人"); app.setSubsidyType("SETTLEMENT"); app.setCreditViolation(false); app.setCriminalRecord(false); app.setReceivedSimilarSubsidy(false); app.setDocumentsComplete(true); return app; }
private CouponRequest couponReq(String userId, String level, double orderAmount, int orderCount, String category, int joinDays, boolean newUser) { CouponRequest r = new CouponRequest(); r.setUserId(userId); r.setMemberLevel(level); r.setOrderAmount(orderAmount); r.setOrderCount(orderCount); r.setCategory(category); r.setJoinDays(joinDays); r.setLastOrderDays(0); r.setAvgOrderAmount(0.0); r.setNewUser(newUser); r.setHasUsedCoupon(false); r.setCampaignActive(false); r.setStockLeft(1000); r.setChannel("APP"); r.setRiskFlag(false); return r; }
private InspectionBatch inspectionBatch(String no, int sampleSize, int critical, int major, int minor, double measured, double target, double tolerance, int surface) { InspectionBatch b = new InspectionBatch(); b.setBatchNo(no); b.setProductCode("P-001"); b.setProductName("精密轴套"); b.setWorkshop("一号车间"); b.setSampleSize(sampleSize); b.setCriticalDefects(critical); b.setMajorDefects(major); b.setMinorDefects(minor); b.setMeasuredSize(measured); b.setTargetSize(target); b.setTolerance(tolerance); b.setSurfaceScore(surface); b.setEquipmentStatus("NORMAL"); b.setFirstArticle(false); return b; }
private Resume resume(String name, String education, String requiredEducation, String schoolTier, int years, int requiredYears, int matched, int required) { Resume r = new Resume(); r.setCandidateName(name); r.setPositionName("Java 开发工程师"); r.setEducationLevel(education); r.setRequiredEducation(requiredEducation); r.setSchoolTier(schoolTier); r.setYearsExperience(years); r.setRequiredYears(requiredYears); r.setSkillMatchCount(matched); r.setRequiredSkillCount(required); r.setChannel("ONLINE"); return r; }}Dependency note: in setUp, the test loads the XML rules of 14 production templates in one pass (risk-admission, post-loan-warning, claim-audit, dynamic-pricing, ticket-sla, performance, tax-compliance, device-alarm, coupon, subsidy-eligibility, quality-inspection, medical-reimbursement, resume-screening, transaction-fraud) and wires the corresponding Services, reusing all models and rule assets from Cases 1-15 with no new files. Run command:
mvn test -Dtest=CoveredScenariosMatrixTestRun & verified results
Section titled “Run & verified results”Run at the project root:
mvn test -Dtest=CoveredScenariosMatrixTestThis test class is the regression entry point for 26 cases: 25 reqXX_* methods are each named after a requirement number and correspond to one “directly covered” requirement in the matrix, and 1 req50_crossScenarioAdmissionAndFraudJointBlock method verifies the dual-engine chain. Every case asserts the final business verdict — for example, Req 1 asserts PASS / grade A / full 300k credit limit, Req 4 asserts the anti-fraud 100-point BLOCK, Req 39 asserts employee tertiary-A class-A reimbursement of 3780 yuan, and the joint-blocking case asserts both the anti-fraud BLOCK (100 points) and the admission REJECT (grade D), composed into JOINT_REJECT. If the engine’s shared pipeline or any reused rule is mis-edited, the corresponding requirement entry point fails immediately.
Together with the 260 cases behind all 40 scenario entry points in USE-CASES.md, it forms the complete regression set; the 26 cases in this class are the only cross-domain entry points organized by requirement number and can be used directly as the CI gate command for engine upgrades and refactoring.
Technical highlights
Section titled “Technical highlights”- One rule asset reused by many requirements: Case 9 device alarms cover Reqs 22/35/47, Case 5 ticket SLA covers Reqs 20/34, Case 11 subsidy review covers Reqs 13/15/17, Case 12 quality inspection covers Reqs 33/38, Case 14 resume screening covers Reqs 24/45, Case 15 anti-fraud covers Reqs 4/11/12, and Case 1 credit admission covers Reqs 1/5/12 — 25 requirements depend on only 14 templates; the reuse rate is the asset value.
- Dual-engine chaining proves rules compose across business domains: the joint-blocking case invokes
TransactionFraudServiceandRiskAdmissionServicein sequence within one test flow, composing the payment-sideBLOCKand the credit-sideREJECTintoJOINT_REJECT, verifying the rule engine’s orchestrability as an embedded capability in real joint-defense pipelines. - Many rule sets coexist in one JVM:
setUploads 14 XML files, creates 14 proxies, and injects 14 Services at once; namespaces are isolated by fully qualified interface name, and proxy caches and concurrent invocations show no cross-talk — direct evidence for embedding multiple business-domain rule sets into a monolith. - Lightweight regression entry with zero new rules: the whole case is a single 856-line test class with no new model, Rules, XML, or Service; the requirement-to-template mapping is frozen in test naming and Javadoc, so maintenance cost is near zero.
- Proxies uniformly assembled and reused: all proxies are created once in
@BeforeviaJQuickJava.create().createApi(...)and shared by all cases, following the performance convention of “reuse proxies, avoid reloading rules” — 26 cases with high-frequency cross-invocation stay lightweight. - Java 8 / domestic-JDK friendly with no external dependencies: the regression set depends only on JUnit and JQuick-Java itself, executes scripts via dynamic proxies and ASM, and can be added directly to pipelines on Oracle JDK 1.8 and domestic (Xinchuang) JDKs without introducing any rule-platform middleware.