Course / Act V · Applications Lecture 20

Uncommon domains.

Cyber, law, health, finance, art, and education do not share a vocabulary. They share a form: a case, a rule, a decision, a trace, and a person who signs the exception.

~40 min read 6 domains 1 interactive lab Updated 2026-11-17

The most useful question about a neurosymbolic system in an institution is not “what model?” It is “what would count as an audit trail?” A decision that cannot name the rule that produced it, the facts it used, and the person who signed the exception is not yet a decision an organization can defend.

00 / CyberThe attack graph is a plan written by the defender.

A log is an observation; an alert is a claim that a pattern matched. Neither explains how an attacker moved. An attack graph does: nodes are security conditions, edges are exploits with preconditions and effects, and a path to a goal composes steps.

Denning’s 1987 model made intrusion detection statistical. Phillips and Swiler (1998) made vulnerability analysis a graph search; Sheyner et al. (2002) generated attack graphs with model checking; Ammann, Wijesekera, and Kaushik (2002) scaled the analysis; MulVAL (Ou, Govindavajhala, and Appel, 2005) expressed generation as a logic program. The vulnerability model is a knowledge base; a path is a plan that composes exploit rules.

A schematic attack graph with one observed alert and one highlighted path Five security conditions: internet, web server user, web server root, database user, and database root. Edges are exploits. One edge carries an observed alert. The orange path from internet to database root composes three exploits; grey edges are alternative routes. SECURITY CONDITIONS · EXPLOITS · ONE OBSERVED ALERT internet initial condition web · user web · root db · user db · root goal condition exploit 0 local privilege exploit 1 exploit 2 exploit 3 observed alert path hypothesis alternative route alert on an edge
Fig. 0.1 A schematic attack graph, not a real network. Nodes are conditions, edges are exploits. The orange path is a hypothesis: if the model and topology are correct, three exploits reach the goal. The teal marker is an observed alert; the grey route is an alternative the alerts do not distinguish.

Alert correlation adds abduction. Ning, Cui, and Reeves (2002) modelled alerts as hyper-alerts with prerequisites and consequences and correlated them into scenarios; Cuppens and Miège (2002) built a cooperative framework. The engine asks which scenario would make the alerts expected; several can, so ranking and ambiguity are part of the output.

Tuple · cyber alert correlation

Neural anomaly and alert classifiers. Symbolic an attack graph from a vulnerability model, plus correlation rules. Interface detector scores become facts; an analyst maps them onto edges. Guarantee if the model is correct, a path is a valid composition of exploits; correlation is a ranked hypothesis, not proof. Limit model drift, graph explosion, and an attacker who need not follow the graph.

Table 0.1 · Three cyber artifacts and what each one actually asserts.
ArtifactWhat it assertsWhat it does not
Log lineAn event was recordedThat it is malicious or complete
AlertA detector matched a patternThat the pattern is an attack
Attack pathA sequence of exploits is possible under the modelThat it occurred, or that no other path exists
Caveat · the graph is a model, not the network

Attack graphs inherit stale feeds, incomplete topology, and exploits absent from the rule base. Correlation inherits detector base rates and false-positive bursts. The guarantee is relative to the model and the alert semantics.

01 / Law & complianceThe statute is a program with an open texture.

Law looks like a production-rule system: conditions, exceptions, definitions, cross-references. It is also interpreted: terms are open-textured, precedents compete, and a principle can defeat a rule that was never written as an exception. The formalization is one reading, made by someone with authority.

Catala is the clearest modern attempt to make that reading explicit: a domain-specific language in which legal text and rule are written together, so the source is a literate document a lawyer can audit. Its semantics includes default logic, giving exceptions a principled place rather than an if-chain. The prototype encodes fragments of French tax and family-benefit law. The guarantee is modest: the result follows the encoded rule and the encoding is readable against the source. It does not certify fidelity, constitutionality, or what a court would do.

Case-based reasoning takes the other route. Ashley’s HYPO (1990) represents cases with dimensions, compares a new case to precedents, and builds three-ply arguments: claim, response, reply. It can show which case is closer and which dimension does the work; it cannot weigh a fact no precedent anticipated. Statutes supply the rule, precedents supply the interpretation, and hard cases live where they meet.

COMPAS is the cautionary case: an actuarial risk instrument used in U.S. criminal-justice settings. ProPublica’s 2016 analysis argued its error rates differed by race; Dressel and Farid (2018) disputed that conclusion and argued the answer depends on which error rate one chooses. This lecture does not adjudicate the dispute. The narrower point: a score is not a trace. It can rank risk and be explained post hoc, but it does not name the rule or facts that produced the ranking. Explainability is not derivation.

Tuple · law and compliance

Neural document classification, clause extraction, and entity resolution. Symbolic a statute or policy encoded as rules with defaults, plus a precedent base with dimensions. Interface extracted facts become predicates; the engine reports the derivation. Guarantee the decision follows from the encoded rule and facts; fidelity, truth, and justice are separate. Limit open texture, conflicting precedent, and the knowledge-acquisition bottleneck.

Caveat · the encoding is a legal act

Rules as code is not neutral translation. Choosing what to encode, which exceptions to make explicit, and which terms to leave to judgment is a policy decision with distributional consequences. A deduction is only as legitimate as the process that produced the rule base. The trace should name the policy version, not only the rule.

02 / HealthThe guideline is a protocol; the patient is not.

Health has the oldest case study, and it predates the neural part. MYCIN (Shortliffe, 1976) was a rule-based consultation system for bacterial infections. Its rules carried certainty factors, a hand-designed calculus for combining evidence that was meant to be inspectable and to avoid demanding priors clinicians lacked. It asked for findings, chained backward, and recommended therapy, but remained a research system. Its lesson is the trace: a recommendation came with the rules that supported it.

INTERNIST-1 (Miller, Pople, and Myers, 1982) and its successor QMR took another route: a large internal-medicine knowledge base and a scoring scheme over competing diagnoses. Curation was expensive, and the score was not a probability. Both teach the same lesson: inspectability costs maintenance.

Modern support usually formalizes the guideline rather than the disease. GLIF, Asbru, PROforma, and EON represent a care pathway as a graph of decisions, eligibility conditions, actions, and temporal constraints; Peleg et al. (2003) compare them on a shared case study. The guideline is a protocol: it says what should happen for a patient who fits its criteria. A constraint or rule engine can check the pathway; it cannot check whether the patient fits.

The neural part enters at the boundary. A classifier can flag a study; a language model can extract findings; a triage model can rank a queue. Esteva et al. (2017) showed a convolutional network matching dermatologists on a skin-lesion benchmark; that is a benchmark result, not a deployment licence. The interface is the pattern: the classifier produces a probability, the guideline decides whether it changes the next step, and a clinician owns the exception.

Tuple · clinical decision support

Neural image, signal, and note models that produce findings with probabilities. Symbolic a guideline with eligibility criteria, actions, and temporal constraints, plus a drug-interaction knowledge base. Interface findings become facts; the guideline derives the next step and can request a missing test. Guarantee the recommendation follows the guideline for the documented facts; it is not a diagnosis or an outcome. Limit outdated guidelines, exceptional patients, and a calibrated model that is still wrong.

Caveat · diagnosis is abductive

Clinical reasoning is also abduction: which condition would explain this constellation of findings? A guideline engine can enumerate hypotheses and check consistency, but it cannot supply the judgment that decides when a guideline does not fit. That judgment is the exception the audit trail must make visible.

03 / Finance & auditEvery posting is a proof obligation.

Finance is where the symbolic side is least optional. Double-entry bookkeeping, described by Pacioli in 1494, is an invariant system: every journal entry has equal debits and credits, and the trial balance follows. A machine can verify these identities exactly; it cannot verify that the amounts are honest, the counterparty exists, or the treatment is appropriate. The invariant is necessary, not sufficient.

Around the ledger sit controls already shaped like rules: segregation of duties, the three-way match between purchase order, receipt, and invoice, duplicate-payment detection, vendor-master review, and materiality thresholds. Decision tables are a good notation; the OMG’s DMN standardizes tables with hit policies such as first, unique, and collect. The hit policy is precedence in disguise, and it needs the same documentation.

The neural part is at the edges: OCR, merchant normalization, entity resolution, anomaly scores, extraction from filings. XBRL gives reporting a symbolic schema; a validator can check an extracted fact against the taxonomy and arithmetic. Benford’s law (1938) is the classic digit-distribution check; it detects, it does not prove fabrication.

The lab isolates the mechanism: a neural component produces facts and flags; a rule engine applies policy in fixed precedence; the decision names the rule that fired; exceptions are policy objects or logged overrides. The audit log lets a reviewer reconstruct the decision without rerunning the model.

Table 3.1 · Four finance controls, what they check, and what they cannot check.
ControlSymbolic checkWhat it catchesWhat it misses
Double entryDebits equal credits; trial balanceAn unbalanced postingA balanced posting that is fraudulent
Segregation of dutiesRequester ≠ approver ≠ payerSelf-approvalCollusion across roles
Three-way matchOrder, receipt, and invoice agreeOverbilling without deliveryA collusive vendor supplying all three
Duplicate detectionSame vendor, amount, and dateResubmission and double paymentSplit invoices below the threshold
Tuple · expense and transaction audit

Neural OCR, merchant normalization, and anomaly ranking. Symbolic ledger invariants, policy rules, decision-table hit policies, and the exception register. Interface extracted facts and flags enter the engine; it writes a decision and a trace; a human reviews the rejected and the ambiguous. Guarantee the decision follows the policy for the recorded facts, and the accounting identities hold; it is not a finding of fraud. Limit false positives, collusion, and a policy that is itself the weakness. Lab rules are fixtures, not accounting advice.

Caveat · the closed-world assumption is a business decision

A rule engine that treats an absent fact as false will reject a claim for a missing receipt that exists elsewhere. That is often the right audit default, but it is an assumption and should be visible: “not found in the case file” differs from “does not exist.” That difference is what review is for.

04 / ArtThe grammar generates; the artist selects.

Art is where the symbolic component is easiest to dismiss and most instructive to keep. AARON, Harold Cohen’s drawing program, is the canonical case. In its classic form its rules were hand-written by Cohen: composition, figures, colour, and the relations that make a drawing hold together. It did not learn a style from a corpus; it encoded a practice. That makes authorship precise: the program explored a space the artist defined, and the artist chose the space.

Generative grammars generalize the idea. Lindenmayer systems (1968), popularized in The Algorithmic Beauty of Plants (Prusinkiewicz and Lindenmayer, 1990), are rewriting rules that produce branching structures; shape grammars produce arrangements of shapes by rule. The symbolic object is a language of possible forms. Membership is checkable; quality is not.

The neural side enters as a critic, stylist, or proposer. Gatys, Ecker, and Bethge (2015) showed that a pretrained convolutional network can separate content and style and recombine them by optimizing a content loss and a style loss. The interesting interface is generate and criticize: the grammar guarantees a candidate is well-formed, a neural critic ranks candidates, and a human selects. The symbolic part guarantees the space; the neural part navigates it.

The guarantee is narrow. The system can guarantee membership in the grammar, satisfaction of a constraint, or reproducibility from a fixed checkpoint. It cannot guarantee that the output is good, original, or authored by anyone in particular. Those are judgments; in the copyright case, legal questions.

Tuple · procedural and neural art

Neural a style model, a generative model, or an aesthetic critic. Symbolic a grammar, a shape rule system, or composition constraints. Interface generate-and-test: the grammar proposes a well-formed candidate; the critic ranks; the human selects. Guarantee membership in the grammar and reproducibility from the recorded rules and checkpoint; no guarantee of quality, originality, or authorship. Limit training-data provenance, critic bias, and “creativity” as a non-benchmark.

Caveat · the artist is not an optimization target

A system that optimizes a learned aesthetic score produces what the score rewards, including shortcuts. A grammar that enumerates a space produces everything in it, including the boring parts. Selection is where intent enters; if the interface hides it, it hides the authorship.

05 / EducationThe hint must be licensed by the next step.

Education has a symbolic core that is easy to miss because it is called a curriculum. A model-tracing tutor represents the domain as a graph of solution paths, including correct and incorrect steps, and compares the student’s step to the graph. Anderson, Corbett, Koedinger, and Pelletier (1995) described the Cognitive Tutors built on this idea: the tutor can say which step is legal next, hint at that step, and update a model of the student’s knowledge. Knowledge tracing (Corbett and Anderson, 1994) adds a latent model of skill acquisition.

The neural part is now the fluent part. A language model can grade a free-text answer, generate a hint, or explain an example. The failure mode is equally fluent: a hint can skip a step, contradict the graph, or solve the problem. The interface that makes the system trustworthy is a verifier: the model proposes, and the tutor checks that the hint is licensed by a valid next step from the student’s state. Without a solution graph, the check becomes a rubric or a constraint, and the guarantee weakens.

Effectiveness is separate. VanLehn’s (2011) meta-analysis compares human tutoring, intelligent tutoring systems, and other tutoring; the literature is more textured than the “two sigma” slogan. This lecture takes no position on the average effect. It claims only that step-level feedback can be checked against a domain model, and that check separates a pedagogical agent from a chatbot with a lesson plan.

Tuple · intelligent tutoring

Neural free-text grading, hint generation, and engagement signals. Symbolic a domain model with solution paths, a student model, and a pedagogical policy. Interface the model proposes; the tutor checks the proposal against the next licensed step and the prerequisite rules. Guarantee the feedback is consistent with the domain model for the student’s state; it is not a guarantee of learning. Limit a wrong graph teaches the wrong procedure, and a fluent hint can still harm.

Caveat · the graph is the curriculum

Model tracing is only as good as its solution graph. If the graph encodes one procedure and the student uses another valid one, a correct step is marked wrong. The symbolic component does not remove the need for interpretability; it relocates it. The graph is now the object to inspect, and teachers should be among the inspectors.

06 / The common patternSix domains, one audit loop.

Put the six cases side by side: the differences are vocabulary, not form. Every one has a normative text, a case, a decision, a trace, and a human who owns the exception. The neural component sits at the boundary; the symbolic component is the part that can say why.

The common audit loop across six domains Normative text feeds a rule base with precedence and exceptions. Case facts, partly produced by a neural extraction step, enter a decision engine. The engine produces a decision and a trace. A human reviews rejected and ambiguous cases and can log an override. The override and the review feed back into the policy as an explicit change. normative text statute · guideline · policy rule base rules · precedence · exceptions decision engine apply · waive · flag case facts extracted · verified neural boundary perceive · extract · rank decision + trace rule · facts · exception human review · sign override is logged; policy change is explicit symbolic neural human override / feedback
Fig. 6.1 The common audit loop. Normative text is encoded once; case facts arrive through a neural boundary; the engine applies the rules and writes a trace; a human reviews the rejected and the ambiguous. The dashed return path is what most systems leave implicit: an override should change policy through an explicit, versioned act, not by quietly editing a threshold.
Table 6.1 · Six domains, one anatomy. The guarantee is what the symbolic part can promise, not what the system hopes.
DomainNormative sourceSymbolic artifactNeural roleGuarantee
CyberVulnerability and topology modelAttack graph and correlation rulesAlert classificationA path is valid under the model
LawStatute, regulation, precedentRules with defaults; cases with dimensionsClause extractionThe decision follows the encoded rule
HealthGuideline and formularyCare pathway and interaction rulesImage and note modelsThe recommendation follows the guideline
FinanceStandards and policyLedger invariants and decision tablesOCR, matching, anomaly rankingThe decision follows the policy
ArtGrammar or composition rulesRewrite or shape grammarStyle and aesthetic criticThe output belongs to the language
EducationCurriculum and solution graphModel-tracing graph and hint policyGrading and hint generationThe hint is licensed by the next step

The pattern explains why a system can impress in a demo and fail in an institution. An institution cares about rejected, reviewed, and overridden cases: who decided, on which rule, with which evidence, and who signed. A system that cannot produce that record has not solved the institutional problem.

Three verdicts, one human act

Accept if no rule blocks, or a named exception waives the only blocker. Reject if a blocking rule fires. Review if a required fact is missing or two rules conflict. Override is not a fourth verdict; it is a human event that accepts a rejected or reviewed case, logged with a reason. Collapsing review into accept or reject hides exactly the cases a human should see.

The rule that fired is not the rule that is right

A rule engine can be sound — it derives only what follows from its rules — and still be wrong about the world, because the rules are wrong. It can be complete relative to its rule set and still miss an exception the rule set never encoded. That is the difference between a guarantee and a certification: the guarantee is conditional on the specification, a human artifact with an owner and a version.

07 / The tupleRead a case as a tuple.

The sections used the anatomy informally; the tuple makes it explicit. Write a case as:

c = (N, F, R, ≺, E, d, τ, o)
(7.1)

N is the normative source and version; F is the case file with provenance; R is the rule set; ≺ is precedence; E is exception policy; d is the decision; τ is the trace; o is the override, a human event with a reason and an owner. The decision is a function of the first five; the trace is the evidence. An override changes the decision and is recorded alongside it, not inside it.

The tuple makes failure attribution possible: a wrong decision can come from a bad fact, rule, precedence, exception, or override. These are different bugs with different owners. A system that reports only “rejected” cannot distinguish them.

Table 7.1 · The trace for one illustrative claim (EXP-1042) under the lab’s default policy. The failure and the exception both remain visible.
OrderRuleResultWhy
1R3 · prohibited categorypassmeal is reimbursable
2R5 · duplicate claimpassno earlier match
3R6 · currency conversionpassUSD; no conversion
4R1 · receipt thresholdpassreceipt attached
5R4 · travel pre-approvalnot applicablenot a travel claim
6R2 · meal capfail$62.00 per attendee > $60.00
—E1 · meal exceptionappliedMEAL-EXC waives R2
—R0 · default acceptskippeddecision came from R2 + E1

The trace is a small proof, but only relative to the rule set and facts. Five properties are worth naming separately:

Table 7.2 · Five properties an audit trail can have. Fairness is not on this ladder; it needs a separate argument.
PropertyWhat it meansHow to test it
DeterministicSame facts, rules, and policy → same decisionReplay the case; compare the trace
SoundThe decision follows from rules and factsRe-derive the decisive rule
CompleteEvery rule was evaluated and reportedCompare trace length with the rule set
AuditableThe record names facts, rules, exceptions, and the humanGive it to a reviewer who did not build the system
ContestableA rejection or review has a path to a human decisionOverride it; confirm the original remains visible

Fairness is deliberately absent. A rule set can have all five properties and still treat people unjustly, because the rules can encode a disparity or the facts can be measured differently across groups. Fairness is a property of the policy and its effects, not of the engine. The audit trail is a prerequisite for asking, not the answer.

08 / RecapWhat you should be able to do now.

01

Explain why an attack path is a symbolic composition, and why correlation is abductive.

02

Distinguish encoded statute, case-based precedent, and an actuarial score.

03

Read MYCIN, INTERNIST/QMR, and a guideline engine as three symbolic cores.

04

Name the finance invariants a rule engine checks exactly — and what they miss.

05

Describe generate-and-criticize in art, and why a grammar cannot guarantee quality.

06

Explain model tracing and the interface that licenses an LLM-generated hint.

07

Write a case as (N, F, R, ≺, E, d, τ, o) and attribute a failure to a slot.

08

State the conditional guarantee: sound relative to rules and facts, not correct about the world.

Exercises

  1. Write five rules. Pick a policy you know. Write five rules, a precedence order, and two exceptions. For each exception, say whether it is a policy rule or a human override.
  2. Break the precedence. Check all claims in the lab’s default policy, then in strict mode. Which verdicts changed because of a rule, and which because of a data gap?
  3. Make an ambiguous case. The lab’s ambiguity is a missing FX rate. Construct one that is ambiguous because two rules conflict. What should the engine do?
  4. Audit an override. Override EXP-1046, then re-check it. What do the trace, the decision, and the audit log each say? Are they consistent?

Check your understanding

Ten questions, with explanations. The score is not the point; the questions you cannot answer are the reading list.

Reading list

The first six are the spine of the lecture. Classics without an arXiv identifier are cited by author, title, venue, and year.

Computer-Based Medical Consultations: MYCIN

Shortliffe · Elsevier, 1976 · rules, certainty factors, and the trace

Health

Internist-1: An Experimental Computer-Based Diagnostic Consultant for General Internal Medicine

Miller, Pople & Myers · NEJM, 1982 · the large knowledge base

Health

An Intrusion-Detection Model

Denning · IEEE Transactions on Software Engineering, 1987 · the statistical baseline

Cyber

Automated Generation and Analysis of Attack Graphs

Sheyner et al. · IEEE Symposium on Security and Privacy, 2002 · paths as counterexamples

Cyber

Constructing Attack Scenarios through Correlation of Intrusion Alerts

Ning, Cui & Reeves · ACM CCS, 2002 · hyper-alerts and abductive scenarios

Cyber

Catala: A Programming Language for the Law

Merigoux, Chataing & Protzenko · ICFP, 2021 · literate law, default logic, executable rules

Law

Modeling Legal Argument: Reasoning with Cases and Hypotheticals

Ashley · MIT Press, 1990 · HYPO, dimensions, and three-ply arguments

Law

Comparing computer-interpretable guideline models: a case-study approach

Peleg et al. · JAMIA, 2003 · GLIF, Asbru, PROforma, EON

Health

Cognitive Tutors: Lessons Learned

Anderson, Corbett, Koedinger & Pelletier · Journal of the Learning Sciences, 1995 · model tracing

Education

The further exploits of AARON, painter

Cohen · Stanford Humanities Review, 1995 · rules, composition, authorship

Art

A Neural Algorithm of Artistic Style

Gatys, Ecker & Bethge · arXiv:1508.06576, 2015 · content, style, and the critic

Art
Source note · what is schematic

The figures are schematic redrawings, not reproductions. The lab’s rules, thresholds, and claims are illustrative fixtures; they are not legal, tax, or accounting advice. Systems are cited by author, title, venue, and year. The lecture makes no claim about deployments or benchmark scores that it does not state.