Employment and staffing
Evaluate role, location and commute radius, shifts and availability, compensation and payment basis, language and visa together, publishing one traceable outcome per cycle.
00SENK MATCHING SYSTEM
Explainable matching engineSupporting enterprise digital transformation
SENK builds an explainable matching engine that turns fragmented business conditions, eligibility rules and resource constraints into traceable, verifiable decision flows. It integrates with existing products and business systems to support enterprise digital transformation.
MATCHING ENGINE FLOW
FACT
RULE
DECISION
EVIDENCE
STATUSVERIFIABLE
DECISION INFRASTRUCTURE01 — 04
01PRODUCTS
Two SENK product directions
WorkFit applies explainable matching to employment; Vovel applies understandable, traceable system principles to participatory narrative.
02INDUSTRY APPLICATIONS
For matching domains with multiple constraints, eligibility gates and limited resources
SENK is not a generic list-ranking tool. It is designed for workflows that must evaluate multiple candidates, hard constraints, explicit unconstrained states, spatial and temporal limits, and resource caps while explaining admission or unmet conditions. WorkFit applies this model to employment first.
Education and training admission
Evaluate prerequisites, time and location, language ability and capacity together while separating genuine condition mismatch from missing evidence.
Business service allocation
Evaluate provider qualifications, coverage, availability, pricing model and capacity together to prevent duplicate admission and resource overuse.
Local and public services
Support allocation governed by jurisdiction, distance, opening hours, eligibility and quotas, with safe stopping when required facts are missing.
03ENGINE INTEGRATION
Connect business facts to a real matching decision chain
Existing account, order and content systems remain in place. Integration is not a matter of flattening fields for equality checks. It defines fact ownership, constrained and explicitly unconstrained conditions, candidate identity and the result contract, then lets the engine perform multi-candidate evaluation, explanation and one final publication per decision cycle.
Freeze decision facts
Freeze source facts such as requests, candidate eligibility, location and radius, time, compensation, language and visa into one versioned decision snapshot while preserving explicit constraints and unconstrained states.
→Build a candidate evaluation window
Evaluate multiple stably identified candidates in one cycle, excluding previously admitted objects, deduplicating them and applying the same condition axes to every candidate.
→Produce judgement and condition-level explanation
The core engine returns pass or fail. The explanation layer marks each condition as matched, unmatched, not applicable or satisfied because it is unconstrained, preserving expected values, actual values and reasons.
→Publish one final outcome
Each cycle publishes exactly one outcome: candidate admission, a genuine condition gap or a safe stop when facts or resources are insufficient. Interfaces only read the result and never judge again.
This is not a field-mapping utility. Integration defines sources of truth, candidate identity, constraint semantics, rule versions, resource gates and the result contract. The existing product retains authority over source data, and users retain the final decision.
04TRUSTED SYSTEM PRINCIPLES
How causality, boundaries and trust are implemented
Trust does not come from the number of rules. It comes from tracing each outcome to source facts, keeping one decision authority and defining explicit failure boundaries.
Traceable causal chain
Link frozen input facts, engine judgement, condition-level explanations and the final route within one decision cycle so every outcome traces to source facts and a rule version.
Single decision boundary
Only the core engine owns matching semantics. Interfaces, projections and screens read outcomes; they cannot compare fields again or rewrite reasons.
Separate explanation from diagnosis
Only genuine condition mismatch is explained to users. Missing candidate identity, spatial truth or evidence is a system diagnosis and cannot be presented as a user shortcoming.
Idempotent resource consistency
Each cycle has one final decision. Reprocessing cannot admit a candidate twice or consume resources twice, and publication stops when resources cannot be confirmed.