Initialized and updated: 2026-10-18. M-001--M-005 document existing implementation choices whose original decision dates/approvals are not yet documented. Later entries record this session's approved research choices or explicit working constraints, with their evidence. Code presence does not validate a research design.
| ID | Date | Decision | Rationale | Evidence/Source | Status |
|---|---|---|---|---|---|
| M-001 | Not yet documented | Use React frontend and Express MVC backend with Sequelize/PostgreSQL. | Three-tier separation described in system thesis/architecture documents; specific technology-selection rationale is Not yet documented. | client/package.json, client/server/package.json, client/server/index.js, client/server/models/index.js, ARCHITECTURE_MAP.md |
Implemented in local path; runtime unverified |
| M-002 | Not yet documented | Track student/KC mastery with Bayesian evidence update and learning transition; permit per-KC parameter overrides. | Utility explicitly describes probabilistic knowledge tracking. | client/server/utils/bktAlgorithm.js, models/knowledgestate.js |
Implemented in Express; calibration unverified |
| M-003 | Not yet documented | Apply time/difficulty/hint/attempt/performance heuristic adjustments after local Bayesian updates when time is supplied. | Code describes contextual adjustments; timing calibration is explicitly aspirational. | client/server/utils/bktAlgorithm.js |
Implemented heuristic; research approval/validation unconfirmed |
| M-004 | Not yet documented | Maintain students, teachers, parents, admins, classroom/student and parent/student relationships with persistent knowledge/content/response models. | Architecture map describes monitoring and multi-role relationships. | ARCHITECTURE_MAP.md, client/server/models/, client/server/migrations/ |
Schema artifacts present; applied database state unknown |
| M-005 | Not yet documented | Provide separate Netlify/Supabase API implementation. | Deployment notes describe serverless hosting; original selection rationale is Not yet documented. | CLAUDE.md, netlify.toml, netlify/functions/api.js |
Code/configuration present; active deployment unverified |
| M-006 | 2026-10-18 | Focus the follow-up paper on whether the implemented fuzzy adjustment adds predictive value to BKT; preserve the log-audit drafts as history. | User selected and authorized the comparative research question. | Manuscript/AAIE2027/main.tex, archive/ and session instructions |
Comparative manuscript drafted; no acceptance claim |
| M-007 | 2026-10-18 | Adopt the explicitly approved input selection, separate historical-validator outcomes, unchanged recorded time, common fixed parameters, prediction-before-update order and paired account-cluster evaluation. | Isolate the implemented adjustment on common observed targets without inventing or repairing data after seeing results. | User approval recorded verbatim in Manuscript/AAIE2027/COMPARISON_PROTOCOL.txt; frozen proposal in archive/comparison-protocol-approved-proposal.txt |
Approved before evaluation and executed; raw CSVs unchanged |
| M-008 | 2026-10-18 | Pin the Express source at 0c407d010d9dd5f1cb2a0eba42b5c0452c22e581, preserve wrapper/session behavior and evaluate through isolated source extraction with inert collections. |
Reproduce the specified implementation without importing the application or initializing a database. | analysis/replay_models.cjs and analysis/comparison_results.json under Manuscript/AAIE2027/ |
Replay fixtures passed; historical online use remains unverified |
| M-009 | 2026-10-18 | Keep fixed exported BKT parameters and report the measured result without post-result fitting, extra variants or threshold tuning. | User-approved protocol fixes the comparison; an interval spanning zero does not justify superiority or equivalence claims. | COMPARISON_PROTOCOL.txt, analysis/run_comparison.py, analysis/comparison_results.json under Manuscript/AAIE2027/ |
Executed; no clear predictive benefit established |
| M-010 | 2026-10-18 | Deliver LaTeX/BibTeX and editable Mermaid sources for manual Overleaf preparation; retain anonymous review-paper sources separately from author-support drafts. | User requested these formats and no new PDF compilation; AAIE public instructions specify anonymous review. | Manuscript/AAIE2027/OVERLEAF.txt, SUBMISSION_FILES.txt, SUBMISSION_READINESS.txt; user instructions |
Source packages prepared; current figures/PDF still require export and inspection |
| M-011 | 2026-10-18 | Use SystematicITS resources for supporting-document organization only. | User explicitly requested reference use; independent SLR findings and declarations do not establish this empirical study's facts. | Manuscript/AAIE2027/REFERENCE_ADAPTATION.txt, source files in Manuscript/SystematicITS/ |
Structural adaptation complete; authorship, funding and approval details still require confirmation |
| M-012 | 2026-10-18 | Preserve application code, raw exports, existing tracked changes and project boundaries; perform no staging, commits, pushes, deletion, restore or live DB operations during closeout. | Explicit user constraints; retain reproducibility and prevent accidental disclosure. | Session instructions and GIT_SAFETY_REVIEW_2026-10-18.txt |
Safety review completed; ignore/cleanup recommendations are not applied actions |
Canonical historical deployment, backend reconciliation, mastery-threshold semantics and empirical parameter/fuzzy calibration remain unresolved. The approved offline experiment above does not resolve those questions. Changes to timing, labels, session behavior, model variants or fitting require separate direction; thesis assertions are not approval records.
Future entries: ID; actual Date; Decision; Rationale; Evidence/Source; Status. Record prospective approvals and reproducibility implications, without inventing historical dates or treating source comments as completed experiments.