Acts as itu learning surface, bukan just a lookup table
What save-algo-results does
Writes algo_category dan algo_reason back to transactions
Marks low-keyakinan rows untuk review
Keeps itu classified output searchable later
Input
Bank penyata row
Huraian, debit, credit, dan source sheet arrive daripada recompile-mbb.
Decision
CategoryAlgo
Match rule, fallback, or review path based on keyakinan.
Ingatan
CatDict
Manual category corrections improve itu next run.
Untuk ini demo, itu URL you should share is https://neuralops.bond/ainna/demos/bank-statement-category-algo/.
06 Langsung Pengelasan Sample
Sedia
07 Algorithm Report
Demo
Algoritma Kategori
Rows
4
Rule Matches
2
Fallback
1
Needs Review
1
Status
Completed
Ini demo is meant to show itu sebenar flow: catdict.php untuk learning/review, CategoryAlgo.php untuk inferens, dan save-algo-results.php untuk persistence.
Log Terminal
[SISTEM] Category algorithm demo sedia
Seni Bina Sistem
IN Input Lapisan
Bank penyata row
Sumber sheet
Debit / credit
AL Algorithm Lapisan
CategoryAlgo.php
Rule match
Fallback logik
LR Learning Lapisan
catdict.php
Manual category save
Review queue
Nota Keselamatan
Review low-keyakinan rows before using itu kategori in production.
CatDict is a learning surface, bukan itu sole source of truth.
Keep itu manual override path available untuk exceptions.