A full week per system
A single system can take a full week of analyst attention before review even begins.
Senior Project · AI Classification · Data Governance · Privacy Readiness
Classify Database Metadata with AI — Review, Approve, Export.
A web platform that takes your Excel or CSV metadata file, classifies every field by confidentiality and personal data status, and produces a professional governance report.
Built as a senior project at the University of Bahrain. Designed for the way data governance work actually happens: fast first pass, human review, structured output.
- Input
- Excel / CSV
- Review
- Human-in-the-loop
- Output
- Multi-sheet report
Built for
- PDPL readiness
- ·
- Governance reviews
- ·
- Metadata audits
- ·
- Consulting engagements
- ·
- Repeatable reporting
Classifying Database Fields Manually Takes Too Long and Misses Too Much
A typical classification engagement covers dozens of systems and thousands of fields. The work today is done in Excel: one analyst, one spreadsheet, one row at a time.
Different analysts, different answers
Two analysts can produce two different classifications for the same fields.
Personal data gets missed
Fields that fall under PDPL obligations are easy to overlook in a spreadsheet-only review.
Documentation goes stale
The output Excel file is often stale within weeks of delivery and disconnected from future updates.
The platform was built to solve exactly this: not data governance in general, but the manual classification step that consumes the most time and produces the least durable output.
02 / Market Need
Bahrain and Saudi Arabia Now Require What Most Teams Cannot Produce
The Bahrain Personal Data Protection Law and the Saudi Personal Data Protection Law both assume that organizations know what personal data they hold and where. Most do not, not because they do not try, but because no practical tool exists between two extremes.
This platform sits in the middle: light enough to deploy in a day, structured enough to produce auditable output.
What the Platform Actually Does
The platform takes one input and produces one output, with structured review in between.
Excel or CSV metadata
An uploaded file containing database metadata such as table names, column names, and data types.
AI classification with review
- Fields are sent in batches to the OpenAI API.
- Each field receives a confidentiality level, classification reason, personal data flag, and personal data reason.
- Results are stored in a database and displayed for review.
- The user corrects AI errors, approves personal data fields, and adds PDPL governance notes.
Formatted governance report
A multi-sheet Excel report with Overall, System Data, Personal Data, and PDPL sheets ready for client delivery or internal documentation.
From system registration to export in five clear steps.
The workflow keeps the AI pass fast while preserving the human review needed for real governance work.
-
01
Register a System
Add the system name, owner, group, and any context notes that help interpret the data.
-
02
Upload Metadata
Drop in an Excel or CSV file containing table and column information.
-
03
Run AI Classification
The platform groups fields by table, sends them to the OpenAI API, and stores the structured response.
-
04
Review and Approve
Use the Edit View to fix AI errors. Use the Personal Data tab to approve fields that genuinely qualify as personal data.
-
05
Export
Download a multi-sheet Excel report. Hand it to a client, use it as documentation, and update it next quarter.
The pieces needed for a repeatable classification workflow.
Each feature supports a practical governance task, from intake to approval to client-ready reporting.
AI Classification Engine
Powered by the OpenAI API with structured JSON responses and batch processing by table.
Confidentiality Detection
Four-level scheme: Public, Internal, Confidential, and Strictly Confidential.
Personal Data Flagging
Each field is independently flagged with reasoning.
Edit View
Override any AI classification; changes persist immediately.
Personal Data Approval
Record-level approve and revoke; only approved records flow to PDPL.
PDPL Governance Page
Add review status and governance notes per personal data field.
Multi-Sheet Excel Export
Overall, System Data, Personal Data, and PDPL sheets, professionally formatted.
System Context
Tag-based notes attached to each system.
System Links
Categorized reference URLs for data dictionaries, documentation, and regulatory references.
Dashboard
Multi-system overview with classification status per system.
What This Project Demonstrates Academically
The project applies four areas of computer science to a single working system.
- Data governance theory: DAMA-DMBOK classification concepts implemented as software.
- AI integration: OpenAI API used inside a structured, persistent, human-in-the-loop workflow.
- Full-stack engineering: Node.js, Express, SQLite, ExcelJS, OpenAI SDK, and session-based authentication.
- Software requirements engineering: Context Diagram, Level 0 and Level 1 DFDs, ERD, Structured English specifications, and twelve unit test cases.
The project is documented as a senior project report under the supervision of Dr. Abdulla Ahmed Alasaadi, with all DFDs, the ERD, and the testing strategy traceable from requirements to implementation.
What Changes When a Team Uses This Platform
Today
With the Platform
Each analyst writes a fresh Excel.
Every system follows the same workflow.
30+ fields per hour at best.
AI processes a full system in minutes.
Personal data fields missed in review.
Personal data routed to a dedicated approval track.
Output format varies by analyst.
Consistent multi-sheet Excel, every time.
File becomes stale and disconnected.
Persistent database; results editable anytime.
The platform does not replace the analyst. It removes the tedious first pass so the analyst can spend time on the parts that need judgment.
Built for the teams responsible for classification and privacy readiness.
Data Governance Consultants
Running multi-system classification engagements.
Internal Governance Teams
Maintaining ongoing data inventories.
Risk and Compliance Officers
Preparing for PDPL review.
Organizations Starting a Governance Program
Needing a structured entry point.
Consulting Firms
Looking for a repeatable workflow across clients.
Where the Project Goes from Here
The senior project version is a working single-tenant application. The path to a commercial product is concrete.
SaaS product
Multi-tenant deployment for governance teams who want a permanent classification tool.
Consulting accelerator
Used by firms internally to deliver classification engagements faster.
Privacy readiness tool
Packaged specifically for PDPL-driven assessments.
Foundation for further features
Direct database connectivity, sector-specific rules, GDPR support, and audit trail.
The market gap between enterprise platforms and manual spreadsheets is the project's product opportunity. The senior project version is the proof that the workflow works.
The platform workflow, screen by screen.
Login Page
Dashboard
Add New System
System Data — Classification View
System Data — Edit View
Personal Data Tab
PDPL Tab
System Context Tab
Excel Export Preview
Developer
Azzam Esam Alquradhi
11 / About the Developer
Azzam Esam Alquradhi
Computer Science, University of Bahrain.
This project came from watching real classification work happen the hard way: in Excel files, across long meetings, without consistent output. The platform is an attempt to solve that problem in a way that is actually usable.
It reflects a focus on data governance, practical AI applications, and software that addresses operational problems rather than hypothetical ones.
- Data Governance
- Practical AI
- Excel Reporting
- Requirements Engineering
12 / Final CTA
See the Platform in Action
The senior project version is live. You can register, create a system, upload metadata, run classification, and download a report: the full workflow, end to end.
Explore the Platform