Introduction

AXON is the entry point of the DEDOMENA.AI platform. It allows users to ingest, clean, validate and standardize data before it is used across the platform — whether for anonymization, synthetic data generation, or analysis. Through automated checks and intelligent workflows, AXON ensures that all datasets entering the platform meet high standards of quality, structure, and compliance.

Flow Overview

1. Upload & Detection – The user uploads a file (CSV, XLSX, etc.) or retrieves one from a connected source. AXON automatically detects encoding, delimiters, headers, and potential format inconsistencies.

2. Automated Cleaning – The system performs an initial repair: fixing quotes, removing corrupted rows, detecting duplicates, and normalizing headers.

3. Validation Setup – Users can define column-specific validation rules (data types, ranges, sensitivity, key structure, etc.) or rely on automatically inferred ones.

4. Metadata Enrichment – Add contextual information (origin, usage, industry, country). Metadata can also be imported from a reference dataset.

5. Diagnosis & Profiling – AXON generates a detailed data quality report including risk score, profiling statistics, and validation summaries.

6. Final Summary – The user can review a summary of applied transformations, profiling insights, and overall quality score.

7. Reference-based Flow(optional) – If a previously validated dataset is selected as reference, all validations and metadata are automatically inherited and applied.

Visual flow

Phase Details

1. Upload & Detection

This is the first step after the user chooses the Upload Files option to connect data. The Select File format option allows the user to select the format of the file. The Browse file option allows the user to browse files from their local system. Once the user selects the file format, the Browse file option becomes available. AXON automatically detects file format, structure, and headers. If the file cannot be parsed, the system produce the corresponding alert.

2. Automated Cleaning

AXON runs multiple automated repair and cleaning steps:

• Fixes misplaced quotes or delimiters

• Detects and removes blank or duplicated rows

• Normalizes headers and aligns column counts

• Checks encoding and delimiter consistency

For each table/dataset a preview will be available and the user can navigate between them. Once the user completes the Import files step, the Next button allows the user to continue the process.

3. Validation Setup

Once the file is ready, users can define validations or review those automatically inferred:

• Data type consistency

• Null value checks

• Semantic validation type

If all validations are feasible and approved, AXON applies the transformations automatically.

4. Metadata

In this step, users enrich the dataset with contextual metadata such as domain, source, country, and intended use. When using a reference dataset, metadata is automatically pre-filled.

5. Diagnosis & Profiling

AXON performs a comprehensive diagnostic:

• Column-level profiling (distribution, outliers, patterns)

• Risk assessment.

• Data quality scoring. A visual summary is generated to help users understand the dataset’s structure and reliability.

6. Final Summary

Users can visualize:

• A report of applied transformations

• Profiling and validation metrics

• Quality and risk evaluation

The processed dataset can then be used directly in other DEDOMENA.AI modules.

Reference-based Flow (optional)

If the user selects a previously validated dataset, AXON automatically:

• Loads and applies the stored validation schema

• Reuses metadata

• Skips manual configuration

Select de reference dataset

The system load the stored rules

Applying automatic fixes

Showing the results

Loading stored metadata

After the Metadata step, the system continues its normal flow, moving through Diagnosis & Profiling and finally reaching the Final Summary, where changes, profiling, and the quality report are consolidated.