Public Media & Sentiment Pipeline

Background & Problem

Public communication signals are distributed across platforms and formats, making it difficult to compare engagement and feedback consistently. The project focuses on creating a reproducible route from raw public records to evidence that can be reviewed alongside communication decisions.

System Strategy

The pipeline separates collection, normalization, text classification, and decision-facing comparison. This keeps source data, preprocessing rules, and analytical outputs distinct, so the workflow can be rerun and interpreted without treating a single dashboard view as the source of truth.

Execution

• Collection layer: Gather public communication and social-platform records into a structured source layer.
• Preparation layer: Standardize fields, dates, and text so engagement data and text feedback can be analyzed on comparable units.
• NLP layer: Classify sentiment and public-feedback signals from the prepared text data.
• Evidence layer: Compare platform and content patterns to produce communication inputs for the partner team.

Current Deliverable

The pipeline is the current deliverable. Findings and performance metrics will be added only after validation with the project partner; the page therefore shows the working system rather than presenting unvalidated impact claims.