Methodology

Below we describe, in plain language, how the Spanish Media Monitor (SMM) processes television news and what you can explore on this site. For a complete technical write-up you can download the full methodology (PDF): readme_smm.pdf

Analytical Pipeline

We collect broadcast news, convert audio to text, and apply automated analyses to identify who is mentioned, what topics appear, and the tone of those mentions. The diagram below gives a visual overview of the main stages from raw videos to the interactive charts on this site.

Media Monitoring Pipeline

How the data are processed

- We extract audio from TV news recordings and transcribe it automatically.
- Transcripts are scanned for mentions of political parties and named politicians. - We group related words and sentences into topics to see which stories each channel covers. - We use language models to estimate the tone (positive, neutral, negative) of each mention, then summarize those scores across time and channels.

Measuring agreement across channels

To check how similar or different channels are on any given day, we compute a measure of agreement across outlets based on the share of airtime devoted to each story. Higher values indicate that channels emphasize different stories (more disagreement); lower values indicate that channels focus on the same stories (more agreement). To avoid misleading values on sparse days, we only compute this measure when at least half of the monitored outlets carry coverage that day.

Political Tone (what the charts show)

Each time a party or politician is mentioned we assign a simple tone label: negative, neutral or positive. We then average those labels over the chosen time window so the charts show overall trends in how channels treat different parties or figures. Monthly averages include uncertainty bands so you can see how robust differences are.

How to use the site — what each tab shows

- Home: High-level summary and latest highlights. Useful to get a quick sense of recent coverage patterns.

- Media Agreement ("coverage.html"): Visualizes which stories and topics receive airtime across channels and shows the agreement/disagreement metric. Use this to identify days when networks focus on the same story or diverge.

- Political Mentions: Two interactive charts show relative mentions by party (top) and by individual political actors (bottom). Use the dropdowns and time selectors inside the charts to filter by channel, date range, or specific parties/people.

- Political Tone: Shows how positive/neutral/negative mentions are for each party across channels and over time. You can compare parties or channels and inspect monthly averages with confidence bands.

- Methodology: This page — concise explanation with a link to the complete technical PDF.

- About Us: Information about the project team and contact details.