Messages from alternative Spanish Telegram channels, 2019-2025

Telegram Channels Dataset

This dataset contains processed data extracted from public Spanish-language Telegram channels using pytopicgram from 2019-12-01 to 2025-05-31. It accompanies the article Multimedia, Engagement, and Diffusion in Alternative Online Communities: Evidence from Spanish Telegram Channels by Gómez-Romero, Cantón-Correa, Fajardo, and Molina-Solana, published in EPJ Data Science (DOI: 10.1140/epjds/s13688-026-00667-w).

The released dataset is anonymized and contains a sample of the data used in the article, including message-level metadata and topic information prepared for analyses of reach, virality, propagation, and thematic patterns across multimedia content types.

Files

The dataset contains these files:

topics.csv: compact topic metadata aligned with the topic IDs present in the released messages file.

messages.csv: sampled message-level data.

Topics (topics.csv)

This file contains the topics used in messages.csv.

Column Name
Description

topic
Topic identifier. -1 is the generic topic for non-assignable messages.

description
Concise topic description generated from the OpenAI representation field.

Messages (messages.csv)

This file contains a sampled subset of messages selected to provide broad topic and channel coverage.

Column Name
Description

channel_id
Channel identifier used to link messages from the same channel.

quarter
Quarter when the message was posted, formatted as YYYY-Q1, YYYY-Q2, YYYY-Q3, or YYYY-Q4.

media_type
Simplified media type.

is_multimedia
Boolean indicating whether the message contains embedded media or a media preview.

reach_bin
Binned number of views recorded for the message.

virality_bin
Binned relative performance score comparing the message with nearby messages from the same channel.

forwards_bin
Binned number of forwards, used as a propagation indicator.

topic
Topic identifier associated with the message.

topic_probability
Topic assignment probability.

Media Type Convention

The media_type field uses these values:

Value
Meaning

txt
Text-only message.

web
Webpage link without a media preview classified as image, audio, or video.

doc
Embedded or previewed document/text/application file.

img
Embedded image/photo attachment.

img-preview
Image/photo detected from a linked webpage preview.

video
Embedded video attachment.

video-preview
Video detected from a linked webpage preview.

audio
Embedded audio attachment.

audio-preview
Audio detected from a linked webpage preview.

Preview values are counted as multimedia through is_multimedia = True. The embedded/preview distinction is retained in media_type itself through values such as img versus img-preview.

Metric Bins

The reach_bin field uses these values: <100, 100-499, 500-999, 1k-4,999, 5k-9,999, 10k-49,999, 50k+.

The forwards_bin field uses these values: 0, 1, 2-4, 5-9, 10-49, 50-99, 100+.

The virality_bin field uses these values: below_average, average_to_2x, 2x_to_5x, 5x_plus.