With human validation at every phase.
We turn TikTok and Instagram videos and comments into data ready for research. With official access to Meta and TikTok. The annotated corpus, the data and the report belong to your team.
Questions this process can answer. Yours can be another one.
How a conversation about a topic is built, who enters it, and which formats carry it.
Which categories appear in a corpus, and how they change across periods or accounts.
Social media discourse analysis. Which emotion, narrative and music travel with it.
How the same topic compares across two moments, or two communities.
A process shared with your team and validated at every phase. Each result can be traced and described in a paper.
We define the questions, state the hypotheses, and agree the limits, the sample, and the success criteria.
We turn the questions into categories and multilabels. They are reviewed before a single data point is classified.
A pilot checks sources, fields, coverage, and quality. The extraction method is validated before it scales.
We collect the full corpus and normalise it so videos, text, interactions, and metadata share one structure.
When the question needs it, we train models to detect patterns, classify content, or produce variables that do not exist in a public dataset.
We annotate between 15% and 20% of the data by hand. That set trains, checks, and measures performance.
We apply the validated system to the rest of the corpus and leave a consistent base, ready to analyse.
We present the method and the metrics. Then come the analyses, the predictions, and the interpretation.
The TikTok and Instagram research data comes from sources you can explain to an ethics board and describe in a paper.
We work with official access to Meta and TikTok, not a slice from an aggregator.
Sources, fields, coverage and quality criteria travel with the deliverable.
The annotated corpus, the data and the report belong to your team. They are not reused for another study.
A closed project, with figures that can be shown.
UOC needed to understand what mental-health content circulates on TikTok. Trendia turned thousands of videos and comments into a structured, validated database ready for analysis. Trendia delivered the validated corpus; the conclusions belong to the UOC team.
The team keeps the material to defend the result and keep working.
The hand-reviewed set that supports training and validation.
DataA clean, normalised base, enriched with the study's variables.
DataThe answer to the hypotheses, with the method and the limits beside it.
ResultsScope changes the detail. Traceability and the handoff do not.
From public sources and official access to Meta and TikTok. Sources, fields and coverage are documented and travel with the deliverable.
We work with public data and a documented protocol that can be presented to the board. If your institution has its own requirements, we review them before we start.
Yes. The annotated corpus, the data and the conclusions belong to your team. We document how the corpus was built so the method can be described in the paper.
It is defined with your team from the hypotheses and reviewed before a single data point is classified.
It depends on the scope. In the UOC project, the data was ready in 14 days.
No. The budget depends on the question, the corpus and the deliverables.
The full project and the rules we follow with data.
Tell us the question, the corpus you have in mind, and the deadline. We will say whether the process fits.
Propose your research question