Latest from MEDIATEXT.

Data paper submissions, sub-corpus releases, codebook updates, and deposit milestones.

7 May 2026

MediaText CEE — peer-review feedback received

Media and Communication (Cogitatio Press) returned peer-review feedback on the MediaText CEE data paper on 7 May. Twenty-five substantive comments across two reviewers; revised manuscript and response letter due 28 May 2026. The paper documents a multilingual, sentence-level corpus of over 1.4 million articles from Czechia, Hungary, Poland, and Slovakia, coded for CAP policy topic, sentiment, and named entities.

22 January 2026

V4 reading portal: NER ingest fixed, sentence-level coded

A patched import script (Lehoczki Richárd) cleared the NER ingest issue at the document level. Sentence-level reconciliation completed overnight. The Overview, Sentiment, Data Explorer, and Dashboard Information tabs of the V4 reading portal are stable.

21 January 2026

V4 reading portal online at mediatext.poltextlab.com

The first MediaText reading portal went online at mediatext.poltextlab.com — Overview, Sentiment, and Dashboard Information tabs covering the V4 sub-corpus. Document and sentence-level row counts reconcile against the source files (modulo two stray empty rows flagged and corrected). NER tab pending an import-script fix.

16 January 2026

One roof: NYT, USA media, V4, EU under MEDIATEXT

Direction set: every poltextLAB news sub-corpus — the NYT release, the original USA-media release, the V4 release, and a planned EU release — sits under a single MEDIATEXT roof with one shared codebook. The V4 release is the first to land under the new mediatext.poltextlab.com subdomain.

5 January 2026

mediatext.org domain acquired (5-year)

mediatext.org registered through Rackhost for five years, charged to the Lendület budget. The project website will eventually move under the new domain; for now the reading portal stays at mediatext.poltextlab.com.

19 December 2025

NYT release: two-tier access model agreed

The NYT data paper settles on a two-tier access model — a public surface exposing only binary text-availability indicators (has_title, has_abstract, …) plus a gated programmatic deposit with full embeddings and the data explorer. The same shape will apply across the MEDIATEXT releases. Codebook variable naming under harmonisation across all three reading portals.

23 December 2025

NYT coding scheme validated by US partners

"Amcsik levalidálták h min menjen a kódolás" — the US team signed off on the variable scheme used for the NYT sub-corpus, with minor pruning and one or two clarifying sentences slated for the article body. Author ORCIDs gathered, including Marcell Kiss's, ahead of the resubmission cycle.