Generating election coverage for CTK
Living up to its motto “Reliability–speed–independence,” the Czech News Agency (CTK) strives to make its work truly efficient and employs state-of-the-art technologies to achieve this goal.
To relieve journalists from the overwhelming amount of plain, repetitive tasks related to covering election results in real time, CTK approached Geneea to collaborate on automatic text generation (NLG). Although the generated texts were still subject to regular editorial processes, the result gave the journalists more time to apply their expertise where it is needed and to leave the uncreative work of reporting numbers to AI.
Challenge
- Too many articles needed in a very short time
- Insufficient time to write accompanying analyses
- Repetitive, monotonous, and time-consuming tasks
- Rewriting and copy-pasting data from tables is error-prone
Solution
- Prepare and test templates in advance
- Respect style guide for formatting numbers, dates, tables, etc.
- Automatically choose appropriate alternative formulations
- Use Geneea’s knowledge base for information about parties, locations, etc.
- Automatically add metadata: publish dates, domiciles, tags, etc.
Results
- Covered election types: presidential (2023), Senate (2020, 2022, 2024), Chamber of Deputies (2021), regional (2020, 2024), local (2022)
- Amount of automatically written reports: 100-530 according to election type
- Peak: one article every second
- No typos, perfect grammar
- First ever commercial use of NLG in Czech
Notice the correct agreement (Kárník a Karen postoupili vs 11 kandidátů usilovalo vs sedmý Žantovský získal vs devátá Kohoutová skončila), prepositions (s 4,88 vs se 3,64), word order, cases, etc.
“The reporting process was significantly accelerated. Our reporters did not have to waste time rewriting and copying results. They could immediately concentrate on the reactions and post-election negotiations.”
Conclusion
CTK considers covering elections to be one of the most challenging events of the year. Implementing Geneea’s first commercial text generation system developed for Czech made it possible for the journalists to apply their minds and time to more complex, stimulating, and satisfying work on election analyses. As a result, CTK’s customers were treated to an even larger number of insightful articles and deeper analyses .
Not just elections
With lower load, but just as important, regular articles reporting on fuel prices, and car accidents are now also automatically generated, leaving even greater space for journalists’ creative work. Articles on fuel prices are reported once a week, and reports on car accidents are published once a month. The system was specifically designed, so that CTK would be able to create its own templates .
About CTK
Founded in 1918, on the same day as Czechoslovakia, the Czech News Agency (CTK) is a national public service news agency in the Czech Republic. It publishes both in Czech and English, and it has an extensive network of branches and correspondents in the main centers across Europe and the USA. The agency is politically and financially independent. CTK also uses the services of other world news agencies, such as Reuters, AP, AFP, dpa, and ITAR–TASS.
The Latest from Geneea
RAG over Large Archives
Last week at the FIBEP Tech Day conference, our Jiri Hana spoke about the practical challenges of building RAG systems for news publishers and media monitoring companies. RAG for news monitoring is a different beast than the typical tutorial examples. We deal with datasets of hundreds of millions or even billions of articles. A few... The post RAG over Large Archives appeared first on Geneea News.
Jirka Hana on Why Semantic Tagging Matters for Publishers
Our co-founder Jirka Hana recently contributed a guest article to the Stibo DX blog as part of our technology partnership with Stibo DX. Geneea’s semantic tagging and RAG-based research assistant are fully integrated into Stibo DX’s CUE enterprise media platform. The article, titled “Why Semantic Tagging Matters for Publishers,” explores the challenges modern publishers face... The post Jirka Hana on Why Semantic Tagging Matters for Publishers appeared first on Geneea News.
The JEDI Project: Rethinking Interview Workflows for Modern Newsrooms
If you have ever conducted an interview, you know the moment when the dynamic shifts. The recorder clicks off, and you’re left with an hour of audio, messy transcripts, and pressure to move quickly without sacrificing accuracy. The value is obvious. The workflow, less so. Interviews are among the richest sources of original reporting and... The post The JEDI Project: Rethinking Interview Workflows for Modern Newsrooms appeared first on Geneea News.