Rewrite the News: Tracing Editorial Reuse Across News Agencies
This paper presents a weakly supervised, cross-lingual method for detecting sentence-level text reuse in multilingual journalism without full translations, revealing that editorial reuse is prevalent, predominantly non-literal, and concentrated in the middle and end of articles rather than in leads.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the global news industry as a massive, bustling kitchen. There are hundreds of chefs (news agencies) in different countries, each speaking their own language, chopping ingredients (facts) and cooking up stories (articles) to serve to their local diners.
This paper is like a food critic who sneaks into the kitchen to answer a very specific question: "How much of the dish served by the Slovenian chef was actually borrowed from the recipes of chefs in Italy, Germany, France, and elsewhere?"
Here is the breakdown of their investigation, translated into everyday language:
1. The Big Problem: Too Much Noise
The researchers noticed that journalists are drowning in information. Every day, thousands of news stories flood in from around the world. A journalist in Slovenia can't possibly read every single story from every foreign country. They need a way to quickly spot, "Hey, this story from a German agency is the same one I need to write about for my Slovenian readers."
2. The Detective Work: Finding the "Recipe"
The team built a digital detective tool. Instead of reading every word, the tool uses AI "smell tests" (multilingual embeddings).
- The Analogy: Imagine you have a recipe for "Chicken Soup." Even if the German chef calls it Hühnersuppe and the Italian chef calls it Zuppa di Pollo, and they use slightly different spices, the tool recognizes they are both making "Chicken Soup."
- The Trick: The tool doesn't need to translate the whole article first. It just compares sentence by sentence to see if they smell similar, even if they are in different languages.
3. The Rules of the Kitchen
To make sure they weren't accusing the wrong person, they set strict rules:
- Time Travel: They checked the timestamps. If the Slovenian article was published before the foreign article arrived, it's impossible the Slovenian chef stole it. That's a "false alarm."
- The Earliest Source: If three different foreign chefs all made "Chicken Soup," the tool assumes the Slovenian chef copied the one who started cooking first.
4. What They Found: The "Middle-Heavy" Theft
The results were surprising and revealed a hidden pattern in how news is made:
- It's Not Just Copy-Paste: The Slovenian journalists aren't just photocopying foreign articles. They are remixing them. They take a fact from a French report, a quote from a German one, and a statistic from a Polish one, and blend them together.
- Analogy: It's like a DJ mixing tracks. They aren't playing one whole song; they are taking the bassline from Song A, the vocals from Song B, and the beat from Song C to make a new hit.
- The "Lead" is Original: The very first paragraph of a news article (the "lead") is usually written fresh by the local journalist. This is the "hook" to grab the reader's attention.
- The "Meat" is Borrowed: The middle and end of the articles are where the borrowing happens.
- Analogy: Think of a news article like a sandwich. The top bun (the headline and first paragraph) is homemade. But the meat, cheese, and veggies in the middle? Those were likely imported from a neighboring country's deli and just rearranged.
5. The Scale of the Operation
- They looked at 1,000+ Slovenian articles and 237,000+ foreign articles.
- They found that 52% of the Slovenian articles contained at least some reused sentences from foreign sources.
- However, because the foreign agencies are so huge, only 1.6% of their articles were ever reused. It's a one-way street: the small agency picks and chooses from the giants.
6. Why This Matters
This study shows that news isn't just "written"; it's assembled.
- For Journalists: It proves that their job is often about curation and synthesis (mixing ingredients) rather than just reporting from scratch.
- For Technology: It shows that simple "copy-paste" detectors don't work. You need smart tools that understand that "The sky is blue" in English and "El cielo es azul" in Spanish are the same idea, even if the words are totally different.
In a nutshell: This paper is a map of how news travels across borders. It reveals that while the "face" of a news story is local, the "body" is often a global collaboration, stitched together by journalists who act like master chefs, blending flavors from around the world to feed their local audience.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.