Researchers from the University of California San Diego have developed an AI-based system SpoilerNet that will save people from accidentally reading spoilers by flagging them in online reviews of books and TV shows.
“Spoilers are everywhere on the internet, and are very common on social media. As internet users, we understand the pain of spoilers, and how they can ruin one’s experience,” said one of the senior authors Ndapa Nakashole.I
The AI tool named ‘SpoilerNet’ is powered by neural networks that automatically detect spoilers. For training it, the team looked for huge datasets of sentences containing spoilers, but found none.
Hence, they created their own spoilers by gathering over 1.3 million book reviews annotated with spoiler tags by book reviewers. The tags feature sentences that had spoilers and hide them behind a ‘view spoiler’ link in the text.
Also, few words also had different semantic meanings in different contexts, thus the study’s first author Mengting Wan admitted that identifying and understanding these differences was challenging. After tests, the system detected spoilers with 89%-92% accuracy.
They also ran the system on a dataset of over 16,000 single-sentence reviews of around 800 TV shows and the tool accurately detected spoilers 74%-80%. The team believes that SpoilerNet could be used for building a browser extension for shielding people from spoilers.