Good news for the victims of spoilers: The new neural network bringing ease in their lives

Internet is a platform providing all sorts of opportunities to its users from finding new jobs to finding entertaining posts to forget about the daily struggles in our lives. Internet is a place where people can enjoy being themselves without the fear of being judged by others.

Internet Spoilers

Sometimes when a new movie or a season or any new episode releases, we wait to get free from all our daily chores to watch the specific part in a very special time. With the growing internet, it really becomes difficult to avoid spoilers. Some people watch the movies or season prior to others and to get attention or to just sarcastically ruin the suspense for others, release the spoiler parts over the internet. Even if people are careful to not let any spoiler come in their way but all it takes is a random tweet or a recommended news item to ruin the plan to watch season finale late.

SpoilerNet - a savior of all your problems

SpoilerNet is a team composed of people who were a victim of spoilers when they tried waiting a week to see infinity war. A new AI that can help them detect spoilers to aware others about it. With this new feature, people can easily spot a spoiler and don’t have to be a victim of it anymore.

Workings of SpoilerNet

The team of SpoilerNet assembles a database of more than a million reviews from Amazon-owned reading community Goodreads. The purpose of this team is to note spoilers from any reviews, analyze them line by line. The neural networks learn the qualities that define a specific image, object as spoilers. The team feeds 1.3 million Goodreads reviews into their system, letting it observe and record the differences between ordinary sentences and ones with spoilers.

Testing the AI agent

After training was complete, the agent was launched on separate sets of sentences which it was able to label like “spoiler” or “non-spoiler” with more than 92% accuracy. With this agent, the spoilers in the content can easily be identified without worrying about anything else at all.

This sort of model helps to analyze the large-scale dataset and to review any sort of sentence-level spoiler tags or any other meta-data. For research communities, this type of information helps with the possibility to analyze real-world review spoilers in details as well as developing deep learning models in this domain. The system should be able to run in real-time soon on user’s computer of course with a specific training too. This new feature opens up the possibility of a plugin or an app that reads reviews before anyone else and hides anything that feels risky or gives a hint of spoilers in it.

Our opinion on this new approach

This kind of approach is still new in our world, it helps a lot to point out the spoilers in any content but along with that it also has a drawback. For instance, this model sometimes mistakes a sentence having spoilers even if other spoiler-ish sentences are adjacent. It’s quite understandable that sometimes the sentences are not good enough to analyze exact spoiler words easily so a computer model may have trouble identifying the difference too. No doubt it would be the best tool for any Amazon or any other sub-businesses to be able to easily mark spoilers in reviews and any other content as well. Until the tool is launched we still have to stick to the old method by trying to ignore all sorts of posts with a hint of spoilers in it.

You can read the complete research paper here.

This AI tool will spot spoilers for you
Photo: timyee Getty Images/iStockphoto

Read next: Amazon’s Alexa May Soon Be Able to Gauge Your Moods
Previous Post Next Post