Facebook's open-source algorithms

Facebook is all set with its attempt to keep its platform safe from abusing content shared over the Internet. The company has planned to open-source two algorithms namely PDQ and TMK+PDQ that can be used to detect child sexual abuse, terrorist propaganda, and graphic violence.

These algorithms are able to identify a harmful piece of content shared over the Internet. Using this technology, the company aims to identify duplicate content and prevent it from sharing with a large audience. It not only identifies duplicate content, but also detects nearly identical pictures as well as videos.

PDQ and TMK+PDQ are developed with an inspiration from existing models and applications, that includes pHash, Microsoft’s PhotoDNA, aHash, and dHash. PDQ is a photo-matching algorithm which is designed after pHash. Whereas, the video-recognizing TMK+PDQF has been designed by the Facebook Artificial Intelligence Research team and academics from the University of Modena and Reggio Emilia in Italy.

Facebook stated that these algorithms are designed based on their experience in identifying ‘abuse’ over billions of Facebook posts. The company hopes to contribute to the community and help many companies safeguard their services.

How do these algorithms work?

PDQ and TMK+PDQ have an effective technique for storing files. They use a short digital hash to identify whether two files are identical. This hash can be easily shared with other companies and non-profit organizations. For instance, if terrorist propaganda is detected on the platform, it can be removed and hashed. Experts believe that these algorithms are sure to pave through a long way in serving to track down exploitation as it occurs, and then obstruct its spread.

John Clark, the President and CEO of the National Center for Missing and Exploited Children (NCMEC) stated that within a single year around 541% of rise was observed in the total number of child sexual abuse videos. Further, he said that he is assured about Facebook’s large contribution of this open-source technology which will sooner or later lead to the identification of such harmful and abusing content, and rescue more child sexual abuse victims.