Encryption Project Uses Embedly Extract to Preprocess Articles
Embedly was selected to present our APIs and offer support as an ambassador at the 2013 hackNY. There was lots of energy in the room, with a focus on projects being awesome, rather than chasing hockey stick growth. They even had a hardware hacking room with 3D printers.
Details of the algorithm are provided on Github.
The algorithm currently adds padding to plaintext in such a way that makes cryptanalysis more difficult. This is because it pads based on a given corpus (chosen by the encrypter) and mimics that frequency. This makes discerning the true plaintext difficult. In order decrypt, you need a series of offsets, for each actual plaintext character. By combining both of these, we have an easy to decrypt and relatively cheap encryption algorithm.
When I saw the project presentation I was curious how Embedly was used in it. Eric got back to us, mentioning that Embedly Extract was used specifically to parse an article and get the content. From Eric:
We used embedly to get corpus’ of a specific kind. We then did frequency analysis(similar to what is done in cryptanalysis) on the corpus in order to determine likely frequency patterns, making it difficult to tell which letters are part of the plaintext and which are part of the ciphertext.
According to Eric, the team doesn’t plan on stopping now, and has some pretty admirable ambitions.
We plan to expand this hack and turn it into a cryptographic standard that makes the internet a safer better place.
When we built Extract, we knew it would be used in natural language processing applications, but we never imagined it would be used for encryption. Pretty cool seeing how people use the Extract API!
HackNY was a great time. If you’re having a hackathon, let us know, we’d love to come by.
Update (04/18/13): Danny got back to us on what it was like using the Extract API:
We only used the extraction API, but it was very nice, easy, simple to use. It returned everything we needed.
No complaints, and it worked quickly, which is always nice at a Hackathon - minimal time was needed to get it doing everything we needed.
Great to hear!