In this demo, we're going to use the Cloud data loss prevention API to redact personally identifiable information like e-mail addresses, phone numbers, that sort of thing, PII data at scale. Again, find all these demos under the Data Engineering demos course folder in our public repository. This one's pretty short. We're just going to be using the web tool, and you can invoke the API and experiment with that yourself. So navigating to that web demo, it's going to copy that link. So imagine in BigQuery, you've got e-mail addresses or something like that inside of your data, how do we actually proactively identify it? It says demo actually takes you to a page that has just a text code file inside of here. Inside the file actually explains a little bit about the DLP API itself. You can have this. It identifies this. It just looks through here. There's unstructured data, and the instructions basically says, "Hey, I'm reading through here. It's all good." Hey, somebody put a phone number inside of their comments. I don't want to have this go to anybody. I'm going to actually flag that as very high likelihood that is this type, it is a phone number, here is the string, here is where it is, and you can inform the model whether or not that's a good result or poor result. So what we can do is we can hide that welcome text, and we can paste in our own examples where we know this is personally identifiable information, and we can see if it catches through. Boom, immediately. Credit card number, very high, US driver's license, very high, e-mail address, part of the e-mail address you also have the domain and where the data is as well. So you imagine this is a trivial demo, this is just four lines here, but imagine you're inherited data-set that has terabytes, how can you automatically end at scale, run-through identifying, and then also use the different components of the API to run hash functions or obfuscation functions to redact that data, so that PII doesn't leak out there? All right. That's DLP, the API.
