Twitter fascinates me. Not only because it’s a platform where many academics disseminate the latest additions to their CVs, but also because it has an Application Programming Interface (API) through which you can automate all kinds of stuff.
One cool thing I’ve come across is bots that automatically retweet and/or favorite tweets that contain specific #keywords. This type of automation is possible through Twitter’s API.
Unfortunately, APIs usually involve a lot of low level technical details that require high levels of effort to use. So developers will create libraries to make APIs more accessible within a specific programming community. Consider the OpenSees framework, whose C++ classes comprise an API to the finite element method. The OpenSeesPy library wraps model building, analysis, and output functions with easy to use Python commands so that earthquake engineering researchers don’t have to deal with low level details of C++.
Knowing Twitter has an API, I Googled how to automatically retweet based on #keywords. It turns out making a bot is not that difficult using Tweepy, a Python library for the Twitter API. All you need is a Twitter developer account and an Amazon Web Services (AWS) account, or some other means of running apps in the cloud. The Twitter developer account is free if all you’re doing is writing a bot, while the AWS account is free for 12 months. That will give you plenty of time to learn how to run OpenSees on AWS (topic of a future post) before having to pay a couple bucks a month.
Following this Tweepy tutorial, I created the @OpenSeesTweets bot in about an hour. The bot monitors the Twitterverse for tweets that contain #OpenSees or #OpenSeesPy, then favorites and retweets those tweets. Follow the bot and its retweets will show up in your feed. Tweet with one of the keywords and your tweet will be retweeted into the feeds of the bot’s followers. You don’t have to follow the bot for it to retweet your tweets.
Now the real question is, how do we combine Tweepy and OpenSeesPy in a useful way? Make your OpenSeesPy script tweet when it finishes a response history analysis? Use the character count from a random tweet to seed your Monte Carlo simulation? Share your ideas in the comments section if you like.