Today we’re announcing the release of Datastreamer 6.5, which includes a number of new APIs and features that provide our clients with the ability to listen to, engage with, classify, as well as analyze social media content.
Datastreamer now provides “engagement” support. By integrating two social products – social content and social engagement – on an integrated platform, Datastreamer allows clients to update shares and likes counts in near real time. This goes beyond just raw number counts; you can sort by descending order, filter results by engagement counts, etc.
Datastreamer updates engagement on about 25k posts per minute If you’re using our Search API then you’ll be able to execute queries, and sort by posts with the most engagement enabling you to find the top content being discussed in social media.
We’ve added the ability to use Datastreamer to analyze and classify a client’s own data utilizing our standalone classifier, which work with any content – not just Datastreamer content. The Classifier API works by taking a large corpus of data (or set of training examples) with labels generated per class, then using a linear classifier to build a model, we are able to mathematically label future input text based upon the original corpus.
Datastreamer now supports “static exports” for our clients. This feature allows Datastreamer to export large amounts of custom data from social media sources. Clients are able to immediately receive large multi-year exports based on time, boolean logic, and other search criteria. This very cost-effective feature allows clients to request custom exports or “dumps” from Datastreamer instead of paying for more complex, expensive solutions.
The Parser API provides a way for clients to request documents and get back parsed metadata around a specific permalink, news article, or blog post and provides clients with API access to content on a more granular basis. If a URL is not currently indexed, or is older, or something Datastreamer might not ordinarily index; this feature allows clients to still “fetch” the desired content, and then subsequently analyze the content, all by using Datastreamer’s content schema and machine learning infrastructure.