Fresh on the heels of its deal with Tumblr for access to the Tumblr “firehose,” social data platform DataSift is putting that data, and more, to good use with the launch of a new API that performs historical analysis across Twitter, Tumblr, and Bit.ly “firehoses,” as well as data pulled from forums, blogs and public Facebook data.
With “Historic Preview,” as DataSift is calling it, developers have the ability to perform historical analysis across all sources combined, using four different analysis types. These include: frequency distribution, numeric statistical analysis, target volume, and word count (a list of up to 100 of the most often used words). The analysis functions can be applied to all 400+ metadata fields, the company says. (More details on that here).
Also shipping today as a part of this new addition to DataSift’s platform are five standard Preview reports, with let developers dig into the data to find out more about the quality and quantity of the data before running a full search.
A Basic report gives you an overview, with things like word clouds and pie charts. The Natural Language Processing Preview offers up sentiment analysis and entities (people, places and things mentioned). A Twitter Preview lets you hone in on Twitter word counts, hourly data volume, mentions, hashtags and more, while the Twitter Demographics Preview report lets you scope out anonymized data detailing things like gender, age, location, profession, likes and interests. And finally, a Links Preview lets you track the hourly volume of links, Twitter card data, Facebook OpenGraph data, and other items.
The company, which competes with other social data API providers like Topsy and Gnip, explains that today conversations are happening across networks, and businesses need tools that let them track more than simple keywords, but rather the social conversations in general. That’s always been DataSift’s strength – that it can not only sort through the billions of social actions out there on the web, but it can then allow its customers to filter those for demographic information, online influence, sentiment and more.
A company blog post describes how these new Preview APIs can be used, explaining how an advertiser could track a social media campaign by watching how many times a YouTube link was shared – even though people may be sharing that link without also including any particular hashtag, for example. That allows the company to better understand who the influencers are on the subject, and what sort of hashtags might work to better promote the message in the future, among other things.
The post also highlighted how complex it can be to track social conversations on today’s web, using another example detailing how a movie release is monitored. It found that Facebook and Twitter mentions peak ahead of the release, but then Instagram peaks after as people start snapping photos of themselves at the event. Meanwhile, historic insights into Bitly link data and Tumblr posts gives an even fuller picture of the kinds of things being said and who’s doing the talking.
More information on the APIs have been posted here for developers.