With each breaking announcement, IBM advances both its agenda as a data-and-services giant and its intent to evolve the Watson machine-learning service into a genuine business tool.
The company's new partnership with Twitter offers customers a way to take advantage of Twitter's data firehose via IBM's Watson Analytics service. Additionally, the linkup may give Watson another source of real-time, real-world data on which to further hone its analytical skills -- and extra payoff for IBM.
At the core of the announcement is news of the direct integration of Twitter's data streams into three IBM services: Watson Analytics, Watson Developer Cloud, and IBM's Bluemix PaaS. By using Twitter as a data source, businesses can mine for "allowing sales, marketing, and customer service professionals to map sentiment and behavior to better engage and support their customers," according to IBM's press release. Other services offered jointly by IBM and Twitter include consulting around specific verticals, such as banking or retail.
This squares with both IBM and Twitter's existing and emerging missions, but one of the big fringe benefits for IBM is access to Twitter's data, further cultivated by its business partners, to enhance Watson's real-world knowledge.
Machine learning isn't merely about algorithms, but rather algorithms plus real-world data, also noted by analytics visualization makers Adatao. The larger and better-curated a data set Watson has to work with, the more answers it can provide -- and the more accurate those answers can be.
The hard part is the same as with any machine-learning system. Feeding the system precleaned data by hand generates accurate results, but the cleaning and feeding process is slow. While the real world is a great source of data, the quality of that data also matters, an issue Google grappled with as part of its quest to build a machine translation system based on data harvested from Web searches.
Many other companies already harvest Twitter as a real-world data source -- doubly so now that Twitter is looking to monetize its business more creatively. Another newly unveiled Watson service, DataWorks, provides a platform for data cleaning that could be used as a front end by IBM for what Watson ingests -- or for what other companies feed Watson.
IBM's long been conscious of -- and banking on -- third-party usage of Watson and the ways it could provide the machine with the fuel it needs to grow. However, in last year's news of Watson's transformation into a cloud service, IBM raised eyebrows with the stipulation it "retains the rights to certain improvements for Watson that emerge because of customer interactions."
The specific leaps and bounds Watson makes due to this deal are tough to predict, but it's clear IBM -- and Twitter -- are ready to turn the results into new revenue streams and business models.