AnnoMarket: A brand spanking new marketplace for natural language processing pipelines
The job-finding wizardry that powers Zonino is based on the text engineering platform GATE. We use it to find jobs and we use it to parse them, too. It's bloody brilliant and I whole heartedly encourage you to give it a try. When it comes to complex rules-based text engineering it's the cream of the proverbial crop. There'll be time later for a detailed discussion of GATE's features and limitations but for now we can focus just one of the latter: It's a teensy bit slow.
Ask any GATE developer you bump into on the street (I know it's scary, but we live among you) and they'll most likely cite GATE's speed as the most significant barrier to scaling. The smart cookies at Sheffield University have come up with some optimisations but the usual approach is to aim for some kind of distributed architecture.
Enter GATE Cloud - an infrastructure-as-a-service solution that allows you to upload GATE pipelines and run huge batch jobs without having to worry about wrestling with AWS. Taking their experience from building this service, and collaberating with a consortium of experts from OntoText, the Press Association and IMR, the team announced the launch of AnnoMarket this month. It provides a marketplace of APIs through which anyone can access those pipelines and enables their creators to monetise the useful stuff that they've built. Helen Lippell from the Press Association sums up the project nicely:
AnnoMarket aims to disrupt the market for text analytics by offering cost effective access to a world-class range of services, all hosted in the cloud.
I rather like this idea. And I have a lot of admiration for the people involved. But it's a shame that the budget didn't stretch to a snappier logo:
But who cares. The point is that this kind of marketplace has never existed before and it'll be really interesting to see what kind of uptake it gets from both sides. Text engineers can finally publish all of those handy pipelines they have gathering dust. These things can take a long time to build and encouraging healthy competition in various domains can only be beneficial.
To get the ball rolling the team are touring conferences and offering free credit to anyone who wants to try out the platform while it's still in beta.
This is still early days and it'll be a while before we see if this is something people want. But in the meantime drop the AnnoMarket folks an email and they'll send you a code for some free credit. You can also try out any of their pipelines (like the rather good tweet parser) online, which is super handy.
Also, be sure to check out current NLP jobs at Zonino and sign up to get alerts on future NLP and Text Engineering jobs at London startups.