MONGOOSE: Ingest, Monitor, Rinse, Repeat

Google Tech Talk
October 23, 2009

ABSTRACT

Presented by Daniel Gruhl.

Currently, data analytics technology is in high demand as people try to extract as much value as possible from their most valuable resource - the information around them, whether in their organizations or freely and publicly available. Unfortunately, though many data analytics efforts are focused a particularly interesting (and often difficult) question, whose answer hopefully lies in the data, these projects tend to spend most of their cycles acquiring and ingesting data. Thus, the focus of these efforts tend to tilt away from data analysis and towards data ingestion. MONGOOSE is 1) A suite of technologies that one can plug domain knowledge cartridges into and that outputs data suitable for OLAP or BI consumption. One plugs in small amounts of domain knowledge that involves pulling in unstructured, semi-structured and structured data, and MONGOOSE converts it all into structured form. 2) A Platform for Worst-Case Scenario Workflow Management. MONGOOSE is built on the assumption that failure happens and it must be handled quickly and seamlessly, such that it does not stop or hinder information ingest. 3) A Platform for Community-Based Information Extraction around specific phenomenon that can be fed into statistical analysis tools.

[mostermand]

Obviously you can. I do it all the time

[ManuelBTC]

14:40 - 15:06
You can't complain about people changing their internal code. They made no guarantees to you and it's not their fault you made yourself dependent.