See our Recent Projects

Epimenides Research's Application for Commodities Market Presented at the IPMI 26th Annual Conference

Development of Flash demos of Epimenides Technology Now Underway

Check out the spec sheet for a test query on Layoff Announcements

Layoff Announcement Query

A Challenge to Other Means of Collecting Forward-Looking Employment Data

Economists struggle with Gray areas; counting layoffs is no Christmas when done by hand

Locating Layoff Announcements in US News Sources: An Example of Precision Finding of Text-Events, using Epimenides NLP

Challenge: Process entire text of 1993 Wall Street Journal (4.3 million words) to find layoff announcements, i.e. layoffs as today's news

Run time: 30 seconds to process all text (Intel PII processor)

Accuracy: Total layoffs found were 430,000, vis-a-vis standard count of 405,000 for year by Challenger Gray Christmas, i.e. a 6% difference; BUT it appears that the extra 25,000 layoffs located were genuine, and were missed by Challenger

Same count could otherwise take a small team of humans about 2 person-years

Factors noted include: percentage of workforce laid off, effective date of layoff, announcement date, number of jobs lost, firm doing the laying off, others

Technology \| Products \| About \| Team \| Sign-In
	Last Update: January 12 2008	Contact Us