Unemployment is a representation of % of people who are currently employed or are currently, actively looking for work. It doesn't (iirc.) count people who are able to work, but have stopped looking for work. The denominator represents the total workforce of the economy being observed. An unemployment of 8% means that 92% of all people who are able and want to work have successfully found employment. What bj is getting at is that because unemployment is dropping, people who had previously tried to find work but were unsuccessful and stopped trying (and therefore dropped out of the denominator) have been recently emboldened by the job prospects and so have restarted their job search, but have yet to have successfully found employment (whether due to latency or, as bj suggests, lack of qualifications). An increase in the denominantor without a consequential increase in the numerator (yet), means that the unemployment% has increased even though there are still job openings which need to be filled.
I can't remember if unemployment accounts for underemployment, but if it doesn't, that would further exacerbate the figure as you could expect people quitting their jobs to pursue a better position on the open market.