29.01.2014 Views

Part 4: Index Construction

Part 4: Index Construction

Part 4: Index Construction

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Further issues with multiple indexes<br />

Sec. 4.5<br />

Collection-wide statistics are hard to maintain<br />

E.g., when we spoke of spell-correction: which of<br />

several corrected alternatives do we present to<br />

the user?<br />

• We said, pick the one with the most hits<br />

How do we maintain the top ones with multiple<br />

indexes and invalidation bit vectors?<br />

• One possibility: ignore everything but the main<br />

index for such ordering<br />

Will see more such statistics used in results<br />

ranking.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!