Building a recommendation Engine from Noisy, Duplicated Data
Jason Hoyt (Mendeley Corporation)
Info
|
Date |
10th November 2009 (week 5, Michaelmas Term 2009) |
|
Time |
11:30 |
|
Place |
478 |
Abstract
Building a recommendation engine from plain text data is a difficult task. Beyond plain text, noisy, inaccurate, and duplicated metadata from text extraction of PDF documents presents an enormous challenge. Mendeley is a reference manager for researchers that that is doing just that. The infrastructure and data mining requirements to build a recommendation engine from text-based PDFs will be discussed.
Further info
|
Related series |