The myth and reality of federated search postprocessing

1
The myth and reality of federated search
post-processing

• Todd Miller
• WebFeat

2
Hot buttons

• De-dupe
• Relevancy
• Parsing

3
The poop on de-dupe
Results downloaded in sets of 10 _at_ 5 seconds per
set 5.97 hours
4
Irrelevancy ranking
Keyword search Hashimotos Encephalopathy
5
Parsing
6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
Why is parsing important?

• Sort by date, title, author, publication
• Link to OpenUR
• Export to bibliographic citation management
  packages
• COUNTER-compliant usage tracking

10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Conclusions

• Federated search engines cannot overcome the laws
  of physics or content providers
• All roads lead to parsing