foundlings on the cathedral steps
TRANSCRIPT
foundlings on the cathedral steps
liis and internet-wide search engines
thomas r. brucedirector, LII (original recipe)
lii-world & google-world
different operating philosophies
self-description•objects should (and do)
describe themselves
•many objects don’t (especially nontextual and acontextual objects)
trust and user-generated data
•user-generated metadata is unreliable
•high-quality metadata is something we do well
<meta>
importance
•popularity is importance
•domain-specific criteria can be more significant
privacy•information wants to be
discovered
•individuals bear the burden of protecting their privacy
•other policy concerns apply, and some are in tension
•publishers can and should act pre-emptively
shared problems, different remedies
context-smashing
•search engines are teleportation devices
•drilling down to detail is easier than jumping up to an overview
ontological mismatch
•mismatch between the way things are searched for and the way they are organized.
•in law, fact patterns vs. legal abstractions and categories
presentation of results
•we still assume the teletype
•we need different presentation for different stages of information-finding
•we want more interactivity, please
illusory completeness
•search engines promise everything , but index much less
unifying worlds: practical tips
improve google-facing data
•end unnecessary dynamic pages
•use <title> fully
•enrich cross-linkage
improve internal search
•restore adjacencies
•break down stovepipes
•expand and improve metadata
•present results in better, richer ways
conclusion: smoothing transitions