Curse of the languages – that people could no longer cooperate
Three lessons
Don’t mess with God
Cooperation
How do we identify that apples are apples and oranges are oranges?
We put our experiences together in order to distinguish the two things
Context
The “database of things” around our experiences
What happens when we don’t agree?
We crash Martian orbiters
What about a computer?
When a computer encounters a piece of data, it has no context
Is the number 20 an hour, a number, the result of an algorithm?
Can’t computers use HTML as a sort of a database to provide context for all the things they look at?
No; HTML was designed for humans to understand; not for computers
People understand relationships between data; computers don’t unless we tell them
How do we get computers to understand and give them context?
How do you explain the semantic web and semantic technologies?
Tagging important data with a context and a vocabulary
Humans frequently don’t speak the same language at all
We have different preconceptions and different filters
We have to build ontologies, vocabularies, and tag everything and connect it to the semantic database so computers can understand it
If you are running predictive analysis on a whole bunch of data, you can use semantic analysis to unify datasets from very distant places and use it to understand what’s predictive
Potential application: healthcare and science
Semantic data makes the data highly valuable, because it allows other groups to “piggyback” on work that is being done without needing a “translation” of the work
Any company can define their own ontology, but if we let a company control them, it becomes hard to use them
Publishing and Web apps
What if we started tagging news semantically so a computer can understand the relationships?
Starts with making pure browsing, search, and discovery better
MS Pivot as an example of data that is semantically tagged
Public health
What if we semantically tagged symptoms for everything across the country?
Could identify the patterns related to disease and illness outbreak
AI
Semantics is the foundation for very robust AI
This is why computers can’t respond to stimuli the same way we do
Example – have people watch something and take semantic data of the emotional response – use those curves to model relationship in a computer
Companies using – Merck, Biogen Idec, GroupM, Chevron
Projects
DBPedia – structured data from Wikipedia
GoodRelations – product information, etc.
Swoogle – semantic resource search engine
Cambridge Semantics – enabled spreadsheets used by Group M
Pivot
What does this all add up to?
A better understanding of our world and the patterns that drive it