This video was originally posted on the Freebase blog. As far as I can see, it could also be re-interpreted as “Topic Maps in a nutshell”. Just replace “entity” with “topic” and you get a pretty good idea of what Topic Maps are all about – the term “map” even pops up in the video itself!
Here’s the key ideas and some comments from a Topic Map-ish point of view (but of course the range of Topic Maps goes well beyond the web):
- The same word can mean so many different things — how do we know what is meant, how do we make sense of the implicit meaning, and especially: how do we make a computer understand?
- When it comes to finding, linking, reconciling, organising multiple layers of information, words are not the best solutions — synonyms, polysemes, homonymes etc. Humans can derive meaning from context, but computers need a more explicit system to manage the semantics
- Build website around entities — topic maps are built around topics, and identity management through attributing subject identifiers is key
- What’s an entity? A singular person, place or thing — a topic represents one real world subject in the map (i.e. it is a subject proxy)
- There are 50 different ways people write UC Berkeley – and they’re all referring to one single place, one entity
- By mapping all those words to a single entity, you can combine all the information about UC Berkeley in one place — that can be achieved by using subject identifiers
- We can build a map to show how entities are related — associations between topics
- You can look for things that share certain attributes — this is also possible by querying a topic map
- Entities are smarter than words — at least when it comes to computers who can’t use human strategies to determine what is meant
- Identifying these entities and mapping out how they’re related and what words other sites use to refer to them
- Switchboard for content on the web
- When your site is built on entities, new things get connected automatically — reliable information integration with the help of subject identifiers and / or through merging
- Linking happens under the hood of the same entity — clustering of information around a topic