[WikiToLearn] Semantic feature ideas

[Alessandro Tundo]

I try to reply to everyone with the same messagge.

Sofia

There are two different approach for extracting information:

- try to query famous knowledge-base using their SPARQL endpoint (easy to
do, but is very difficult to extract that kind of information)
- create an internal data-set

The second one is obviously more interesting. I have to model our domains,
define what are "entities" and what are their "properties" and so on.
Obviously I can use template formats to find them in the pages.

Feature you have suggested is interesting too. I should make a topic
extraction and collect pages with same or similar topic, after hint them in
a box or something like that.
Most likely I add it to my use cases.

Gianluca

I don't know if is possible to represent math this way.

Davide

There are two main unit of work:

- extract & collect
- visualize

The second one most likely will be an extension, the first one depends on a
big problem: automatic annotation or human annotation?

If the system will be all automatic I think we have to create a web-service
that extract information and store them into a triple-store, if we want to
introduce a kind of "semi-automatic" system we have think about how people
could annotate entities while they edit pages.

Regarding "index equations", I still don't understand. Now, it isn't
important to say "how make something", but "What we want to represent and
why".
So, how "indexing" equations could be useful? Search them? Which "format"?
Would you like they are automatically recognized into pages and linked to
pages where they are explained?

Sorry for my bad english!

Bye,

Alessandro
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/wikitolearn/attachments/20151217/5b304bbc/attachment.html>