◀ 7.1. Semantic Text Composition with Seed

7.1.1. Basics

The SEmantic EDitor (Seed) recognizes entity mentions in a text and annotates them with (i.e. maps them to) things in the user's PIMO or to entities from public knowledge sources such as Freebase or DBPedia.

By analyzing texts, Seed directly adds semantic information to a text while the user is still writing. He/She may then modify, confirm or reject individual annotations. A simple color code signals an annotation as follows:

* For the sake of simplicity we speak of a "term" here, but actually refer to one or more words
in the text that form a term, e.g. "Edinburgh", "Queen Elizabeth II", "Summer Olympic Games" ... etc.

The following video illustrates these basic functionalities.

7.1.2. Import new concepts into the PIMO

If no information about an entity is found, it is not highlighted at all. The user may then select that entity and add it as a new thing to his/her PIMO within Seed. The following video shows an example.

Note taking is already preserving!

Note taking is very important for us humans. It helps us write down what was, what is, what has to be done, what should not be forgotten, etc. To be more precise, it is as easy as that: Note taking is already a human made preservation act, and a very good one indeed, as we humans are trained to do that right from school on.

It is obvious, that notes are a very good candidate to be preserved themselves (not all of them, but probably many). In addition, many notes describe relationships between (other) important things. As such, these notes are perfect context providers for these things. They are perfect because, they even describe the relationships in human readable form. How can context get any better?

The only problem is to: not all of the notes taken by a human are really important. We need to find good measures to estimate this as automatically as possible. Users' text editing interactions may yield sufficient evidences in that direction, but experiments are not finished here.

Seed Online Demo (w/o PIMO)

An online evaluation experiment demonstrating the use of Seed for annotating existing texts or composing new ones. It also includes a questionnaire for assessing some of its features and its usability.