A lot of new cool stuff has been released in the new version of OpenAmplify, the 2.0 Version: Text-level Polarity, Guidance, Decisiveness and Temporality; NER - Named Entity Recognition; PinPoint to name some.
One thing that is not mentioned in any release note, is the fact that we try to enhance our Signals whenever we can, wherever we can. Smaller changes, like resource and list updates, are made basically every day. The OpenAmplify codebase is completely depending on linguistic resource files (says I, being a linguist), we maintain several hundreds of resource files. That is what is so thrilling about being a linguist and working on a project like OpenAmplify.
Just think about verbs.
In my world, the world of linguistics, a verb is a part of speech that denotes a state of being or an action.
According to Martha Palmer's Verbnet the English language holds about 5000 unique verbs. We at OpenAmplify recognize the frustration the end user might experience, trying to deal with the large number and wide range of different verbs and senses. Actions are based upon the verbs identified in the analysed text, and the range of values that
can be returned is very broad. Our solution to help make automated decisions upon Actions possible, is ActionType. With ActionTypes we add a layer to the output that classifies Actions into useful categories.
So, what is a useful category then? Have you considered how many types of actions or verbs there is?
If you were asked to list some, which types would you start with?
Run and Hide?
Laugh and Cry?
Speak and Listen?
When I think of verbs, the first that comes to my mind are things like "jump", "run", "eat" and so on; typically physical Actions, but there is so much more in Actions:
OpenAmplify currently identifies 50 different ActionTypes.
Go take a look! Maybe the ones you first thought of aren't listed?
Which groups or types verbs describe your universe?
Posted
3 Feb 2010 9:33 PM
by
Alexandra Stalnacke