* Add Topic model
* Add PDB post-mortem debugging
* Add logger to plugins (`self.log`)
* Add NLTK resource auto-download
* Force installation of requirements even if adding doesn't work
* Add a method to find files in several possible locations. Now the plugin.open
method will try these locations IF the file is to be opened in read mode.
Otherwise only the SENPY_DATA folder will be used (to avoid writing to the
package folder).
* Fixed Options for extra_params in UI
* Enhanced meta-programming for models
* Plugins can be imported from a python file if they're named
`senpy_<whatever>.py>` (no need for `.senpy` anymore!)
* Add docstings and tests to most plugins
* Read plugin description from the docstring
* Refactor code to get rid of unnecessary `.senpy`s
* Load models, plugins and utils into the main namespace (see __init__.py)
* Enhanced plugin development/experience with utils (easy_test, easy_serve)
* Fix bug in check_template that wouldn't check objects
* Make model defaults a private variable
* Add option to list loaded plugins in CLI
* Update docs
* Refactored BaseModel for efficiency
* Added plugin metaclass to keep track of plugin types
* Moved plugins to examples dir (in a previous commit)
* Simplified validation in parse_params
* Added convenience methods to mock requests in tests
* Changed help schema to use `.valid_parameters` instead of `.parameters`,
which was used in results to show parameters provided by the user.
* Improved UI
* Added basic parameters
* Fixed bugs in parameter handling
* Refactored and cleaned code
We've changed the way plugins are activated, and removed the notion of
deactivated plugins.
Now plugins activate asynchronously.
When calling a plugin, it will be activated if it wasn't, and the call will wait
for the plugin to be fully activated.
This lines up the names in the conversion plugins with the [emotionML suggested vocab](https://www.w3.org/TR/emotion-voc/#dimensions).
emoml has different names for the 4-dimensional fsre scheme and the 3-dimensional vad scheme, which this pull request has added.
I've added the "unpredictability" dimension and mapped big6:surprise to it's maximum value. The assumption is that surprise varies between 5 and 10 to be in line with the other dimensions (no such thing as negative surprise, so no values less than 5). I see that arousal also has all values >5 (so no negative arousal). Ideally, surprise mappings for V, A and D should be calculated empirically - I think there'll be some arousal and possibly slightly lowered dominance.
I wonder if we should use another colon in the emoml names, eg: "emoml:fsredim:valence" or "emoml:big6:happiness", since the [emoml suggested vocab](https://www.w3.org/TR/emotion-voc/xml) only specifies names like "happiness" in a category "big6" (ie: it's hard to know which is the category in "big6happiness").
We'd have to go through the example plugins and make sure they also conform...
open to discussion on this btw...
ps: apologies for multiple changes in this one pr..
I've used euclidean metric instead of taxicab as I feel it makes more sense (taxicab has bizzare unintuitive effects for points far from the centroids).
Does a weighted average of centroids.
If intensity sums to zero for a category, a 'neutral' category is used or 0 if it's not present. I'm not 100% sure this is the best approach, and the name of the "neutral" category perhaps should use some convention?
Note that if there are no categories present, then no VAD (or other dimensional) estimate is returned. It may be better to use the neutral centroid if it's present in this case also.