[frankinstien]

I landed on a multi-tiered emotions model, where it borrows from psychology and neurology.  There Arousal & Valence levels as well as chemical signatures. So for a little background: the charts shown in the pics are from a descriptor modeling approach I developed. Its an object-oriented modeling scheme that does use inheritance. So if you change the type of field properties or add more properties all elements that inherit from the base class will get updated automatically. Below are some pics of the tool I developed to manage the data generated using this model:



As you can see the Object Arm has many parts and those parts have parts! So an object oriented approach works well as a word descriptor schema.



From this image you can see what classes the emotion "Adoration" inherits from.



The three tiers of the emotional hierarchy totals 112 separate emotions.

This post continues to the next one.

[frankinstien]

Continuing:



By selecting an emotion on the left side of the panel causes the right hand side charts to come up. From the image above we can see that Primary emotions comprise of Surprise, Love, Fear, Joy, Anger and Sadness. The lower panel show the Chemical vector and the Arousal & Valence vector for a particular emotion. By selecting an emotion on the right hand chart causes the lower Chemical Vectors and Arousal and Valence vectors.



By Selecting a base class of Secondary Emotions displays the hierarchy and list of all secondary  emotions in doughnut chart.



Here is the full hierarchy with all the tertiary emotions in the doughnut chart.



By selecting an emotion will display that emotions hierarchy of the primary, secondary and tertiary emotions. As you cans see Surprise is a primary emotion that has secondary emotions which are listed in the doughnut chart. By selecting an emotion on the doughnut chart it will populate the chemical vectors and arousal valenece vectors.



When you select one of the lower radar charts it will populate the left hand side with track bars where you can modulate each of those vectors.

[frankinstien]

I added and made some improvements on the coding of emotions, inclusive of beautifying the charts. I've included facial muscle vectors and sensory nerves for body maps that are to indicate where emotions are felt:





Below is the vector editor I created to build these data structures. Its a breeze to update because of its object oriented nature, all I need to is add the vector to a base class and all classes that inherit from it get updated immediately. To build the sensory nerve vectors I simply listed the nerves in note pad set up their ranges then just pasted the list onto the chart and it builds the vector structure. I'm a lazy man...

[frankinstien]

Just want to add that adding and modifying data is pretty easy with the tool, its just a matter of dragging and dropping to a generalized type. So it could be vectors or images or video or audio or url, etc: And again; because it's an object oriented approach, once you draged an item to a class any classes that inherit from it gets updated as well without any intervention.

[frankinstien]

If you're not familiar with object-oriented concepts when a class inherits from another class the datapoint; property, field, method, or attribute does get inherited but each class takes its own instance of the data where if it gets modified it does not affect any other class that's within the hierarchy. I did implement the solution so the modification of the base class changes by adding or removing a feature or if the data type is changed will change the hierarchy.

[infurl]

There are many different ways to implement and support object oriented programming. Some are better than others and some are downright counter-productive. For some languages like Smalltalk, OOP is the only programming paradigm and given how fanatical the Smalltalk programmers I've known were about it, it must have been a good implementation though I haven't used it myself. Other languages like Python have been broken from the start and every now and then its architects come out with a whole new version that's supposed to fix it and which only seems to succeed in breaking backward compatibility with everything. I much prefer the way Common Lisp does it. It is the most complete, thorough, and powerful implementation of object-oriented programming, and also the most robust and reliable.

[frankinstien]

There are many different ways to implement and support object oriented programming. Some are better than others and some are downright counter-productive. For some languages like Smalltalk, OOP is the only programming paradigm and given how fanatical the Smalltalk programmers I've known were about it, it must have been a good implementation though I haven't used it myself. Other languages like Python have been broken from the start and every now and then its architects come out with a whole new version that's supposed to fix it and which only seems to succeed in breaking backward compatibility with everything. I much prefer the way Common Lisp does it. It is the most complete, thorough, and powerful implementation of object-oriented programming, and also the most robust and reliable.

Excellent point, however, I would not call what I've implemented a programming language but a object oriented data model. Where the model has abilities to be aware of changes and take advantage of object oriented concepts that make building up data and forming relationships more efficient as well as effective.

As far as object oriented languages go I am a c# fan and while c# doesn't allow for multiple class inheritance my data model does. 

[ivan.moony]

This could be a basis of forming response to stimulus, right?

Stimulus -> input emotion -> AI algorithm -> output emotion -> response

AI algorithm could make a track of input sequence to form an output sequence.

You could group a set of all primary emotions to one, ultimate emotion called "life", right?

[frankinstien]

This could be a basis of forming response to stimulus, right?

Stimulus -> input emotion -> AI algorithm -> output emotion -> response

AI algorithm could make a track of input sequence to form an output sequence.

You could group a set of all primary emotions to one, ultimate emotion called "life", right?

The beauty of using this kind of approach is it can be culture-centric, so if you wanted to group all primary emotions under that kind of generalization you can. Effectively the relationships that the AI forms through its interactions with other AIs or people could allow for creating, modifying or extending the word descriptions.