A - What does your customer's emotional expression mean in the context of your business?
Most affective computing systems today include (at least) six dimensions -- anger, sadness, fear, disgust, happiness and surprise. Because people can mix multiple expressions together, that would mean there is a combinatorial explosion of possible expressive mixtures (729, in fact).
But what do these more than 700 expressions actually mean? Building software to recognize a customer's emotional expression is a complicated challenge. But the power in doing so comes from understanding what the expressions actually mean, and thereby knowing how the software should respond.
EMOTION CHIP technology addresses this issue, giving rigorous meanings to emotional expressions customizable for your business.
B - What emotional expression does the customer think your software is expressing?
Whether or not your interactive software is designed to express emotions, it is often treated by the user as if it is an expressive agent.
It would be nice to know what emotional expression your software appears to be expressing. Often you won't know, because you didn't design your software to express emotions at all. And even if you did, it is important to understand what expression the user takes your software to be expressing, which might well differ from your intent.
EMOTION CHIP technology solves this problem: emotional expressions have rich semantic meaning with multiple components, and one concerns the user’s interpretation of what he believes the other agent (in this case your software) expressed he wants.
Emotion recognition software with the EMOTION CHIP can thereby not only read the user’s expression, but can infer what emotional expression the user believes was expressed by the interactive software agent. (And this includes the nine semantic facets described in (A).)
The software can then use this information to modulate the software's behavior (maybe an advertisement) on the next interaction. But without this information, the software would not know what its own software even expressed in the opinion of the user.
C - What class of algorithms underlies emotional artificial intelligence?
True intelligence is not about solving math and puzzles. What people really mean by intelligence is… social intelligence. Dope, fool, idiot, imbecile, moron, simpleton, etc. These imply social incompetency. An "idiot-savante" has genius level skills in some area, but lacks social competency. Intelligence is our capacity to be polite, to negotiate, to strategize, to raise the stakes, to push back, to compromise. With true social intelligence, humans are able to cooperate, compete, and even lead.
We all have to know our boundaries; how far we can "push" each individual in each kind of interaction. Learning these boundaries is what gives us an awareness of others. In the future, AI's fate -- like ours -- will be to navigate an ever-buffeting sea of crafty beings, and will accordingly need the other-awareness utterly missing in AI today.
It's not enough to have shallow face-recognition attribution and a veneer of human facial expression. AI must fully grasp the language of emotion: understand what each emotion means (see (A) and (B)), reason about what reply to make, and know which emotion signals that reply.
EMOTION CHIP technology allows the construction of AI algorithms with emotional intelligence, and with customizable personalities that vary along socio-emotionally sensible dimensions.
Such algorithms enable AI not only to comprehend emotional expressions, but also to negotiate, grasp etiquette, possess an awareness of status, and understand the manner in which signaling emotions commits one to bets of reputation (e.g., a trash talker is at greater risk than a polite person).
D - How to summarize an emotional expression in a low dimensional space, e.g., for extracting emotions from text?
To extract emotional expressions from text -- such as from customer support complaints -- traditional machine learning would require training on at least six dimensions, one for each of the six independent expressions: anger, sadness, fear, disgust, happiness and surprise.
That leads to two big difficulties.
First, because people mix multiple expressions together, that means there is a combinatorial explosion of possible expressive mixtures (729, in fact). It is therefore extremely cumbersome to manually tag text with the right emotion mixture, and difficult for machine learning to handle it even if manual tagging were possible.
Second, even if the many expressive mixtures can be recognized from text, with more than 700 qualitatively distinct expressions, what do they all mean in customer support terms? (See (A) above.)
EMOTION CHIP technology accommodates emotional expressions in a much lower dimensional space than traditional theories, requiring only two dimensions for tagging and learning.
But it is nevertheless theoretically rich, with 81 qualitatively distinct emotional expressions, each recognizable, with a rigorous quantitative meaning, and with the appropriate machinery underlying it.