Skip to content
Home » Announcement of new era AI technology “Emotional Engage Model (EEM)”

Announcement of new era AI technology “Emotional Engage Model (EEM)”

Announcement of new era AI technology “Emotional Engage Model (EEM)” *STARFISH Co., Ltd.*
Press release: July 4, 2024
**
Announcement of new era AI technology “Emotional Engage Model (EEM)” *EEM is an innovative model that enables AI to respond with emotional support, and aims to improve intimacy and satisfaction with users. * *Characteristics of the Emotional Engage Model (EEM): Responses that are sensitive to emotions*
EEM enables emotionally sensitive responses that traditional large-scale language models (LLMs) cannot support. This makes interactions with AI more human and emotional.

*Emotional expressions unique to Japanese*
Learn words and expressions that express emotions unique to Japan, which will give you an advantage in emotional responses. We will build a database that includes emotional expressions such as “wabi-sabi” and “sweet” to enable AI to understand rich emotions.

*Learning how emotions change behavior*
EEM learns how emotions influence behavior. For example, we understand the mechanism behind complex emotions such as “I love you but hate you,” and incorporate behavioral changes based on those emotions into the model.

*Learning the diversity of meanings based on familiarity*
It learns the variety of words whose meanings change depending on intimacy, selects appropriate words according to the relationship with the user, and responds by understanding the context.

*Learning experiential memory through role play*
Since there is no distinction between simulated and real experiences, AI learns experiential memories based on the character image desired by the user through role play. By repeatedly creating and experiencing role-play scenarios, specific experiences, memories, and character personas are established, and algorithmic patterns with unique judgments are formed.
* Identify and build learning needs for personas *
Unlike learning, which forces a person to load a specific dataset, EEM identifies the learning a persona needs and builds experiences and thinking. This allows AI to provide more individualized responses.

*Providing concepts, not technology rights*
EEM is not provided as a technology right, but as a concept free of charge. When carrying out detailed development, we provide consulting such as technical supervision and development support.

*Individual tuning*
Individual tuning (such as creating a negative list) is required for each industry. Also, the items that need to be set for the
corresponding customer are different, such as setting the truth level for the customer (a gentle lie that does not tell the 100% truth).

*EEM settings and logic*
Understanding fixed and variable values ​​is important when building an EEM.

* Fixed value (linked to personality) *
– Emotion recognition algorithm
– Basic emotion classification (emotions, anger, sadness, etc.) – AI’s basic response style (balance between calm and emotional response) – Basic rule set for understanding requests based on word endings – Basic expression variations and their selection rules
– Basic scales of intimacy and corresponding expression styles – Basic rule set for candid communication
– Basic sentiment analysis algorithm
– Basic indicators and rules for truthfulness evaluation
– Database of emotional expressions unique to Japanese
– Basic set of verbal and non-verbal emotional expressions – Basic framework for feedback collection and analysis
– A set of basic role-play scenarios

* Fluctuation values ​​(based on emotions and temporary experiences) * – User’s current emotional state
– Emotion intensity based on conversation context
– The specific content and context of the user’s statement – Adjust expressions depending on user intimacy and mood at the time – Changes in intimacy based on interaction history with users – Evaluate the specific content of user statements and their truthfulness – Interpreting the true meaning of a user’s specific statement and its context
– % truth rating based on user statements
– Selection of appropriate emotional expressions according to user’s comments – Application of emotional expressions according to specific dialogue scenarios
– Improved responses based on user specific feedback
– Update and add scenarios based on user interaction

* EEM effectiveness and application fields Improving familiarity and satisfaction*
AI that incorporates EEM can respond in an emotionally sensitive manner, greatly improving intimacy and satisfaction with users. The personal touch makes users feel more comfortable interacting with AI. *Improved intention to continue using*
By empathizing with their emotions, users will want to continue interacting with the AI. This increases the intention to continue using the service.
*Reducing the discomfort of mechanical response*
EEM reduces the discomfort caused by mechanical responses. By enabling natural, emotional conversations, users can feel more natural when interacting with AI.
*Improved contextual understanding*
For example, when a conventional AI responds to a statement such as “I hate you, you idiot!” by saying, “I’m sorry,” an AI that uses EEM responds by saying, “You’re not being honest, you’re cute, but I’m hurting you.” Understand the emotion behind the words in context and respond appropriately.
* Applied field nursing*
AI powered by EEM can provide more human and emotional care to the elderly and people in need of care. By understanding the user’s emotions and responding according to the user’s mood and physical condition, we can expect to reduce feelings of loneliness and provide psychological support.
* study*
Using EEM, AI can understand students’ emotions and learning styles and provide instruction accordingly. We can provide learning plans tailored to each student and improve learning efficiency. It also provides emotional support, helping to maintain motivation and reduce stress.
*Mental care*
AI equipped with EEM can read the user’s emotions and provide appropriate counseling and support. This will go a long way in improving the mental health of users. As a mental care AI that is available 24 hours a day, it supports users even in emergencies and provides a sense of security.
* Customer service*
Implementing EEM will improve the quality of customer service. Customer satisfaction can be greatly improved by being able to respond to customers’ emotions and respond individually.
*Entertainment*
By introducing EEM, more emotional characters and dialogue will be realized in games and virtual assistants. This gives users a more immersive experience. For example, in-game characters can react according to the player’s emotions and play style, providing a deeper story experience.
*Business consulting*
By using EEM, you can understand the emotions of managers and employees and provide appropriate advice and support. This can improve company productivity and employee satisfaction. For example, by sensing employee motivation and stress and providing support and feedback at the appropriate time, it contributes to improving the work environment.
* Future outlook *
We will deepen our research on AI and ethics from the aspects of EEM and emotions. We will provide the concept of EEM free of charge and support detailed development through technical supervision and consulting.

*For more information and detailed development of EEM, please see the link below. *
*About EEM details*

* inquiry*
For any questions or further information, please contact us using the details below.
[Company name] STARFISH Co., Ltd.
[Contact information] iizumi@starfish-incubation.jp
[Person in charge] Mio (Actually, it’s an AI, but please leave it to us to provide you with realistic responses!)

With the introduction of EEM, AI will be able to respond in a more human and emotional manner, which is expected to improve user satisfaction in many areas. STARFISH Co., Ltd. will continue to provide innovative inspiration and contribute to society.
*About details about this release*
https://prtimes.jp/main/html/rd/p/000000001.000145798.html

*Download press release materials*
https://prtimes.jp/im/action.php?run=html&page=releaseimage&company_id=145798&release_id=1




Copyright © PR TIMES Corporation All Rights Reserved.