World's First: Technology Developed to Estimate Team's Shared Recognition (SMM) in Real-Time from Chat Messages and Visualize Team State Changes
NTT Docomo and NAIST have jointly developed the world's first technology to estimate team's shared understanding (SMM) in real-time from business chat using AI. It contributes to productivity improvement by visualizing team states.
📋 Article Processing Timeline
- 📰 Published: April 10, 2026 at 20:00
- 🔍 Collected: April 11, 2026 at 00:26 (4h 26m after Published)
- 🤖 AI Analyzed: April 20, 2026 at 02:40 (218h 13m after Collected)
NTT Docomo, Inc. (hereinafter, "Docomo") and the National University Corporation Nara Institute of Science and Technology (NAIST) have jointly developed the world's first technology (hereinafter, "this technology") that uses AI to estimate Shared Mental Model (SMM), which is important for teams to achieve high performance for business growth through efficient work styles, in real-time from messages on daily business chats such as Slack and Microsoft Teams. A paper on the research and development of this technology has also been accepted by the prestigious international conference "ACM CHI conference on Human Factors in Computing Systems (hereinafter, "CHI 2026")" in the field of Human Computer Interaction.
In today's era where business operations in each company are becoming more sophisticated and complex, it is considered important for teams to continuously grasp and maintain a high level of SMM score – an indicator of how much common understanding team members have about "work methods," "role division," and "each other's strengths and weaknesses" – in order to perform efficiently. On the other hand, the main methods for grasping SMM scores are assessments by experts and questionnaire surveys. These methods require a certain amount of time and effort for preparation, implementation, and analysis, and tend to have a time lag before results are obtained, making it difficult to continuously grasp the day-to-day changing team state.
This technology, jointly developed by both parties, analyzes messages from chat tools such as Slack and Microsoft Teams used in daily business operations by Docomo's proprietary Graph Neural Network (hereinafter, "SMM Estimation Engine") to estimate the team's SMM score in real-time and visualize changes in the team's state. Messages from chat tools are classified into 11 categories such as "information sharing," "question," and "gratitude" by multiple AIs (LLMs), and then the SMM Estimation Engine analyzes the directionality of how specific categories of messages are transmitted within the team, from whom to whom, to automatically calculate the team's SMM score. Since this technology does not require survey questionnaires like traditional methods, it reduces the burden of costs, time, and effort for team members, enabling real-time continuous monitoring. Furthermore, as it does not involve human intervention like expert assessments, it analyzes the content of team members' statements without individual review or evaluation by others, and its high confidentiality protects privacy.
This technology can be utilized by team managers and leaders to grasp a decline in their team's SMM score in real-time, allowing them to implement communication measures such as sharing prerequisite knowledge or reconfirming objectives at an appropriate time to improve the SMM score. Additionally, when new members join a team, it can also be used to understand whether the SMM score is trending upwards over time, indicating successful onboarding.
For the development and validation of this technology, an actual demonstration experiment using business chat data was conducted within Docomo. It was confirmed that high-precision SMM score estimation is possible by utilizing this technology, which analyzes the content, category, and directionality of messages together. Moving forward, through demonstration experiments in other companies and organizations besides Docomo,
In today's era where business operations in each company are becoming more sophisticated and complex, it is considered important for teams to continuously grasp and maintain a high level of SMM score – an indicator of how much common understanding team members have about "work methods," "role division," and "each other's strengths and weaknesses" – in order to perform efficiently. On the other hand, the main methods for grasping SMM scores are assessments by experts and questionnaire surveys. These methods require a certain amount of time and effort for preparation, implementation, and analysis, and tend to have a time lag before results are obtained, making it difficult to continuously grasp the day-to-day changing team state.
This technology, jointly developed by both parties, analyzes messages from chat tools such as Slack and Microsoft Teams used in daily business operations by Docomo's proprietary Graph Neural Network (hereinafter, "SMM Estimation Engine") to estimate the team's SMM score in real-time and visualize changes in the team's state. Messages from chat tools are classified into 11 categories such as "information sharing," "question," and "gratitude" by multiple AIs (LLMs), and then the SMM Estimation Engine analyzes the directionality of how specific categories of messages are transmitted within the team, from whom to whom, to automatically calculate the team's SMM score. Since this technology does not require survey questionnaires like traditional methods, it reduces the burden of costs, time, and effort for team members, enabling real-time continuous monitoring. Furthermore, as it does not involve human intervention like expert assessments, it analyzes the content of team members' statements without individual review or evaluation by others, and its high confidentiality protects privacy.
This technology can be utilized by team managers and leaders to grasp a decline in their team's SMM score in real-time, allowing them to implement communication measures such as sharing prerequisite knowledge or reconfirming objectives at an appropriate time to improve the SMM score. Additionally, when new members join a team, it can also be used to understand whether the SMM score is trending upwards over time, indicating successful onboarding.
For the development and validation of this technology, an actual demonstration experiment using business chat data was conducted within Docomo. It was confirmed that high-precision SMM score estimation is possible by utilizing this technology, which analyzes the content, category, and directionality of messages together. Moving forward, through demonstration experiments in other companies and organizations besides Docomo,