• 

  Psychological Safety Dashboard

• 

  In-depth Analysis of Conversation Style

• 

  Detection of Problematic Behaviour

Conversate: Psychological Safety Dashboard

Inspiration

The inspiration for Conversate came from observing the challenges faced in modern workplaces, where diverse communication styles often lead to misunderstandings and conflicts. With remote work becoming more prevalent, the need for tools that ensure psychological safety and promote healthy communication has never been greater.

What We Learned

Throughout the development of Conversate, we learned several key lessons:

• The Importance of Psychological Safety: We gained a deeper understanding of how crucial psychological safety is for team productivity and cohesion.
• Natural Language Processing (NLP): We expanded our knowledge of NLP techniques, especially in the context of sentiment and emotion analysis.
• User Privacy: We learned about the ethical considerations and technical challenges involved in handling sensitive communication data securely.
• Team Dynamics: We explored how different communication styles impact team interactions and the overall work environment.

How We Built Conversate

1. Conceptualization:

   • Identified the problem of diverse and potentially harmful communication styles in the workplace.
   • Defined the features needed to address these issues: sentiment analysis, emotion detection, and behavior identification.

2. Technology Stack:

   • Language: Python
   • NLP: Mistral AI API
   • Frontend: Streamlit
   • Deployment: Hugging Face Spaces

3. Analysis:

   • Implemented sentiment analysis to measure positivity, negativity, and neutrality in messages.
   • Used Mistral AI for emotion detection to identify key emotions like happiness, anger, and frustration and to detect harmful behaviors such as sexism, racism, toxicity, spam, and harassment.

4. Insights & Feedback:

   • Created personalized feedback reports for each team member, highlighting their communication strengths and areas for improvement.
   • Designed a dashboard to provide an overview of the team’s communication health, including detailed visualizations of sentiment and behavior trends.

Challenges We Faced

1. Data Noise:

   • Slack conversations contain a lot of irrelevant data (e.g., system messages, links). Cleaning this noise was a challenge.
   • Solution: Data cleaning scripts to filter out irrelevant information.

2. Privacy and Security:

   • Handling sensitive communication data raised privacy concerns.
   • Solution: Not storing any messages in the application.

Conclusion

Conversate is more than just a tool; it’s a step towards creating healthier, more supportive workplaces.

Built With

• hfspaces
• mistral
• streamlit