Background
- The Apollo-bot team are passionate about integrating break-through machine learning tools to lower the barrier of entry and increase the user experience of accessing NASA Earth data.
Achievements
- Apollo-bot is social personal assistant integrated into Facebook Messenger as a Chatbot to be universally accessible. Apollo-bot accepts requests 24/7 on-demand and automatically does the legwork of navigating, searching, and generating visualisations of NASA Earth datasets to improve the users' quality of life on Earth by saving time.
- Users can use Facebook Messenger to query the Apollo-bot for specific NASA Earth datasets captured over a timeframe
- Bot responds quickly to User query and automatically generates and returns a statistical data visualisation image representing the data
- User can set a Humour percentage to sustain user engagement. higher value increases the likelihood that any response is accompanied by a joke
Resources
- Common Metadata Repository (CMR) (replaces NASA's Earth Observing System (EOS) Clearing House (ECHO))
- Waffle.io Project Management to sustain ongoing development and contributions to the project
- GitHub to collaborate and share project code https://github.com/ltfschoen/Apollo-bot
Challenges Faced
- Creating Wit.ai story flows for User-Bot interface
- Developing Bot server and logic using the Flask framework and Wit.ai Python SDK, including retrieving data in CSV format using NASA ECHO API, building statistical data visualisations using Seaborn
- Conversion of Bot response from text to speech using eSpeak text-to-speech software synthesiser
- Integrating Bot server with Facebook Messenger API
- Generation of engaging jokes for Bot integration
Note: Wit.ai is a machine learning tool that **trains and learn with usage to improve confidence in its accuracy with its design using Wit (built with NLU algorithms upon proven Facebook technologies including Facebook's AI and NLP platforms: DeepText and FBLearner Flow) that integrates with Facebook's Messenger API and GUI.**