2. INTRODUCTION
Who are we?
• Jari Ferguson - graduated Information Sciences student
• Kim Bosman - 2nd year Artificial Intelligence student
The ICT4D course at the VU:
• Master course Artificial Intelligence/Computer Science
• Combining theory with practical
• Use cases for Mali
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
3. USE CASE
Mali:
• Land area: 1.2 million km2 (36 times larger than The Netherlands)
• 14.5 million inhabitants
• 66% of gross domestic product from agriculture
• Around 40 different languages, French is the main language
• High percentage of illiteracy (66%)
• Limited to no Internet access
• Limited access to power-grid
• Lack of reliable and accessible weather forecasts, even though the global
climate changes are noticeable
• Telecommunications: 112,000 fixed lines vs 14,6 million mobile phones
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
4. USE CASE
Requirements:
• Autonomous system
• Allow farmers to get actual, region and town
specific weather forecasts
• Support multiple-day forecasts
• Provide spoken forecasts in plain language
• Support multiple languages
• Keypad operated menus
• Low power consumption
(solar power and batteries)
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
5. KASADAKA PLATFORM
• Small, cheap and low-powered hardware
• Web-like information sharing made possible for the
rural poor
• Multiple ways of accessing information
- Through voice-, SMS-based, visual or textual
interfaces
- Through the Web, using mesh networks or over
the GSM network
• Allows easy development of information services
relevant to local communities in their preferred
language
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
6. PROTOTYPE #1
• Call-in system
• Support for French and English
• Up to five days of weather forecasts
• Four forecasts per day (night, morning, afternoon and evening)
• Forecasts included: rainfall, temperature, wind direction and
wind speed
• Meteorological data obtained from OpenWeatherMap
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
7. KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
“In the night it will be warm
with light rain with a gentle
breeze from the southwest.”
“In the afternoon it will be hot
with no rain with a gentle
breeze from the west.”
8. Numerical data to verbal expressions
Numerical data from OpenWeatherMap.org is translated into verbal
expressions, for instance:
• Wind direction in degrees to “north, northeast, east etc.”
• Wind speed in m/s to Beaufort ”gentle breeze, strong breeze, gale
etc.”
• Temperature in degrees Fahrenheit to “cold, warm, hot, very hot
etc.”
• Rainfall in mm to “dry, moderate, heavy etc.”
Text-to-speech is not suitable: expensive and does not cover the many (40)
languages in Mali or other potential countries:
• Solution: create audio clips and compose sentences at runtime
10. FEEDBACK FROM MALI
• Reduce number of forecasts to two per day:
Daytime
Night time
• Support for more languages
• Translation of numerical temperature to verbal
expressions not consistent with local standards
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
11. PROTOTYPE #2
• Web based User Interface for administration and configuration
• Ability to add more languages
• Ability to add / modify countries, regions and places
• Two forecasts per day (night, day), instead of four
• Export usage data for analysis
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
12. Add, modify and remove supported languages
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
13. Add, modify and remove supported countries and assign languages
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
14. Assign regions to the countries
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
15. Assign places to countries and regions
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
16. Upload audio clips in order to support additional languages
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
17. Configure the system and download usage data
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON
19. QUESTIONS?
Thank you for your attention
We will provide a demonstration of the Kasadaka Weather Service later
KASADAKA WEATHER SERVICE – BOSMAN & FERGUSON