FRDCSA

Bad diet is a universal, severe and yet preventable health problem that leads to more preventable deaths than smoking, not to mention myriad other complications. Nutrition is a complex field and for most people, especially in today's fast paced world, the information is simply unavailable. A very comprehensive, intelligent and easy to use meal planner would definitely improve the situation. Another obstacle which must be combatted is the harmful myths surrounding meal planners that cause people to think they are not useful. Hopefully, with the release of the first truly user friendly and all-inclusive GPLed meal planner we can combat these problems.

Furthermore, chronic bad diet touches on, affects, and excites many other problems. For instance - depression, violent behaviour (due to low blood sugar) - really the list would be too long to mention, such is the depth of the issue. People tend to desire a certain level of pleasure in their life, and if their meals are not providing a lot of this, they will turn elsewhere. It's very sad.

There are many similar products, but gourmet is a superior meal planner in at least two ways. It is the only program in its class which is completely free and open source software, and its distribution is unlimited. Secondly, it benefits from advanced ai techniques such as Knowledge Based Systems and Collaborative Filtering.

The following features are planned to be supported.

• Ease of Use
  • A novel system is incorporated which estimates the amount of time that the user must spend to accomplish various tasks.
• Inventory Management
  • Cut spendings up to 4 times through better utilization and selection of groceries
    • Optimize via stochastic meal planning over price/nutrition/palatability
    • Automatically plans for food expiration in various storage conditions
  • SHOPS Inventory Management system
    • Real-time inventory management
    • Barcode/RFID scanning support
    • Generate shopping lists (Automatic Reordering)
• Nutritional Information
  • SR20 USDA nutrition database
  • Additional sophisticated axiomatizations of nutrional requirements for multiple diet profiles
  • Detailed nutrional records
  • Modular medical reasoning system for specific dieting, allergies, etc.
  • Preferences modules system
• Recipe Management
  • Large, open source, machine readable (RecipeML) recipe database with multiple levels of granularity
  • Base on www.recipesource.com
  • Ability to rate recipes and food preferences
  • Ratings-based decision strategy for recipe exploration versus exploitation tradeoff.
• Meal planning
  • Automatic learning of food preferences (machine learning/collaborative filtering).
  • Generate comprehensive meal plans with minimal user effort
  • Interactive planning, scheduling and execution of cooking process by semi-automatically translating steps portion of recipes into PDDL2.2 or similar temporal planning with resources domains

To do this we envision the integration of a P2P knowledge representation and trust management architecture, which when connected to by gourmet allows users to distribute empirically observed health information in this way. Lemmas and theorems are proposed by knowledge-sharing clients to the P2P network, and the knowledge-receiving clients must then estimate the truth value of the theorems. What is interesting is that trust is not monolithic or even wrt a given subject matter. Trust must boil down to a judgement by the receiver based on the truth value of the individual statement. Therefore audit-based justifications, a.k.a. proofs, must be given. The proof may be checked for validity and then the client might believe the information given, and a user mediated approach might also be interesting. Using "trusted" information, the system then proceeds to make dieting inferences, abductions, etc.