Kitchen Appliances

Commonsense Inference Engine

KitchenSense is a kitchen-related daily knowledge subset of
OpenMind: a verbal database containing over 700,000 utterances
about everyday life. The OpenMind database contains sentences
describing common sense facts. Kitchensense uses concepts about
kitchen activities and dangers as well as hints and recipes. These
facts are compared to statements created for KitchenSense based
on sensor data. KitchenSense connects concepts based on their
probability of being related based on how similar their verbal
descriptions are. For example, the Augmented Refrigerator
recognizes the proximity sensor as I walk to the fridge and the
micro-switches are annotated as I open the refrigerator or I
open the freezer in order to trigger related events.
KitchenSense seeks to infer what will be useful or productive to
the user through a Goal-Oriented Interface, so that when a
person opens the freezer and goes to the microwave,
KitchenSense decides he probably wants to defrost food. The
system constantly infers causal and temporal relationship of
events by using spreading activation algorithm. Events
are decomposed into a directed graph of action nodes. Each action
node is linked to one another based on the probability of the
action that might occur next. Spreading activation of semantic
association is used for injecting energy to certain nodes and
checking if some nodes are activated under such situations. The
ConceptNet database tell us that people get hurt by hot water
and the kitchen is a dangerous place for children. KitchenSense
associates hot water and child with danger. When a child
approaches a pot of boiling water, the Augmented Reality Kitchen
flickers the lights and play an audiovisual animation.
KitchenSense offers a safety monitor, an augmented recipe and
persuasive reminders.