# Simple two state example that runs through my proposal and thesis. # This is a thinly disguised version of the tiger problem, though not # as symetric. values: reward discount: 0.95 # There are two types of people that visit our web site states: executive teenager # We can either go for the hard sell of either ties or skates or # watch what type of interests they have by the pages they access. actions: monitor adv-ties adv-skates # Our web pages are either golf related items or video game related # items. observations: golf video # 7 out of 10 people that visit the web are teens. start: 0.3 0.7 ###################################################################### # State Transitions # Monitoring doesn't change who is at the site. We assume they will # stay there as long as we just monitor. T: monitor identity # Once we decide to advertise a product, it either sells or it # doesn't, and regardless we are done with this customer and we # get a different customer. We assume that 7 out of 10 visitors # to the web site are teenagers. T: adv-ties 0.3 0.7 0.3 0.7 T: adv-skates 0.3 0.7 0.3 0.7 ###################################################################### # Observation Probabilities # Assume 7 out of 10 executives look at golf product pages O: monitor : executive : golf 0.7 O: monitor : executive : video 0.3 # Assumes that only 1 out of 10 teenagers look at the golf web # pages O: monitor : teenager : golf 0.1 O: monitor : teenager : video 0.9 O: adv-ties uniform O: adv-skates uniform ###################################################################### # Immediate Rewards # Monitoring will cost us nothing but our daily operating costs which # we set to be 10 cents per time step R: monitor : * : * : * -0.10 # Assume that a tie have a profit of $10 and half of the executives, # when given the ad, will buy one. R: adv-ties : executive : * : * 1.0 # Assume that no teenagers will buy the ties. R: adv-ties : teenager : * : * -6.0 # Assume that skates have a profit of $20 and that 1 out of 50 # executives will buy them. R: adv-skates : executive : * : * -4.60 # Assume that 3 out of 10 teens will buy the skates R: adv-skates : teenager : * : * 2.0