Instead of predicting the business type, our prediction algorithm, P-DPA, predicts a set of stores that could be visited by a targeted client, and then, uses the predicted stores to generate a list of advertisements for the targeted client.
AdNext uses the business type of the last two locations visited by a targeted client to predict the business type of the next location for that client.
Consequently, if a targeted client has a plan to only visit one or two stores, then, AdNext will not be able to make any predictions.
Another group of researchers designed a framework called LASA, Location Aware Shopping Advertisement, that uses an ontology based formulation of clients and products profiles to generate a list of ads related to the selection history of a targeted client [5].
Kim et al. [3] developed a system called, AdNext, that uses Bayesian networks to build a transition matrix for the shopping mall clients in order to predict the business type of the next location that a targeted client could visit.
Consequently, failing to discover the correct location of a targeted client will hinder the process of determining the list of products' titles that should be sent to that client.
In addition, P-DPA algorithm can predict the first store to visit by a targeted client while AdNext will need to wait for the targeted client to visit at least two stores in order to be able to make any predictions.
Perhaps you want to expand your business, go after another client segment or launch a new product.
Instead, focus on finding and serving a targeted segment of customers, with a unique set of needs, and tailor your product and service experience specifically for them.
Thus, Mobil could use a fairly simple set of metrics (share of targeted customer segments and a summary score from the mystery shoppers) for its consumer objectives.
For example, the head of a large commercial organization wanted his people to become more focused on targeted customer segments as a way of improving revenue.
