These PUBSUB algorithms can be used to create databases and database managers that track consumers' changing interests and show them products that match their wants and needs. In publish/subscribe systems, a consumer's interest constitutes a "subscription" and the offer or sales advertisement constitutes a "publication." The subscription triggers a conditional query associated with an action, which can be short or long-lived. For example, if a consumer searches for flights to London that leave on a Friday in May or June and cost less than $2,000, the consumer's web browser will display new flights that match these criteria even when the consumer is not actively engaged in a search. Publish/subscribe algorithms are protocols for distributing information. Publishers sort the new content into classes and the users receives this new content that match their specific criteria based on subscriptions. Amazon, Orbitz, YouTube and other websites use publish/subscribe systems to provide their customers with relevant product information. Researchers at the University of Florida created PUBSUB algorithms that can be applied to a variety of data structures with different optimal settings. Their scalability enables them to handle large data sets. The technology could capture a significant portion of the $17 billion market for middleware software.
Algorithms that permit faster response times in publish/subscribe systems
Publish/subscribe systems are used in a wide range of Web operations because they offer greater flexibility than other data distribution processes. BE-Tree and Siena are two front-runners in the market. The PUBSUB algorithms developed by UF researchers outperformed both of these available programs in several ways. Unlike BE-Tree, the new algorithms are a heterogeneous system with a variety of data structures. The best solution is chosen based on data type and size. The new algorithms also enable better data clustering methods than BE-Tree. In tests, the new algorithms performed searches much faster than Siena.