Browse Prior Art Database

An Automated Predictive Notification System Disclosure Number: IPCOM000015851D
Original Publication Date: 2002-Jul-21
Included in the Prior Art Database: 2003-Jun-21

Publishing Venue



An internet-scale notification system can provide eCommerce customers to be informed about the most updated information through publish/subscribe communication paradigm. The information a customer needs can be classified into different event categories. Events are generated by event producers. Events can be stored into event repositories and retrieved by either notification systems or other tools. Events can be delivered as a notification to event consumers who have subscribed to the corresponding event category. By using notification systems, we can achieve asynchronous, flexible, and loosely-coupling communication between event producers and consumers. However, there are several problems that need to be solved: As the Internet becomes the preferred media for business transactions, the number of business transactions is increasing every moment and internet is flooded with good and services. Buyers enjoy locating the products, service providers and information sources globally from their desktop computers conveniently in a few minutes without a need to travel long distances. Sellers, on the other hand, display their goods and services and are able to sell 24 hours a day everyday in their virtual shops without worrying about real estate and maintainance. Although, there is a lot of information flowing over the internet that can be used to make predictions for future trends, there exists no system or mechanism that would generate predictions for future trends and notify both the buyers and the sellers to prepare in advance for future events. Buyers would like to have the ability to predict the trends of the products and services that they are interested in investing. In other words, they would like to know what type of services or products will be available for purchasing within 3 months, 6 months, a year etc and . They could use this information to redirect their investments wisely. From the seller's point of view, it is very important to be able to predict market demands, customer preferences and portfolios. In addition to predicting what events will occur in a particular domain, the sellers demand the ability to predict new domains in the market. The ability to predict is especially valuable when a seller is aware of new category of market that need to be paid attention to so new goods and services can be customized to satisfy the needs of the predicted domain. The problems of predicting future events and future event categories are solved in the invention that presents a method and apparatus for the predictive notification system. Mainly, the system adopts data mining techniques to predict the behavior of the event generation and the evolution of event categories. Two major prediction techniques are used in the invention: (1) event instance prediction; and (2) event category generation. Taxonomically, systems or processes which produce events can be referred to as "event producers" and systems or processes which should be notified of the event can be called "event consumers". A group of events with similar characteristics can be grouped into an "event category". A basic structure of an event notification system is illustrated in Fig.1. The system 10 comprises the components manifesting the functions of the event notification service. The event producer 30 can publish event categories and/or generate events to the event notification system through flow 72. Similarly, the event consumer 20 can subscribe event categories and/or receive/consume events 1 transported from the event notification system 10 through flow 71. 10