Personalized information delivery

Peter W. Foltz, Susan T. Dumais
1992 Posters and short talks of the 1992 SIGCHI conference on Human factors in computing systems - CHI '92  
Preface While most of our day-by-day interactions with information systems are of pull-based nature, the availability of mobile information appliances like smart phones, PDAs, or even laptop computers demands permanent and personalized information supply without explicitly querying the information sources. This shift from pull-to push-based information delivery hides a tremendous potential as well as the risk of information flooding. Thus the need of systems providing a push-based and
more » ... ed information service is obvious. In general we may distinguish four classes of personalized information services. The lowest class applies to broadcasting, where information is simply distributed to every consumer device. Either the end user or an appropriate application on his/her side is responsible for filtering and further processing. The next class of data dissemination provides a channel-based addressing scheme, where users register explicitly for certain channels. Every information fed to the selected channels by any information producer is forwarded to the end user without further filtering. Sample application scenarios may be seen in news-or stock-tickers. The third class of personalized information services may be considered the content-based filtering scheme provided by data notification services. In this context, the user may specify a predicate addressing the relevant information with regard to a certain channel, thus providing extensive information filtering. All these concepts covered by broadcast, dissemination or notification preserve the original messages produced by the publisher. Any further processing like combining information from multiple information sources or aggregating numerical or textual information is not supported. Therefore the class of subscription systems allows users to register complex queries in combination with specific delivery criteria. Once messages are produced by publishing components, they are integrated into a publication-consistent global database and propagated to the appropriate subscribers. The application context, the resulting requirements, the existing technologies, and advanced strategies from a modeling as well as implementional perspective are the focus of this report. The collection of papers is subdivided into four parts. The first two articles give a general overview and a comprehensive introduction into the publish/subscribe area. While the first paper emphasizes the interface between publish/subscribe systems and classical database Preface 4 concepts, the second paper provides a general overview of the subscription system PubScribe to build an open information market where information may be freely published and users may specify their personalized delivery strategies. The second part of this report contains a comprehensive discussion of personalized information delivery in the context of data warehousing and eBusinnes application areas. The first paper in this part, 'Revealing Real Problems in Real Data Warehouse Applications', pinpoints several requirements with regard to a successful data warehouse installation. The following contribution extends the classical demand-driven data warehouse architecture by an additional subscription component providing flexible data integration as well as push-based data delivery. The use of subscription technology in the context of contract negotiation in electronic business environments is discussed in the following article. Subscription technology is used during an informational match making phase between the selling and buying side. The third part includes a collection of papers addressing technological issues from several directions. The first contribution discusses maintenance strategies to synchronize materialized views with changes to the base tables implemented in IBM DB2/UDB. Efficient synchronization techniques may be considered the core of an efficient propagation mechanism of incoming messages to the corresponding subscribers. While the second paper addresses the producer of a subscription system by reviewing web site scraping technologies and proposes a new iterative mechanism called XWeb, the third article in this part gives an example of possible target systems of subscription delivery in the context of footprint database systems running on handheld devices. The last part finally provides an in-depth explanation of the research prototype PubScribe. In general the PubScribe system aims at implementing a highly scalable and open subscription system based on relational database technology. The first contribution introduces the general architecture at a microscopic and at a macroscopic level. The second paper emphasizes the different modeling perspectives from a structural and dynamic perspective. The last article closes the description by sketching implementational issues like mass query optimization and mapping schemes to relational structures.
doi:10.1145/1125021.1125024 dblp:conf/chi/FoltzD92 fatcat:ddxdxdx35zazhcwut6tgnb53ru