Research and Development of Techniques for Data Indexing, Querying, and Browsing

¡@

Due to the rapid advance in communication and wireless network technologies, mobile (or nomadic) computing has become a hot research field. In this environment, the user can retrieve information by a portable computer at any time and any place. In this sub-project, we will investigate three sub-issues.

The first sub-issue is to explore the on-demand processing issues of a mobile information server, which includes the usage of the hot broadcasting data, the design of the on-demand query language, the transmission of query results, and the query processing with local cache. We explore the impact of access frequency on the indexing of broadcasting data, design the index structure of replicated data, and consider the ways of broadcasting under different data models. We will design a query language and reduce the repeated processing efforts by grouping techniques, including the optimal decision of the time to collect those similar queries. Next, we will transmit the query results by applying some techniques, which include broadcasting, multicasting, data-sharing and scheduling. Furthermore, the locating issue for data transmission will also be solved by data mining technique, which can derive the hot regions of user showup. We will explore the preprocessing approaches of unpredictable query processing to apply the caching techniques.

The second sub-issue is to consider resource management, query processing and optimization in a mobile environment, and design a query computing model. We will consider to manage resources using centralized or distributed methods, process location dependent queries, and combine traditional databases and location databases to process queries. Besides, we have design a Flow Control Language for query processing in a multidatabase system. We will design a query computing model by combining the concept of intelligent message and the technique of FCL.

The third sub-issue is to explore data mining, prefetching and caching techniques based on a WWW structure, and design a partitioned application model. We can derive hot sequences by data mining, which provides users some useful information. Furthermore, we will provide the language for resource discovery to let users query the hot sequence and the related information on their demands. Hot sequences can be used as the basis of browsing prefetching and cache decision. We also consider the sharing of prefetching and caching in a multi-users environment. Besides, how to migrate the storage and computation of user applications to powerful workstations, such as the information filtering service, is also an important research topic in a wireless environment.