A Database Interface for Mobile Computer

Current Trends

Computer-based personal information service is evolving beyond simple applications such as retrieval of phone numbers to include interaction with large, geographically distributed information bases. Concurrently, small, pen-based, mobile computers are becoming the machine of choice for personal computing. These two trends place conflicting demands on the design of database interfaces. The latter trend suggests simple interfaces that are easy-to-use, avoid keyboard use, and are suited for the small screens and small (relatively speaking) memory sizes of mobile machines. The former trend, however, suggests an increased sophistication in database interfaces, so as to provide access to the larger databases that are now part of a personal information service. We describe a pen-based graphical database language that begins to combine these conflicting demands for simplicity and sophistication. We compare this language with previous work on graphical user interfaces designed for workstations. A prototyping effort has recently begun in our lab, and we provide a summary of its status.

End-User Working Styles

As computer users begin to shift more of their work to pen-based mobile computers (mobilestations), the demands placed on such machines grow. While the typical user's initial expectations are for inter-personal communication (e.g., electronic mail), simple computations (e.g., spreadsheets), and basic record-keeping (e.g., calendars), very soon they try to carry out on their mobilestations the same tasks which they are accustomed to perform on their workstations. An important application for many computer users is accessing information stored in a database management system (DBMS).

Access Methods

There are two important ways to access such information. These are querying the DBMS directly via a data manipulation language such as SQL, and accessing the data indirectly by using an application program with embedded database commands. Both of these interfaces are supported by most of the commercial DBMS's of which we are aware. As applications are ported to mobilestations, users can continue to access DBMS information in the usual manner. However, the obvious approach for direct DBMS access which is using the handwriting recognition software of the mobilestations to translate pen strokes into, say, SQL commands have several problems.

First, the state of the art in handwriting recognition is such that most users will quickly decide that the effort required to query a DBMS in this manner is excessive. Secondly, the size of the typical mobilestation is sufficiently smaller than that of a normal workstation (48 square inches vs. 165 square inches for some of our local computers), that one must devise new strategies for displaying information. Lastly, many users are interested in accessing data having a complex schema, and thus require a more powerful interface than is possible by supporting solely SQL commands.

While better algorithms and more CPU power will improve current character recognition capabilities, even if perfect recognition was available one could argue that most users would rather not compose SQL queries but would instead prefer a more graphically-oriented interface. (A preference for these types of interfaces also hold for normal workstations, as argued by much of the work referenced in the next section). Even if one claims that mobile CPU's will get faster, their memory and disks larger, and communication more reliable, it is difficult to argue that portable displays will change radically in size. Thus, we feel that a useful DBMS interface for mobile environments must

Research Objectives

The main objective of this paper is to present our research on new interfaces for mobile computers that meet the conditions listed above. Clearly, much more than a well-designed interface is needed to support mobile database access. For example, the increased likelihood of disconnection for mobile computers forces us to renew research on database partitions, and the lesser computational capabilities of such machines lead to a reconsideration of query optimization techniques. However, we will not discuss these or other issues in this paper, and instead concentrate on interface issues. In the remainder of this paper we first present a short guide to the relevant database literature. We also provide a brief description of our interface design, and present some of the details of our current prototype implementation. We conclude with a few observations on our current work, and present our plans for future research in this area.

Copyright notice: the contents of this page were extracted from the original article written by the real author; the information here provided consists either an abstract or a summary of the original article.

Click here for original article

[Index Page] [Next]