The growing body of experience of the many educational projects that use telecommunications, including those in the Testbed for Telecollaboration, points to the importance of automated services to support teachers and students in their work. The first automated service created by the Testbed is automated data sharing, the ability of many sites to collect data in an identical format, submit the data to a designated server, query the data, and download subsets of the data for analysis. Originally designed to work only with the Alice Network Software client and e-mail clients, more recently the automated data sharing has been configured to be readily accessible via Web clients.
This paper will describe the general design problem for data sharing in education and the solution that the Testbed project is proposing. The goal of the paper is to encourage discussion among groups in education facing similar issues, in order to agree on common standards and solutions. Specifically, it is our goal to see an approach established that will enable the interoperability of a variety of client and server applications. This priority reflects the situation in education of greatly varying kinds of hardware and software, and widely disparate levels of connectivity, from school to school.
Earlier work in the Testbed focused on the development of an easy-to-use client software (Alice Network Software); current work focuses on automated data sharing. This service will become widely available to the educational community during the next year. The general plan is to develop an automated data sharing service available in two ways: via email -- therefore available to the widest range of classrooms; and via http -- accessible to the small but increasing number of schools with IP connections. In this way, the service is designed to be available to the widest possible range of users, from low-end to high-end.
The problem we are exploring has two major elements. The first element are the problems related to submitting data. These problems include: who can submit data -- user authentication ensuring that data is submitted to the correct table permitting users to correct errors -- identification of data with class which submitted it specifying the format for submitted data
The second element are the problems related to retrieving data. These problems include: querying the data table transferring data and a template for the data to the client making maximum use of the client data types
These last two points recognize that the client software that will be used for analyzing the data will have very different data types. Most simply, a spreadsheet recognizes text and numbers, and numbers can be displayed in various formats (integers, percentages, decimals to a set number of places, date and/or time). However, specialized data analysis software such as mapping software allows for latitude and longitude, and the Alice Network Software allows for categorical data (e.g., urban/rural/suburban) in addition to latitude and longitude.
We are looking, therefore, for a general descriptive language that describes a data set, column by column, giving the key information: column title, column width, column type, and information needed for that particular column type (e.g., for decimal numbers, the number of decimal places to be displayed; for categorical information, the range of categories possible).