Page to be reviewed:
Developing interfaces between the World Wide Web and Statistical Applications
Author: Robert Kabacoff
This paper gives a good overview of the potential for interfaces between the World Wide Web and statistical programs and databases. The paper initially discusses the format of the Web and its abilities (e.g. requesting and displaying documents, incorporation of selectable links, fill-out forms, and image maps). These features are used to provide a platform on which extensions to statistical software and databases can be developed.
The paper goes on to examine the steps involved in creating Web based applications, which include setting up a server, creating the program code, and providing users with access. Different servers are available, and they vary with regard to their cost, speed, platform availability, formatting extensions and security options. The applications that are created for the Web consist of HTML (HyperText Markup Language) documents, CGI (Common Gateway Interface) scripts and optional multimedia resources.
The paper discusses two examples which facilitate the readers understanding of the topics presented in the text. The first example concerns the admissions records for mental health clinics and involves fill-out forms being used to request a report with specified characteristics so patients meeting criteria for research protocols can be identified. This example illustrated how the creation of a Web based user interface to an existing statistical software package occurs. The second example involved a SAS program that was written to process the usage log file of a popular Web site and update several cumulative SAS datasets on a daily basis. This is an advantage as the SAS datasets are much smaller than the cumulative access log would be. This example illustrated how SAS programs can be used to generate Web documents.
The paper presents some limitations (as of 1996) of developing interfaces between the Web and statistical applications. One is that the HTML specification has somewhat primitive formatting capabilities (making only some browsers able to support tables etc.), and this is a problem since statistical applications often return data in tabular form. Another limitation is that web browsers require in-line images in a certain format, and thus high resolution graphical output in many statistical packages cannot be saved. However, as this paper was published in 1996, these limitations do not apply, or at least are not so much of a problem today. Thus the World Wide Web is very useful in the dissemination of database information and when used with interfaces to existing software.