Most commercial web services are generally quite simple things. Here at Scientio we are trying to move three very complex applications onto an entirely web service footing. This means our web services have to be quite sophisticated

 

In order to use XmlMiner as a web service there are various development tools that are needed. It's true that everything used to control XMLMiner is XML based, so one should be able to develop a solution with just an editor. In practice there are few people other than Scientio staff liable to take on such a job. 

 

So we need online development tools. We were considering Ajax, but happily Silverlight 1.1 and Netikatech came along, and the quality of the user interface is much better and much more consistent than an AJax solution. Also we've been able to write everything in C#.

 

We've created a series of silverlight applications that each live on a single web page, as part of the web service interface.  They all use the same keys and IDs as the web services, so you can for instance edit a rule set via a web page, and use the same rule set for inference via the web service.  The web pages are formatted so that they fit easily into frames within host web pages, including SharePoint.

 

Talking of SharePoint, we've created a custom frame webpart that will make connections with other web parts, and transmit those values to the application within the frame.

 

That's how we've got the linkage between the rule set selection webpart and the rule editor working here.

 

So far we have a rule set editor and a processing map editor working on Silverlight, so we have the minimum set to create rule based applications. In the pipeline we've got a Lacuna interface controller, A data mining spec generator and a questionnaire page.

 

We'll also be adding workflow capabilities to help you through the process of going from mineable data to usable web service.

 

Happy new Year,

 

Andy