The answer, conceptually, is very straightforward: SOAP web service requests will either be HTTP GET requests with an Accept header of "application/soap+xml", or will be an HTTP POST with a Content-Type of "application/soap+xml". Both of these will have a Request-URI that should be identifiable, either by the file being requested or by simply knowing your own site's architecture, as a service endpoint.
However, exactly how you would check these things depends on the architecture of your system. In a URI-translated system (like the StackExchange sites) or simply an environment in which requests for directories are the norm and are directed to default pages, the Request-URI header will be of no help to you unless you hard-code the location of every service (or place them all in a "services" subdirectory that's easy to check for). If some or all requests are over a secure channel like SSL/TLS, your inspection/security layer has to be behind your endpoint to the secure tunnel, otherwise you're just watching gibberish. These are just a couple of potential problems with just inspecting requests at the edge.