It depends a lot on what capabilities this shipyard brings. One of the primary advantages could be in tremendous volume savings - by having a dedicated spot to do the majority of the spacecraft assembly, one could send up raw (read: dense) materials to feed into the assemblers on this station/array. The array could then process these into the required components through refining and sub-assembly. This doesn't result in any weight savings, however. This could also be useful for processing any asteroid materials, but that presents another suite of problems and issues.
Another method, and one that is close to being tested, is sending up the raw material required for the largest structures on a spacecraft while sending up the assembled spacecraft bus. An assembly drone could then use the raw material to build the extra structure on top of the pre-built satellite bus (this being what contains the most complicated elements). This would be highly advantageous for building large orbital telescopes, reflectors, or antennae as there is no great way to pack large structures like this into launch fairings (look at the complexity involved in the JWST deployment for example). A few companies have received SBIR grants for concepts related to this (the company I work for being one of them - I'm too leery of saying what our tech is though).
So, in summary - the capabilities of the array matter. Assembly alone at the array would provide for some advantages. Assembly and some degree of fabrication extends these capabilities and the worth of such an array, and inclusion of refining allows for usage of materials acquired in space. Further capabilities in manufacturing electronics, silicates, organics, plastics, and so on would advance its worth to be tremendous (and I may have misidentified things here, my apologies).
Some interesting links:
https://www.nasa.gov/spacetech/niac/2013phaseII_hoyt.html#.V5pxeLgrKUk
https://ti.arc.nasa.gov/news/IRG-Dragonfly-robot-UI/ - currently underway, attempt to build a large antenna in orbit using robotic methods