how to hide something that i don't want to look in Xtext

I am using Sirius and Xtext to do a graphic-text two-way synchronization work, probably after I use Sirius to draw a graphic to generate Xtext text, and modifying the text will modify the graphic synchronously. When I draw with Sirius, the model will store a lot of necessary information, as shown below [enter image description here][1] I will get the XML likes this :

<siriusModel:ElementFactory>
  <Elements> type="A" deviveName="A" logicalEntity="AbstractA"
    <Elements> type="B" deviceName="B" logicalEntity="AbstractB"
    </Elements>
  </Elements>
</siriusModel:ElementFactory>

My xtext grammer is like this:

ElementFactory returns ElementFactory:
{ElementFactory}
'ElementFactory'
'{'
  Elements+=Element*
'}';

Element returns Element:
{Element}
'{'
'[''type''='type=STRING']'
'[''deviceName''='deviceName=STRING']'
'[''logicalEntity''='logicalEntity=STRING']'
'}';

And my DSL looks like this:

ElementFactory{
  Element{
    [type="A"]
    [deviceName="A"]
    [logicalEntity="AbstractA"]
  }
  Element{
    [type="B"]
    [deviceName="B"]
    [logicalEntity="AbstractB"]
  }
}

but i don't want to show the [deviceName="A"] and [logicalEntity="AbstractA"] , I only want to this:

ElementFactory{
  Element{
    [type="A"]
  }
  Element{
    [type="B"]
  }
}

is there any way i can do it？ Any help i would appreciate! thanks! [1]: https://i.sstatic.net/qh5Qy.png

Your problem is that you want your textual concrete syntax (Xtext grammar) to not represent your whole semantic model. As far as I know this is not really possible for Xtext: in general the whole file is the whole model. In Sirius you can more easily choose to not represent some of the semantic informations and its OK.

I would advise you to devise a "real, complete" abstract syntax for your language, then also create an Xtext-facing abstract syntax to use for the textual parts, and synchronize between the "real" model and the "made for text" one. I believe Eclipse OCL works like this as well with several metamodels with transformations in-between in order to both accomodate the Xtext technology and the OCL standard/API it is supposed to conform to.