This specification is a very useful document that potentially can be used for a wide range of application domains. It 
can be considered as a first step toward a full information dissemination and exchange model. I see the following strong
 sides of this document:
o	It is based on the publish / subscribe and subscribe to publish paradigms that already proved very useful in multiple 
domains. It can be considered as a further improvement of these paradigms.
o	It explicitly includes the use of vocabularies and vocabulary management infrastructure in a very useful way. While 
vocabularies are implicitly included in all information related models, very few, if at all, include vocabularies 
explicitly in a useful way.
o	It explicitly includes mutual filtering as a symmetric technique for the matching a discrimination of publishers and 
subscribers. While matching and discrimination are well known (expressions of interest, access control list, etc.) 
symmetric approach to filtering, or mutual filtering, is not well known.

With all these strengths in mind, I would like to make a few suggestions for further improvements:
o	The use of entity as information about an object is rather unusual and possibly confusing. In the case when objects 
are information objects (documents) information (data) about documents (data) is usually called metadata. The use of 
entity as equivalent to metadata is rather confusing.
o	There are places in the specification where entity is used instead of object and places where resource is used instead
 of object. I think that a clarification of how these three terms (entity, object, and resource) are used would help to 
understand the model described. I actually think you do not need all three
o	The role of subscribers is not clear and might lead to confusion. They have no explicit interface and produce no 
visible actions. The registry that acts as a subscriber and matchmaker handles subscriptions. I recommend removing 
subscriber from the specification until some further versions where the subscriber will have a face (interface).
o	The use of verbs Create, Register, Drop, and Unregister seems to need more consistency. For example, CreatePubliser 
actually registers a publisher, and UnregisterVocabulary actually unregisters the data vocabulary (I guess?)
o	EPR needs to be in the glossary and possibly explicitly defined in the first place it is used. This abbreviation is 
used not only the end point reference.