- From: Seaborne, Andy <andy.seaborne@hp.com>
- Date: Mon, 17 May 2004 16:12:53 +0100
- To: RDF Data Access Working Group <public-rdf-dawg@w3.org>
I would be happy to publish this version as our working document. Andy ::Comments:: Section 1: Intro Para2: "in industry and in open source projects" - some things are both. Would s/industry/commercial/ be better? Talks of standardization - doesn't W3C do "recommendations" so as to be clear they have no legal significance (c.f. safety standards). Para4: "less urgent objectives" I see the objectives as being things that are more difficult to determine whether we have met them or not. e.g. Human-friendly Syntax (well, app-writer friendly!) is very important. Section 2: Use Cases Very few of the UCs have "3.1 graph matching" as a requirement yet it probably applies to nearly all of them. Quite a few UCs have "Bookmarkable queries (3.8)" when I think it is just the fact the query can be written down in some syntax. The queries may be regualrly issued over SOAP/POST. 2.2: Finding Information about Motorcycle Parts (Supply Chain Management) Add motivates 3.4 subgraph results because the mounting bracket includes details of the screews needed. The query could not be written to anticipate every single possible extra piece of information that could be returned. I don't understand why "User-specifiable Serialization (4.4)" is motivated. 2.3: Finding Unknown Media Objects 2.4: Monitoring News Events. Both these are examples of bookmarkable queries being about recording the query. s/bookmarkable queries (3.8)/Human-friendly Syntax (4.1)/ 2.4: Add motivates "Aggregate Queries" (combine results). 2.5 Avoiding Traffic Jams Motivates: 3.3 Extensible Value testing I didn't see the connection to "Result Limits (3.10)" if we have "bandwidth-efficient protocol" Not sure this is a good example of the "Aggregate Queries" because different quesies go to different sources based what they are known to contain. This makes it different to 2.4 where the same vocabulary is present at each source. 2.6: Discovering What People Say about News Stories s/bookmarkable queries (3.8)/Human-friendly Syntax (4.1)/ 2.8: Sharing Vacation Photos Don't see the connection ot "Non-existent Triples (4.3)" I would not be worried if some techncial requirments are not directly connected to UC because the need for the requirment may not clearly be visible at the application/UI level. 2.9 Finding Input and Output Document for Test Cases Motivates: "3.1 RDF Graph Pattern Matching" and "3.2 Variable Binding Results", but not "User specifiable serialization" (I wasn't assuming the one line per result stuff was done by the query system). 3 Requirements I would s/mandatory/to be expected/ 3.9 Bandwidth-efficient Protocol This seems to be a difficult-to-quantify aspect and so is a design objective. I am not unhappy if it remains in the requriments. 4 Design Objectives 4.4 User-specifiable Serialization If this is saying RDF/XML vs application/N3 or whatever, it doesn't seem to be very strong. If this is saying that it could be XHTML as the serialization, then I think it gets outside of RDF data access - that is not to say that it isn't worthwhile but I wonder if the WG has the time to cover it. 4.5 Aggregate Query I suggest making this clearly about combining results: s/against/on each of the/ """ that is, the result of an aggregate query is the merge of the results of executing the query on each of the two or more graphs. """
Received on Monday, 17 May 2004 11:13:13 UTC