Requirements In Scope For Quality

From Data on the Web Best Practices
Jump to: navigation, search

Which requirements are relevant for Quality and Granularity vocabulary?

Parent page: https://www.w3.org/2013/dwbp/wiki/Quality_Requirements_From_UCR

Source: http://www.w3.org/TR/2015/NOTE-dwbp-ucr-20150224/#requirements-1

Table format: id of requirement, assessment of relevance for quality scope (+, -, +/-), comment, Intrinsic/Extrinsic metadata


Req RA Comment I/E
R-AccessBulk (-) I don't think general access reqs are relevant for Q&G voc E
R-AccessLevel (+/-) depends on whether the Q&G metadata should indicate whether and how general metadata requirements have been implemented or not E
R-AccessRealTime, R-Access Up to date (+) info about how up-to-date the data is very important. R-AccessRealTime could be considered just to be plain (technical) metadata, not quality metadata. E
R-APIDocumented (+/-) depends on whether the Q&G metadata should indicate whether and how general metadata requirements have been implemented or not. E
R-DataEnrichment (+/-) info about whether the data has been enriched is very important. But this specific requirement seems to be about making enrichment possible. I or E??
R-FormatLocalize, R-FormatOpen (+/-) depends on whether the Q&G metadata should indicate whether and how general data requirements have been implemented or not. E
R-FormatMachineRead, R-FormatMultiple, R-FormatStandardized (+/-) depends on whether the Q&G metadata should indicate whether and how general data requirements have been implemented or not. Info about how structured the data (or the various formats) is can be very useful (it's one of the LOD stars). Finding indicators may be non trivial, though. E
R-UniqueIdentifier (+/-) depends on whether the Q&G metadata should indicate whether and how general data requirements have been implemented or not. I
R-DataIrreproducibility (+) seems a natural fit for Q&G: this is explicitly about flagging that the data can't be reproduced. So there's also provenance in it too. I
R-DataLifecyclePrivacy (+/-) see open issues from the BP analysis
R-DataLifecycleStage (+/-) it could give some hint about quality, if this is some sort of provenance related to R-DataProductionContext. This must be check against the life cycle digram that was once in the BP document E
R-VocabDocum, R-VocabOpen, R-VocabVersion, R-VocabReference (+/-) depends on whether the Q&G metadata should indicate whether and how general data requirements have been implemented or not.
R-SLAAvailable (+/-) depends on whether the Q&G metadata should indicate whether and how general metadata requirements have been implemented or not. E
R-LicenseAvailable, R-LicenseLiability (+/-) depends on whether the Q&G metadata should indicate whether and how general metadata requirements have been implemented or not. E
R-DataProductionContext (+) all provenance is probably may not be in scope for quality, but certainly the production context is. I
R-GeographicalContext (+/-) it's unclear whether this is related to granularity I
R-MetadataAvailable, R-MetadataDocum, R-MetadataMachineRead, R-MetadataStandardized (+/-) depends on whether the Q&G metadata should indicate whether and how general metadata requirements have been implemented or not. E
R-PersistentIdentification (+) the act of using persistent identifier is part of the data creating itself, though maintaining persistence is more an extrinsic aspect I
R-DataVersion (+/-) depends on whether the Q&G metadata should indicate whether and how general data requirements have been implemented or not. E
R-ProvAvailable (+/-) maybe it depends on whether it's about data production or other aspects, which would be more extrinsic. But it general it would seem in scope for Q&G
R-SensitivePrivacy, R-SensitiveSecurity (-) These dimensions seem very extrinsic to the data, i.e., depending on legislation, who is allowed to see which data E
R-DataMissingIncomplete, R-QualityCompleteness (+) what's the difference between the two? I
R-QualityComparable (+) but is it really different from the requirement for standardized? Standardized would imply comparable, to me. Different standards exists, which are not directly comparable, but in general the comparison can be made indirectly (using e.g., conversion rules) I
R-QualityMetrics (+) the requirement doesn't hint to specific metrics. Is it only about having a general framework for expressing any quality measure? I
R-QualityOpinions (+) E
R-GranularityLevels (+) The requirement doesn't say much about documenting the granularity, just about the availability of different granularity level. I
R-Citable (+/-) This seems to be about data publication, not the data itself. On the other hand such high-level indicator for quality of publication could be in scope. the problem is how to measure it. E
R-TrackDataUsage (-) E
R-IncorporateFeedback (-) if this is not R-QualityOpinions then it should be out-of-scope. But The relationship between the two reqs is not clear E