Portal:Complex Systems Digital Campus/Project-Team on Referencing Committee

Portal:Complex_Systems_Digital_Campus/Project-Team_on_Referencing Committee Some of the issues that need to be considered in setting up identifiers are: ‘functional granularity’ – the rules around assigning identifiers at different levels of granularity when things need to be separately identified; ‘co-reference’ – when different identifiers become assigned to the same ‘thing’ and ‘ambiguous identifiers’ when the same identifier gets assigned to different ‘things’. Real world identifier systems usually suffer from ‘co-reference’ and ‘ambiguous identifiers’ despite having seemingly incorruptible assignment procedures, so it’s best to start with some acceptance of this and design ways of handling it. Of course identifiers (and their referents) may suffer from all kinds of ‘rot’ such as the ‘semantic rot’ described so succinctly in the ARK documentation and of course ‘link rot’. Schemes such as DOI / handle and ARK have a special focus in providing tools to manage link rot of course, through the provision of tools for ‘social infrastructure’ around the identifier. Interestingly, some alternative views seem to exist, saying effectively that some degree of ‘rot’ is important in that it can decrease clutter to ‘let go’ irrelevant information. There’s probably some way of reconciling these views through optimisation, the well trodden path, the high value prize vs the tortuous path the the low value data. It should be very easy to construct identifier systems for articles, hardware, parties etc., or rather to adopt the appropriate existing identifier systems. What’s much more interesting is, creating some kinds of (perhaps semantic) links between different identified resources such as a digital microscope and the data it produces. This is interesting because an extensive infrastructure will be needed to capture data about the setup, the operator, the task, the wider team etc. Of course it’s vital to keep a clear separation between on the one hand the identifiers (and their referents) vs on the other hand the relationships that link the identified entities to each other. In fact there is a new ISO standard being created ‘International Standard Link Identifier’ or ISLI that might be useful in identifying these links, however it should be possible to construct a proprietary system using ARK, or so it would seem. It’s quite a hard problem to establish a clean framework of relationship types that might form the basis for links between things. Whenever I think about this I have the feeling of relational database analysis, which goes back to the ‘80s (or even earlier with things like Codd). Databases have moved on so far though. But formal identifier systems - assignment rules, granularity, relators, link identifiers etc – seem to be mainly designed in the manner of librarians designing cataloguing systems, while all the work on doing this ‘by induction’ etc in search engines is all quite secret within companies like Google.