Project:Monasticon Hibernicum database

From CODECS Wikibase
Jump to navigation Jump to search

This project represents a first attempt at transforming the online Monasticon Hibernicum available from https://monasticon.celt.dias.ie into Linked Open Data using the possibilities of a Wikibase instance as hosted on Wikibase Cloud. This includes a number of extensions of the Wikibase family and a SPARQL endpoint.

Aims

  • The aim is not to build a convenient interface for querying the data, but to offer the data as LOD and the tools (via a SPARQL endpoint) with which third parties can build their own queries and query interfaces.
  • To help overcome the problem of data siloes: to serve as a bridge for data resources that currently live in isolation.

Wikibase structure

Strategies

Preserving the original CSV data

All of the original CSV data will be included unaltered as string values along with the Monasticon Hibernicum identifier as qualifiers of that resource. This makes it possible to easily check against information from the original resource.

Authority records

Linked Data, at least in the sense envisaged by this project, requires the presence of authority records we can link to (as opposed to a string value). For instance, in order to record that a particular site X is located in a particular parish, that parish must have a Wikibase item can we link to. Creating and maintaining those records are projects in their own rights seeing as their uses and benefits extend beyond the present scope.

  • Project:Mapping Ireland (name undecided) - the general project intended to create records for provinces, counties, baronies, parishes, counties and townlands. Townlands are so numerous (60000+) that it would be impractical to wait until every single one of them has a record.
  • Project:Saints - general project for creating entities about saints. Not yet started.

Confidence levels

It is inevitably in the nature of the sources that any statement should be treated as a scholarly attempt at the best possible answer, with some being more secure than others. MonHib makes a rudimentary distinction in that it adds a question mark when the identification is particularly insecure, while still being worth suggesting.

How do we translate this status to the database? After all, the process of reconciliation with the database does not make a distinction between identifications with and without a question mark. What we can do, however, to add a qualifier to each property-value relationship. A property has confidence level will be added, with at least two possible values:

  • "reasonable" in the absence of a question mark
  • "uncertain" in the presence of question mark


Multiple values/statements

There are multiple ways in which table cells can be multi-valued. Separators encountered include:

  • ";" - multiple statements, each of which can be qualified as "reasonable" or "uncertain"
  • "/" - multiple possibilities, sometimes accompanied by one or more question marks
  • "&" - only incidentally and it just means "and"
  • " " - a space, very rarely, probably because the separator was omitted
  • Wikibase supports arrays of multiple values and qualifiers are possible for each individual value.
  • There is built-in method to change the meaning of an array, such as 'one of these statement is (probably) true'.
  • For the purpose of a SPARQL query, it is important that each statement can be queried and retrieved.

Territorial and administrative

People

Sources

Retrieved from "https://codecs.wikibase.cloud/w/index.php?title=Project:Monasticon_Hibernicum_database&oldid=6358"