Types of Data
The Finding the Celtic website holds only metadata: metadata is data which describes the primary data resources held on other websites, which are usually called "content providers." FtC is thus like a sophisticated index that contains information about resources that exist elsewhere on the internet and pointers to those resource.
Primary digital assets are the digital representations of the objects in the collections: scanned images representing artifacts, architecture and manuscripts, textual transcriptions representing the texts of sources, webpages relating information about an artifact, and so on.
Each entry in the FtC database can (but does not necessarily need to) have a corresponding webpage, plain-text source, full-size image and thumbnail image. An entry without any of these is essentially just a bibliographic entry, as it will require that the reader find the item in a library or other collection. The entry becomes more useful if it can be used online (i.e., if it has its own webpage, plain-text source, etc). Older websites that use frames thwart the ability to locate a single item with a specific URL.Metadata is information about those primary digital assets which is meant to help the user to locate, browse through, analyze and visualize those primary digital assets. The FtC metadata schema is a modification and extension of the RDF (Resource Description Framework) used by NINES.
All contributors must adhere to the metadata guidelines as carefully as possible. The mechanisms in FtC that allow users to search, browse and visualize will not be able to compensate for poor metadata entries. As the old computer adage goes, "Garbage in, garbage out."
You and your organization can keep your primary digital assets on your own servers: all that FtC needs is the metadata that describes those assets and points to their locations on your servers with a unique and stable URL (Universal Resource Locator). This scheme of storing pointers on one server to resources on others is called a federated approach to data management.
In other words, (apart from the RDF files) your data will continue to reside on your own servers; the only item to deliver to us is the RDF file. See the Files to Submit section at the bottom of this page.
Corporate Identity
If your organization is contributing RDF files, you will be assigned an Archive code value that must appear in the <ftc:archive> field of each file you submit.
You should also supply an icon to represent your institution's archive (see the Collex browser for some current examples). The icon will appear for all items from your archive which don't have their own specific thumbnail image. It can be either in JPEG or GIF format and be a maximum of 100 x 100 pixels in size (at 72 dpi, and compressed as efficiently as possible).
Granularity of Objects
As a contributor, you are free to define and subdivide items as you wish. For example, you might enter all of the articles of a conference proceedings as a single item corresponding to the published conference proceedings; it would be much more useful, however, to create a separate file for each individual article.
The question of how items are defined as independent entities is often referred to as one of granularity. Please adhere to the following principles when deciding the level of granularity of your contributions:
- Define an item at the largest level of granularity that allows a single set of metadata elements with values to describe it.
- Consider the interests of your users/readers: what unit will they want to collect and annotate? (e.g., books vs. chapters, etc)
- Break down objects that have parts with special relationships between them (see below)
For example, if a poetry collection has poems by different authors, it might be preferable to create each poem as an individual object since a single role:AUT element entry could not describe the multiple authors.
Relationships between Objects
The Parent and Children metadata tags in the schema allow you to create a set of relationships between objects. This hierarchical scheme was originally designed to express the component relationships between books and chapters, journals and articles, and so on. However, you can extend the hierarchy to express the following kinds of relationships:
- Archeological sites and artifacts: If there are a number of artifacts from or buildings at an important archaeological site, you might want to define the Site as the parent object and all of the artifacts/buildings objects as its Children.
- Manuscripts and texts: A Manuscript should be defined as an object in itself; you may wish to create Text items for all of the edited texts that derive from the manuscript. If they are being defined together, you may want to link these all together in parent-child relationships.
You should also ensure that the names used as element values are consistent with the Title elements of any object in the datastore. For example, if a role:AUT element has "Seumas MacGriogair" as its value, an item defining this author should have "Seumas MacGriogair" in the Title element (rather than "James MacGregor").
Range Overspill
There are cases when the value for an item's element exceeds a prescribed definition or spans more than one. For example, the Period element has a value Early Medieval, and by definition spans the years 400-950 CE. You might need, however, to describe a Manuscript that was inscribed over a long period, such as 850-1100 CE, or a Site that was occupied between 700 and 1700 CE, thus spilling over a single value range. The Period element can only take a single value, however.
There are at least a couple of ways to address such situations:
- Decide the value that represents the most characteristic value for the object. For example, in the case of a Site, choose the most characteristic or defining period of the object. You can still put the entire range in the Date element
- Create multiple objects for the different stages or forms of the object.
Personal Names
The difficulty with using the modern common format Surname, Forename is that it pertains only to the Modern period and still is not indicative of naming conventions in many areas (such as Gaelic-speaking Scotland). Because of these limitations and inconsistencies, it is recommended that personal names start with the forename and follow their own cultural conventions as best as possible. For example:
- Modern English and Lowland Scots: Forename Surname (such as Matthew Arnold)
- Modern Gaelic-speaking Scotland and Ireland: Forename patronymic Surname (such as Donnchadh mac Iain Caimbeul) or Forename attributes Surname (such as Donnchadh Bàn nan Òran Mac an t-Saoir)
It is also recommended that a name be given in the same language as the item when possible.
Best Practice Guidelines
The following best practices guidelines should be followed for these item types.
Multiple titles (and languages)
It is suggested that the most common name of an object be used for the Title element and that other forms (such as other languages) be given in the the AlternativeTitle element. This principle can be applied to manuscripts, texts, artifacts, sites (i.e., placenames), and so on.
In the case of Manuscripts, the library and catalog index (as well as alternative titles) should be given in the AlternativeTitle element. For example, the Book of the Dean of Lismore has these values:
<dc:title>The Book of the Dean of Lismore</dc:title>
<dcterms:alternative>Leabhar Deathan Lios Mòir</dcterms:alternative>
<dcterms:alternative>NLS Adv. 72.1.37</dcterms:alternative>
Patrons of the arts
The role:Publisher element can be semantically extended (especially before the Modern period) to refer to patrons of the arts or other works. For example, the Book of the Dean of Lismore appears to have been commissioned by the Campbells of Glenorchy, so the following element would be appropriate:
<role:PUB>Campbells Of Glenorchy</role:PUB>
Likewise, artifacts or buildings that were created under the patronage of specific kindreds, kings, churches, etc, should have that information indicated with a <role:PUB> entry.
Texts: Original texts, translations, articles, etc
The Text ItemType covers a broad range of textual items: original texts (often in edited form), translations of original texts, articles about Celtic Studies, and so forth. Each of these kinds of texts should be considered an artifact in its own right. For example, the metadata for a modern English translation of an ancient Gaelic text should describe the translation (the translator, the year it was done, etc), rather than referring to the original text.
The Country element refers to the subject of the item. So, for example, even if an article was produced by an American scholar in the US, if the article is about medieval Ireland, its Country element should be IE, not US.
The Relations element can be used in all versions of a text (original and translations) to maintain pointers to one another.
Objects that are born digital (such as ItemTypes Definition and Event) should have their material value set to Digital. Otherwise, entries that describe texts such as articles and translations should describe the text in its original physical form (Paper for books and journals) even if the text has been digitized since its original appearance.
It is recommended that you supply a <ftc:text> element containing a summary of the item if it is not available online. The summary text can then be indexed by the search engine and located in phrase searches even if the item is not otherwise in digital form.
Files to Submit
Each submission entry will consist of an RDF file. The RDF file must have the extension .rdf and be encoded as a utf-8 text file. Do not submit until the file's syntax has been validated by the online RDF validator. This validator only checks the syntax of the file; it does not know whether or not your data is valid.
You should store any thumbnail images for your objects on your own server as a JPEG or GIF file. The image cannot be larger than 100 pixels on either axis at 72 dpi. It is recommended that thumbnails be as close to 100 pixels in each dimension as possible, even if much of that space is empty white space. The URL to your item must be specified in your RDF file.