Differences between BIBFRAME and BFFI
This section of the Wiki aims to detail the various differences between BFFI and BIBFRAME.
Arguably the most notable difference to the original BIBFRAME data model is that BFFI adheres to the WEMI model employed by RDA content description rules. Accordingly, we have split bf:Work into two disjoint classes bffi:Work and bffi:Expression. This also required the division of subclasses and properties of bf:Work similarly. In our initial testing, the approach seems robust and the conversion back to regular BIBFRAME is simple.
Several of the bf:Work subclasses required a corresponding subclass for both work and expression level. However, some of the subclasses are only present in one of them: Dataset, MixedMaterial, Kit, Object, StillImage and Multimedia are subclasses of bffi:Expression while Integrating and Manuscript are subclasses of bffi:Work only. (See Image 1)
We have mainly used identifiers and English labels in accordance with BIBFRAME. However, in some cases RDA takes priority. An example of this is the class bffi:Manifestation which is set as equivalent with the BIBFRAME class bf:Instance. Another example is bf:genreForm whose label in BIBFRAME is "Genre/form" while we went with the RDA version of "category of work".
For the Finnish labels, we have mostly used labels from the RDA registry or from a corresponding field or subfield of the Finnish translation of the MARC format. We also plan to add Swedish labels at a later date.
Our implementation does not include some of the properties and classes of BIBFRAME due to various reasons.
BFFI follows the new relationship model of BIBFRAME so that the types of relationships (value vocabulary for bffi:Relationship) are taken from the Metadata Thesaurus. For this reason, the subproperties of bffi:relatedTo are omitted.
We omit the subclasses of bf:Identifier due to using of a controlled vocabulary from our Metadata Thesaurus. Likewise, we do not include the subclasses of bf:Title for the same reason. We also left out some properties and classes associated with MARC fields that are not used in Finnish cataloguing. For a full list of omitted classes and subclasses, see "List of BIBFRAME classes and prperties not used in the Finnish BFFI data model" (kts. lähde BFFI mallista poistettuja BIBFRAME ja BFCL entiteettejä )
We added several classes and properties to our data model. Almost all of these are directly based on RDA with a couple of additions that have to do with Finnish cataloguing conventions. These are listed separately here: List of classes and properties in the BFFI datamodel which are not in the BIBFRAME datamodel
BIBFRAME does not define domains and ranges for many properties opting to instead use "Suggested use - With Work or Instance" in the comment. Since we have a more strictly defined use case of catering to Finnish cataloguing, we decided to have explicit domain and range definitions for all properties.
BIBFRAME has six subclasses of bf:Title. Additionally, bf:variantType allows for more variants to be defined as needed. In order to comply more closely with RDA, we chose to modify the BIBFRAME approach for titles slightly.
In the BFFI model, we have only bffi:Title whose instances are located in four value vocabularies based on RDA properties for titles, e.g., "preferred title for work" or "variant title for manifestation". We also have the property bffi:titleNote for additional remarks that takes its values from MTS as a note on title, e.g., "cover title". Properties such as bffi:mainTitle, bffi:subtitle, bffi:partNumber and bffi:partName work the same as the corresponding properties in BIBFRAME.
For a more complete description on title in BFFI, see Titles in BFFI.
We have modelled the RDA aggregates as faithfully as possible to a BIBFRAME based model. An aggregate in RDA is a manifestation that embodies multiple expressions, e.g., a monograph with supporting content such as illustrations or an index, a monograph with multiple short stories by different authors, or a series of publications. An aggregating work is the plan to arrange several expressions of one or more works and embody them in a single manifestation and an aggregating expression is the realization of the aggregating work. This was a relatively simple addition requiring only two new classes, namely bffi:AggregatingWork and bffi:AggregatingExpression, as well as two new properties in the form of bffi:aggregates and bffi:aggregatedBy, linking the aggregating expression to the individual expressions and vice versa.
BIBFRAME uses bf:Hub for somewhat similar purposes. Due to some differences in implementation we decided to omit bf:Hub from BFFI.
Due to the cataloguing conventions in Finland, we left out bf:Ensemble, bf:Instrument, bf:Voice etc. and their related properties from the BFFI data model opting to use bffi:MusicMedium instead.
For a more complete description on music in BFFI, see (UNDER CONSTRUCTION).