Our Reference Model

Donna Bergmark

December 2, 1999

Introduction

The terms citation and reference are used interchangeably in the literature, which causes confusion when one is actually trying to build tools for performing reference linking (also called citation linking). This short note spells out the model we will be using in the CDLRG. In this note, I'll be referring to documents, but you may think of them as objects (digital and otherwise).

References and Citations

Taking a document centric view, a document is initially created with 0 or more references in it, typically gathered together at the end and called "References". Over time, the document may also accrue citations from other documents.

We have decided to unify the concept of reference and citation into one concept, the cite-ref, which has two fields: from and to. A document then is an object that contains a list of 0 or more cite-refs that is non-decreasing in length over time. A document's references are those of its cite-refs whose "from" field refers to itself; a document's citation is a cite-ref whose "to" field is that document. Consider the following picture:

citation relationships

The cite-ref shown belongs to both documents A and B, but for A it is a "reference" and for B it is a "citation". It is also a relationship between the two documents, since it is a citation of B by A.

In projects that involve citation analysis, you often find the term "Related Documents". Quite often, this really means documents that co-reference a third document. Two or more items that have similar sets of references are also said to be bibliographically coupled[1].  The following figure shows some second-order relationships:

document relationships



Part of a document's metadata is its list of cite-refs, shown in angle brackets near each document in the above. Using Fedora as an intermediate implementation between the universe of documents and the various views one can have of these documents, we hope to manipulate cite-refs. For example, a Fedora object could be made to yield up its list of cite-refs, or just the list of its references, etc.

[1]  M.M.Kessler, "Bibliographic Coupling Between Scientific Papers", American Documentation, vol. 14, no. 1, Jan. 1963, pp. 10-25.


Donna Bergmark
1999-12-02