The Open Citation Project - Reference Linking and Citation Analysis for Open Archives

Enhancing OAI Metadata for Eprint Services: two proposals

The Open Citation project is funded by the Joint NSF - JISC International Digital Libraries Research Programme

Version history of this paper

• This version final v. 1.0 (17 August 2001), accepted for the Experimental OAI-based Digital Library Systems Workshop to be held in conjunction with the 5th European Conference on Research and Advanced Technology for Digital Libraries (ECDL) September 8, 2001.
• First draft v. 0.1 (10 July 2001) as submitted.
• Version 0.1 is also available in PDF.

Abstract: The Open Archives Initiative has always maintained a distinction between data providers and service providers. This works at a functional level; some current projects are showing that it is less significant at an operational level. The Open Citation project harvests reference data from full-text eprint archives for reference linking, citation analysis and citation-ranked search. These are regarded as service provider functions where the end-user interacts directly with the services. The aim is to make these data available for export back to the full-text archives and to other service providers. Thus OpCit becomes a data provider too. The current Open Archives Protocol for Metadata Harvesting says nothing about full-text data harvesting for services such as these, nor about the export of processed data. This short paper outlines two proposals for progress on these issues.
 

1 Introduction

The OAI Protocol for Metadata Harvesting (Van de Sompel and Lagoze 2001) mandates simple, Dublin Core based metadata to achieve interoperability between archives with low overhead for archive maintainers. This focus on simplicity would appear to be vindicated by the apparent demise of NCSTRL, a forerunner of OAI as a collection of distributed archives. (Krichel and Warner 2001b) It suggests, however, the expectation of data transfer between services at a fairly low level of functionality unless OAI data and service providers can supplement the basic metadata for particular application areas, as allowed in the OAI protocol.

An example of enhanced metadata is the proposed Academic Metadata Format (AMF), a parallel metadata set that can be deployed with basic OAI metadata and which is designed to be used by the eprint archiving community, or more generally 'to advance scholarly communication over the Internet'. (Krichel and Warner 2001a)

This is a welcome development. Once an OAI service provider itself becomes a data provider, it is inevitable that data output in this case will not be simple document metadata as offered by the original document archives (which might be a better distinction than 'data provider'), otherwise, what service would the service provider be offering?

This paper considers the enhancement of OAI metadata from the perspective of the Open Citation (OpCit) project, which is demonstrating reference linking and citation analysis services by working with the largest OAI-compliant archive, the Los Alamos physics archives (arXiv). The approach described is intended to be generalised for other Open Archives, and is presented here to promote discussion and participation in the process.

2 Reference linking and citation analysis

An early demonstrator of OpCit reference linking was described by Hitchcock et al. (2000). Since then the reference database continues to be kept-up-to-date and has been restructured to store richer data, improving accuracy and robustness. This has enabled the service to be extended, providing forward linking citation analysis as well as a Google-like search service that ranks results according to citations or hits.

The project has achieved this by working in partnership with the arXiv maintainers. To extend this approach to other archives, the project confronts two questions:

1. How can reference lists be extracted from Open Archives within the framework of the OAI?
2. How can processed data be exported back to the original archives so that it can be visible to the users of the archives (rather than as a standalone demonstrator)?

In the OpCit application the schematic in Figure 1 shows data input and output in the context of the OAI. The only parts of this schematic mandated by the OAI are the target paper archives and OAI metadata output. OpCit is responsible for the citation database and specified user services. The subject of the questions above are the production of the reference list and the data export (stages A and D, respectively, in the schematic).

2.1 Extracting and parsing reference lists

A series of Perl modules is available from http://arabica.ecs.soton.ac.uk/code/doc/ref/readme.html, and includes two modules for extracting reference strings from TeX source files:

• Extract::Reference

  inserts mark-up 'xxxOpCit' at the beginning of each reference in a TeX source file, converts TeX/LaTeX to a DVI file, then DVI to text by 'dvitype' (Unix command), and parses the text file to produce a list of references

• Parser::Citation

  extracts metadata (authors, journal, volume, issue, etc.) from a reference string

• Parser::arXivCite

  a special case of "Parser::Citation", processes arXiv IDs (e.g. hep-th/0001001) found in references

• Parser::arXivRef

  processes multiple citations within a single reference entry, as often found in physics papers

2.2 Citation database

Within the project this database, called cite-base, is also used to serve the cite-baseSearch engine (http://cite-base.ecs.soton.ac.uk/cgi-bin/search) that ranks results according to citations or hits, as selected by the user.
 
 

a

b

2.3 Data export

OpCit's immediate requirement is for a metadata format to express the reference lists (and hence the referenced articles), along with fields such as title, abstract, etc. This can be expressed using Dublin Core metadata, specifically the relation attribute. An in-house format, "opcit_dc" (Figure 3), augmenting Dublin Core with a citation attribute, was developed to store structured data for references (title, author, year of publication, and separate identifier).

The development of  the AMF offers the chance to extend this approach for OpCit and OAI archives that handle research papers. AMF is a relational model for data, e.g. two documents are related by a reference. These relations can be expressed in either direction, e.g. AMF can express all the papers by an author, or all the authors of a paper.

Within AMF will be new identification systems for authors and institutions, beyond OAI's current document identifiers. Using unique identifiers for authors, for example, will allow users to find the output of a particular researcher. Current systems cannot differentiate two authors with the same name.

It is planned to use AMF to transfer reference data between OpCit and arXiv. Another Open Archive, RePEc (Krichel and Warner 2001b), which is a database of papers on economics, plans to use AMF to implement the OAI protocol, and such a rich information resource could be used by others to implement, e.g. academic Web portals.

3 Conclusion

In the case of the Open Citation project and arXiv, service and data providers together recognise the need to  supplement the basic OAI metadata to improve functionality and performance.

Two proposals for the OAI community to consider emerge from this brief description of the reference linking and citation analysis work of the Open Citation project:

• Richer formats are required to supplement the basic OAI metadata and to expose data for transfer between service providers and archive maintainers. The Academic Metadata Format appears to be a good foundation for this, and others are encouraged to participate in the development and review of AMF so that it might be adopted with some consensus.
• Reference lists need to be available for automated download from archives. Modular software aimed at archive maintainers, which automates the extraction and mark-up of references from submitted papers, is available for free download from the OpCit project.

References