You are browsing the archive for authors.


Thomas Krichel - February 17, 2011 in authors, identification, linking

Objective The overall objective is to invert bibligraphic data from its tradtional format where each record describes a document. We want to create to a cv-style format that has authors as the heading and the documents written by the author underneeath it. This allows for a navigation of the bibliographical space by author. It also prepare for performance evaluation of authors. Inputs There are two sources. One is a set of simple document data from the OKFN sponsored 3lib project. 3lib. These data are de facto open because they contain only factual descriptions of documents, titles, authornames, identifiers. The document data describe scientific articles and preprint. The other source is a set of author profiles that are openly available from AuthorClaim. Method Authors are referenced in bibliographic information by names. Names are ambiugous. There are many ways to write the name of a single person. We call these "name expressions". Serveral persons may share valid name expressions. Since names don't identify authors, AuthorProfile can not do a reliable job. The AuthorClaim project allows authors to claim documents. Only a very small part of documents are subject to author claims at this time. These are the people for authoritative publication lists are available. For the others we have to use name expressions. We look at bibliographic data records containing such author name expressions, and create files, one for each author name expression. We call this process "auversion". The system will have list of author pages as top-entry navigation. Author pages can only be constructed for authorclaim registrants. However most AuthorClaim registrants have co-authors, and most of these are not yet registered. These non-registered co-authors then provide entry points to author name expressions, etc. Thus a substantial part of "auverted" bibliographic data can be linked from the authors. System In addition to navigating a set of authors (not implemented yet), we plan two navigational features. First, we want to link from an auverted author name page to the closest registered. By "closest" we mean by shortest intermediate author name expression path through co-authorship. This is partly implemented on our test set system. We call this "vertical integration". Second, we want to provide links between related author name expression. Assume for example, we have the author name J. Griffin, but we also have James Griffin, we want to create a link from J. Griffin author name expression to James Griffin author name expression. We want to do a similar thing for diacritics, linking from expressions with diacritics to those without and back. We links between author name expressions that may refer to the same person as "horizontal integration". Current state A debugging/testing demonstrator of the system is available Eligibility The 3lib dataset includes the IUCR data from the JISC funded open bibliography project. However since the data is very small, it is not likely to be seen in the actual demonstrator that we have running.