The Shanghai ranking and other works based on bibliometics have drawn attention to the importance of accurate transcription of French addresses. At the same time, the process by which the address furnished by the author of a scientific article culminates in bibliometric indicators passes through three intermediaries:

• the publisher of the journal in which the article appears,
• the publisher of the bibliographic database that establishes the bibliographic notice of the article,
• the bibliometers and other users who make use of the information contained in the notices.

The object of this initiative by OST is to improve the quality of the treatment of French addresses in Thomson Reuter's Web of Science®. Two types of actions must result from the project if this goal is to be met: a standardisation of the way researchers write their address on manuscripts submitted for publication; and a familiarisation by indicator users with the various calculation methods used by indicator producers.

First of all, French researchers should be helped to understand the importance of full and complete addresses on articles they submit. This is an essential step, when the necessary information for identifying an article, the article's authors and their home institutions are first collected and published. Among other things, researchers should be aware that publishers often limit the number of characters for an address transcription.

The next step is the creation of the bibliographic notice. When Thomson Reuters registers an article in the Web of Science® database, it does not eliminate any institution appearing in the addresses, but it is occasionally led to modify the wording of the institution for the purpose of standardisation. Similarly, the database publisher may modify the order of the information, whether among institutions, or among entities within the same institution (institution, department, laboratory, etc.). The treatment of the addresses of associated or jointly-operated ("mixed") laboratories present particular problems for the editor, and the solution found is not always the best possible one.

Further downstream, the production and the interpretation of bibliometric indicators require  rigorous methods. For this reason OST works line by line with each institution to identify its articles while taking into account the multiple affiliations of mixed laboratories, which it differentiates from co-authorship. Many international studies, however, are based on automatic data treatment options that only identify certain institutions or certain forms of a name, or that limit the article's identification to the first institution named in the address. These options may be better -- or not so well -- adapted to a particular objective, but the importance lies in a thorough understanding of the characteristics of whatever treatment techniques have been used for a particular set of indicators, in order to be able to interpret correctly the resulting indicators.

To top