Co-link Web Indicators Technical Report
Co-link Web Indicators of the European Research Area is the technical report of the WISER project work package 2.1. The data sets on which the report is based are available from the data sets item in the table of contents on the lefthand side of the page.
Executuve Summary
This technical report describes in detail the methods that were used to gather co-link data and to prepare co-link indicators of 862 natural science and engineering group web sites at 50 major European universities.
Two documents are considered to be co-linked if the links to them occur in the same web document. Citation theory says that if a pair of documents is frequently co-cited then the contents of the documents are related in some manner. This property is discussed within the framework of co-linked web documents and then it is used to build example co-link web indicators. The examples illustrate how co-link indicators can used to investigate questions such as 'How frequently are ERA university group web documents co-linked with web documents on servers in other ERA countries or on US servers?' and 'Are linguistic and cultural biases present in web indicators data like they are in bibliometric data?'
Google was used to identify source documents that contained links to university group web sites. A custom web crawler was used to collect links from source documents and the links from these documents were used to build co-link indicators. Google has two limitations that hinder its usefulness as a tool for easily identifying and collecting links to source documents. These limitations are discussed and a technique is provided to overcome one of the limitations.
Indicators were constructed using geographical information about the location of hosts of the source documents and co-linked documents. These indicators showed that ERA group web sites were most often linked to by source documents that reside on European (48%) and US (39%) hosts. They showed that documents in ERA group web sites tended to be co-linked most frequently with documents residing on European hosts. ERA university group web sites tended to be most often co-linked with documents on servers in their own country and the second most frequent co-link are with documents on servers in the UK or Germany. These findings are discussed as being potential indicators of European cohesion and language and cultural biases on the Internet and/or Google.
Interpreting co-link indicators and web indicators can be problematic. Unlike bibliometric indicators such things as basic units for counting and the limitations of these units are not well-defined for web indicators. Also, unlike scientific publications there is little, if any, quality control on the content of web pages and the content can change with time. We are just beginning to understand the usefulness and difficulties of creating and interpreting web indicators for innovation systems. The information gained from this project will help clarify the problems associated with producing and using co-link web indicators of science, engineering and innovation research and provide a focus for future research.
|
