COUNTER Filter
The aim of the COUNTER Filter project was to develop a data filter that would deal with the possible inflation of usage statistics due to interface effects, assess its implications for the application of unique article identifiers, such as DOIs, and to propose how the filter could be implemented by vendors.
Executive Summary
The overall aim translated into three specific objectives:
- the development of a filter to be applied to usage data that would dampen or compensate for the inflationary effects of certain vendor interface configurations
- an assessment of current vendor practice regarding implementation of unique article identifiers, such as DOIs
- propose how the filter could be implemented by vendors and recommend appropriate modifications to the COUNTER Code of Practice to the COUNTER Executive Committee
In the early stages of the project it was decided to test two data filters, as no experimental evidence existed to indicate if either approach would work in practice. The two filters tested were:
- the ‘unwanted html’ filter, designed to filter out requests for full-text html articles deemed not to have been specifically requested by the user
- the ‘unique article filter’ (UAF), designed to provide a count for the number of requests for unique articles (irrespective of format) in a given session
The project was divided into three Phases and the overall approach taken had two main strands, which were carried forward in parallel. One strand involved the development and testing of the data filters on real vendor usage data. The second strand involved a survey of vendors to investigate how unique article identifiers (UAIs) are currently applied. The project team considered that there was no real alternative to testing the filters developed on real data and did not feel that a survey or interviews alone would have been adequate for this purpose. Only by testing the filter with a range of vendors could we assess its viability and how generally applicable it would be. As far as determining how UAIs are being applied, it was felt that a survey should include telephone interviews, as the responses were unlikely to be straightforward and there were issues that the team wished to explore further with each vendor.
The main results of the project were:
- the successful development of a UAF that will be recommended to the COUNTER Executive Committee for inclusion in the next Release of the Code of Practice. The UAF will compensate for the inflation of usage statistics, by providing a new metric – the number of successful unique article requests in a session
- evidence that an ‘unwanted html’ filter would not be viable due, among other reasons, to limitations imposed by the time it takes to download an article from a browser
- an overview of current vendor implementations of unique article identifiers and recommendations to the COUNTER Executive Committee to increase consistency among vendors for the implementation of unique article identifiers
We concluded from this project that while it is possible to further enrich the COUNTER usage data by the application of appropriate data filters, there will be limits to this in practice. These are determined on the one hand by the limitations in the data generating process, and on the other hand by the unwillingness of vendors to invest further in these processes unless there is a clear benefit in terms of the quality and value of the data thus generated.