Arachne usage analysis
The research conducted by Andreas Geißler and supervised by Prof. Dr. Manfred Thaller and Prof. Dr. Förtsch, aims to find out, whether the anonymized record and evaluation of research processes in historical and cultural-scientific object databases may be used to automatically enlarge the contextualization of these databases.
A basic requirement is a solid pool in terms of a well used and contextualized research database.
The increase of the average attendance per week, from 500 visitors in 2008 to 1500 visitors, which was observed after the conversion to Arachne 3.0 since the beginning of 2009, has successfully continued in 2010. Therefore Arachne offers the ideal conditions for the research processes listed in the following:
By searching for one or more terms, most of the n IR-systems applied today offer users an aligned amount of search results.
Contextualized object databases, like Arachne for example, allow the user to navigate from one object, which he has chosen from the original search result, to other objects, which are linked to it (i. e. contextualized). This browsing or navigating within contexts forms an integral part of research processes and can lead to the user finding objects displayed that no longer even contain the originally requested search terms and therefore were not included in the actual search results.
Even though the objects do not contain the used search terms, one can still expect them to have at least some relation to each other. A detailed record may also permit conclusions as to the extent of analogy between the search terms and the displayed objects. The cohesion between these factors is to be surveyed with the help of a detailed record of research processes, by using the example of the Arachne-database.
Studies that have been conducted so far:
In order to estimate whether research processes in contextualized object-databases have enough depth for this research project, each user’s movements in Arachne were anonymized and recorded for a limited amount of time during a first test run.
The results showed, among other things, that during the time the test was conducted, in more than 25% of the sessions more than 10 datasets were viewed. This and further studies have proven that the research processes in fact provide enough depth for the aspired research project. The present results were displayed to an international audience of specialists on the CCAA2009 in Williamsburg in April 2009.