The main objective of the project is to design and develop an open source simulation framework for evaluating the frontend and information retrieval system of digital libraries using cost/gain user models. Furthermore, the benefit of this framework will be demonstrated by using it in a real digital library. SINIR is funded within the DFG program for Scientific Library Services and Information Systems (LIS)External link.
Access to digital libraries usually happens via interactive user interfaces with search functions. The software itself as well as the indexed contents are constantly updated to provide better results. While web services with large user bases are able to conduct A/B tests to evaluate changes, this procedure is not feasible for digital libraries with their small number of users.
In order to be able to conduct A/B tests in the context of digital libraries, user interactions such as requests, clicks and other interactions are to be generated using a simulation framework. This way, arbitrary changes to the front- and backend of the system can be evaluated using different user models. The training of the models can be based on real user interactions from logs and can be additionally supplemented by artificially generated behavioral patterns.
SINIR is a collaboration with partners of the Data Science ChairExternal link at the University of Passau and the Leibniz Information Centre for Economics (ZBW)External link.