Establishing Contextual Dataset Retrieval - transferring concepts from document to dataset retrieval (ConDATA)

• Applicant: Dr. Philipp Mayr-Schlegel
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: from 2018 to 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 410845317

As the number of datasets has increased in recent years, many researchers discern a need for an effective way of dataset retrieval satisfying a user’s interest and information need. A challenging task in the current dataset retrieval field is to ensure the access, the reuse and the reproducibility of existing datasets by researchers. Especially researchers need to find datasets to test their approaches and to validate their preliminary findings. In order to support a user in fulfilling his or her information need, retrieval systems are tending to collect information about the user’s interests in order to tailor the results accordingly. In this project we aim to include the user context such as issued queries, reformulated queries, and seen datasets in order to provide the user with datasets relevant to his or her interests based on user-centred contextualized approaches. To achieve this goal, a five-step methodology will be performed in the project: (1) observing the dataset retrieval-seeking behaviour of users, (2) studying existing dataset representation formats and proposing an enriched representation, (3) developing a user profile modelling approach which is targeting the context, (4) building a contextual dataset retrieval system based on different contextualisation approaches and (5) evaluating the different implemented approaches in a real-life system. This development process will lead to the proposition of an integrated dataset retrieval system that employs advances from document retrieval and transfers these to the novel field of dataset retrieval. This requires an investigation to which extent the contextualised approaches (e.g. user profiling, content similarity, collaborative similarity, popularity, etc.) can be transferred to the dataset retrieval field. After identifying the appropriate approaches for dataset contextualisation in ConDATA, they will be integrated into our reference system GESIS search portal in order to compare the different contextualisation concepts and to recommend the best performing ones. In addition, the contextualised approaches could be used in various research areas having similar datasets such as other social sciences or Economics. The overall contribution of the project will be an implementation and a recommendation of successful techniques, approaches, concepts and implementations of contextualised dataset retrieval. The contextualisation will be developed using state-of-the-art methods which are widely used in the information retrieval field and thus can easily be reproduced by other researchers.

• DFG Programme: Research data and software (Scientific Library Services and Information Systems)
• Co-Investigator: Dr. Ameni Kacem Sahraoui, Ph.D.