The Resource Description Framework (RDF) is a triple based representation of directed graphs with labelled edges. With the emergence of RDF graphs special databases, called RDF stores, were developed. In order to query graphs, which are stored in these RDF stores, the query language SPARQL Protocol And Query Language (SPARQL) is used.
To help in overcoming a shortage of citation information for the German social sciences, we contribute an approach for extracting author names from reference sections. Instead of relying on small amounts of manually labeled data, we use a distantly supervised approach in combination with the widely used probabilistic framework of conditional random fields.
Our objective is to identify and assess gender bias related to professions in the results of collaborative community work. We present the results of studies in which we characterize the gender inequality present over the three dimensions: redirections, images, and people mentioned in the articles.
During the development of a distributed database for big data, small datasets are used to test the implementation and to evaluate alternative solutions. These small datasets have the advantage that evaluation results are produced more quickly and the implementation progresses faster. When the database is tested with large datasets, several design decisions based on the experiments with the small datasets can lead to a poor performance or unstable database. In this talk some of these wrong design decisions during the implementation of the distributed RDF store Koral are presented and how they could be improved.
How can we recognize social roles of users, given a completely novel social network? We present a transfer learning approach based on feature transformation, in order to learn knowledge from networks that we know well, and apply the knowledge to predict user roles in other networks that are new to us. We implement our approach and evaluate it with real networks.
Interaction on the web or mobile platform is done by humans on devices designed by humans using interfaces that too has the human touch to it. And thats why this talk focus on understanding how the human signal responds to different interaction methods and how we can use these measurements to improve the existing framework or create interfaces or platforms for those who are challenged to use the conventional interaction methods.
PiGrid is a citizen science application based on the BOINC open-source network for volunteer computing from the Berkeley university. "Gridcoin" is a blockchain token that serves as a refund for giving computing power to research projects. PiGrid realizes an easy to use device based on a handy and smart microcontroller that realizes the services of a crypto-currency wallet. The device is also capable of directing its computing-power to work units in BOINC.
The business model of credit bureaus is to collect and provide relevant information for assessing creditors’ risks when interacting with consumers or companies. Central tasks in managing such information are to search and identify entities for which a credit report is requested and to predict and explain the future behaviour of these entities.
Data science which is nowadays strongly interconnected with Big Data analysis, machine learning and statistics is a key issue in Knowledge Discovery (KD) techniques from sources such as Databases (KDD) or Text (KDT). This presentation aims to discuss results of Evolutionary K-Means towards large scale bibliographic data analysis.
Crowdsourcing has become an extremely powerful and useful thing for many people.It will limit the improvement of crowdsourcing platforms if only focusing on the crowdsourcing platform itself. It is also important to listen to crowd workers and requesters in social media.