Our Research Projects
For topics of the future such as text mining, definition of key performance indicators for social media sources or the automatic identification of influencers, pressrelations has launched and joins several research projects: For instance, we cooperate with Düsseldorf’s Heinrich Heine University and we are involved in a European research project. Detailed information on our current projects can be retrieved below
MULTISENSOR is a three-year FP7-STReP-research project launched in 2014 and funded by the European Commission with a total budget of €4 million. The consortium consists of 9 participants from various universities, research organizations and companies: These include the Centre for Research and Technologies Hellas Information Technologies Institute (CERTH-ITI) (coordination, multimedia technology), Universitat Pompeu Fabra (language technology), Fundació Barcelona Media - Yahoo! Research (social media analysis), Ontotext (Semantic technologies), Linguatec (language technology), everis (information technology), pressrelations (media monitoring service), Deutsche Welle (German public broadcasting service) and PIMEC (SME network). The aim of the project is to combine advanced technologies with emotional, social and spatiotemporal skills, so that multimedia and multilingual information can be interpreted, combined, translated and summarized. These technologies are intended to serve journalists, international media monitoring service providers and SMEs as a decision guidance in internationalization strategies.
In a research cooperation project dubbed Anfragenbasierte GENerierung von Textzusammenfassungen (request based generation of text summaries, AGENT) that launched spring 2014, we cooperate for the second time with the Chair of Databases and Information Systems at Heinrich-Heine University Dusseldorf. It is also the second research cooperation project, which is funded by the German government’s Central Innovation Programme for SMEs (ZIM) nearly with the maximum amount. The research project AGENT is aimed to automatically summarize the core information of a text in the form of a summary – as a human being would create it. The special feature here is that the summaries are created based on requests, i.e. by using customized keywords. This approach ensures that the information contained therein is relevant with regard to the selected topic. “Much of the working time of our media analysts accounts for reading and simultaneous filtering of information that is irrelevant to the current objects of research”, say Pashutan Modaresi and Dr. Philipp Gross, both developers pressrelations and members of the research team. “Empirically, at best 35% of the edited articles filtered by the editors deal primarily or exclusively with the research objects. In the remaining 65% of news pieces, the relevant issues are mentioned only parenthetically. The AGENT project builds on this point and meets the demand for information filtering within particular texts. This workflow can be optimized simultaneously and help to realize new, diverse and sophisticated analysis products.”
The research project AGENT is based on the research project ATOM (Automatische Themenverfolgung und Opinion-Mining für eine Medienresonanzanalyse / Automatic Topic Tracking and Opinion Mining for a Media Content Analysis, project ended in December 2013), which has previously already been funded by the German government’s ZIM. The project continues the successful technology transfer between pressrelations and the University of Dusseldorf.
ATOM - Automatic Topic Tracking and Opinion Mining for a Media Content Analysis
In cooperation with the Institute of Computer Science at the Heinrich Heine University in Dusseldorf, pressrelations launched a first joint research project in July 2011. The research project entitled “ATOM - Automatic Topic Tracking and Opinion Mining for a Media Content Analysis” was aimed to improve significantly the IT-based detection and analysis of media content of all kinds. Thus, the hitherto still predominantly manual and thus labor-intensive coding process of media reports was aimed to be supported successively by digital, automated methods for text analysis.
The research project, which has been successfully completed by the end of 2013, was funded with the maximum amount for cooperation projects between SMEs and research institutions by the German government’s Central Innovation Programme (ZIM).
Here you find further information about the data set and the coding guidelines.