RISC Software: Technological Milestones for the Media Industry19. June 2022
Using artificial intelligence for natural language processing opens up new digital opportunities for the newspaper industry. New technologies relating to the fully automatic structuring of the data behind quality news articles form the basis for maintaining the independence and diversity of publishers of newspapers and periodicals in a digital environment. The two research centres Software Competence Center Hagenberg GmbH (SCCH) and RISC Software GmbH have now made significant progress on this issue on a research project carried out with the tech company Newsadoo.
The newly developed technologies enable publishing houses to use databases for their work in digital channels, drawing on them to create new functions and monetization models in-house. The TIDE (Transparent Intelligent Diverse European News-Recommendation Algorithm) project used various methods from the field of artificial intelligence to work on algorithms that enable fully automatic recognition of relevant news content for private users and commercially relevant applications. Publishers can use particular components of this technology in the background to significantly improve the quality of their own database by automating work steps such as tagging and raising them to a uniformly high quality standard. In this way, content on related topics and local editions can be created automatically and publishers can place content in niches that best serve its monetization.
Fully automated specialized newspapers are created
Alongside functions that are available to users of the Newsadoo news platform free of charge, public sector companies and institutions can also have newspapers with a particular thematic focus created on the basis of the algorithms developed and can integrate them into their own apps, websites or intranet. The possibilities range from fully automatic digital corporate newspapers to digital municipal or party newspapers and publications for fields such as tourism and sport or interest groups. Integrated publishers that cooperate with Newsadoo benefit here, too. Newsadoo CEO David Böhm stresses the necessity of the research project for the media industry:
"“Only on the basis of technologies and structured data will it be possible to maintain the independence and diversity of media in Europe. So far, these technologies have only come from the USA, China or Japan. We feel it is important to develop our own solutions for the industry according to European principles and that these solutions offer new, digital forms of monetization and guarantee the publishers’ independence.”"
Next research topics already identified
On the strength of their fruitful collaboration and the considerable technological progress that has been made, the research partners have already started a new project: NEEED (News-Extracted Evolving European Datasphere) takes the potential uses of artificial intelligence in the European media industry to the next level. The Austrian Research Promotion Agency (FFG), the government of Upper Austria and Business Upper Austria already supported the TIDE research project, and are now doing the same for the new NEEED project. Newsadoo was presented with the Upper Austrian Innovation Award last year by government member Markus Achleitner.
"“Newsadoo is currently working on development of a continuously evolving news datasphere that will serve as a European alternative to approaches and solutions from America and China. SCCH is supporting Newsadoo with all its expertise on the topics of data extraction and knowledge graphs.”"
Another research partner is RISC Software GmbH which specializes in solving problems relating to mathematics and software and develops AI-based solutions for various fields. For approximately three years, research work on natural language processing has focused on the use of neural networks or classic machine-learning models to automatically process and analyse texts.
"“We contribute our expertise in the fields of data engineering and natural language processing to the project. Specifically, the aim is to develop prototypical textual networks to create models of relationships between tags and topics in news items.”"