Ahead of public votings and general elections, media coverage increases substantially. For politicians is what media covers a primary source to understand public opinion and sentiment. On the other hand, Journalists write about what they think might be of interest to their audiences, based on the Information they obtain from news agencies, tips or public sources. But is there a way to collect more reliable information for more informed decisions?
The way to understand public opinion is to turn to specialized research institutes, which conduct polls on a regular base. While this is a highly costly approach, the question is, if there might be a cheaper way and one, which delivers information nearly in real-time?
Ahead of the German general election and the public voting in Switzerland in 2017, we used natural language processing and concept detection to dig through thousands of content pieces on news sites, message boards, video platforms, Facebook, Twitter and blogs to investigate topic trends. While standard tools available keywords based frequency counts, such approaches are highly biased. Keywords might appear in texts, but that doesn’t say that such content pieces are all about those subjects. Instead, it could be just the opposite. To minimize bias and to achieve a reliable outcome we used a method we developed to detect concepts in texts.
In addition to published content, we evaluated discussions on social media to test the hypothesis that there is a gap between what media covers (what it thinks is relevant) and what people communicate about to, in order to avoid the media bias. We also weighted the sources to take their impact into account. A blog post read by only a handful of people but a higher frequency of keywords shouldn’t have more weight and influence than a top news site. A typical problem, which standard social listening tools are not able to solve.
The knowledge about such gaps are helpful to adjust political campaigns to make sure, only central questions of the general public are in focus. During the Swizz voting, we were able to discover that “women” and “young people” had much more impact on the overall opinion making without having to wait for a public polling. With the methods we developed, we outperformed results from social media monitoring tools, which can't weight findings to get a more accurate picture of reality.
At LIQUID NEWSROOM we believe that artificial intelligence can help us to excel in providing relevant information to our audiences, which help them to live either a better life or deliver outstanding results at work. That’s why we heavily invest in R&D and why we have developed AI-driven methods to challenge current marketing approaches. We want to assist companies in understanding the hidden mechanics. We want to accompany them in decoding their market DNA. Do you now know what we could do for you? Let’s get started! Click the button below!
We’ll get in touch with you
soon via email.
CEO @ LIQUID NEWSROOM