Traditional media monitoring involved people doing lots and lots of paper sifting and indexing. Machine learning and artificial intelligence (AI) promise to change this. At the leading edge, much of the hassle of manual labor is in the past, while accuracy is dramatically up. There is still a long way to go, however.
A hybrid media monitoring approach
Leading media monitors like Newsmeter have boosted performance by embracing a hybrid approach. This is obviously faster with online publications. The still-plentiful paper publications and periodicals need workers to scan them. Optical character recognition (OCR) programs then process the scans, essentially rendering them into the same format as online content. Algorithms search both online and paper media for specific instances like mentions of a brand.
At the current state of play, human workers have to scrutinize the ‘long lists’ algorithms provide to eliminate the chaff and end up with the hard residue falling within clients’ terms of reference.
Humans still rule, for instance, to ensure that favorite of recent discourse, fake news, does not get beyond the analysis stage. True, some algorithms also have machine learning capabilities, meaning they will learn to distinguish real from fake in time. But even they cannot put a piece of news into context. Hence, experienced media monitoring professionals will continue validating data for some time to come.
Big Data getting bigger
While algorithms are learning, however, the information avalanche is continuing apace. Manual media monitoring was an uphill task in the past; it is impossible today. Companies like Newsmeter are facing up to using machine intelligence to deal with the sheer scale of growing information. Even a single post on a personal blog can prompt a torrent of comments that have the potential of ruining a brand. Intelligent automated news scraping is helping extract accurate data from thousands and even millions of such sources. An algorithm can look for specific sources, keywords, even emotional expressions.
The sheer amount of collected data is also an issue. Media monitors compile literally tons of data on diverse subjects and clients. Making use of them requires machine intelligence able to track historical data. This highlights fascinating developments and can end up predicting how consumers will react to specific brand moves, making media monitoring proactive.
Enter predictive analytics
Predictive analytics can spot current and predict future trends. Coupled with machine learning, it can give early warning of trending customer concerns. Adaptive media monitoring algorithms can spot emerging news among a myriad of corporate and personal blogs. While pouring through data and hinting whether chance online conversations might evolve into lasting trends, it would spot fake or automatically generated comments.
In a sense, much of the above is what media monitoring has always claimed to provide. If the industry puts its hand to heart, it will admit it has rarely been 100% successful in providing it. Yet, as AI matures, it will undoubtedly bring further benefits to media monitoring – many of them ones we can scarcely dream of now and greatly improving the overall media monitoring performance.