Automation and media monitoring 

Media monitoring can involve many steps. Basically, we can break it down to:

  1. choosing relevant sources;
  2. finding relevant information within those;
  3. tracking the analytics of importance;
  4. creating summaries;
  5. analysing the data.

You can imagine that this process can be difficult and exhausting if people needed to manually go through all these steps. Actually, media monitoring was such a task before the boom in AI technology. Back then, analysts had to scan for articles in print or online media, do manual newspaper clipping if needed, and do everything by hand.

While this was possible in a world with no internet, now we have millions of sources of information – from online news to social media comments, all of which emerge by the minute, sometimes even seconds. It is nearly impossible for brands to monitor what people are saying about them if they rely on manual work only. And here is where automation comes in. It proves to be helpful in those types of tasks that are repetitive and effortful.

How does automation help media monitoring?

Q: Where does automation help the most in media monitoring? 
A: The manual, repetitive tasks, which take a lot of time and effort without increasing productivity.

These days, automation is an inseparable part of the media monitoring process. It helps with tasks such as tracking mentions in articles, finding relevant sources of information, and even sentiment analysis. 

Let’s look into automation’s pros and cons for some of the tasks in media monitoring:

Sentiment analysis

Natural language processing (NLP) can be used in sentiment analysis. The technology can automatically detect negative or positive sentiments based on how charged the words are. But unfortunately, NLP is not always reliable. As mentioned previously, the human language is very complex, and it’s hard for a machine to determine all nuances people use in their everyday language or written articles.

Text summarisation

AI provides help with text summarisation as well. There are two basic types of summarisations: extractive and abstractive. The extractive one gets sentences directly from the text to form the summary. That is possible by identifying the most important sentences and combining them to form a shorter and coherent version.

Abstractive summarisation, on the other hand, is much more complex. Instead of taking portions of the text, the abstractive summary is based on interpreting a text using advanced natural language processing, create a condensed adaptation. It’s very similar to the way people write short versions of articles by getting the most crucial information out of the text and rephrasing it. While this may sound simple, it’s one of the most challenging parts of natural language processing. So far, abstractive summarisation is still far from human-like quality. For example, we can often see grammatical errors or even insufficient information in such summaries.

“Automation is a powerful tool, but it’s not relevant for every step of the process,” said Vladimir Petkov, CTO at A Data Pro, in an interview for the AIBEST podcast.

Still, in the context of media monitoring, automation is not everything. While it saves enormous amounts of time for the analyst, automation can’t do all the work. As we mentioned earlier, media monitoring analysts also have to cover sentiment – whether the tone of the article/social media post is positive, neutral, or negative. 

Language is complex, and we can sometimes use positive words in a negative context, with sarcasm being a prime example. Machines don’t always recognise that. That’s why the most efficient thing we can do as professionals is to work alongside AI.

We call this – human-in-the-loop machine learning.

Will AI take jobs in the Media Monitoring industry?

According to recent data, millions of Americans have lost their jobs due to the pandemic. They are now replaced by robots and AI faster than ever. So, the question stands – where does automation stand in media monitoring?

To put it simply, it all depends on the difficulty of the task we have to perform for the client. If the task is simple enough – if the client wants simple answers to their research, it can all be done by an AI trained model to do that particular task. 

According to Vladimir Petkov, A Data Pro’s technical director, the main question we have to ask is which processes should be automated. The answer to that question is hidden in the so-called value chain. If we have to perform repetitive, manual tasks, automation will be crucial to stay competitive. An example of such a task would be if your job was to find articles online and clip them into an Excel sheet. For tasks that involve more complex roles (up in the value chain), we need people.

Such tasks in media monitoring include coding or sentiment analysis.

As we said earlier, media monitoring and automation go hand in hand. That’s why for a lot of tasks, human analysts check and correct the AI to deliver the best quality to their clients.

But to answer the question: will automation take jobs? Honestly, automation takes over manually demanding, tedious jobs, so that people can focus their attention on more meaningful tasks. That’s why our technical director, Vladimir Petkov, advises that people should work on their skills to make sure that their job qualities stay in demand.

You can find the full interview with Vladimir Petkov here.

Here at A Data Pro, we believe that human potential is invaluable. Therefore, we focus our effort on the continuous development of the skills people need to perform in the best possible way. We have a dedicated technology team that works on innovation and automation which can free our colleagues from repetitive manual work, optimise processes and help us deliver more value to our customers.