| Customer Communication Tomorrow

Security on the Social Web: Risk Management with Artificial Intelligence

Security on the Social Web: Risk Management with Artificial Intelligence

User-generated content is an important part of the marketing mix. However, it doesn’t just provide for interactions and reach, but also potentially for problems. Namely, when illegal content is posted. What options do companies have at their disposal to stay in control of content on their social media channels?

The example of Facebook shows how important it’s become to continuously monitor social media content: The US company has bolstered its team in this area up to a total of 30,000 positions in 2019. These are resources that only a few companies can muster. In comparison: The energy service provider Vattenfall and the perfumery chain Douglas each employ a total of around 20,000 people.

In principle, there are three ways to manage your own social media channels. The first is manual review of texts, images, and videos for legality and netiquette. Admittedly, the number of posts that can be reviewed in this way is limited, and can usually only be done at random. However, since the volume of user-generated content is more likely to increase than decrease as a result of growing user numbers online, this approach would only be scalable through even higher personnel deployment.

The second approach is the use of technology – and artificial intelligence in particular – to filter out unwanted or dangerous content based on rules or algorithms. These systems are very powerful: With the aid of machine learning, YouTube was able to review content for which they would have had to employ 180,000 people within one year. However, in spite of all this progress, purely technological solutions are still too error-prone. You can test how well an AI currently works in image recognition, for example, using Microsoft’s CaptionBot. This program can achieve impressive results and has a high hit rate, but always has its limits.

Hybrid approach for the best results

The solution is to combine both approaches. A hybrid approach offers the greatest possible security, as it allows both automated checking of large volumes of content and situation-specific or context-specific checking.

But the artificial intelligence has to be trained first. This requires as large a data set as possible, one that’s already been classified, as “illustrative material.” The AI then reviews new posts on this basis. Regular manual sampling by employees then shows how well it works. In practice, Arvato CRM Solutions uses a three-stage process: First, contributions are statically filtered according to certain terms or web addresses, for example. Then the classification is carried out by the AI, which has been trained for the client’s respective requirements. The third step is the manual sampling, made necessary by content that’s often ambiguous. Does an image really show something offensive, or do the image contents just happen to result in a certain shape? Is a text post a simple statement, or does the rhetoric hide another message? This is where even the most cutting-edge systems still reach their limits. The results of the review are incorporated into the AI’s ongoing training. This “supervised learning” approach leads to increasingly reliable artificial intelligence.

Opportunities through artificial intelligence

Using artificial intelligence, supported by an experienced team, companies are tapping valuable opportunities for their social media channels.

  • Increased security: Immediate reaction to content minimizes risks for users and operators of social media sites and platforms.
  • Efficient use of resources: Because the AI system and an expert team are taking care of the moderation, the client’s social media team can concentrate on activities that add value and promote interaction.
  • Optimization of the customer experience: Users are protected from illegal, negative, or offensive content, ensuring a positive online experience.
  • Transparent handling of user-generated content: An open dialogue on how to deal with negative or harmful content is encouraged. The risks faced by organizations and users through social media activity are minimized.

Artificial intelligence can optimize the interaction between companies and customers, but not just on social media. The additional advantages for companies that can result from the use of artificial intelligence are shown, for example, by the intelligent search function eSearch, chatbots in customer service, and applications for voice assistants.

Author: Editorial team Future. Customer.
Image: © Production Perig – AdobeStock

Tags for this article Artificial Intelligence (80) Digitization (162) Social Media (22)

58000+

People

30

Countries

60+

Languages

12+

Years - avg. client partnership

400+

Clients worldwide

We're global so you can be close

We're global

we believe in innovation. here’s the latest

what's new

Social Media Newsroom

    We speak your language. and we’d love to hear from you