Case Study Campfire + WebPurify

Background: Campfire created an award-winning promotional campaign for the Cinemax Show Hunted which included an interactive microsite, where users were invited to join the show’s private intelligence organization. This “recruitment” process involved conducting covert psychological tests, including the creation of a child’s drawing. This is where the team at WebPurify came into play for medium user-generated content in the form of pictures.

 

Our solutions:

Completely. All of the submitted images reviewed with our standard criteria: partial nudity, nudity, violence, hate or hate crimes, offensive gestures or language and drugs. Users with inappropriate images were prompted to take the test again.

Creative. Added the “wow” factor for users by categorizing and labeling acceptable images. Immediately after a user completes their drawing, the “system” will identify what it was (for example a cat or tree), and who has identified the user’s personality and qualifications to join the program’s elite spy team Was indicated infast. Completed the image moderation and grading process in less than one minute to meet the rapid response needs of this real-time marketing campaign.

 

Flexible. At any given time, there is a sudden increase.

Flexible. At any given time, there is a sudden increase in traffic and quick response to deliver a top user experience to more than 1,000 users performing a personality test on the site.Results: WebPurify successfully created over 1 million images in two months. The campfire won AdAge’s Gold Award for the Small Agency Campaign of the Year and the Web Marketing Association’s Internet Advertising Competition for its Best Integrated Campaign.

 

Government pressure on censors

Some recent changes on the largest social media platforms have been inspired by government efforts to control the information distributed by terrorist organizations. Such governmental steps can be seen as closing the stationary door after the horse bends.

In many cases, the problems with social unrest are simply that – social problems caused by government policies, wars and exploitation in world politics. However, the fact remains that the spread of unrest is widespread on social media. The viral nature of shocking images and information is a problem for those working with maintaining law and order.

 

# 1 – Written / Spoken Speech and AI Content Moderation

Machine learning is active only as training received for a specific task. When it comes to moderating interactions between humans on a global scale, such as on the Internet, the rules have to be ridiculously complex to work.

Just add to the challenge of the many forms and dialects within the English language, and that it takes a lifetime for humans to learn their local style. This shows that the work ahead of the machine is a deeply complex and challenging one. To develop a tool that can work in every form of communication, language and every dialect.

 

Natural language processing (NLP) tools are trained.

Natural language processing (NLP) tools are trained by humans to understand the type of text they use. However, this understanding is limited to simple rule-making that has a difficult time accounting for the complexities of human communication.

Even in single countries, there are different accents and dialects. These accents and dialects are each unique and different. You should be born and brought up in a specific place, which is being discussed in or around you. Phrases and their intended meanings may vary from street to street in some streets and neighborhoods – as can the pronunciation. The whole thing can be a mine of communication misunderstanding.

 

The current reality of AI-driven content moderation

The current reality of AI-driven content moderation and NLP is that understanding the multiple, linguistic differences of these subtle and any machines will require an inadvertently large amount of input and training data. And you have to find humans to understand it first.

Comprehensive NLP tools that operate within all areas of the web and monitor all languages ​​and communications are not possible. AI would have to be more advanced, and training of each tool would have to be more localized, specialized and expensive.

 

We all know that the possibilities of misunderstandings

We all know that the possibilities of misunderstandings in the real world are largely between people from different cultures. Online, the potential for crime increases exponentially as communication occurs immediately and is exacerbated by the fact that discussions, especially in public forums, are conducted among thousands of users simultaneously.

These participants may be from different geographical locations and have different belief systems and opposite moral or civil codes. It is a powder keg that requires very subtle restraint and great understanding and attention when things are hot, and passion is high.

Updated: May 31, 2020 — 7:55 am

Leave a Reply

Your email address will not be published. Required fields are marked *