Technique and Process
Our server crawl every minute for new illegally shared content as photography, videos or audio. Against a general search engine which is searching the whole internet our server focus on known websites which offer content in a one-click download for a whole and general group of people. While focussing just on over hundreds of websites, forums and blogs we identify content shared on p2p networks, cyberlockers/file lockers or tubes and be able to report this content just-in-time.
If our crawler spot new content our four-step analysis starts like following:
- The found content will be compared with a huge data archive about image similarity which makes it quick and sure to identify relevant content.
- If there was no identification our service forward the written content of blogs, websites and forums to a text analyse heuristic to identify content by keyword monitoring.
- The most powerful engine is our Image and Logo Identification which finds and reads logos and text on images and classify the content. This process is very complex and unique with a success rate of over 95%.
- Finally, our human service checks all media and verifys the correctness of analysed content. The selection of our support has still the highest rank but gives our engines always a new chance to learn from selections. So the system learns over time to classify photography, videos and music with each new found and chosen content.
We analyse detected content normally within 24 hours and report the content directly to the hosting companies for deletion. After reporting the content is still monitored. If the content is not deleted within 72 hours we resend a deletion demand with a stark warning to see the need of legal aid.