What is the main task of a filter? The most important task of filters for the protection of minors is to provide a reliable barrier preventing access to completely unacceptable content, which is dangerous or morally damaging to the development of minors. Ideally, at the same time, content suitable for children and young people should not be affected (i.e. filter should have high blocking rate with low overblocking).
Filtering in the context of parental control also concerns outgoing streams, for instance to prevent young children from disclosing their name, address, school name or parents' credit card number. This type of filtering is particularly important for young users and there is freeware available to deal exclusively with this issue, for example at: http://www.logprotect.net/
How do filters work?
Filters are based on just a few principles: - Editorial classification Positive lists only permit access to content suitable for children, which an editorial team has inspected and selected. Conversely, negative lists block all content, which a manual inspection has classified as relevant to the protection of minors. - Automatic classification In this method, algorithms decide whether the content can be accessed or not. The simplest method for automated evaluation of the typical features of a website is keyword blocking. This blocks all websites containing specific forbidden words. "More intelligent" filter systems evaluate different typical features of websites (e.g. statistical analysis of the full text for a website) in order to categorise them automatically. The best automatic classification tools do not use text-only analysis but also include image analysis, for instance detecting a predominance of skin texture in an image. - Self-classification by the provider Here, every provider marks his websites with a virtual content classification label. Software installed on the user's computer reads the label and decides whether or not to display the content based on the user's settings.
What kind of filters are available?
Recommended lists with content suitable for children So-called "walled gardens" or "white lists" are lists of selected content, which only permit the user to surf to these selected websites.
Blacklists of internet addresses with content relevant to the protection of minors Filter systems based on blacklists take the opposite approach. They permit access to all internet content and attempt to filter out morally damaging or dangerous content. However, editing the entire internet is utterly impossible, so the blacklisting of internet content will always be catching up with developments in the internet.
Blocking websites with forbidden words The simplest filter systems block internet content using a list of forbidden words. These keyword lists are easy to produce and simple to maintain.
Filtering based on automatically classified contents The second generation of automatic classification systems evaluate the complete text contained in a website. They use statistical methods e.g. spam filters.
Self-classification by the providers Providers use a list of questions to describe what can be seen on their websites (e.g. nudity, violent behaviour resulting in serious injury or death. This description is inserted into the source code for the website or saved on the web server in a standardised form so that filter systems can evaluate the classification and block all content which does not conform to the user's settings. Currently the effectiveness of the filters using labelling is still very low because only a few providers have taken the trouble to classify their pages themselves.
Combinations of filtering methods The basic approaches to filtering are now combined in many ways in order to increase their effectiveness but also to permit age-differentiated access.
What filtering software is available and can be recommended?
At the moment, it is impossible to recommend a specific filter for the protection of minors because no program of this type has proved its effectiveness under comparable conditions. There are no reliable benchmark figures relating to the effectiveness of filter systems. There are no uniform standards or test procedures. The European Commission has commissioned a study for "an objective evaluation of filtering software and services currently available on the market, independent of the seller or provider and conducted by experts".The results of this evaluation are available at the sip-bench website.
What can I do until better filters are developed?
The filtering of internet content can play a supporting role in the media education of children and young people. However, it cannot be the sole protector as it is currently only effective in the World Wide Web and ignores other communication technologies. Educational supervision, both at home and at school, is the best way forward. Banning children and young people from using technologies is never the best solution - dialogue and a shared understanding of the risks and benefits are needed. In any circumstances, the filtering of internet content must be subordinate to the education of young people and be integrated into an overall concept for ensuring the protection of minors.
What can be expected in the future
The filters currently available for the protection of minors are still unsatisfactory in their overall performance. The future lies in the development of classification mechanisms, which take greater account of contextual information, and in the intelligent combining of various filter concepts. We can expect better results from systems in which automatic classifications are supplemented by editorially compiled lists and intelligently linked to self-classification by the providers. The onus here is particularly on content providers and operators of filter systems to offer age-differentiated options for accessing their content and take better account of the requirements for the protection of minors.
Now go to your national node's website to find useful links and further information.