 |
WebCleaner |
|
WebCleaner |
This document defines the data and mechanisms used to rate internet pages. This can be used to enable surfing sessions that are limited to specified contents, and is especially useful for allowing children to experience the web in a safe manner.
Unless otherwise noted, character data is assumed to be in Unicode.
A rating is a dictionary filled with name/value items. Each name/value item is an instance of a defined rating format.
A rating format defines a unique, case insensitive name, and a range of valid values along with their interpretation.
A given interpretation allows comparing two ratings against each other, to test if one exceeds the other. Comparisons are used to decide wether to allow a rated page or not.
A URL rating relates a rating to one or more URLs. It consists of a rating, an URL and an optional generic flag specifying that all URLs with the given URL as base path apply to the rating. It also has an optional comment explaining the rating.
A rating service defines a unique name and the (sub-)set of supported rating formats. The service name should be an URL with human-viewable information about the service.
The service should store and/or deliver URL ratings of the formats it defines.
Several recommended storage and delivery methods are explained below, however they are not mandatory.
Several different rating delivery methods are defined. They allow content providers flexibility when providing rating data. Since rating data is in Unicode, it must be encoded if the delivery method does not support Unicode.
When answering a HTTP request, rating data can be submitted as HTTP headers. All headers start with X-Rating and are defined as follows:
submits the service this rating applies to
submits one name/value item of the rating dictionary
HTTP header names are defined case insensitive. This is the reason why rating names, which are used as part of HTTP header names, are also defined case insensitive.
The order in which headers are sent does not matter. No duplicate X-Rating headers are allowed, but a X-Rating-XYZ header can occur multiple times, in which case it is service dependent on how to interpret the multiple values.
A HTML file can be rated by supplying data in a meta tag. The format is basically the same as the HTTP header format, using X-Rating in the name attribute and the value in the content attribute.
There are several things to consider when storing URL ratings for a service.
Since rating data is in Unicode, it must be encoded if the storage method does not support Unicode.
Rating data will be stored and processed on lots of different computers and platforms. All algorithms should be platform independent.
The storage format should be easily readable by humans. A readable format is automatically editable with a text editor, which also guarantees platform independence for editing operations.
As you might have guessed by now, the WebCleaner software delivers both a full featured rating service, and supports to deploy and enforce it on your computer. You can even submit your own ratings to the service. Below you will find a short description of the service. For more information, visit the WebCleaner rating service page (XXX todo).
The following rating formats are defined by WebCleaner:
| Name | Values | Interpretation |
| WC-Agerange | INT “-” INT? | Content is only suited for people of an age in the given range. If the second range value is missing, it is considered infinite. |
| WC-Violence WC-Sex WC-Language | “none” or “mild” or “heavy” | Content contains violence, sex or foul language in the given degree: either none, some or a lot of it. |
WebCleaner stores URLs normalized and absolute.
WebCleaner supports both HTTP header and HTML meta data ratings.
HTTP header example:
X-Rating: http://imadoofus.org/service/
X-Rating-WC-Agerange: 10-
HTML meta data example:
<meta name="X-Rating" content="http://imadoofus.org/service/">
<meta name="X-Rating-WC-Agerange" content="10-">
The WebCleaner rating web interface allows to query and send ratings for specific URLs. Sent ratings will undergo a review process before being admitted into the official rating database.
The WebCleaner proxy web interface will allow to send local ratings for review to the rating service URL. It will also be possible to update the local rating database with the official rating database.
URL ratings are stored as a number of fields. Each field begins with a tag, such as URL or Comment (case insensitive), followed by a colon, and the body of the field. Certain fields may be multiple lines in length, with subsequent lines indented by whitespace.
The format grammar assumes that the values of the URL rating are given as the -value terminals. Strings are case insensitive.
url-rating ::= url generic? rating+ comment?
url ::= "Url: " url-value
generic ::= "Generic: " ("true" | "false")
rating ::= rating-name ": " rating-value
comment ::= "Comment: " indented-comment-value
Example:
Url: http://www.imadoofus.org/
generic: true
wc-Agerange: 6-
Comment: This is just a dandy site, suitable for
all kinds of people.
Now with more extras!
The parser accepts malformed entries: the fields can be in any order, and whitespace will be normalized.