In today's world, it is of paramount importance to protect the privacy of customers, which is also regulated by governments in order to reduce the possibility of abuse of sensitive data. Companies, however, need to provide this information to third parties in some cases, so they have to hide all the information that could be used to indentify their customers. One way to do this is anonymisation, which transforms the data so that it does not lose its information content, but it can not be inferred from who the particular data relates to.
Anonymisation is usually done statically, running the algorithm on an entire database. However, in systems where data is constantly generated and used for data mining purposes, it may be necessary to have an algorithm capable of immediate anonymization of the data provided in the stream.
In my thesis, I will show the theoretical basis of anonymization, in which I will present the algorithm of k-anonymity in more detail. Then with this knowledge I will create my own model for the processing of continuous data, and will work out an own method of k-anonymization. To test the results I will make the implementation of the algorithm and measure the correctness of the method.
My goal is to develop a method that works with a lower computing need, because that is extremely important in real-time systems, but which distorts the output data to a lesser extent.