Data Stream Mining Addresses Research Issues Addressed by the Data Mining Community

The above approximation algorithm can guarantee the error range. The sampling technique uses the same concept as used in VFML. Data mining results on wireless networks with limited bandwidth Transmission of knowledge structure representation is another important research topic. Once the models and patterns are extracted locally from the data stream generator, you need to transfer these structures to the user. Modeling changes in mining results over time In some cases, users are not interested in mining the results of a data stream, but they are not interested in how these results change over time.

Data mining is one of the most difficult research themes today, which is entering an era where data becomes rich. This situation requires a computationally intensive learning algorithm that can be extended to handle large data streams. In addition, data streams are often dynamic and do not follow certain predictable distribution of data. Flexible machine learning algorithms with self-organizing characteristics can cope with any change in data flow, so it is expected that this situation will be overcome. Evolving Intelligent System (EIS) is a recent initiative of the Computing Intelligence Society (CIS) for data stream mining tasks. It has an open architecture that allows you to use empty rulebase or first trained rulebase from scratch. Fuzzy rules are automatically generated with reference to the contribution and novelty of the data stream. The direction of the project is to solve the data flow uncertainty problem.

Like other technologies, data mining also has privacy and ethical issues. There is much debate about how to solve privacy issues. Some people think that data mining is morally neutral, but as advertising companies purchase customer's expenditure data and actions at the expense of privacy, the use of current data mining is getting a lot of attention . Data mining can destroy privacy in many ways. First of all, data mining requires a large amount of data preparation, which may reveal unknown information and patterns. For example, you can combine numerous data sets from various sources for analysis (called data aggregation). A threat occurs if a user with access to this data can identify or track a particular individual.