An Alternating Algorithm for Mining Redescriptions

Title: An Alternating Algorithm for Mining Redescriptions

Research Question: How can we develop a novel algorithm to find subsets of data that can be described in multiple ways, also known as redescriptions?

Methodology: The researchers introduced a new data mining task called redescription mining and proposed a novel tree-based algorithm called CARTwheels for this task. CARTwheels is an alternating algorithm that grows two trees in opposite directions, matching their leaves. This approach helps in finding subsets of data that can be described in multiple ways.

Results: The researchers presented experimental results that demonstrated the effectiveness of the CARTwheels algorithm. They showed that the algorithm could efficiently find redescriptions with high Jaccard's coefficients, indicating strong similarity between the descriptors.

Implications: The CARTwheels algorithm can be applied in various fields where data can be described in multiple ways. This can help in integrating multiple forms of characterizing data and harnessing high-level abstractions for uncovering cryptic and subtle features of data. The algorithm can be particularly useful in domains like bioinformatics, where genes can be studied in a variety of ways.

Link to Article: https://arxiv.org/abs/0311048v1 Authors: arXiv ID: 0311048v1