Data Mining pp 135-152 | Cite as

Association Pattern Mining: Advanced Concepts



Association pattern mining algorithms often discover a large number of patterns, and it is difficult to use this large output for application-specific tasks. One reason for this is that a vast majority of the discovered associations may be uninteresting or redundant for a specific application. This chapter discusses a number of advanced methods that are designed to make association pattern mining more application-sensitive:
  1. 1.

    Summarization: The output of association pattern mining is typically very large. For an end-user, a smaller set of discovered itemsets is much easier to understand and assimilate. This chapter will introduce a number of summarization methods such as finding maximal itemsets, closed itemsets, or nonredundant rules.

  2. 2.

    Querying: When a large number of itemsets are available, the users may wish to query them for smaller summaries. This chapter will discuss a number of specialized summarization methods that are query friendly. The idea is to use a two-phase approach in which the data is preprocessed to create a summary. This summary is then queried.

  3. 3.

    Constraint incorporation: In many real scenarios, one may wish to incorporate application-specific constraints into the itemset generation process. Although a constraint-based algorithm may not always provide online responses, it does allow for the use of much lower support-levels for mining, than a two-phase “preprocess-once query-many” approach.

These topics are all related to the extraction of interesting summary information from itemsets in different ways. For example, compressed representations of itemsets are very useful for querying. A query-friendly compression scheme is very different from a summarization scheme that is designed to assure nonredundancy. Similarly, there are fewer constrained itemsets than unconstrained itemsets. However, the shrinkage of the discovered itemsets is because of the constraints rather than a compression or summarization scheme. This chapter will also discuss a number of useful applications of association pattern mining.


Frequent Pattern Pattern Mining Frequent Itemsets Inverted Index Frequent Pattern Mining 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Copyright information

© Springer International Publishing Switzerland 2015

Authors and Affiliations

  1. 1.IBM T.J. Watson Research CenterYorktown HeightsUSA

Personalised recommendations