Itemset mining: A constraint programming perspective

Abstract

The field of data mining has become accustomed to specifying constraints on patterns of interest. A large number of systems and techniques has been developed for solving such constraint-based mining problems, especially for mining itemsets. The approach taken in the field of data mining contrasts with the constraint programming principles developed within the artificial intelligence community. While most data mining research focuses on algorithmic issues and aims at developing highly optimized and scalable implementations that are tailored towards specific tasks, constraint programming employs a more declarative approach. The emphasis lies on developing high-level modeling languages and general solvers that specify what the problem is, rather than outlining how a solution should be computed, yet are powerful enough to be used across a wide variety of applications and application domains.This paper contributes a declarative constraint programming approach to data mining. More specifically, we show that it is possible to employ off-the-shelf constraint programming techniques for modeling and solving a wide variety of constraint-based itemset mining tasks, such as frequent, closed, discriminative, and cost-based itemset mining. In particular, we develop a basic constraint programming model for specifying frequent itemsets and show that this model can easily be extended to realize the other settings. This contrasts with typical procedural data mining systems where the underlying procedures need to be modified in order to accommodate new types of constraint, or novel combinations thereof. Even though the performance of state-of-the-art data mining systems outperforms that of the constraint programming approach on some standard tasks, we also show that there exist problems where the constraint programming approach leads to significant performance improvements over state-of-the-art methods in data mining and as well as to new insights into the underlying data mining problems. Many such insights can be obtained by relating the underlying search algorithms of data mining and constraint programming systems to one another. We discuss a number of interesting new research questions and challenges raised by the declarative constraint programming approach to data mining.

The Top-k Frequent Closed Itemset Mining Using Top-k SAT Problem

“…Note that several data mining tasks are closely related to the itemset mining problem such as the ones of association rule mining, frequent pattern mining in sequence data, data clustering, etc. Recently, De Raedt et al in [24,25] proposed the first constraint programming (CP) based data mining framework for itemset mining. This new framework offers a declarative and flexible representation model.…”

Finding relational redescriptions

“…As each example has at most one associated transaction, the support of an itemset here is simply the number of distinct transactions containing it, that is, a frequent itemset covers at least γ transactions. Similarly to FIMPaths, FIMGraphs (line 6) combines the transactions T with a specification of this frequency constraint and passes them to the system of Guns et al (2011) to obtain the result. We do not include connectivity constraints, as those cannot be enforced at the time of mining here.…”

“…The system interprets the task as a constraint program, to which it finds all solutions by calling the constraint solver. Algorithm FIMPaths thus simply combines the transactions T with a specification of our support and connectivity constraints, passes them to the system of Guns et al (2011) to obtain the result, and transforms each itemset in the result into the corresponding path query as discussed above.…”

“…Any algorithm that solves this mining task could be used here. Our solution exploits a declarative approach to mining patterns under constraints (Guns et al, 2011), as the corresponding system allows for user-defined constraints on frequent itemset mining tasks. This is important for our approach, as the definition of support employed when mining for frequent path queries in T differs from the usual: the support of an itemset is not the number of transactions that contain it, but instead the number of distinct corresponding example pairs.…”

“…Furthermore, in order to maintain connectivity, we require that some item corresponding to an edge constraint must be present for each backbone edge. More specifically, the approach of Guns et al (2011) combines a declarative task specification language with a generic constraint programming solver. The user provides the transactions to mine itemsets from and specifies the desired constraints on itemsets in this language.…”