Aída Jiménez

Mining induced and embedded subtrees in ordered, unordered, and partially-ordered trees

Many data mining problems can be represented with non-linear data structures like trees. In this paper, we introduce a scalable algorithm to mine partially-ordered trees. Our algorithm, POTMiner, is able to identify both induced and embedded subtrees and, as special cases, it can handle both completely ordered and completely unordered trees (i.e. the particular situations existing algorithms address).

The design and use of the TMiner component-based data mining framework

This paper provides some practical guidelines for the design of data mining frameworks. It describes the rationale behind some of the key design decisions that guided the design, development, and implementation of the TMiner component-based data mining framework. TMiner is a flexible framework that can be used as a stand-alone tool or integrated into larger business intelligence (BI) solutions. TMiner is a general-purpose component-based system designed to support the whole KDD process into a single framework and thus facilitate the implementation of complex data mining scenarios.

Using trees to mine multirelational databases

This paper proposes a new approach to mine multirelational databases. Our approach is based on the representation of multirelational databases as sets of trees, for which we propose two alternative representation schemes. Tree mining techniques can thus be applied as the basis for multirelational data mining techniques, such as multirelational classification or multirelational clustering. We analyze the differences between identifying induced and embedded tree patterns in the proposed tree-based representation schemes and we study the relationships among the sets of tree patterns that can be discovered in each case. This paper also describes how these frequent tree patterns can be used, for instance, to mine association rules in multirelational databases.

POTMiner: mining ordered, unordered, and partially-ordered trees

Non-linear data structures are becoming more and more common in data mining problems. Trees, in particular, are amenable to efficient mining techniques. In this paper, we introduce a scalable and parallelizable algorithm to mine partially-ordered trees. Our algorithm, POTMiner, is able to identify both induced and embedded subtrees in such trees. As special cases, it can also handle both completely ordered and completely unordered trees.

Constraint-Based Pattern Mining in Multi-relational Databases

We propose a new framework for constraint-based pattern mining in multi-relational databases. Distinguishing features of the framework are that (1) it allows finding patterns not only under anti-monotonic constraints, but also under monotonic constraints and closed ness constraints, among others, expressed over complex aggregates over multiple relations, (2) it builds on a declarative graphical representation of constraints that links closely to data models of multi-relational databases and constraint networks in constraint programming, (3) it maps multi-relational pattern mining tasks into constraint programs. Our framework builds on a unifying perspective of multi-relational pattern mining, relational database technology and constraint networks in constraint programming. We demonstrate our framework on IMDB and Finance multi-relational databases.

Mining transposed motifs in music

The discovery of frequent musical patterns (motifs) is a relevant problem in musicology. This paper introduces an unsupervised algorithm to address this problem in symbolically-represented musical melodies. Our algorithm is able to identify transposed patterns including exact matchings, i.e., null transpositions. We have tested our algorithm on a corpus of songs and the results suggest that our approach is promising, specially when dealing with songs that include non-exact repetitions.

Mining frequent patterns from XML data: Efficient algorithms and design trade-offs

XML documents are now ubiquitous and their current applications are countless, from representing semi-structured documents to being the de facto standard for exchanging information. Viewed as partially-ordered trees, XML documents are amenable to efficient data mining techniques. In this paper, we describe how scalable algorithms can be used to mine frequent patterns from partially-ordered trees and discuss the trade-offs that are involved in the design of such algorithms.

Interestingness Measures for Association Rules within Groups

The study of association rules within groups of individuals in a database is interesting to define their characteristics and their behavior. In this paper, we define group association rules and we study interestingness measures for them. These evaluation measures can be used to rank groups of individuals and also rules within each group.

Hierarchical program representation for program element matching

Many intermediate program representations are used by compilers and other software development tools. In this paper, we propose a novel representation technique that, unlike those commonly used by compilers, has been explicitly designed for facilitating program element matching, a task at the heart of many software mining problems.

Mining patterns from longitudinal studies

Longitudinal studies are observational studies that involve repeated observations of the same variables over long periods of time. In this paper, we propose the use of tree pattern mining techniques to discover potentially interesting patterns within longitudinal data sets. Following the approach described in [15], we propose four different representation schemes for longitudinal studies and we analyze the kinds of patterns that can be identified using each one of the proposed representation schemes. Our analysis provides some practical guidelines that might be useful in practice for exploring longitudinal datasets.