Comparing Intended and Real Usage in Web Portal Temporal Logic and Data Mining.

Constraint%2Dbased Data Mining.

Data Mining Query Languages.
Data mining is the discovery of knowledge and useful information from the large amounts of data stored in databases. The emerging data mining tools and systems lead to the demand of a powerful data mining query language. The concepts of such a language for relational databases are discussed in [11]. With the increasing popularity of object-oriented databases, it is important to design a data mining query language for such databases.
Co%2Dclustering numerical data under user%2Ddefined constraints.
Abstract: In the generic setting of objects × attributes matrix data analysis, co-clustering appears as an interesting unsupervised data mining method. A co-clustering task provides a bi-partition made of co-clusters: each co-cluster is a group of objects associated to a group of attributes and these associations can support expert interpretations. Many constrained clustering algorithms have been proposed to exploit the domain knowledge and to improve partition relevancy in the mono-dimensional clustering case (e.g. using the must-link and cannot-link constraints on one of the two dimensions). Here, we consider constrained co-clustering not only for extended must-link and cannot-link constraints (i.e. both objects and attributes can be involved), but also for interval constraints that enforce properties of co-clusters when considering ordered domains. We describe an iterative co-clustering algorithm which exploits user-defined constraints while minimizing a given objective function. Thanks to a generic setting, we emphasize that different objective functions can be used. The added value of our approach is demonstrated on both synthetic and real data. Among others, several experiments illustrate the practical impact of this original co-clustering setting in the context of gene expression data analysis, and in an original application to a protein motif discovery problem. © 2009 Wiley
Advances in Intelligent Data Analysis VIII

Agglomerating local patterns hierarchically with ALPHA.
To increase the relevancy of local patterns discovered from noisy relations, it makes sense to formalize error-tolerance. Our starting point is to address the limitations of state-ofthe-art methods for this purpose. Some extractors perform an exhaustive search w.r.t. a declarative specification of error-tolerance. Nevertheless, their computational complexity prevents the discovery of large relevant patterns. Alpha is a 3-step method that (1) computes complete collections of closed patterns, possibly error-tolerant ones, from arbitrary n-ary relations, (2) enlarges them by hierarchical agglomeration, and (3) selects the relevant agglomerated patterns.
Extraction de motifs ferm%E9s dans des relations n%2Daires bruit%E9es.

Construction de descripteurs pour classer %E0 partir d%27exemples bruit%E9s.

Application%2DIndependent Feature Construction from Noisy Samples.
Abstract. When training classifiers, presence of noise can severely harm their performance. In this paper, we focus on “non-class ” attribute noise and we consider how a frequent fault-tolerant (FFT) pattern mining task can be used to support noise-tolerant classification. Our method is based on an application independent strategy for feature construction based on the so-called δ-free patterns. Our experiments on noisy training data shows accuracy improvement when using the computed features instead of the original ones. 1
Discovering Relevant Cross%2DGraph Cliques in Dynamic Networks.
Abstract. Several algorithms, namely CubeMiner, Trias, andData-Peeler, have been recently proposed to mine closed patterns in ternary relations. We consider here the specific context where a ternary relation denotes the value of a graph adjacency matrix at different timestamps. Then,wediscusstheconstraint-basedextractionofpatternsinsuchdynamic graphs. We formalize the concept of δ-contiguous closed 3-clique and we discuss the availability of a complete algorithm for mining them. It is based on a specialization of the enumeration strategy implemented in Data-Peeler. Indeed, clique relevancy can be specified by means of a conjunction of constraints which can be efficiently exploited. The added-value of our strategy is assessed on a real dataset about a public bicycle renting system. The raw data encode the relationships between the renting stations during one year. The extracted δ-contiguous closed 3-cliques are shown to be consistent with our domain knowledge on the considered city. 1
Application%2DIndependent Feature Construction from Noisy Samples.
Abstract. When training classifiers, presence of noise can severely harm their performance. In this paper, we focus on “non-class ” attribute noise and we consider how a frequent fault-tolerant (FFT) pattern mining task can be used to support noise-tolerant classification. Our method is based on an application independent strategy for feature construction based on the so-called δ-free patterns. Our experiments on noisy training data shows accuracy improvement when using the computed features instead of the original ones. 1


Discovery Science
Introduction  Discovery has always attracted people and has been a major motiveforce  of progress in human history. It has long been discussed in the philosophyof  science but it is only after 1960&#039;s that the problem of discovery became properly addressed by new philosophy of science group  6)  . In the last two decades, there has also been a considerable progress, in the field of artificial intelligence, of a domain often considered the realm of genius - empirical discovery  5)  . Historically, discovery tends to be viewed as a difficult form of learning from observation although learning and discovery convey rather different meanings  4)  . The former suggests a gradual process, while the latter suggests a more rapid mental event, often involving some form of insight. Learning may lead to an unconscious change in knowledge, while one is always aware that a discovery has been made. The results of learning can be declarative or procedura
A Parameter%2DFree Associative Classification Method.

Co%2Dclassification sous contraintes par la somme des r%E9sidus quadratiques.

If Constraint%2DBased Mining is the Answer What is the Constraint%3F %28Invited Talk%29.

Parameter Tuning for Differential Mining of String Patterns.
Constraint-based mining has been proven to be extremely useful for supporting actionable pattern discovery. However, useful conjunctions of constraints that support domain driven mining tasks generally need to set several parameter values and how to tune these parameters remains fairly open. We study this problem for substring pattern discovery, when using a conjunction of maximal frequency, minimal frequency and size constraints. We propose a method, based on pattern space sampling, to estimate the number of patterns that satisfy such conjunctions. This permits the user to probe the parameter space in many points, and then to choose some initial promising parameter settings. Our empirical validation confirms that we efficiently obtain good approximations of the number of patterns that will be extracted. 1.
Actionability and Formal Concepts A Data Mining Perspective.
Abstract. The last few years, we have studied different set pattern mining techniques from binary data. It includes the computation of formal concepts to support various knowledge discovery processes. For instance, when considering post-genomics, we can exploit Boolean data sets that encode a relation between some genes and the proteins that may regulate them. In such a context, it appears interesting to exploit the analogy between a putative transcriptional module (i.e., a typically important hypothesis for gene regulation understanding) and a formal concept that holds within such data. In this paper, we assume that knowledge nuggets can be captured by collections of formal concepts and we discuss the challenging issue of mining/selecting actionable patterns from these collections, i.e., looking for relevant patterns that really support knowledge discovery. Therefore, a major issue concerns the computation of complete collections of formal concepts that satisfy user-defined constraints. This is useful not only to avoid the computation of too small patterns that might be due to noise (e.g., using size constraints on both their intents and extents) but also to introduce some fault-tolerance. We discuss the pros and the cons of some recent proposals in that direction. 1
Feature Construction Based on Closedness Properties Is Not That Simple.
Abstract. Feature construction has been studied extensively, including for 0/1 data samples. Given the recent breakthrough in closedness-related constraint-based mining, we are considering its impact on feature construction for classification tasks. We investigate the use of condensed representations of frequent itemsets (closure equivalence classes) as new features. These itemset types have been proposed to avoid set counting in difficult association rule mining tasks. However, our guess is that their intrinsic properties (say the maximality for the closed itemsets and the minimality for the δ-free itemsets) might influence feature quality. Understanding this remains fairly open and we discuss these issues thanks to itemset properties on the one hand and an experimental validation on various data sets on the other hand. 1
Constrained Co%2Dclustering of Gene Expression Data.
In many applications, the expert interpretation of coclustering is easier than for mono-dimensional clustering. Co-clustering aims at computing a bi-partition that is a collection of co-clusters: each co-cluster is a group of objects associated to a group of attributes and these associations can support interpretations. Many constrained clustering algorithms have been proposed to exploit the domain knowledge and to improve partition relevancy in the mono-dimensional case (e.g., using the so-called must-link and cannot-link constraints). Here, we ... Here, we consider constrained co-clustering not only for extended must-link and cannot-link constraints (i.e., both objects and attributes can be involved), but also for interval constraints that enforce properties of co-clusters when considering ordered domains. We propose an iterative coclustering algorithm which exploits user-defined constraints while minimizing the sum-squared residues, i.e., an objective function introduced for gene expression data clustering by Cho et al. (2004). We illustrate the added value of our approach in two applications on gene expression data. 

Data Peeler Contraint%2DBased Closed Pattern Mining in n%2Dary Relations.


SQUAT A web tool to mine human%2C murine and avian SAGE data.

Pattern%2Dbased decision tree construction.
Learning classifiers has been studied extensively the last two decades. Recently, various approaches based on patterns (e.g., association rules) that hold within labeled data have been considered. In this paper, we propose a novel associative classification algorithm that combines rules and a decision tree structure. In a so-called δ-PDT (δ-Pattern Decision Tree), nodes are made of selected disjunctive δstrong classification rules. Such rules are generated from collections of δ-free patterns that can be computed efficiently. These rules have a minimal body, they are nonredundant and they avoid classification conflicts under a sensible condition on δ. We show that they also capture the discriminative power of emerging patterns. Our approach is empirically evaluated by means of a comparison to stateof-the-art proposals. 1.
Knowledge Discovery in Inductive Databases

Construction it%E9rative d%27un mod%E8le de connaissance par l%27exploitation de r%E8gles d%27association.

Introducing Softness into Inductive Queries on String Databases.

Feature Construction and delta%2DFree Sets in 0%2F1 Samples.

Feature Construction and delta%2DFree Sets in 0%2F1 Samples.

Utilisation des r%E9seaux bay%E9siens dans le cadre de l%27extraction de r%E8gles d%27association.

Am%E9lioration des indicateurs techniques pour l%27analyse du march%E9 financier.

Iterative Bayesian Network Implementation by Using Annotated Association Rules.
Abstract. This paper concerns the iterative implementation of a knowledge model in a data mining context. Our approach relies on coupling a Bayesian network design with an association rule discovery technique. First, discovered association rule relevancy isenhanced by exploiting the expert knowledge encoded within a Bayesian network, i.e., avoiding to provide trivial rules w.r.t. known dependencies. Moreover, the Bayesian network can be updated thanks to an expert-driven annotation process on computed association rules. Our approach is experimentally validated on the Asia benchmark dataset. 1
Mining a New Fault%2DTolerant Pattern Type as an Alternative to Formal Concept Discovery.
Abstract. Formal concept analysis has been proved to be useful to support knowledge discovery from boolean matrices. In many applications, such 0/1 data have to be computed from experimental data and it is common to miss some one values. Therefore, we extend formal concepts towards fault-tolerance. We define the DR-bi-set pattern domain by allowing some zero values to be inside the pattern. Crucial properties of formal concepts are preserved (number of zero values bounded on objects and attributes, maximality and availability of functions which “connect” the set components). DR-bi-sets are defined by constraints which are actively used by our correct and complete algorithm. Experimentation on both synthetic and real data validates the added-value of the DR-bi-sets. 1
Towards Constrained Co%2Dclustering in Ordered 0%2F1 Data Sets.
Abstract. Within 0/1 data, co-clustering provides a collection of biclusters, i.e., linked clusters for both objects and Boolean properties. Beside the classical need for grouping quality optimization, one can also use user-defined constraints to capture subjective interestingness aspects and thus to improve bi-cluster relevancy. We consider the case of 0/1 data where at least one dimension is ordered, e.g., objects denotes time points, and we introduce co-clustering constrained by interval constraints. Exploiting such constraints during the intrinsically heuristic clustering process is challenging. We propose one major step in this direction where bi-clusters are computed from collections of local patterns. We provide an experimental validation on two temporal gene expression data sets. 1
Looking for monotonicity properties of a similarity constraint on sequences.
Constraint-based mining techniques on sequence databases have been studied extensively the last few years and efficient algorithms enable to compute complete collections of patterns (e.g., sequences) which satisfy conjunctions of monotonic and/or anti-monotonic constraints. Studying new applications of these techniques, we believe that a primitive constraint which enforces enough similarity w.r.t a given reference sequence would be extremely useful and should benefit from such a recent algorithmic breakthrough. A non trivial similarity constraint is however neither monotonic nor anti-monotonic. Therefore, we have studied its definition as a conjunction of two constraints which satisfy the desired monotonicity properties: a pattern is called similar to a reference pattern x when its longest common subsequence with x (LCS) is large enough (i.e., a monotonic part) and when the number of deletions such that it becomes the LCS is small enough (i.e., an anti-monotonic part). We provide an experimental validation which confirms the added value of this approach on a biological database. Classical issues like scalability and pruning efficiency are discussed. 1.
Mining Bi%2Dsets in Numerical Data.
Abstract. Thanks to an important research effort the last few years, inductive queries on set patterns and complete solvers which can evaluate them on large 0/1 data sets have been proved extremely useful. However, for many application domains, the raw data is numerical (matrices of real numbers whose dimensions denote objects and properties). Therefore, using efficient 0/1 mining techniques needs for tedious Boolean property encoding phases. This is, e.g., the case, when considering microarray data mining and its impact for knowledge discovery in molecular biology. We consider the possibility to mine directly numerical data to extract collections of relevant bi-sets, i.e., couples of associated sets of objects and attributes which satisfy some user-defined constraints. Not only we propose a new pattern domain but also we introduce a complete solver for computing the so-called numerical bi-sets. Preliminary experimental validation is given. 1
Looking for monotonicity properties of a similarity constraint on sequences.
Constraint-based mining techniques on sequence databases have been studied extensively the last few years and efficient algorithms enable to compute complete collections of patterns (e.g., sequences) which satisfy conjunctions of monotonic and/or anti-monotonic constraints. Studying new applications of these techniques, we believe that a primitive constraint which enforces enough similarity w.r.t a given reference sequence would be extremely useful and should benefit from such a recent algorithmic breakthrough. A non trivial similarity constraint is however neither monotonic nor anti-monotonic. Therefore, we have studied its definition as a conjunction of two constraints which satisfy the desired monotonicity properties: a pattern is called similar to a reference pattern x when its longest common subsequence with x (LCS) is large enough (i.e., a monotonic part) and when the number of deletions such that it becomes the LCS is small enough (i.e., an anti-monotonic part). We provide an experimental validation which confirms the added value of this approach on a biological database. Classical issues like scalability and pruning efficiency are discussed. 1.
Supporting bi%2Dcluster interpretation in 0%2F1 data by means of local patterns.
Clustering or co-clustering techniques have been proved useful in many application domains. A weakness of these techniques remains the poor support for grouping characterization. As a result, interpreting clustering results and discovering knowledge from them can be quite hard. We consider potentially large Boolean data sets which record properties of objects and we assume the availability of a bi-partition which has to be characterized by means of a symbolic description. Our generic approach exploits collections of local patterns which satisfy some user-defined constraints in the data, and a measure of the accuracy of a given local pattern as a bi-cluster characterization pattern. We consider local patterns which are bi-sets, i.e., sets of objects associated to sets of properties. Two concrete examples are formal concepts (i.e., associated closed sets) and the so-called δ-bi-sets (i.e., an extension of formal concepts towards faulttolerance). We introduce the idea of characterizing query which can be used by experts to support knowledge discovery from bi-partitions thanks to available local patterns. The added-value is illustrated on benchmark data and three real data sets: a medical data set and two gene expression data sets. 1
Local Pattern Detection
 Separate-and-conquer or covering rule learning algorithms may be viewed as a technique for using local pattern discovery for generating a global theory. Local patterns are learned one at a time, and each pattern is evaluated in a local context, with respect to the number of positive and negative examples that it covers. Global context is provided by removing the examples that are covered by previous patterns before learning a new rule. In this paper, we discuss several research issues that arise in this context. We start with a brief discussion of covering algorithms, their problems, and review a few suggestions for resolving them. We then discuss the suitability of a well-known family of evaluation metrics, and analyze how they trade off coverage and precision of a rule. Our conclusion is that in many applications, coverage is only needed for establishing statistical significance, and precision is the metric that should be optimized for rules. The main problem with optimizing precision is its unreliability for low example sizes, which is mainly caused by overfitting. We then report some preliminary experiments that addresses this problem by meta-learning a predictor for the true accuracy of a rule based on its coverage on the training set.  
Constraint%2DBased Mining and Inductive Databases

Towards Fault%2DTolerant Formal Concept Analysis.
Abstract. GivenBooleandatasetswhichrecordpropertiesofobjects, Formal Concept Analysis is a well-known approach for knowledge discovery. Recent application domains, e.g., for very large data sets, have motivated new algorithms which can perform constraint-based mining of formal concepts (i.e., closed sets on both dimensions which are associated by the Galois connection and satisfy some user-defined constraints). In this paper, we consider a major limit of these approaches when considering noisy data sets. This is indeed the case of Boolean gene expression data analysis where objects denote biological experiments and attributes denote gene expression properties. In this type of intrinsically noisy data, the Galois association is so strong that the number of extracted formal concepts explodes. We formalize the computation of the so-called δ-bisets as an alternative for capturing strong associations between sets of objects and sets of properties. Based on a previous work on approximate condensed representations of frequent sets by means of δ-free itemsets, we get an efficient technique which can be applied on large data sets. An experimental validation on both synthetic and real data is given. It confirms the added-value of our approach w.r.t. formal concept discovery, i.e., the extraction of smaller collections of relevant associations. 1
Approximation de collections de concepts formels par des bi%2Densembles denses et pertinents.

From Local Pattern Mining to Relevant Bi%2Dcluster Characterization.
Abstract. Clustering or bi-clustering techniques have been proved quite useful in many application domains. A weakness of these techniques remains the poor support for grouping characterization. We consider eventually large Boolean data sets which record properties of objects and we assume that a bi-partition is available. We introduce a generic cluster characterization technique which is based on collections of bi-sets (i.e., sets of objects associated to sets of properties) which satisfy some userdefined constraints, and a measure of the accuracy of a given bi-set as a bi-cluster characterization pattern. The method is illustrated on both formal concepts (i.e., “maximal rectangles of true values”) and the new type of δ-bi-sets (i.e., “rectangles of true values with a bounded number of exceptions per column”). The added-value is illustrated on benchmark data and two real data sets which are intrinsically noisy: a medical data about meningitis and Plasmodium falciparum gene expression data. 1
Constraint%2DBased Mining of Fault%2DTolerant Patterns from Boolean Data.
Abstract. Thanks to an important research effort the last few years, inductive queries on local patterns (e.g., set patterns) and complete solvers which can evaluate them on large data sets have been proved extremely useful. The more we use such queries on real-life data, e.g., biological data (and thus intrinsically dirty and noisy), the more we are convinced that inductive queries should return fault-tolerant patterns. In this work, we consider user-defined constraints for a declarative specification of faulttolerance. We discuss the design of such constraints on bi-sets extracted from Boolean data sets. Our starting point is the fundamental limitation of formal concept discovery (i.e., closed set mining) from noisy data and we propose a constraint-based mining approach for relevant faulttolerant bi-set mining. Formalizing three recent proposals, our framework enables a better understanding of the needed trade-off between extraction feasibility, completeness, relevancy, and ease of interpretation of these fault-tolerant patterns. An original empirical evaluation on both synthetic and real-life medical data is given. It enables a comparison of the various proposals and it motivates further directions of research. 1
A Bi%2Dclustering Framework for Categorical Data.
Abstract. Bi-clustering is a promising conceptual clustering approach. Within categorical data, it provides a collection of (possibly overlapping) bi-clusters, i.e., linked clusters for both objects and attribute-value pairs. We propose a generic framework for bi-clustering which enables to compute a bi-partition from collections of local patterns which capture locally strong associations between objects and properties. To validate this framework, we have studied in details the instance CDK-Means. It is a K-Means-like clustering on collections of formal concepts, i.e., connected closed sets on both dimensions. It enables to build bi-partitions with a user control on overlapping between bi-clusters. We provide an experimental validation on many benchmark datasets and discuss the interestingness of the computed bi-partitions. 1
Constraint%2DBased Data Mining.

Data Mining Query Languages.
Data mining is the discovery of knowledge and useful information from the large amounts of data stored in databases. The emerging data mining tools and systems lead to the demand of a powerful data mining query language. The concepts of such a language for relational databases are discussed in [11]. With the increasing popularity of object-oriented databases, it is important to design a data mining query language for such databases.
Constraint%2Dbased concept mining and its application to microarray data analysis.
data analysis
Machine Learning ECML 2004

Knowledge Discovery in Databases PKDD 2004

Assessment of discretization techniques for relevant pattern discovery from gene expression data.
In the domain of gene expression data analysis, various researchers have recently emphasized the promising application of pattern discovery techniques like association rule mining or formal concept extraction from boolean matrices that encode gene properties. To take the most from these approaches, a needed step concerns gene property encoding (e.g., over-expression) and its need for the discretization of raw gene expression data. The impact of this preprocessing step on both the quantity and the relevancy of the extracted patterns is crucial. In this paper, we study the impact of discretization parameters by a sound comparison between the dendrograms, i.e., trees that are generated by a hierarchical clustering algorithm, computed from raw expression data and from the various derived boolean matrices. Thanks to a new similarity measure and practical validation over several gene expression data sets, we propose a method that supports the choice of a discretization technique and its parameters for each specific data set. 1.
Contribution to Gene Expression Data Analysis by Means of Set Pattern Mining.

A Survey on Condensed Representations for Frequent Sets.
Abstract. Solving inductive queries which have to return complete collections of patterns satisfying a given predicate has been studied extensively the last few years. The specific problem of frequent set mining from potentially huge boolean matrices has given rise to tens of efficient solvers. Frequent sets are indeed useful for many data mining tasks, including the popular association rule mining task but also feature construction, association-based classification, clustering, etc. The research in this area has been boosted by the fascinating concept of condensed representations w.r.t. frequency queries. Such representations can be used to support the discovery of every frequent set and its support without looking back at the data. Interestingly, the size of condensed representations can be several orders of magnitude smaller than the size of frequent set collections. Most of the proposals concern exact representations while it is also possible to consider approximated ones, i.e., to trade computational complexity with a bounded approximation on the computed support values. This paper surveys the core concepts used in the recent works on condensed representation for frequent sets. 1
Inductive Databases and Multiple Uses of Frequent Itemsets The cInQ Approach.
Abstract. Inductive databases (IDBs) have been proposed to afford the problem of knowledge discovery from huge databases. With an IDB the user/analyst performs a set of very different operations on data using a query language, powerful enough to perform all the required elaborations, such as data preprocessing, pattern discovery and pattern postprocessing. We present a synthetic view on important concepts that have been studied within the cInQ European project when considering the pattern domain of itemsets. Mining itemsets has been proved useful not only for association rule mining but also feature construction, classification, clustering, etc. We introduce the concepts of pattern domain, evaluation functions, primitive constraints, inductive queries and solvers for itemsets. We focus on simple high-level definitions that enable to forget about technical details that the interested reader will find, among others, in cInQ publications. 1
Query Languages Supporting Descriptive Rule Mining A Comparative Study.

A Methodology for Biologically Relevant Pattern Discovery from Gene Expression Data.
Abstract. One of the most exciting scientific challenges in functional genomics concerns the discovery of biologically relevant patterns from gene expression data. For instance, it is extremely useful to provide putative synexpression groups or transcription modules to molecular biologists. We propose a methodology that has been proved useful in real cases. It is described as a prototypical KDD scenario which starts from raw expression data selection until useful patterns are delivered. Our conceptual contribution is (a) to emphasize how to take the most from recent progress in constraint-based mining of set patterns, and (b) to propose a generic approach for gene expression data enrichment. The methodology has been validated on real data sets. 1
Mining Formal Concepts with a Bounded Number of Exceptions from Transactional Data.
Abstract. We are designing new data mining techniques on boolean contexts to identify a priori interesting bi-sets (i.e., sets of objects or transactions associated to sets of attributes or items). A typical important case concerns formal concept mining (i.e., maximal rectangles of true values or associated closed sets by means of the so-called Galois connection). It has been applied with some success to, e.g., gene expression data analysis where objects denote biological situations and attributes denote gene expression properties. However in such real-life application domains, it turns out that the Galois association is a too strong one when considering intrinsically noisy data. It is clear that strong associations that would however accept a bounded number of exceptions would be extremely useful. We study the new pattern domain of α/β concepts, i.e., consistent maximal bi-sets with less than α false values per row and less than β false values per column. We provide a complete algorithm that computes all the α/β concepts based on the generation of concept unions pruned thanks to anti-monotonic constraints. An experimental validation on synthetic data is given. It illustrates that more relevant associations can be discovered in noisy data. We also discuss a practical application in molecular biology that illustrates an incomplete but quite useful extraction when all the concepts that are needed beforehand can not be discovered. 1
Boolean Property Encoding for Local Set Pattern Discovery An Application to Gene Expression Data Analysis.
Abstract. In the domain of gene expression data analysis, several researchers have recently emphasized the promising application of local pattern (e.g., association rules, closed sets) discovery techniques from boolean matrices that encode gene properties. Detecting local patterns by means of complete constraint-based mining techniques turns to be an important complementary approach or invaluable counterpart to heuristic global model mining. To take the most from local set pattern mining approaches, a needed step concerns gene expression property encoding (e.g., over-expression). The impact of this preprocessing phase on both the quantity and the quality of the extracted patterns is crucial. In this paper, we study the impact of discretization techniques by a sound comparison between the dendrograms, i.e., trees that are generated by a hierarchical clustering algorithm on raw numerical expression data and its various derived boolean matrices. Thanks to a new similarity measure, we can select the boolean property encoding technique which preserves similarity structures holding in the raw data. The discussion relies on several experimental results for three gene expression data sets. We believe our framework is an interesting direction of work for the many application domains in which (a) local set patterns have been proved useful, and (b) Boolean properties have to be derived from raw numerical data. 1
Constraint%2DBased Mining of Formal Concepts in Transactional Data.
Abstract. We are designing new data mining techniques on boolean contexts to identify a priori interesting concepts, i.e., closed sets of objects (or transactions) and associated closed sets of attributes (or items). We propose a new algorithm D-Miner for mining concepts under constraints. We provide an experimental comparison with previous algorithms and an application to an original microarray dataset for which D-Miner is the only one that can mine all the concepts.
Using Classification and Visualization on Pattern Databases for Gene Expression Data Analysis.

Optimizing subset queries a step towards SQL%2Dbased inductive databases for itemsets.

Introduction.
The use of monads to structure functional programs is described. Monads provide a convenient framework for simulating e ects found in other languages, such as global state, exception handling, output, or non-determinism. Three case studies are looked at in detail: how monads ease the modi cation of a simple evaluator; how monads act as the basis of a datatype of arrays subject to in-place update; and how monads can be used to build parsers.  
Proceedings of the Second International Workshop on Inductive Databases

Using transposition for pattern discovery from microarray data.
We analyze expression matrices to identify a priori interesting sets of genes, e.g., genes that are frequently co-regulated. Such matrices provide expression values for given biological situations (the lines) and given genes (columns). The frequent itemset (sets of columns) extraction technique enables to process di#cult cases (millions of lines, hundreds of columns) provided that data is not too dense. However, expression matrices can be dense and have generally only few lines w.r.t. the number of columns. Known algorithms, including the recent algorithms that compute the so-called condensed representations can fail. Thanks to the properties of Galois connections, we propose an original technique that processes the transposed matrices while computing the sets of genes. We validate the potential of this framework by looking for the closed sets in two microarray data sets.
Comprehensive Log Compression with Frequent Patterns.
Abstract. In this paper we present a comprehensive log compression (CLC) methodthat uses frequent patterns andtheir condensedrepresentations to identify repetitive information from large log files generated by communications networks. We also show how the identified information can be usedto separate andfilter out frequently occurring events that hide other, unique or only a few times occurring events. The identification can be done without any prior knowledge about the domain or the events. For example, no pre-defined patterns or value combinations are needed. This separation makes it easier for a human observer to perceive andanalyse large amounts of log data. The applicability of the CLC methodis demonstratedwith real-worldexamples from data communication networks. 1

Mining Concepts from Large SAGE Gene Expression Matrices.
Abstract. One of the crucial needs in post-genomic research is to analyze expression matrices (e.g., SAGE and microarray data) to identify a priori interesting sets of genes, e.g., sets of genes that are frequently co-regulated. Such matrices provide expression values for given biological situations (the lines) and given genes (columns). The inductive database framework enables to support knowledge discovery processes by means of sequences of queries that concerns both data processing and pattern querying (extraction, post-processing). We provide a simple formalization of a relevant pattern domain (language of patterns, evaluation functions and primitive constraints) that has been proved useful for specifying various analysis tasks. Recent algorithmic results w.r.t. the efficient evaluation (constraint-based mining) of the so-called inductive queries are emphasized and illustrated on a 90 × 12 636 human SAGE expression matrix. 1

GO%2DSPADE Mining Sequential Patterns over Datasets with Consecutive Repetitions.
Abstract. Databases of sequences can contain consecutive repetitions of items. This is the case in particular when some items represent discretized quantitative values. We show that on such databases, a typical algorithm like the SPADE algorithm tends to loose its efficiency. SPADE is based on the used of lists containing the localization of the occurrences of a pattern in the sequences and these lists are not appropriated in the case of data with repetitions. We introduce the concept of generalized occurrences and the corresponding primitive operators to manipulate them. We present an algorithm called GO-SPADE that extends SPADE to incorporate generalized occurrences. Finally we present experiments showing that GO-SPADE can handle sequences containing consecutive repetitions at nearly no extra cost.
Constraint%2DBased Mining of Sequential Patterns over Datasets with Consecutive Repetitions.
Abstract. Constraint-based mining of sequential patterns is an active research area motivated by many application domains. In practice, the real sequence datasets can present consecutive repetitions of symbols (e.g., DNA sequences, discretized stock market data) that can lead to a very important consumption of resources during the extraction of patterns that can turn even efficient algorithms to become unusable. We propose a constraint-based mining algorithm using an approach that enables to compact these consecutive repetitions, reducing drastically the amount of data to process and speeding-up the extraction time. The technique introduced in this paper allows to retain the advantages of existing state-of-the-art algorithms based on the notion of occurrence lists, while permitting to extend their application fields to datasets containing consecutive repetitions. We analyze the benefits obtained using synthetic datasets, and show that the approach is of practical interest on real datasets.
Mining Frequent Sequential Patterns under Regular Expressions A Highly Adaptive Strategy for Pushing Contraints.

A Comparison between Query Languages for the Extraction of Association Rules.
Abstract. Recently inductive databases (IDBs) have been proposed to afford the problem of knowledge discovery from huge databases. With an IDB the user/analyst performs a set of very different operations on data using a special-purpose language, powerful enough to perform all the required manipulations, such as data preprocessing, pattern discovery and pattern post-processing. In this paper we present a comparison between query languages (MSQL, DMQL and MINE RULE) that have been proposed for association rules extraction in the last years and discuss their common features and differences. We present them using a set of examples, taken from the real practice of data mining. This allows us to define the language design guidelines, with particular attention to the open issues on IDBs. 1
Free%2DSets A Condensed Representation of Boolean Data for the Approximation of Frequency Queries.
Abstract. Given a large collection of transactions containing items, a basic common data mining problem is to extract the so-called frequent itemsets (i.e., sets of items appearing in at least a given number of transactions). In this paper, we propose a structure called free-sets, from which we can approximate any itemset support (i.e., the number of transactions containing the itemset) and we formalize this notion in the framework of ɛ-adequate representations (H. Mannila and H. Toivonen, 1996. In Proc. of the Second International Conference on Knowledge Discovery and Data Mining (KDD’96), pp. 189–194). We show that frequent free-sets can be efficiently extracted using pruning strategies developed for frequent itemset discovery, and that they can be used to approximate the support of any frequent itemset. Experiments on real dense data sets show a significant reduction of the size of the output when compared with standard frequent itemset extraction. Furthermore, the experiments show that the extraction of frequent free-sets is still possible when the extraction of frequent itemsets becomes intractable, and that the supports of the frequent free-sets can be used to approximate very closely the supports of the frequent itemsets. Finally, we consider the effect of this approximation on association rules (a popular kind of patterns that can be derived from frequent itemsets) and show that the corresponding errors remain very low in practice.
A Comparison between Query Languages for the Extraction of Association Rules.
Abstract. Recently inductive databases (IDBs) have been proposed to afford the problem of knowledge discovery from huge databases. With an IDB the user/analyst performs a set of very different operations on data using a special-purpose language, powerful enough to perform all the required manipulations, such as data preprocessing, pattern discovery and pattern post-processing. In this paper we present a comparison between query languages (MSQL, DMQL and MINE RULE) that have been proposed for association rules extraction in the last years and discuss their common features and differences. We present them using a set of examples, taken from the real practice of data mining. This allows us to define the language design guidelines, with particular attention to the open issues on IDBs. 1
Extraction de motifs s%E9quentiels sous contraintes de similarit%E9.

Signatures de situations boursi%E8res repr%E9sent%E9es par des s%E9quences d%27%E9v%E9nements.

Mining Frequent Sequential Patterns under a Similarity Constraint.
Abstract. Many practical applications are related to frequent sequential pattern mining, ranging from Web Usage Mining to Bioinformatics. To ensure an appropriate extraction cost for useful mining tasks, a key issue is to push the user-defined constraints deep inside the mining algorithms. In this paper, we study the search for frequent sequential patterns that are also similar to an user-defined reference pattern. While the effective processing of the frequency constraints is well-understood, our contribution concerns the identification of a relaxation of the similarity constraint into a convertible anti-monotone constraint. Both constraints are then used to prune the search space during a levelwise search. Preliminary experimental validations have confirmed the algorithm efficiency. 1
Using Condensed Representations for Interactive Association Rule Mining.
Abstract. Association rule mining is a popular data mining task. It has an interactive and iterative nature, i.e., the user has to refine his mining queries until he is satisfied with the discovered patterns. To support such an interactive process, we propose to optimize sequences of queries by means of a cache that stores information from previous queries. Unlike related works, we use condensed representations like free and closed itemsets for both data mining and caching. This results in a much more efficient mining technique in highly correlated data and a much smaller cache than in previous approaches.
Constraint%2DBased Discovery and Inductive Queries Application to Association Rule Mining.
Abstract. Recently inductive databases (IDBs) have been proposed to afford the problem of knowledge discovery from huge databases. Querying these databases needs for primitives to: (1) select, manipulate and query data, (2) select, manipulate and query “interesting ” patterns (i.e., those patterns that satisfy certain constraints), and (3) cross over patterns and data (e.g., selecting the data in which some patterns hold). Designing such query languages is a long-term goal and only preliminary approaches have been studied, mainly for the association rule mining task. Starting from a discussion on the MINE RULE operator, we identify several open issues for the design of inductive databases dedicated to these descriptive rules. These issues concern not only the offered primitives but also the availability of efficient evaluation schemes. We emphasize the need for primitives that work on more or less condensed representations for the frequent itemsets, e.g., the (frequent) δ-free and closed itemsets. It is useful not only for optimizing single association rule mining queries but also for sophisticated post-processing and interactive rule mining. 1
Optimization of association rule mining queries.
Levelwise algorithms (e.g., the Apriori algorithm) have been proved effective for association rule mining from sparse data. However, in many practical applications, the computation turns to be intractable for the user-given frequency threshold and the lack of focus leads to huge collections of frequent itemsets. To tackle these problems, two promising issues have been investigated during the last four years: the efficient use of user defined constraints and the computation of condensed representations for frequent itemsets, e.g., the frequent closed sets. We show that the benefits of these two approaches can be combined into a levelwise algorithm. It can be used for the discovery of association rules in difficult cases (dense and highly correlated data). For instance, we report an experimental validation related to the discovery of association rules with negations.
Query%2DDriven Knowledge Discovery via OLAP manipulations.

Towards the Tractable Discovery of Association Rules with Negations.
Abstract. Frequent association rules (e.g., A∧B ⇒ C to say that when properties A and B are true in a record then, C tends to be also true) have become a popular way to summarize huge datasets. The last 5 years, there has been a lot of research on association rule mining and more precisely, the tractable discovery of interesting rules among the frequent ones. We consider now the problem of mining association rules that may involve negations e.g., A ∧ B ⇒ ¬C or ¬A ∧ B ⇒ C. Mining such rules is difficult and remains an open problem. We identify several possibilities for a tractable approach in practical cases. Among others, we discuss the active use of constraints. We propose a generic algorithm and discuss the use of constraints to mine the generalized sets from which rules with negations can be derived. 1
Mining Free Itemsets under Constraints.
Computing frequent itemsets and their frequencies from large boolean matrices (e.g., to derive association rules) has been one of the hot topics in data mining. Levelwise algorithms (e.g., the APRIORI algorithm) have been proved effective for frequent itemset mining from sparse data. However, in many practical applications, the computation turns to be intractable for the user-given frequency threshold and the lack of focus leads to huge collections of frequent itemsets. The last three years, two promising issues have been investigated: the use of user defined constraints and closed sets mining. To the best of our knowledge, combining these two frameworks has not been studied yet. In this paper, we show that the benefit of these two approaches can be combined into levelwise algorithms. An experimental validation related to the discovery of association rules with negations is reported. 1.
Association Rule Discovery in Highly%2DCorrelated Data a Case Study in Web Usage Mining.

Using Constraints during Frequent Set Mining a Generic Approach.

Towards the Tractable Discovery of Association Rules with Negations.
Abstract. Frequent association rules (e.g., A∧B ⇒ C to say that when properties A and B are true in a record then, C tends to be also true) have become a popular way to summarize huge datasets. The last 5 years, there has been a lot of research on association rule mining and more precisely, the tractable discovery of interesting rules among the frequent ones. We consider now the problem of mining association rules that may involve negations e.g., A ∧ B ⇒ ¬C or ¬A ∧ B ⇒ C. Mining such rules is difficult and remains an open problem. We identify several possibilities for a tractable approach in practical cases. Among others, we discuss the active use of constraints. We propose a generic algorithm and discuss the use of constraints to mine the generalized sets from which rules with negations can be derived. 1
Frequent Closures as a Concise Representation for Binary Data Mining.
Frequent set discovery from binary data is an important  problem in data mining. It concerns the discovery of a concise representation  of large tables from which descriptive rules can be derived,  e.g., the popular association rules. Our work concerns the study of two  representations, namely frequent sets and frequent closures. N. Pasquier  and colleagues designed the    algorithm that provides frequent sets  via the discovery of frequent closures. When one mines highly correlated  data,    algorithms clearly fail while    remains  tractable. We discuss our implementation of    and the experimental  evidence we got from two real-life binary data mining processes. Then,  we introduce the concept of almost-closure (generation of every frequent  set from frequent almost-closures remains possible but with a bounded  error on frequency). To the best of our knowledge, this is a new concept  and, here again, we provide some experimental evidence of its add-value.
Approximation of Frequency Queris by Means of Free%2DSets.
Abstract. Given a large collection of transactions containing items, a basic common data mining problem is to extract the so-called frequent itemsets (i.e., set of items appearing in at least a given number of transactions). In this paper, we propose a structure called free-sets, from which we can approximate any itemset support (i.e., the number of transactions containing the itemset) and we formalize this notion in the framework of ɛ-adequate representation [10].We show that frequent free-sets can be efficiently extracted using pruning strategies developed for frequent itemset discovery, and that they can be used to approximate the support of any frequent itemset. Experiments run on real dense data sets show a significant reduction of the size of the output when compared with standard frequent itemsets extraction. Furthermore, the experiments show that the extraction of frequent free-sets is still possible when the extraction of frequent itemsets becomes intractable. Finally, we show that the error made when approximating frequent itemset support remains very low in practice. 1
Query Driven Knowledge Discovery in Multidimensional Data.
We study KDD (Knowledge Discovery in Databases) processes on multidimensional data from a query point of view. Focusing on association rule mining, we consider typical queries to cope with the pre-processing of multidimensional data and the post-processing of the discovered patterns as well. We use a model and a rule-based language stemming from the OLAP multidimensional representation, and demonstrate that such a language ts well for writing KDD queries on multidimensional data. Using an homogeneous data model and our language for expressing queries at every phase of the process appears as a valuable step towards a better understanding of interactivity during the whole process.  1 Introduction  Discovering knowledge from data appears as a complex iterative and interactive process containing many steps: understanding the data, preparing the data set (also called pre-processing), discovering potentially interesting patterns (mining phase), post-processing of discovered patterns and na...
Modeling KDD Processes within the Inductive Database Framework.
One of the most challenging problems in data manipulation in the future is to be able to e ciently handle very large databases but also multiple induced properties or generalizations in that data. Popular examples of useful properties are association rules, and inclusion and functional dependencies. Our view of a possible approach for this task is to specify and query inductive databases, which are databases that in addition to data also contain intensionally de ned generalizations about the data. We formalize this concept and show howitcan be used throughout the whole process of data mining due to the closure property of the framework. We show that simple query languages can be de ned using normal database terminology. We demonstrate the use of this framework to model typical data mining processes. It is then possible to perform various tasks on these descriptions like, e.g., optimizing the selection of interesting properties or comparing two processes. 
Query Languages for Knowledge Discovery in Databases.

Spreadsheet Generation from Rule%2DBased Specifications.
Experimental studies have pointed out that operational spreadsheets contain a lot of errors. In this paper, we propose to use a multidimensional rule-based data manipulation language to specify spreadsheets. This language allows formal declarative de nition of spreadsheet tabular data treatments. The main contribution of this paper is to show that from textual rule-based speci cations, we can generate the corresponding spreadsheet. This generation of a spreadsheet from its logical de nition is a promising way to reduce encoding errors. The technique presented has been implemented in a crude running prototype. 1
Querying Inductive Databases A Case Study on the MINE RULE Operator.
. Knowledge discovery in databases (KDD) is a process that can include steps like forming the data set, data transformations, discovery of patterns, searching for exceptions to a pattern, zooming on a subset of the data, and postprocessing some patterns. We describe a comprehensive framework in which all these steps can be carried out by means of queries over an inductive database. An inductive database is a database that in addition to data also contains intensionally defined generalizations about the data. We formalize this concept: an inductive database consists of a normal database together with a subset of patterns from a class of patterns, and an evaluation function that tells how the patterns occur in the data. Then, looking for potential query languages built on top of SQL, we consider the research on the MINE RULE  operator by Meo, Psaila and Ceri. It is a serious step towards an implementation framework for inductive databases, though it addresses only the association rule mi...
Towards the Reverse Engineering of Denormalized Relational Databases.
This paper describes a method to cope with denor-malized relational schemas in a database reverse en-gineering process. We propose two main steps to irn-prove the understanding of data semantics. Firstly we extract inclusion dependencies by analyzing the equi-join queries embedded in application progranis and hy querying the database extension. Secondly we show how to discover only functional dependencies uhicli in-fluence the way attributes should be restructured. The method is interactive since an expert user has to val-idate the presumptions on the elicited dependencies. Moreover, a restructuring phase leads to a relational schema an third normal form provided with key con-straints and referential integrity constraints. Finally, we sketch how an Entity-Relationship schema can be derived from such information. 1
Abduction et d%E9duction de structures d%27objets une int%E9gration multiparadigmes.

Une approche multi%2Dparadigmes our le test d%27applications BDOO.

Using Queries to Improve Database Reverse Engineering.

F%2DLogic Programming and Terminological Constraints.

Vers une typologie des s%E9mantiques op%E9rationnelles pour les extensions de Prolog vers les objets.

Checking and Debugging of Two%2Dlevel Grammars.

STARLET An Affix%2DBased Compiler Compiler Designed as a Logic Programming System.

AFFLOG une implantation de grammaires %E0 deux niveaux pour l%27%E9tude de la Programmation Grammaticale Logique.

STARLET un langage pour une programmation logique fiable.

