Pattern sampling algorithms produce interesting patterns with a probability proportional to a given utility measure. Utility changes need quick repreprocessing when sampling patterns from large databases. In this context, existing sampling techniques require storing all data in memory, which is costly. To tackle these issues, this work enriches D. Knuth’s trie structure, avoiding 1) the need to access the database to sample since patterns are drawn directly from the enriched trie and 2) the necessity to reprocess the whole dataset when utility changes. We define the trie of occurrences that our first algorithm TPSpace (Trie-based Pattern Space) uses to materialize all of the database patterns. Factorizing transaction prefixes compresses the transactional database. TPSampling (Trie-based Pattern Sampling), our second algorithm, draws patterns from a trie of occurrences under a length-based utility measure. Experiments show that TPSampling produces thousands of patterns in seconds.

This paper describes a new semi-supervised clustering algorithm as part of a more general framework of interactive exploratory clustering, that favors the exploration of possible clustering solutions so that an expert tailors the best clustering according to her domain knowledge and preferences. Co ntrary to most existing approaches, the novel algorithm considers the feature space as a first class citizen for the exploration of alternative solutions. Our proposal represents and integrates quantitative preferences on attributes that will guide the exploration of possible solutions by learning an appropriate space metric. It also achieves a compromise clustering based on expert confidence, between a data-driven and a user-driven solution and converges with a good complexity. We show experimentally that our method is also able to deal with irrelevant user preferences and correct those choices in order to achieve a better solution. Experiments show that the best results may be achieved only with the addition of preferences to traditional metric learning algorithms and that our approach performs better than state-of-the-art algorithms.

In recent years, the field of pattern mining has shifted to user-centered methods. In such a context, it is necessary to have a tight coupling between the system and the user where mining techniques provide results at any time or within a short response time of only few seconds. Pattern sampling is a non-exhaustive method for instantly discovering relevant patterns that ensures a good interactivity while providing strong statistical guarantees due to its random nature. Curiously, such an approach investigated for itemsets and subgraphs has not yet been applied to sequential patterns, which are useful for a wide range of mining tasks and application fields. In this paper, we propose the first method for sequential pattern sampling. In addition to address sequential data, the originality of our approach is to introduce a constraint on the norm to control the length of the drawn patterns and to avoid the pitfall of the "long tail" where the rarest patterns flood the user. We propose a new constrained two-step random procedure, named CSSampling, that randomly draws sequential patterns according to frequency with an interval constraint on the norm. We demonstrate that this method performs an exact sampling. Moreover, despite the use of rejection sampling, the experimental study shows that CSSampling remains efficient and the constraint helps to draw general patterns of the "head". We also illustrate how to benefit from these sampled patterns to instantly build an associative classifier dedicated to sequences. This classification approach rivals state of the art proposals showing the interest of constrained sequential pattern sampling.

Pattern mining in numerical data remains a challenging task due to the pattern search space that becomes potentially infinite with real-valued dimensions. Most approaches reluctantly reduced the expressiveness of mined patterns to make possible extraction. Despite this expressiveness loss, they do not provide results within a short response time of a few seconds. This paper addresses the instant discovery of patterns in numerical data based on sampling techniques. Instead of splitting each dimension into intervals, we use a metric to introduce the density as new interestingness measure, and to define neighborhood patterns. The language of neighborhood patterns is semantically rich but in return, its size is infinite. We then present a new exact and non-enumerative random procedure to sample this infinite language according to density. An experimental study demonstrates the good compromise between precision and diversity of neighborhood patterns. Finally, in the context of associative classification, we show that a sample of neighborhood patterns is as accurate as traditional methods that traverses the entire search space.

This paper proposes a new semi-supervised clustering framework to represent and integrate quantitative preferences on attributes. A new metric learning algorithm is derived that achieves a compromise clustering between a data-driven and a user-driven solution and converges with a good complexity. We observe experimentally that the addition of preferences may be essential to achieve a better clustering. We also show that our approach performs better than the state-of-the art algorithms.

Recommending database queries is an emerging and promising field of investigation. This is of particular interest in the domain of OLAP systems where the user is left with the tedious process of navigating large datacubes. In this paper we present a framework for a recommender system for OLAP users, that leverages former users' investigations to enhance discovery driven analysis. The main idea is to recommend to the user the discoveries detected in those former sessions that investigated the same unexpected data as the current session.

An OLAP analysis session can be defined as an interactive session during which a user launches queries to navigate within a cube. Very often choosing which part of the cube to navigate further, and thus designing the forthcoming query, is a difficult task. In this paper, we propose to use what the OLAP users did during their former exploration of the cube as a basis for recommending OLAP queries to the user. We present a generic framework that allows to recommend OLAP queries based on the OLAP server query log. This framework is generic in the sense that changing its parameters changes the way the recommendations are computed. We show how to use this framework for recommending simple MDX queries and we provide some experimental results to validate our approach.

Most research on preference elicitation, preference reasoning and preference query languages design focus mainly on preferences over single objects represented by relational tuples. An increasing interest on preferences over more complex structures like sets of objects has arised in recent papers. However, most recent applications deal with more sophisticated complex structured objects, like sequences, trees and graphs. In this paper, we introduce TPref, a formalism to reason with qualitative conditional preferences over sequences of objects indexed in time. TPref generalizes the CP-Nets formalism by allowing temporal conditional preferences besides the static rules used in CP-Nets. An algorithm for computing optimal sequences of objects satisfying a set of temporal constraints is also presented.

OLAP users heavily rely on visualization of query answers for their interactive analysis of massive amounts of data. Very often, these answers cannot be visualized entirely and the user has to navigate through them to find relevant facts.In this paper, we propose a framework for personalizing OLAP queries. In this framework, the user is asked to give his (her) preferences and a visualization constraint, that can be for instance the limitations imposed by the device used to display the answer to a query. Given this, for each query, our method computes the part of the answer that respects both the user preferences and the visualization constraint. In addition, a personalized structure for the visualization is proposed.