The article presents a simple yet optimal approach to compute aggregates of window queries over data streams. The proposal is built as a fully-fledged pipeline of operators that are literal transcripts of mathematical definitions. Main features are an application of the prefix sums and a well-founded un-/slicing technique. The overall process is linear in the number of events and windows, and it takes quasi-constant space for a mix of periodic windows, still applicable to multiple deterministic window queries. The limitations are twofold: the events come in-order and the aggregation function is a left-cancellative monoid.

Streaming systems often use slices to share computation costs among overlapping windows. However they are limited to instantaneous events where only one point represents the event. Here, we extend streams to events that come with a duration, denoted as spanning events. After a short review of the new constraints ensued by event lifespan in a temporal sliding-window context, we propose a new structure for dealing with slices in such an environment, and prove that our technique is both correct and effective to deal with such spanning events.

In clustering, consensus clustering aims at providing a single partition fitting a consensus from a set of independently generated. Common procedures, which are mainly statistical and graph-based, are recognized for their robustness and ability to scale-up. In this paper, we provide a complementary and original viewpoint over consensus clustering, by means of algebraic definitions which allow to ascertain the nature of available inferences in a systematic approach (e.g. a knowledge base). We found our approach on the lattice of partitions, for which we shall disclose how some operators can be added with the aim to express a formula representing the consensus. We show that adopting an incremental approach may assist to retain significant amount of aggregate data which fits well with the set of input clustering's. Beyond that ability to model formulae, we also note that its potential cannot be easily captured through such a logical system. It is due to the volatile nature of handling partitions which finally impacts on ability to draw some valuable conclusions.

In this paper, we present the concept of Time Sequence Summarization to support chronology-dependent applications on massive data sources. Time sequence summarization takes as input a time sequence of events that are chronologically ordered. Each event is described by a set of descriptors. Time sequence summarization produces a concise time sequence that can be substituted for the original time sequence in chronology-dependent applications. We propose an algorithm that achieves time sequence summarization based on a generalization, grouping and concept formation process. Generalization expresses event descriptors at higher levels of abstraction using taxonomies while grouping gathers similar events. Concept formation is responsible for reducing the size of the input time sequence of events by representing each group created by one concept. The process is performed in a way such that the overall chronology of events is preserved. The algorithm computes the summary incrementally and has reduced algorithmic complexity. The resulting output is a concise representation, yet, informative enough to directly support chronology-dependent applications. We validate our approach by summarizing one year of financial news provided by Reuters.

Sharing huge, massively distributed databases in P2P systems is inherently difficult. As the amount of stored data increases, data localization techniques become no longer sufficient. A practical approach is to rely on compact database summaries rather than raw database records, whose access is costly in large P2P systems. In this paper, we consider summaries that are synthetic, multidimensional views with two main virtues. First, they can be directly queried and used to approximately answer a query without exploring the original data. Second, as semantic indexes, they support locating relevant nodes based on data content. Our main contribution is to define a summary model for P2P systems, and the appropriate algorithms for summary management. Our performance evaluation shows that the cost of query routing is minimized, while incurring a low cost of summary maintenance.

Major media companies such as The Financial Times, the Wall Street Journal or Reuters generate huge amounts of textual news data on a daily basis. Mining frequent patterns in this mass of information is critical for knowledge workers such as financial analysts, stock traders or economists. Using existing frequent pattern mining (FPM) algorithms for the analysis of news data is difficult because of the size and lack of structuring of the free text news content. In this article, we demonstrate a comprehensive Streaming TEmporAl Data (STEAD) analysis framework for mining frequent patterns in financial news. In this demonstration, we show how the mining task is supported by the use of a Time-Aware Content Summarization algorithm (TACS). This summary generates a concise representation of large volume of data by taking into account the expert's peculiar interest while preserving the news arrival temporal information which is essential for FPM algorithms. We experimented the whole framework on a set of news data from Reuters.

The database summarization system coined S aint E ti Q provides multi-resolution summaries of structured data stored into acentralized database. Summaries are computed online with a conceptual hierarchical clustering algorithm. However, most companies work in distributed legacy environments and consequently the current centralized version of S aint E ti Q is either not feasible (privacy preserving) or not desirable (resource limitations).

SaintEtiQ (Saint-Paul et al., 2005) is an on-line linguistic summarization system of tables and/or views. Our approach considers a first normal form relation R(A1,...,An) in the relational database model, and constructs a new relation R*( A1,...,An), in which tuples z are summaries and attribute values are linguistic labels describing a set of tuples Rz, sub-table of R. Thus, the SaintEtiQ system identifies statements of the form "Most tuples of R are (a11or a12...or a1m1) and (a21...or a a2m1)".

In this paper, we propose a general framework to explore and analyze database summaries built from massive data sets. Summaries are multidimensional vague concepts, represented by a set of fuzzy terms provided on each attribute of a database relation as domain knowledge. We define a logical multidimensional data model called summary partition, as data cubes do in OLAP systems, in order to provide the end-user with a reduced and well-suited presentation of the data. The overall on-line summarization process is then intended to support a new approach to OLAP. To achieve this we introduce a collection of algebraic operators over the space of summary partitions : relational, granularity and restructuring operators are defined as the core algebra for online analytical processing of summarized versions of the data.

The data summarization techniques are usually not provided with tools for endusers to efficiently use the produced summaries. In a previous work, a querying tool for the SaintEtiQ summarization model has been outlined. It takes advantage off the hierarchical structure of those summaries to efficiently describe tuples that satisfy some selection criteria. This querying mechanism can be seen both as a Boolean one over the summaries, and as a flexible one over the underlying relational tuples. This paper mainly investigates the case of null answers. It is shown how, in this case, alternative queries with non-null answers can be proposed to the user

The increasing number of satellite and cable television channels is resulting in a soaring number of broadcast programs available to viewers. To alleviate this problem, Personal Video Recorders (multimedia platforms which record TV programs on a hard disk) should integrate a recommender system, which purpose is to filter programs according to their relevance. These systems are based on a user profile, acting as a representative for the user's interests. An important research issue resides in going beyond explicitly user-defined profiles. This paper presents a TV recommender system using fuzzy linguistic summarization technique, which enables automatic learning of the user profile. The logical architecture of the recommender system based on the SAINTETIQ model, as well as the main ideas of the filtering task are introduced in this communication.

A new approach for the summarization of image databases is described. A linguistic description of images is first generated from its low level features such as color. An original incremental summary process, called SAINTETIQ is then applied to the images, leading to a comprehensive hierarchically organized set of summaries of parts of the database. The summary process is driven by data and relies on the background knowledge represented by a fuzzy relational thesaurus. It provides both a proximity measure between terms of color domain, and a way to produce generalized descriptions of summaries. Finally, this article points out usages of the summaries. A new method to browse image databases is then presented as well as a way to produce relevant prototypes from groups of images. Those tasks take advantage of the linguistic descriptions of summaries provided by SAINTETIQ.

In this paper, some important features of a new approach to data summarization are introduced. Our model named SAINTETIQ produces summaries of groups of database records with different granularities. A summary is represented on each attribute by fuzzy sets associated to linguistic descriptors. One major feature of the SAINTETIQ system is the intensive use of background knowledge (BK) in the summarization process. BK is built a priori on each attribute. It supports both a translation step of descriptions of database tuples into a user-defined vocabulary, and a generalization step providing synthetic intents of summaries. Furthermore, the fuzzy set-based representation of summaries allows the system to improve robustness and accuracy of summary descriptions.

We present an original fuzzy-based heuristic measure, the partition quality, highly integrated to a general database summarization framework. Our system SAINTETIQ is based on an incremental conceptual clustering algorithm, building a concept hierarchy from database observations. Concepts of the hierarchy are summaries of a part of the database, at different levels of abstraction. The algorithm roughly consists in applying learning operators on the hierarchy as new observations arrive into the system. The decision function named partition quality that we introduce, reflects contrast and typicity of a concept partition and allows the system to select the best operator to apply during the construction of the concept hierarchy.