International audience

Several distinct strategies have been proposed for recommending relevant items to groups of users. However, comparing the performance of these systems remains challenging due to the lack of representative datasets that capture the diversity of scenarios where group recommendation is needed, including variations in group composition and user behavior. To address this limitation and enable in-depth benchmarking, this article makes two main contributions. The first is an empirical analysis of widely used real-world GRS datasets, based on an adaptabilitydriven characterization of user behavior when interacting with items in group settings. This analysis compares users' individual preferences with those expressed during group interactions, while also examining the diversity of items involved. Building on these findings, we introduce BeHAVE, the first configurable dataset generator specifically designed for group recommendation. On top of a domain knowledge graph, BeHAVE allows controlled simulation of group interactions through behavioral parameters such as user adaptability and item diversity. Experimental results demonstrate that BeHAVE can replicate a range of behavioral patterns observed in real-world datasets and serves as a flexible benchmarking tool for evaluating GRS models under varying group dynamics.

International audience

Given an arbitrary group of computers, how to identify abnormal changes in their communication pattern? How to assess if the absence of some communications is normal or due to a failure? How to distinguish local from global events when communication data are extremely sparse and volatile? Existing approaches for anomaly detection in interaction streams, focusing on edge, nodes or graphs, lack flexibility to monitor arbitrary communication topologies. Moreover, they rely on structural features that are not adapted to highly sparse settings. In this work, we introduce MAD, a novel Multi-scale Anomaly Detection algorithm that (i) allows to query for the normality/abnormality state of an arbitrary group of observed/non-observed communications at a given time; and (ii) handles the highly sparse and uncertain nature of interaction data through a scoring method that is based on a novel probabilistic and multi-scale analysis of sub-graphs. In particular, MAD is (a) flexible: it can assess if any time-stamped subgraph is anomalous, making edge, node and graph anomalies particular instances; (b) interpretable: its multi-scale analysis allows to characterize the scope and nature of the anomalies; (c) efficient: given historical data of length N and M observed/non-observed communications to analyze, MAD produces an anomaly score in O (NM); and (d) effective: it significantly outperforms state-of-the-art alternatives tailored for edge, node or graph anomalies.

Linguistic summaries are used in this work to provide personalized exploration functionalities on massive relational data. To ensure a fluid exploration of the data, cardinalities of the data properties described in the summaries are estimated from statistics about the data distribution. The proposed workflow also involves a vocabulary inference mechanism from these statistics and a sampling-based approach to consolidate the estimated cardinalities. The paper shows that soft computing techniques are particularly relevant to build concrete and functional business intelligence solutions.

In the spirit of explainable AI approaches, this paper introduces a new strategy whose aim is to linguistically describe the inner structure of a dataset. Instead of removing irregular points and focusing on the analysis of regular points, the proposed approach relies on a unified data structure, an isolation forest, to both separate regular from irregular points and to identify their inner structure using a data-driven similarity measure. In addition, clusters of regular and irregular points are then linguistically described so as to help users focus on the most characteristic properties of each cluster and to possibly understand the reason why some points are irregular.

This paper proposes two original approaches to the processing of fuzzy queries in a relational database context. The general idea is to use views, either materialized or not. In the first case, materialized views are used to store the satisfaction degrees related to user-defined fuzzy predicates, instead of calculating them at runtime by means of user functions embedded in the query (which induces an important overhead). In the second case, abstract views are used to efficiently access the tuples that belong to the α-cut of the query result, by means of a derived Boolean selection condition.

The linguistic summarization of a dataset is a process whose complexity depends linearly on the size of the dataset and exponentially on the size of the fuzzy vocabulary. To efficiently summarize large datasets stored in Relational DataBases, reliable estimated cardinalities can be derived from statistics about the data distribution maintained by the RDB Management System, with no expensive data scans. This paper proposes to improve the precision of such estimated summaries while preserving their computational efficiency, by enriching the statistics-based approach with local scan-based corrections when needed: the proposed FRELS method provides efficient strategies both for identifying the needs and performing the corrections. Experiments conducted on real data show that FRELS remains incomparably more efficient than data-scan-based approaches to data summarization and offers a better precision than purely statistics-based approaches. The generation of estimated linguistic summaries takes a couple of seconds, even for datasets containing millions of tuples, with a reliability of more than 95%.

Summarizing data with linguistic statements is a crucial and topical issue that has been largely addressed by the soft computing community. The goal of summarization is to generate statements that linguistically describe the properties observed in a dataset. This paper addresses the issue of efficiently extracting these summaries and rendering them to the final user, in the case where the data to be summarized are stored in a relational data base: it proposes a novel strategy that leverages the statistics about the data distribution maintained by the database system. This paper shows that reliable summaries can be very efficiently estimated based on these statistics only and without any costly data access. Additionally, it proposes a visualization of the set of extracted summaries that offers a fruitful interactive exploration tool to the user. Experiments performed on two real data bases show the relevance and efficiency of the proposed approach: with a negligible loss of accuracy, we provide the first linguistic summarization approach whose processing time does not depend on the size of the dataset. The generation of estimated linguistic summaries takes less than one second even for dataset containing millions of tuples.

Extracting useful and interpretable knowledge from raw data is a crucial issue that has been largely addressed by the data mining community especially. In this paper we provide an interactive data exploration approach that relies on two steps. First, a personalized linguistic summary of the data set concerned is built and displayed as a tag cloud. Then, exploration functionalities are provided on top of the summary to let the user discover interesting properties in the data as frequent/atypical/diversified associations between properties.

In this paper, a novel approach is introduced to let users extract knowledge from a raw dataset in an intuitive way and using their own vocabulary. The inner structure of a raw data set is first identified using a clustering algorithm, structure on which specificity-driven measures are defined to extract the most informative knowledge. To let domain experts interact with the cluster-based structure and its embedded knowledge, a graphical visualisation is proposed as well as dedicated query operators.

Graph databases have aroused a large interest in the last years thanks to their large scope of potential applications. Defining a language allowing a flexible querying of graph databases may greatly improve usability of data. In this paper, we present a system for querying graph databases in a flexible way. The preferences are based on fuzzy set theory and may concern i) the content of the vertices and ii) the structure of the graph.

In this work, the first stones of a flexible approach to linked data querying based on fuzzy set theory are laid. Flexibility refers to the capability of expressing flexible queries over a RDF model containing gradual information.

Graph databases have aroused a large interest in the last years thanks to their large scope of potential applications (e.g. social networks, biomedical networks, data stemming from the web). In a similar way as what has already been proposed in relational databases, defining a language allowing a flexible querying of graph databases may greatly improve usability of data. This paper focuses on the notion of fuzzy graph database and describes a fuzzy query language that makes it possible to handle such database, which may be fuzzy or not, in a flexible way. This language, called FUDGE, can be used to express preference queries on fuzzy graph databases. The preferences concern i) the content of the vertices of the graph and ii) the structure of the graph. The FUDGE language is implemented in a system, called SUGAR, that we present in this article. We also discuss implementation issues of the FUDGE language in SUGAR.

A large amount of linked data is now available, but to retrieve knowledge from these data, queries have to be formulated using formal query languages. While expressive query languages are developed, their use by end users, generally not familiar with formal languages, is limited. Keyword-based search is considered as a convenient and intuitive way for users to express their information needs. Keyword search over structured data is thus an interesting alternative but which raises challenging issues. The main challenge is to determine the meaning of a keyword query in order to translate it into a target formal query language, SPARQL in our case. In this paper, we address this challenge and propose a novel approach that relies on user interactions to determine the correct interpretation of the keyword query. The principle is to first ask the user to define a coarse keyword query, then to suggest candidate interpretations expressed in an explicit, thus unambiguous, and human readable form. Once the correct interpretation has been selected, the query may be refined with aggregate functions and comparatives. Experiments conducted on a large knowledge base show the effectiveness and the efficiency of the proposed approach.

To improve the expressivity and accuracy of database query interfaces, a keyword-based constrained query language is introduced to let users explicitly express the intent of their search using keywords linked by meaningful grammatical connectives. Individually, keywords and connectives correspond to textual descriptions attached to components of the database graph schema, and as a whole, a so-called connected keywords query corresponds to a textual description of an SQL query. The translation process of such a query into SQL is mainly composed of two steps: first, the syntactic structure of the keyword query is analyzed to exhibit projection and selection statements using predefined graph patterns, then non explicit joins are deduced to obtain a complete translation of the keyword query as a meaningful connected subgraph. Experimentations show the relevance of the approach in terms of expressivity and efficiency.

AGGREGO SEARCH offers a novel keyword-based query solu-tion for end users in order to retrieve precise answers from se-mantic data sources. Contrary to existing approaches, AGGREGO SEARCH suggests grammatical connectors from natural languages during the query formulation step in order to specify the meaning of each keyword, thus leading to a complete and explicit definition of the intent of the search. An example of such a query is name of person at the head of company and author of article about “busi-ness intelligence". In order to help users formulate such connected keywords queries, a specific autocompletion strategy has been de-veloped. A translation of the user keyword query into SPARQL is performed on-the-fly during the interactive query construction pro-cess. For this demonstration, we show how AGGREGO SEARCH has been integrated on top of a mediation system to let users intu-itively define explicit and precise keyword queries in order to ex-tract knowledge distributed in heterogeneous large semantic data sources. 1.

This paper proposes a notion of fuzzy graph database and describes a fuzzy query algebra that makes it possible to handle such database, which may be fuzzy or not, in a flexible way. The algebra, based on fuzzy set theory and the concept of a fuzzy graph, is composed of a set of operators that can be used to express preference queries on fuzzy graph databases. The preferences concern i) the content of the vertices of the graph and ii) the structure of the graph. In a similar way as relational algebra constitutes the basis of SQL, the fuzzy algebra proposed here underlies a user-oriented query language and an associated tool implementing this language that are also presented in the paper.

Clustering methods are of a particular interest to discover and to summarize the structure of a data set. However, interpreting clusters may be abstruse for unexperienced users who most of the time possess their own vocabulary to describe data and properties. In this article, an approach is proposed to determine and quantify how appropriate a user-defined vocabulary is regarding the structure captured on the data distribution using a clustering method. Two measures of vocabulary appropriateness based on clustering are proposed and tested on artificial data.

In this paper, we present an implementation strategy for a fuzzy querying system embedded in a regular DBMS. This system relies on the language SQLf that makes it possible to express a great variety of fuzzy queries. Experiments show that this implementation strategy induces performance gains with respect to existing strategies based on a loose (or milder) coupling between a fuzzy querying layer and a DBMS, that necessitate an external postprocessing so as to compute the result in the form of a fuzzy relation. We also describe a user-friendly interface aimed at helping nonexpert users express their fuzzy queries in an intuitive manner.

In this demonstration we present a complete fuzzy-set-based approach to preference queries that tackles the two main questions raised by the introduction of flexibility and personalization when querying relational databases: i) how to efficiently execute preference queries? and, ii) how to help users define preferences and queries? As an answer to the first question, we propose PostgreSQL_f, a module implemented on top of PostgreSQL to handle fuzzy queries. To answer the second question, we propose ReqFlex an intuitive user interface to the definition of preferences and the construction of fuzzy queries.

In this paper, we describe an approach to database preference queries based on the notion of fuzzy outranking, suited to the case where partial preferences are incommensurable. This model constitutes an alternative to the use of Pareto order. Even though outranking does not define an order in the strict sense of the term, we describe a technique which yields a complete pre-order, based on a global aggregation of the outranking degrees computed for each pair of tuples.