International audience

We address the problem of detecting data glitches in ML training sets, specifically mislabeled and anomalous samples. Detection of data glitches provides insights into the quality of the data sampling. Their repair may improve the reliability and the performance of the model. The proposed methodology is based on exploiting influence functions that estimate how much the loss of the model (or a given sample) is affected when a sample is removed from the training set. We introduce three novel signals for detecting, characterizing, and repairing data glitches in a training set based on sample influences. Influence-based signals form an explainable-by-design data glitch detection framework, producing intuitively explainable signals of the actual predictive model built. In contrast, specialized algorithms that are agnostic to the target ML model (e.g., anomaly detectors) replicate the work of fitting the data distribution and may detect glitches that are inconsistent with the decision boundary of the predictive model. Computational experiments on tabular and image data modalities demonstrate that the proposed signals outperform, in some cases up to a factor of 6, all existing influence-based signals, and generalize across different datasets and ML models. In addition, they often outperform specialized glitch detectors (e.g., mislabeled and anomaly detectors) and provide accurate label repairs for mislabeled samples.

Currently, the growing proliferation of information systems generates large volumes of data continuously, stemming from a variety of sources such as web platforms, social networks, and multiple devices. These data, often lacking a defined schema, require an initial process of consolidation and cleansing before analysis and knowledge extraction can occur. In this context, Entity Resolution (ER) plays a crucial role, facilitating the integration of knowledge bases and identifying similarities among entities from different sources. However, the traditional ER process is computationally expensive, and becomes more complicated in the streaming context where the data arrive continuously. Moreover, there is a lack of studies involving fairness and ER, which is related to the absence of discrimination or bias. In this sense, fairness criteria aim to mitigate the implications of data bias in ER systems, which requires more than just optimizing accuracy, as traditionally done. Considering this context, this work presents TREATS, a schema-agnostic and fairness-aware ER workflow developed for managing streaming data incrementally. The proposed fairness-aware ER framework tackles constraints across various groups of interest, presenting a resilient and equitable solution to the related challenges. Through experimental evaluation, the proposed techniques and heuristics are compared against state-of-the-art approaches over five real-world data source pairs, in which the results demonstrated significant improvements in terms of fairness, without degradation of effectiveness and efficiency measures in the streaming environment. In summary, our contributions aim to propel the ER field forward by providing a workflow that addresses both technical challenges and ethical concerns.

Numerous real-world data contain missing values, while in contrast, most Machine Learning (ML) algorithms assume complete datasets. For this reason, several imputation algorithms have been proposed to predict and fill in the missing values. Given the advances in predictive modeling algorithms tuned in an Automated Machine Learning context (AutoML) setting, a question that naturally arises is to what extent sophisticated imputation algorithms (e.g., Neural Network based) are really needed, or we can obtain a descent performance using simple methods like Mean/Mode (MM). In this article, we experimentally compare six state-of-the-art representatives of different imputation algorithmic families from an AutoML predictive modeling perspective, including a feature selection step and combined algorithm and hyper-parameter selection. We used a commercial AutoML tool for our experiments, in which we included the selected imputation methods. Experiments ran on 25 binary classification real-world incomplete datasets with missing values and 10 binary classification complete datasets in which synthetic missing values are introduced according to different missingness mechanisms, at varying missing frequencies. The main conclusion drawn from our experiments is that the best method on average is the Denoise AutoEncoder on real-world datasets and the MissForest in simulated datasets, followed closely by MM. In addition, binary indicator variables encoding missingness patterns actually improve predictive performance, on average. Last, although there are cases where Neural-Network-based imputation significantly improves predictive performance, this comes at a great computational cost and requires measuring all feature values to impute new samples.

Explainable machine learning techniques (XAI) aim to provide a solid descriptive approach to Deep Neural Networks (NN). In Multi-Variate Time Series (MTS) analysis, the most recurrent techniques use relevance attribution, where importance scores are assigned to each TS variable over time according to their importance in classification or forecasting. Despite their popularity, post-hoc explanation methods do not account for causal relationships between the model outcome and its predictors. In our work, we conduct a thorough empirical evaluation of model-agnostic and model-specific relevance attribution methods proposed for TCNN, LSTM, and Transformers classification models of MTS. The contribution of our empirical study is threefold: (i) evaluate the capability of existing post-hoc methods to provide consistent explanations for high-dimensional MTS (ii) quantify how post-hoc explanations are related to sufficient explanations (i.e., the direct causes of the target TS variable) underlying the datasets, and (iii) rank the performance of surrogate models built over post-hoc and causal explanations w.r.t. the full MTS models. To the best of our knowledge, this is the first work that evaluates the reliability and effectiveness of existing XAI methods from a temporal causal model perspective.

Abstract In recent years, we have witnessed the proliferation of knowledge graphs (KG) in various domains, aiming to support applications like question answering, recommendations, etc. A frequent task when integrating knowledge from different KGs is to find which subgraphs refer to the same real-world entity, a task largely known as the Entity Alignment. Recently, embedding methods have been used for entity alignment tasks, that learn a vector-space representation of entities which preserves their similarity in the original KGs. A wide variety of supervised, unsupervised, and semi-supervised methods have been proposed that exploit both factual (attribute based) and structural information (relation based) of entities in the KGs. Still, a quantitative assessment of their strengths and weaknesses in real-world KGs according to different performance metrics and KG characteristics is missing from the literature. In this work, we conduct the first meta-level analysis of popular embedding methods for entity alignment, based on a statistically sound methodology. Our analysis reveals statistically significant correlations of different embedding methods with various meta-features extracted by KGs and rank them in a statistically significant way according to their effectiveness across all real-world KGs of our testbed. Finally, we study interesting trade-offs in terms of methods’ effectiveness and efficiency.

There is an urgent call to detect and prevent "biased data" at the earliest possible stage of the data pipelines used to build automated decision-making systems. In this paper, we are focusing on controlling the data bias in entity resolution (ER) tasks aiming to discover and unify records/descriptions from different data sources that refer to the same real-world entity. We formally define the ER problem with fairness constraints ensuring that all groups of entities have similar chances to be resolved. Then, we introduce FairER, a greedy algorithm for solving this problem for fairness criteria based on equal matching decisions. Our experiments show that FairER achieves similar or higher accuracy against two baseline methods over 7 datasets, while guaranteeing minimal bias.

International audience

Numerous algorithms have been proposed for detecting anomalies (outliers, novelties) in an unsupervised manner. Unfortunately, it is not trivial, in general, to understand why a given sample (record) is labelled as an anomaly and thus diagnose its root causes. We propose the following reduced-dimensionality, surrogate model approach to explain detector decisions: approximate the detection model with another one that employs only a small subset of features. Subsequently, samples can be visualized in this low-dimensionality space for human understanding. To this end, we develop PROTEUS, an AutoML pipeline to produce the surrogate model, specifically designed for feature selection on imbalanced datasets. The PROTEUS surrogate model can not only explain the training data, but also the out-of-sample (unseen) data. In other words, PROTEUS produces predictive explanations by approximating the decision surface of an unsupervised detector. PROTEUS is designed to return an accurate estimate of out-of-sample predictive performance to serve as a metric of the quality of the approximation. Computational experiments confirm the efficacy of PROTEUS to produce predictive explanations for different families of detectors and to reliably estimate their predictive performance in unseen data. Unlike several ad-hoc feature importance methods, PROTEUS is robust to high-dimensional data.

Entity Resolution (ER) aims to identify different descriptions in various Knowledge Bases (KBs) that refer to the same entity. ER is challenged by the Variety, Volume and Veracity of entity descriptions published in the Web of Data. To address them, we propose the MinoanER framework that simultaneously fulfills full automation, support of highly heterogeneous entities, and massive parallelization of the ER process. MinoanER leverages a token-based similarity of entities to define a new metric that derives the similarity of neighboring entities from the most important relations, as they are indicated only by statistics. A composite blocking method is employed to capture different sources of matching evidence from the content, neighbors, or names of entities. The search space of candidate pairs for comparison is compactly abstracted by a novel disjunctive blocking graph and processed by a non-iterative, massively parallel matching algorithm that consists of four generic, schema-agnostic matching rules that are quite robust with respect to their internal configuration. We demonstrate that the effectiveness of MinoanER is comparable to existing ER tools over real KBs exhibiting low Variety, but it outperforms them significantly when matching KBs with high Variety.

Understanding the sensing context of raw data is crucial for assessing the quality of large crowdsourced spatio-temporal datasets. Detecting sensing contexts in the wild is a challenging task and requires features from smartphone sensors that are not always available. In this paper, we propose three heuristic algorithms for detecting sensing contexts such as in/out-pocket, under/over-ground, and in/out-door for crowdsourced datasets that are destined for human mobility mining. These are unsupervised binary classifiers with a small memory footprint and execution time. Using a segment of the Ambiciti real dataset - a feature-limited crowdsourced dataset - we report that our algorithms perform equally well in terms of balanced accuracy (within 4.3%) when compared to machine learning (ML) models reported by an AutoML tool.

Entity Resolution (ER) aims to identify different descriptions in various Knowledge Bases (KBs) that refer to the same entity. ER is challenged by the Variety, Volume and Veracity of descriptions published in the Web of Data. To address them, we propose the MinoanER framework that fulfills full automation and support of highly heterogeneous entities. MinoanER leverages a token-based similarity of entities to define a new metric that derives the similarity of neighboring entities from the most important relations, indicated only by statistics. For high efficiency, similarities are computed from a set of schema-agnostic blocks and processed in a non-iterative way that involves four threshold-free heuristics. We demonstrate that the effectiveness of MinoanER is comparable to existing ER tools over real KBs exhibiting low heterogeneity in terms of entity types and content. Yet, MinoanER outperforms state-of-the-art ER tools when matching highly heterogeneous KBs.

Entity resolution aims to identify descriptions of the same entity within or across knowledge bases. In this work, we provide a comprehensive and cohesive overview of the key research results in the area of entity resolution. We are interested in frameworks addressing the new challenges in entity resolution posed by the Web of data in which real world entities are described by interlinked data rather than documents. Since such descriptions are usually partial, overlapping and sometimes evolving, entity resolution emerges as a central problem both to increase dataset linking, but also to search the Web of data for entities and their relations. We focus on Web-scale blocking, iterative and progressive solutions for entity resolution. Specifically, to reduce the required number of comparisons, blocking is performed to place similar descriptions into blocks and executes comparisons to identify matches only between descriptions within the same block. To minimize the number of missed matches, an iterative entity resolution process can exploit any intermediate results of blocking and matching, discovering new candidate description pairs for resolution. Finally, we overview works on progressive entity resolution, which attempt to discover as many matches as possible given limited computing budget, by estimating the matching likelihood of yet unresolved descriptions, based on the matches found so far.

Entity resolution aims to identify descriptions of the same entity within or across knowledge bases. In this work, we present the Minoan ER platform for resolving entities described by linked data in the Web (e.g., in RDF). To reduce the required number of comparisons, Minoan ER performs blocking to place similar descriptions into blocks and executes comparisons to identify matches only between descriptions within the same block. Moreover, it explores in a pay-as-you-go fashion any intermediate results of matching to obtain similarity evidence of entity neighbors and discover new candidate description pairs for resolution.

International audience

In the Web of data, entities are described by interlinked data rather than documents on the Web. In this work, we focus on entity resolution in the Web of data, i.e., identifying descriptions that refer to the same real-world entity. To reduce the required number of pairwise comparisons, methods for entity resolution perform blocking as a pre-processing step. A blocking technique places similar entity descriptions into blocks and executes comparisons only between descriptions within the same block. We experimentally evaluate blocking techniques proposed for the Web of data and present dataset characteristics that determine the effectiveness and efficiency of such methods. Furthermore, we analyze the characteristics of the missed matching entity descriptions and examine different types of links that blocking techniques can potentially identify.

There exists a large body of work on online drift detection with the goal of dynamically finding and maintaining changes in data streams. In this paper, we adopt a query-based approach to drift detection. Our approach relies on a drift index, a structure that captures drift at different time granularities and enables flexible drift queries. We formalize different drift queries that represent real-world scenarios and develop query evaluation algorithms that use different materializations of the drift index as well as strategies for online index maintenance. We describe a thorough study of the performance of our algorithms on real-world and synthetic datasets with varying change rates.

This demonstration presents MeowsReader, a real-time news ranking and filtering prototype. MeowsReader illustrates how a general class of continuous top-k queries offers a suitable abstraction for modeling and implementing real-time search services over highly dynamic information streams combining keyword search and realtime web signals about information items. Users express their interest by simple text queries and continuously receive the best matching results in an alert-like environment. The main innovative feature are dynamic item scores which take account of information decay, real-time web attention and other online user feedback. Additionally, a trends detection mechanism automatically generates trending entities from the input streams, which can smoothly be added to user profiles in form of keyword queries.

We show that the evaluation of SPARQL algebra queries on various notions of annotated RDF graphs can be seen as particular cases of the evaluation of these queries on RDF graphs annotated with elements of so-called spm-semirings. Spm-semirings extend semirings, used for positive relational algebra queries on annotated relational data, with a new operator to capture the semantics of the non-monotone SPARQL operator OPTIONAL. Furthermore, spm-semiring-based annotations ensure that desired SPARQL query equivalences hold when querying annotated RDF. In addition to introducing spm-semirings, we study their properties and provide an alternative characterization of these structures in terms of semirings with an embedded boolean algebra (or seba-structure for short). This characterization allows to construct spm-semirings and to identify a universal object in the class of spm-semirings. Finally, we show that this universal object provides a concise provenance representation and can be used to evaluate SPARQL queries on arbitrary spm-semiring-annotated RDF graphs.

With the increasing use of Web 2.0 to create, disseminate, and consume large volumes of data, more and more information is published and becomes available for potential data consumers, that is, applications/services, individual users and communities, outside their production site. The most representative example of this trend is Linked Open Data (LOD), a set of interlinked data and knowledge bases. The main challenge in this context is data governance within loosely coordinated organizations that are publishing added-value interlinked data on the Web, bringing together issues related to data management and data quality, in order to support the full lifecycle of data production, consumption, and management. In this article, we are interested in curation issues for RDF(S) data, which is the default data model for LOD. In particular, we are addressing change management for RDF(S) data maintained by large communities (scientists, librarians, etc.) which act as curators to ensure high quality of data. Such curated Knowledge Bases (KBs) are constantly evolving for various reasons, such as the inclusion of new experimental evidence or observations, or the correction of erroneous conceptualizations. Managing such changes poses several research problems, including the problem of detecting the changes (delta) between versions of the same KB developed and maintained by different groups of curators, a crucial task for assisting them in understanding the involved changes. This becomes all the more important as curated KBs are interconnected (through copying or referencing) and thus changes need to be propagated from one KB to another either within or across communities. This article addresses this problem by proposing a change language which allows the formulation of concise and intuitive deltas. The language is expressive enough to describe unambiguously any possible change encountered in curated KBs expressed in RDF(S), and can be efficiently and deterministically detected in an automated way. Moreover, we devise a change detection algorithm which is sound and complete with respect to the aforementioned language, and study appropriate semantics for executing the deltas expressed in our language in order to move backwards and forwards in a multiversion repository, using only the corresponding deltas. Finally, we evaluate through experiments the effectiveness and efficiency of our algorithms using real ontologies from the cultural, bioinformatics, and entertainment domains.

In this work we are interested in the scalable processing of content filtering queries over text item streams. In particular, we are aiming to generalize state of the art solutions with non-homogeneous scoring functions combining query-independent item importance with query-dependent content relevance. While such complex ranking functions are widely used in web search engines this is to our knowledge the first scientific work studying their usage in a continuous query scenario. Our main contribution consists in the definition and the evaluation of new efficient in-memory data structures for indexing continuous top-k queries based on an original two-dimensional representation of text queries. We are exploring locally-optimal score bounds and heuristics that efficiently prune the search space of candidate top-k query results which have to be updated at the arrival of new stream items. Finally, we experimentally evaluate memory/matching time trade-offs of these index structures. In particular we experimentally illustrate their linear scaling behavior with respect to the number of indexed queries.

The explosion of published information on the Web leads to the emergence of a Web syndication paradigm, which transforms the passive reader into an active information collector. Information consumers subscribe to RSS/Atom feeds and are notified whenever a piece of news (item) is published. The success of this Web syndication now offered on Web sites, blogs, and social media, however raises scalability issues. There is a vital need for efficient real-time filtering methods across feeds, to allow users to follow effectively personally interesting information. We investigate in this paper three indexing techniques for users' subscriptions based on inverted lists or on an ordered trie. We present analytical models for memory requirements and matching time and we conduct a thorough experimental evaluation to exhibit the impact of critical workload parameters on these structures.

Query optimization in RDF Stores is a challenging problem as SPARQL queries typically contain many more joins than equivalent relational plans, and hence lead to a large join order search space. In such cases, cost-based query optimization often is not possible. One practical reason for this is that statistics typically are missing in web scale setting such as the Linked Open Datasets (LOD). The more profound reason is that due to the absence of schematic structure in RDF, join-hit ratio estimation requires complicated forms of correlated join statistics; and currently there are no methods to identify the relevant correlations beforehand. For this reason, the use of good heuristics is essential in SPARQL query optimization, even in the case that are partially used with cost-based statistics (i.e., hybrid query optimization). In this paper we describe a set of useful heuristics for SPARQL query optimizers. We present these in the context of a new Heuristic SPARQL Planner (HSP) that is capable of exploiting the syntactic and the structural variations of the triple patterns in a SPARQL query in order to choose an execution plan without the need of any cost model. For this, we define the variable graph and we show a reduction of the SPARQL query optimization problem to the maximum weight independent set problem. We implemented our planner on top of the MonetDB open source column-store and evaluated its effectiveness against the state-of-the-art RDF-3X engine as well as comparing the plan quality with a relational (SQL) equivalent of the benchmarks.

The ability to compute the differences that exist between two RDF/S Knowledge Bases (KB) is an important step to cope with the evolving nature of the Semantic Web (SW). In particular, RDF/S deltas can be employed to reduce the amount of data that need to be exchanged and managed over the network in order to build SW synchronization and versioning services. By considering deltas as sets of change operations, in this article we introduce various RDF/S differential functions which take into account inferred knowledge from an RDF/S knowledge base. We first study their correctness in transforming a source to a target RDF/S knowledge base in conjunction with the semantics of the employed change operations (i.e., with or without side-effects on inferred knowledge). Then we formally analyze desired properties of RDF/S deltas such as size minimality, semantic identity, redundancy elimination, reversibility, and composability, as well as identify those RDF/S differential functions that satisfy them. Subsequently, we experimentally evaluate the computing time and size of the produced deltas over real and synthetic RDF/S knowledge bases.

Preference queries are crucial for various applications (e.g. digital libraries) as they allow users to discover and order data of interest in a personalized way. In this paper, we define preferences as preorders over relational attributes and their respective domains. Then, we rely on appropriate linearizations to provide a natural semantics for the block sequence answering a preference query. Moreover, we introduce two novel rewriting algorithms (called LBA and TBA) which exploit the semantics of preference expressions for constructing progressively each block of the answer. We demonstrate experimentally the scalability and performance gains of our algorithms (up to 3 orders of magnitude) for variable database and result sizes, as well as for preference expressions of variable size and structure. To the best of our knowledge, LBA and TBA are the first algorithms for evaluating efficiently arbitrary preference queries over voluminous databases.

In this paper, we measure and analyze the graph features of semantic Web (SW) schemas with focus on power-law degree distributions. Our main finding is that the majority of SW schemas with a significant number of properties (respectively, classes) approximate a power law for total-degree (respectively, the number of subsumed classes) distribution. Moreover, our analysis revealed some emerging conceptual modeling practices of SW schema developers: (1) each schema has a few focal classes that have been analyzed in detail (that is, they have numerous properties and subclasses), which are further connected with focal classes defined in other schemas, (2) class subsumption hierarchies are mostly unbalanced (that is, some branches are deep and heavy, while others are shallow and light), (3) most properties have as domain/range classes that are located high at the class subsumption hierarchies, and (4) the number of recursive/multiple properties is significant. The knowledge of these features is essential for guiding synthetic SW schema generation, which is an important step toward benchmarking SW repositories and query language implementations.

Semantic Web (SW) technology aims to facilitate the integration of legacy data sources spread worldwide. Despite the plethora of SW languages (e.g., RDF/S, OWL) recently proposed for supporting large-scale information interoperation, the vast majority of legacy sources still rely on relational databases (RDB) published on the Web or corporate intranets as virtual XML. In this article, we advocate a first-order logic framework for mediating high-level queries to relational and/or XML sources using community ontologies expressed in a SW language such as RDF/S. We describe the architecture and reasoning services of our SW integration middleware, termed SWIM, and we present the main design choices and techniques for supporting powerful mappings between different data models, as well as reformulation and optimization of queries expressed against mediator ontologies and views.

Online Curriculum Portals aim to support networks of instructors and learners by providing a space of convergence for enhancing peer-to-peer learning interactions among individuals of an educational institution. To this end, effective, open and scalable e-learning systems are required to acquire, store, and share knowledge under the form of learning objects (LO). In this paper, we are interested in exploiting the semantic relationships that characterize these LOs (e.g., prerequisite, part-of or see-also) in order to capture and access individual and group knowledge in conjunction with the learning processes supported by educational institutions. To achieve this functionality, Semantic Web (e.g., RDF/s) and declarative query languages (e.g., RQL) are employed to represent LOs and their relationships (e.g., LOM), as well as, to support navigation at the conceptual e-learning Portal space. In this way, different LOs could be presented to the same learners, according to the traversed schema navigation paths (i.e., learning paths). Using the Apache Jetspeed framework we are able to generate and assemble at run-time portlets (i.e., pluggable web components) for visualizing personalized views as dynamic web pages. Last but not least, both learners and instructors can employ the same Portal GUI for updating semantically described LOs and thus support an open-ended continuum of learning. To the best of our knowledge, the work presented in this paper is the first Online Curriculum Portal platform supporting the aforementioned functionality.

The emerging paradigm of electronic services promises to bring to distributed computation and services the flexibility that the web has brought to the sharing of documents. An understanding of fundamental properties of e-service composition is required in order to take full advantage of the paradigm. This paper examines proposals and standards for e-services from the perspectives of XML, data management, workflow, and process models. Key areas for study are identified, including behavioral service signatures, verification and synthesis techniques for composite services, analysis of service data manipulation commands, and XML analysis applied to service specifications. We give a sample of the relevant results and techniques in each of these areas.

This paper focuses on the optimization of the navigation through voluminous subsumption hierarchies of topics employed by Portal Catalogs like Netscape Open Directory (ODP). We advocate for the use of labeling schemes for modeling these hierarchies in order to efficiently answer queries such as subsumption check, descendants, ancestors or nearest common ancestor, which usually require costly transitive closure computations. We first give a qualitative comparison of three main families of schemes, namely bit vector, prefix and interval based schemes. We then show that two labeling schemes are good candidates for an efficient implementation of label querying using standard relational DBMS, namely, the Dewey Prefix scheme [6] and an Interval scheme by Agrawal, Borgida and Jagadish [1]. We compare their storage and query evaluation performance for the 16 ODP hierarchies using the PostgreSQL engine.

Real-scale Semantic Web applications, such as Knowledge Portals and E-Marketplaces, require the management of large volumes of metadata, i.e., information describing the available Web content and services. Better knowledge about their meaning, usage, accessibility or quality will considerably facilitate an automated processing of Web resources. The Resource Description Framework (RDF) enables the creation and exchange of metadata as normal Web data. Although voluminous RDF descriptions are already appearing, sufficiently expressive declarative languages for querying both RDF descriptions and schemas are still missing. In this paper, we propose a new RDF query language called RQL. It is a typed functional language (a la OQL) and relies on a formal model for directed labeled graphs permitting the interpretation of superimposed resource descriptions by means of one or more RDF schemas. RQL adapts the functionality of semistructured/XML query languages to the peculiarities of RDF but, foremost, it enables to uniformly query both resource descriptions and schemas. We illustrate the RQL syntax, semantics and typing system by means of a set of example queries and report on the performance of our persistent RDF Store employed by the RQL interpreter.

Modern applications (Web portals, digital libraries, etc.) require integrated access to various information sources (from traditional DBMS to semistructured Web repositories), fast deployment and low maintenance cost in a rapidly evolving environment. Because of its flexibility, there is an increasing interest in using XML as a middleware model for such applications. XML enables fast wrapping and declarative integration. However, query processing in XML-based integration systems is still penalized by the lack of an algebra with adequate optimization properties and the difficulty to understand source query capabilities. In this paper, we propose an algebraic approach to support efficient XML query evaluation. We define a general purpose algebra suitable for semistructured on XML query languages. We show how this algebra can be used, with appropriate type information, to also wrap more structured query languages such as OQL or SQL. Finally, we develop new optimization techniques for XML-based integration systems.

In the past few years, query languages featuring generalized path expressions have been proposed. These languages allow the interrogation of both data and structure. They are powerful and essential for a number of applications. However, until now, their evaluation has relied on a rather naive and inefficient algorithm.In this paper, we extend an object algebra with two new operators and present some interesting rewriting techniques for queries featuring generalized path expressions. We also show how a query optimizer can integrate the new techniques.

Structured documents (e.g., SGML) can benefit a lot from database support and more specifically from object-oriented database (OODB) management systems. This paper describes a natural mapping from SGML documents into OODB's and a formal extension of two OODB query languages (one SQL-like and the other calculus) in order to deal with SGML document retrieval.