Enabling and automating interoperability in the Semantic Web of Things involves complex semantic reasoning tools to process knowledge graphs. To improve performance and energy efficiency, such tools should be deployed as close as possible to the devices, ideally on any available edge node. However, edge nodes often lack sufficient resources, especially memory. We propose a method to distribute reasoning in this context. We then evaluate three algorithms to plan the distribution over a network of heterogeneous nodes. These algorithms take into account architectural constraints such as the position of sensors and actuators and the available resources on each node, while minimizing costly data exchanges on the network.

International audience

Navigating large knowledge bases made of billions of triples is very challenging. In this demonstration, we showcase Fouilla, a topical Knowledge Base browser that offers a seamless navigational experience of DBpedia. We propose an original approach that leverages both structural and semantic contents of Wikipedia to enable a topic-oriented filter on DBpedia entities. We devise an approach to drastically reduce the query time and to ensure a seamless browsing experience to the end user. We demonstrate how our system offers a novel and meaningful experience of DBpedia browsing by challenging the user to search for relevant information within the Knowledge Base in different use cases.

International audience

Hady Elsahar, Christophe Gravier, Frederique Laforest. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.

Automatic generation of questions has recently received attention, as an indirect consequence of the renewed interest in Question Answering systems. Yet, most Automatic Question Generation systems focused on tasks like question selection, verbalization, distractor generation and difficulty assessment. In this paper, we come up with a novel task for Question Generation systems: thematic question generation. Inspired by a famous trivia board game, we aim at solving the problem of generating meaningful questions and their distractors for common knowledge topics. In this paper, we develop an end-to-end system that tackles these issues. We use the Wikipedia structure and content to determine the topics and their boundaries. We developed a template based approach to generate questions, allowing complex questions generation from binary and n-ary statements. To automatically generate templates, we developed an approach that reverts templates used for questions answering, allowing us to import more than 2000 templates. Our experimental campaign reports a success in topic assignment of 0.69 and very high scores (>0.9) for questions and distractors quality along with high inter-rater agreements.

Most people need textual or visual interfaces in order to make sense of Semantic Web data. In this paper, we investigate the problem of generating natural language summaries for Semantic Web data using neural networks. Our end-to-end trainable architecture encodes the information from a set of triples into a vector of fixed dimensionality and generates a textual summary by conditioning the output on the encoded vector. We explore a set of different approaches that enable our models to verbalise entities from the input set of triples in the generated text. Our systems are trained and evaluated on two corpora of loosely aligned Wikipedia snippets with triples from DBpedia and Wikidata, with promising results.

Docker is the most popular open-source software for managing software containers. Docker is widely used in the industry and is becoming the de-facto standard. In this paper, we tackle the issue of bringing Docker into the Linked Open Data Cloud to enhance docker images descriptions and search. We first present the construction of the ontology that describes Docker images. This ontology is interconnected with DBPedia and reuses SIOC and prov vocabularies. We then describe the automated process that automatically extracts data from the Docker Hub Registry to populate the aforementioned ontology. We also present a Web platform that provides a SPARQL endpoint as well as analytical insights into the dataset.

Top-k queries over data streams is a well studied problem. There exists numerous systems allowing to process continuous queries over sliding windows. At the opposite, non-append only streams call for ad-hoc solutions, e.g. tailor-made solutions implemented in a mainstream programming language. In the meantime, the Stream API and lambda expressions have been added in Java 8, thus gaining powerful operations for data stream processing. However, the Java Collections Framework does not provide data structures to safely and conveniently support sorted collections of evolving data. In this paper, we demonstrate Upsortable, an annotation-based approach that allows to use existing sorted collections from the standard Java API for dynamic data management. Our approach relies on a combination of pre-compilation abstract syntax tree modifications and runtime analysis of bytecode. Upsortable offers the developer a safe and time-efficient solution for developing top-k queries on data streams while keeping a full compatibility with standard Java.

This paper proposes a new approach for the the incremental evaluation of RDF graph streams over sliding windows. Our system, called "SPECTRA", combines a novel formof RDF graph summarisation, a new incremental evaluation method and adaptive indexing techniques. We materialise the summarised graph from each event using vertically partitioned views to facilitate the fast hash-joins for all types of queries. Our incremental and adaptive indexing is a byproduct of query processing, and thus provides considerable advantages over offline and online indexing. Furthermore, contrary to the existing approaches, we employ incremental evaluation of triples within a window. This results in considerable reduction in response time, while cutting the unnecessary cost imposed by recomputation models for each triple insertion and eviction within a defined window. We show that our resulting system is able to cope with complex queries and datasets with clear benefits. Our experimental results on both synthetic and real-world datasets show up to an order of magnitude of performance improvements as compared to state-of-the-art systems.

The advent of semantic data on the Web requires efficient reasoning systems to infer RDF and OWL data. The linked nature and the huge volume of data entail efficiency and scalability challenges when designing productive inference systems. This paper presents Inferray, an implementation of RDFS, ρdf, and RDFS-Plus inference with improved performance over existing solutions. The main features of Inferray are 1) a storage layout based on vertical partitioning that guarantees sequential access and efficient sort-merge join inference; 2) efficient sorting of pairs of 64-bit integers using ad-hoc optimizations on MSD radix and a custom counting sort; 3) a dedicated temporary storage to perform efficient graph closure computation. Our measurements on synthetic and real-world datasets show improvements over competitors on RDFS-Plus, and up to several orders of magnitude for transitivity closure.

Real-time recommendation of Twitter users based on the content of their profiles is a very challenging task. Traditional IR methods such as TF-IDF fail to handle efficiently large datasets. In this paper we present a scalable approach that allows real time recommendation of users based on their tweets. Our model builds a graph of terms, driven by the fact that users sharing similar interests will share similar terms. We show how this model can be encoded as a compact binary footprint, that allows very fast comparison and ranking, taking full advantage of modern CPU architectures. We validate our approach through an empirical evaluation against the Apache Lucene’s implementation of TF-IDF. We show that our approach is in average two hundred times faster than standard optimized implementation of TF-IDF with a precision of 58%.

This paper presents a research survey on the Home Health Care. It is a way of caring for patients who suffer from complex, non-serious diseases and especially evolutionary ones. At their homes, they benefit from medical and paramedical cares that can be only provided by hospitals. This form of treatment has a positive impact on both the patients and the state. To study its evolution as well as its development and to understand the difference in this new mode of care, we have devoted the first part of this investigation for the home health care history. In the second part, based on official data published on the websites of the European and American States, we have analyzed home health care centers, the most recognized ones at the national and internal levels, to distinguish similarities and differences between them and especially to determine their specifications and limitations. To overcome the latter, it is necessary to dig new leads of research; that is why the founders of these institutions have resorted to scientists. The third section covers the state of the art since the 70s. In this part, we have depicted the research work that contributed to resolve home health care problems.

The Semantic Web has gained substantial momentum over the last decade. It contributes to the manifestation of knowledge from data, and leverages implicit knowledge through reasoning algorithms. The main drawbacks of current reasoning methods over ontologies are two-fold: first they struggle to provide scalability for large datasets, and second, the batch processing reasoners who provide the best scalability so far are unable to infer knowledge from evolving data. We contribute to solving these problems by introducing Slider, an efficient incremental reasoner. Slider goes a significant step beyond existing system, including i) performance, by more than a 70% improvement in average compared to the fastest reasoner available to the best of our knowledge, and ii) inferences on streams of semantic data, by using intrinsic features that are themselves streams-oriented. Slider is fragment agnostic and conceived to handle expanding data with a growing background knowledge base. It natively supports pdf and RDFS, and its architecture allows to extend it to more complex fragments with a minimal effort. In this demo a web-based interface allows the users to visualize the internal behaviour of Slider during the inference, to better understand its design and principles.

Microblogging websites such as Twitter produce tremendous amount of data each second. Consequently, real-time recommendation systems require very efficient algorithm to quickly proceed this massive amount of data. In this paper we propose a scalable and extensible way of building content-based user profiles. Scalability refers to the relative complexity of algorithms involved in building the users profiles with respect to state-of-the-art solutions. Extensibility consider avoiding to recompute the model for newcomers. We present a tractable algorithm to build user profiles out of their tweets. Our model is a graph of terms cooccurency, driven by the fact that user sharing similar interests will share similar terms. We then present how this model can be encoded as a binary footprint, hence boosting comparison of users. We provide an empirical study to measure how the distance between users in the hash space differs from distance between users using standard Information Retrieval techniques. This experiment is based on a Twitter dataset we crawled, and represents 25K users and 1 million tweets. Our approach is driven by real-time analysis requirements and is thus oriented on a trade-off between expressivity and efficiency. Experimental results shows that our approach outperforms vector space model by three orders of magnitude, with a precision of 58%.

One of the leading challenges for pervasive computing is to ease the application development to smoothly handle the surrounding environment. We consider the case where the environment produces heterogeneous and continuous data, e. g. temperature readings, car positions... We have defined a scenario for containers transportation tracking in a medical context involving the transportation of fragile biological matter in sensor-enhanced containers. This scenario has been simulated as a testbed and offers a very nice setting to measure the agility of data-centric application development.

Querying non-conventional data is recognized as a major issue in new environments and applications such as those occurring in pervasive computing. A key issue is the ability to query data, streams and services in a declarative way. Our overall objective is to make the development of pervasive applications easier through database principles. In this paper, through the notion of virtual attributes and binding patterns, we define a data-centric view of pervasive environments: the classical notion of database is extended to come up with a broader notion, defined as relational pervasive environment, integrating data, streams and active/passive services. Then, the so-called Serena algebra is proposed with operators to homogeneously handle data and services. Moreover, the notion of stream can also be smoothly integrated into this algebra. A prototype of Pervasive Environment Management System has been implemented on which first experiments have been conducted to validate our approach.

XML semi-structured documents provide a flexible and natural way to capture data. More often, users exchange documents with others or reuse their data in a wide range of tasks. A direct consequence is a growing need to manage semi-structured documents and to gain conventional database functionality. The aim of our research is the development of a comprehensive capturing and mapping data system that ensures flexible and well-adapted information capture and at the same time efficient information retrieval. A transformation process allows the mapping between the semi-structured document model and the traditional database model. It handles data extraction and data restructuring and deals with data mapping rules to feed traditional databases.

Electronic documents have become a universal way of communication due to Web expansion. But using structured information stored in databases is still essential for data coherence management, querying facilities... We thus face a classical problem known as "impedance mismatch" in the database world: two antagonist approaches have to collaborate. Using documents at the end-user interface level provides simplicity and flexibility. But it is possible to take documents as data sources only if helped by a human being: automatic documents analysis systems have a significant error rate. Databases are an alternative as semantics and format of information are strict: queries via SQL provide 100% correct responses. The aim of this work is to provide a system that associates document capture freedom with database storage structure. The system we propose does not intend to be universal. It can be used in specific cases where people usually work with technical documents dedicated to a particular domain. Our examples concern medicine and more explicitly medical records. Computerization has very often been rejected by physicians because it necessitates too much standardization, and form-based user interfaces are not adapted to their daily practice. In this domain, we think that this study provides a viable alternative approach. This system offers freedom to doctors: they would fill in documents with the information they want to store, in a convenient order and in a more free way. We have developed a system that allows users to fill in a database quasi automatically from document paragraphs. The database used is an already existing database, that can be queried in a classical way for statistical studies or epidemiological purposes. In this system, the document fund and the database containing extractions from documents coexist. Queries are sent to the database, answers include data from the database and references to source documents.