Section: New Results

Vocabularies, Semantic Web and linked data based knowledge representation

SPARQL Template Transformation Language

Participants : Olivier Corby, Catherine Faron-Zucker, Fabien Gandon, Fuqi Song.

We designed and developed a generic software environment to generate Semantic Web Servers and Linked Data Navigators (http://corese.inria.fr ) on top of the STTL SPARQL Template Transformation Language. We designed STTL transformations from RDF to HTML that enable to set up hypertext Linked Data Navigators on local or remote (e.g. DBpedia) triple stores. This work was published at ISWC, WebIST, LNBIP and IC [26] , [25] , [45] , [39] .

We extended STTL in order to perform rule based constraint checking. Templates return boolean true (resp. false) when constraint cheking succeeds (resp. fails). We applied this extension on OWL profile conformance checking and we tested with success OWL RL, OWL EL and OWL QL profiles.

SPARQL Function Language

Participants : Olivier Corby, Catherine Faron-Zucker.

We started the design of a Function Language on top of SPARQL filter language. We added the function statement that enables users to define extension functions directly in the filter language. We added statements to the filter languages such as let local variables, for loop and list datatype and we integrated select and construct queries in the language. Extension functions are directly available into SPARQL queries. This solves the problem of extension function interoperability. We were able to design custom datatypes such as roman numbers, custom aggregates such as median and standard deviation, extension functions to compute the week day of a given date, approximate search functions, recursive functions with the service clause, etc. [50] .

Graph Pattern Matching

Participants : Olivier Corby, Fuqi Song.

We proposed a heuristics-based query planning approach which allows reducing SPARQL query executing time. This approach has been developed and integrated to Corese plaform. The relevant work and results have been published at conference KES 2015 [35] .

We developed a component that can improve the storage capacity of Corese software, generally speaking this approach stores large RDF literals into the file system instead of in memory. The experiments are performed based on the data set of BSBM [54] and the results suggested that with this component, it can save up to 40% RAM space without slowing down the query execution time.

We implemented and integrated similarity measurement algorithms to Corese software in order to enable approximate semantic search. The main objective is to return approximate results when there are no results in the data source corresponding to the query.

Dynamic Application Scheme Composition

Participant : Isabelle Mirbel.

Dynamic service composition has emerged as a promising approach to build complex runtime-adaptable applications. In this context, new approaches for bottom-up opportunistic assembly of services have emerged. However, these approaches may lead to meaningless and useless compositions. Therefore, we advocate an approach in which bottom-up discovery of services is coupled with top-down user's requirements elicitation.

In our approach, application schemes publish available behaviors from basic component assembly. Our user's requirements elicitation framework, based on previous work, offers the capability to capture high-level end-user's requirements in an iterative and incremental way and to turn them into queries to retrieve application scheme's descriptions. We adopt semantic Web languages and models as a unified framework to deal with end-user's requirements and application scheme's descriptions in order to take advantage of their reasoning and traceability capabilities. We extended previous work about requirement's modeling by providing means to represent and reason on AND and OR operators as well as contextual data. Moreover, relying on the STTL language (see Section 7.3.1 , we proposed two transformations for runtime composition: the first transformation aims at detecting the possible compositions with regards to the available applications schemes; the second one aims at building a BPMN modeling to achieve user's requirements.

Semantic Web Languages And Techniques for Digital Humanities

Participants : Serena Villata, Elena Cabrio, Catherine Faron-Zucker, Franck Michel.

In the framework of the Zoomathia project, we conducted three complementary works. Their results have been published at the SW4SH international workshop [22] [37] [38] . First, together with Cécile Callou, Chloé Martin and Johan Montagnat (UNS), we started working on the construction of a thesaurus to support multi-disciplinary studies on the transmission of zoological knowledge throughout historical periods, combining the analysis of ancient literature, iconographic and archaeozoological resources. We constructed a SKOS thesaurus based on the TAXREF taxonomical reference designed to support studies in Conservation Biology.

Second, together with Molka Tounsi (UNS), and Arnaud Zucker (UNS), we have introduced a methodology to (i) extract pertinent knowledge from medieval texts using Natural Language Processing methods, (ii) semantically enrich semi-structured zoological data and publishing it as an RDF dataset and its vocabulary, linked to other relevant Linked Data sources, and (iii) reason on this linked RDF data to help epistemologists, historians and philologists in their analysis of these ancient texts.

Third, together with Arnaud Zucker, we have proposed to adopt argumentation theory together with Semantic Web languages and techniques to provide an overall view of conflicting critiques over ancient texts, and to detect what are the different competing viewpoints and what are the strongest arguments emerging from the debate. An ontology for argumentative documents is used to annotate ancient texts, and an example of such annotation is provided about the topic of the Eternity of the species in Aristotle.

Moreover, together with Ahmed Missaoui (UNS) and Sara Tonelli (FBK Trento, Italy), we have presented the process performed to map the metadata from the Verbo-Visual-Virtual Project to the Linked Open Data cloud and the related data enrichment. Although the work was largely inspired by past efforts by other cultural heritage institutions, they face new challenges, partly related to the small size of the collection, with little-known artists and few information available from other online sources, and partly to the integration of Natural Language Processing techniques to enrich the metadata. The results of this research have been published at the AIUCD international conference.

Autonomous Learning of the Meaning of Objects

Participants : Valerio Basile, Elena Cabrio, Fabien Gandon.

The goal of ALOOF (CHIST-ERA) project (http://www.dis.uniroma1.it/~aloof/ ) is to enable robots to tap into the ever-growing amount of knowledge available on the Web, by learning from there about the meaning of previously unseen objects, expressed in a form that makes them applicable when acting in situated environments. By searching the Web, robots will be able to learn about new objects, their specific properties, where they might be stored and so forth. To achieve this, robots need a mechanism for translating between the representations used in their real-world experience and those on the Web.

In this direction, we are building a machine reading pipeline to extract formally encoded knowledge from unstructured text. By combining linguistic and semantic analysis of natural language with entity linking and formal reasoning techniques, our system is capable of extracting meaningful knowledge about entities with URIs in the Linked Open Data (e.g., from DBpedia) and their relationships, encoded in standard Semantic Web fashion, i.e., RDF triples. We then employ the machine reading software to harvest the Web, targeting informative natural language resources such as educational Websites, to create a large-scale meaning bank of common sense knowledge.

Social Media Intelligence and Linked Knowledge

Participants : Farhad Nooralahzadeh, Elena Cabrio, Fabien Gandon.

Automated Natural Language Processing (NLP), Web Open Data (Linked Open Data) and social networks are the three topics of the SMILK ANR LabCom including their coupling studied in three ways: texts and Linked Data, Linked Data and social resources, texts and social resources. It is a Joint laboratory between the Inria research institute and the VISEO company to develop research and technologies on the one hand, retrieve, analyze, and reason about linking data from textual Web resources and other to use open Web data taking into account the social structures and interactions in order to improve the analysis and understanding of textual resources.

In this context, we have developed the entity discovery tools by adopting the semantic spreading activation, then we integrated it with the SMILK framework. The goal of this work was to semantically enrich the data by linking the mentions of named entities in the text to the corresponding known entities in knowledge bases. In our approach multiple aspects are considered: the prior knowledge of an entity in Wikipedia (i.e. the keyphraseness and commonness features that can be precomputed by crawling the Wikipedia dump), a set of features extracted from the input text and from the knowledge base, along with the correlation/relevancy among the resources in Linked Data. More precisely, this work explores the collective ranking approach formalized as a weighted graph model, in which the mentions in the input text and the candidate entities from knowledge bases are linked using the local compatibility and the global relatedness. Experiments on the datasets of the Open Knowledge Extraction (OKE) (https://github.com/anuzzolese/oke-challenge ) challenge with different configurations of our approach in each phase of the linking pipeline reveal its optimum mode. We investigate the notion of semantic relatedness between two entities represented as sets of neighbors in Linked Open Data that relies on an associative retrieval algorithm, with consideration of common neighborhood. This measure improves the performance of prior link-based models and outperforms the explicit inter-link relevancy measure among entities (mostly Wikipedia-centric). Thus, our approach is resilient to non-existent or sparse links among related entities.

Ontology-Based Workflow Management Systems

Participants : Tuan Anh Pham, Nhan Le Thanh.

The main objective of this PhD work is to develop a Shared Workflow Management System (SWMS) using ontology engineering. Everybody can share a semi-complete workflow which is called “Workflow template”, and other people can modify and complete it to use it in their system. This customized workflow is called “Personalized workflow”. The challenges of a SWMS are to be simple, easy to use, user friendly and not too heavy. But it must have all functions of a WMS. There are three major challenges in this work: How to allow the users to customize the workflow template to correspond to their requirements, with changes compliant with the predefined rules in the workflow template? How to build an execution model to evaluate step by step a personalized workflow [34] [33] .

Semantic Mappings with a Control Flow-Based Business Workflow

Participants : Thi Hoa Hue Nguyen, Nhan Le Thanh.

The aim of this PhD work is to improve the Coloured Petri Nets (CPNs) and Ontology engineering to support the development of business process and business workflow definitions of various fields. To realize this objective, we first propose an ontological approach for representing business models in a meta-knowledge base. We introduce four basic types of manipulation operations on process models used to develop and modify business workflow patterns. Second, we propose a formal definition of semantic constraints and an O(n3)-time algorithm for detecting redundant and conflicting constraints. By relying on the CPN Ontology and sets of semantic constraints, workflow processes are semantically created. Finally, we show how to check the semantic correctness of workflow processes with the SPARQL query language [34] .