The Semantic Web – ISWC 2012

The LOD2 Stack is an integrated distribution of aligned tools which support the whole life cycle of Linked Data from extraction, authoring/creation via enrichment, interlinking, fusing to maintenance. The LOD2 Stack comprises new and substantially extended existing tools from the LOD2 project partners and third parties. The stack is designed to be versatile; for all functionality we define clear interfaces, which enable the plugging in of alternative third-party implementations. The architecture of the LOD2 Stack is based on three pillars: (

1

) Software integration and deployment using the Debian packaging system. (

2

) Use of a central SPARQL endpoint and standardized vocabularies for knowledge base access and integration between the different tools of the LOD2 Stack. (

3

) Integration of the LOD2 Stack user interfaces based on REST enabled Web Applications. These three pillars comprise the methodological and technological framework for integrating the very heterogeneous LOD2 Stack components into a consistent framework. In this article we describe these pillars in more detail and give an overview of the individual LOD2 Stack components. The article also includes a description of a real-world usage scenario in the publishing domain.

The success of pervasive computing depends on the ability to compose a multitude of networked applications dynamically in order to achieve user goals. However, applications from different providers are not able to interoperate due to incompatible interaction protocols or disparate data models. Instant messaging is a representative example of the current situation, where various competing applications keep emerging. To enforce interoperability at runtime and in a non-intrusive manner,

mediators

are used to perform the necessary translations and coordination between the heterogeneous applications. Nevertheless, the design of mediators requires considerable knowledge about each application as well as a substantial development effort. In this paper we present an approach based on ontology reasoning and model checking in order to generate correct-by-construction mediators automatically. We demonstrate the feasibility of our approach through a prototype tool and show that it synthesises mediators that achieve efficient interoperation of instant messaging applications.

To realize the Smart Cities vision, applications can leverage the large availability of open datasets related to urban environments. Those datasets need to be integrated, but it is often hard to automatically achieve a high-quality interlinkage. Human Computation approaches can be employed to solve such a task where machines are ineffective. We argue that in this case not only people’s background knowledge is useful to solve the task, but also people’s physical presence and direct experience can be successfully exploited. In this paper we present UrbanMatch, a Game with a Purpose for players in mobility aimed at validating links between points of interest and their photos; we discuss the design choices and we show the high throughput and accuracy achieved in the interlinking task.

In this paper we discuss our experience with the design, development and deployment of the

ourSpaces

Virtual Research Environment.

ourSpaces

makes use of Semantic Web technologies to create a platform to support multi-disciplinary research groups. This paper introduces the main semantic components of the system: a framework to capture the provenance of the research process, a collection of services to create and visualise metadata and a policy reasoning service. We also describe different approaches to support interaction between users and metadata within the VRE. We discuss the lessons learnt during the deployment process with three case study groups. Finally, we present our conclusions and future directions for exploration in terms of developing

ourSpaces

further.

Many industrial use cases, such as machine diagnostics, can benefit from

embedded reasoning

, the task of running knowledge-based reasoning techniques on embedded controllers as widely used in industrial automation. However, due to the memory and CPU restrictions of embedded devices like programmable logic controllers (PLCs), state-ofthe- art reasoning tools and methods cannot be easily migrated to industrial automation environments. In this paper, we describe an approach to porting lightweight OWL 2 EL reasoning to a PLC platform to run in an industrial automation environment. We report on initial runtime experiments carried out on a prototypical implementation of a PLC-based

$\mathcal{EL}$

+-reasoner in the context of a use case about turbine diagnostics.

Semantic annotation of patient data in the skeletal dysplasia domain (e.g., clinical summaries) is a challenging process due to the structural and lexical differences existing between the terms used to describe radiographic findings. In this paper we propose an ontology aimed at representing the intrinsic structure of such radiographic findings in a standard manner, in order to bridge the different lexical variations of the actual terms. Furthermore, we describe and evaluate an algorithm capable of mapping concepts of this ontology to exact or broader terms in the main phenotype ontology used in the bone dysplasia domain.

This paper proposes to apply semantic technologies in a new domain, Field research. It is said that if “raw data” is openly available on the Web, it will be used by other people to do wonderful things. But, it would be better to show a use case together with that data, especially in the dawn of LOD. Therefore, we are proceeding with both of LOD content generation and its application for a specific domain. The application addresses an issue of information retrieval in the field, and the mechanism of LOD generation from the Web might be applied to the other domain. Firstly, we demonstrate the use of our mobile application, which searches a plant fitting the environmental conditions obtained by the smartphone’s sensors. Then, we introduce our approach of the LOD generation, and present an evaluation showing its practical effectiveness.

Diagnosis, or the method to connect causes to its effects, is an important reasoning task for obtaining insight on cities and reaching the concept of sustainable and smarter cities that is envisioned nowadays. This paper, focusing on transportation and its road traffic, presents how road traffic congestions can be detected and diagnosed in quasi real-time. We adapt pure Artificial Intelligence diagnosis techniques to fully exploit knowledge which is captured through relevant semantics-augmented stream and static data from various domains. Our prototype of semantic-aware diagnosis of road traffic congestions, experimented in Dublin Ireland, works efficiently with large, heterogeneous information sources and delivers value-added services to citizens and city managers in quasi real-time.

Despite decades of effort, intelligent object search remains elusive. Neither search engine nor semantic web technologies alone have managed to provide usable systems for simple questions such as “find me a flat with a garden and more than two bedrooms near a supermarket.”

We introduce

deqa

, a conceptual framework that achieves this elusive goal through combining state-of-the-art semantic technologies with effective data extraction. To that end, we apply

deqa

, to the UK real estate domain and show that it can answer a significant percentage of such questions correctly.

deqa

achieves this by mapping natural language questions to

Sparql

patterns. These patterns are then evaluated on an RDF database of current real estate offers. The offers are obtained using

OXPath

, a state-of-the-art data extraction system, on the major agencies in the Oxford area and linked through

Limes

to background knowledge such as the location of supermarkets.

In this paper, we present QuerioCity, a platform to catalog, index and query highly heterogenous information coming from complex systems, such as cities. A series of challenges are identified: namely, the heterogeneity of the domain and the lack of a common model, the volume of information and the number of data sets, the requirement for a low entry threshold to the system, the diversity of the input data, in terms of format, syntax and update frequency (streams vs static data), and the sensitivity of the information. We propose an approach for incremental and continuous integration of static and streaming data, based on Semantic Web technologies. The proposed system is unique in the literature in terms of handling of multiple integrations of available data sets in combination with flexible provenance tracking, privacy protection and continuous integration of streams. We report on lessons learnt from building the first prototype for Dublin.

Biomedical ontologies have become a mainstream topic in medical research. They represent important sources of evolved knowledge that may be automatically integrated in decision support methods. Grounding clinical and radiographic findings in concepts defined by a biomedical ontology, e.g., the Human Phenotype Ontology, enables us to compute semantic similarity between them. In this paper, we focus on using such similarity measures to predict disorders on undiagnosed patient cases in the bone dysplasia domain. Different methods for computing the semantic similarity have been implemented. All methods have been evaluated based on their support in achieving a higher prediction accuracy. The outcome of this research enables us to understand the feasibility of developing decision support methods based on ontology-driven semantic similarity in the skeletal dysplasia domain.

BioPortal is a repository of biomedical ontologies—the largest such repository, with more than 300 ontologies to date. This set includes ontologies that were developed in OWL, OBO and other languages, as well as a large number of medical terminologies that the US National Library of Medicine distributes in its own proprietary format. We have published the RDF based serializations of all these ontologies and their metadata at

sparql.bioontology.org

. This dataset contains 203M triples, representing both content and metadata for the 300+ ontologies; and 9M mappings between terms. This endpoint can be queried with SPARQL which opens new usage scenarios for the biomedical domain. This paper presents lessons learned from having redesigned several applications that today use this SPARQL endpoint to consume ontological data.

Our work is settled in the context of the public administration domain, where data can come from different entities, can be produced, stored and delivered in different formats and can have different levels of quality. Hence, such a heterogeneity has to be addressed, while performing various data integration tasks. We report our experimental work on publishing some government linked open geo-metadata and geo-data of the Italian Trentino region. Specifically, we illustrate how 161 core geographic datasets were released by leveraging on the geo-catalogue application within the existing geo-portal. We discuss the lessons we learned from deploying and using the application as well as from the released datasets.

Robust solutions for ambient assisted living are numerous, yet predominantly specific in their scope of usability. In this paper, we describe the potential contribution of semantic web technologies to building more versatile solutions — a step towards adaptable context-aware engines and simplified deployments. Our conception and deployment work in hindsight, we highlight some implementation challenges and requirements for semantic web tools that would help to ease the development of context-aware services and thus generalize real-life deployment of semantically driven assistive technologies. We also compare available tools with regard to these requirements and validate our choices by providing some results from a real-life deployment.

It has become common to use RDF to store the results of Natural Language Processing (NLP) as a graph of the entities mentioned in the text with the relationships mentioned in the text as links between them. These NLP graphs can be measured with Precision and Recall against a ground truth graph representing what the documents actually say. When asking conjunctive queries on NLP graphs, the Recall of the query is expected to be roughly the product of the Recall of the relations in each conjunct. Since Recall is typically less than one, conjunctive query Recall on NLP graphs degrades geometrically with the number of conjuncts. We present an approach to address this Recall problem by hypothesizing links in the graph that would improve query Recall, and then attempting to find more evidence to support them. Using this approach, we confirm that in the context of answering queries over NLP graphs, we can use lower confidence results from NLP components if they complete a query result.

Questions often explicitly request a particular type of answer. One popular approach to answering natural language questions involves filtering candidate answers based on precompiled lists of instances of common answer types (e.g., countries, animals, foods, etc.). Such a strategy is poorly suited to an open domain in which there is an extremely broad range of types of answers, and the most frequently occurring types cover only a small fraction of all answers. In this paper we present an alternative approach called TyCor, that employs soft filtering of candidates using multiple strategies and sources. We find that TyCor significantly outperforms a single-source, single-strategy hard filtering approach, demonstrating both that multi-source multi-strategy outperforms a single source, single strategy, and that its fault tolerance yields significantly better performance than a hard filter.

Semantic Web allows us to model and query time-invariant or slowly evolving knowledge using ontologies. Emerging applications in Cyber Physical Systems such as Smart Power Grids that require continuous information monitoring and integration present novel opportunities and challenges for Semantic Web technologies. Semantic Web is promising to model diverse Smart Grid domain knowledge for enhanced situation awareness and response by multi-disciplinary participants. However, current technology does pose a performance overhead for dynamic analysis of sensor measurements. In this paper, we combine semantic web and complex event processing for stream based semantic querying. We illustrate its adoption in the USC Campus Micro-Grid for detecting and enacting dynamic response strategies to peak power situations by diverse user roles. We also describe the semantic ontology and event query model that supports this. Further, we introduce and evaluate caching techniques to improve the response time for semantic event queries to meet our application needs and enable sustainable energy management.

Usability and user satisfaction are of paramount importance when designing interactive software solutions. Furthermore, the optimal design can be dependent not only on the task but also on the type of user. Evaluations can shed light on these issues; however, very few studies have focused on assessing the usability of semantic search systems. As semantic search becomes mainstream, there is growing need for standardised, comprehensive evaluation frameworks. In this study, we assess the usability and user satisfaction of different semantic search query input approaches (natural language and view-based) from the perspective of different user types (experts and casuals). Contrary to previous studies, we found that casual users preferred the form-based query approach whereas expert users found the graph-based to be the most intuitive. Additionally, the controlled-language model offered the most support for casual users but was perceived as restrictive by experts, thus limiting their ability to express their information needs.

This paper presents an evaluation of state of the art black box justification finding algorithms on the NCBO BioPortal ontology corpus. This corpus represents a set of naturally occurring ontologies that vary greatly in size and expressivity. The results paint a picture of the performance that can be expected when finding all justifications for entailments using black box justification finding techniques. The results also show that many naturally occurring ontologies exhibit a rich justificatory structure, with some ontologies having extremely high numbers of justifications per entailment.

Linked Stream Data, i.e., the RDF data model extended for representing stream data generated from sensors social network applications, is gaining popularity. This has motivated considerable work on developing corresponding data models associated with processing engines. However, current implemented engines have not been thoroughly evaluated to assess their capabilities. For reasonable systematic evaluations, in this work we propose a novel, customizable evaluation framework and a corresponding methodology for realistic data generation, system testing, and result analysis. Based on this evaluation environment, extensive experiments have been conducted in order to compare the state-of-the-art LSD engines wrt. qualitative and quantitative properties, taking into account the underlying principles of stream processing. Consequently, we provide a detailed analysis of the experimental outcomes that reveal useful findings for improving current and future engines.

Testbeds proposed so far to evaluate, compare, and eventually improve SPARQL query federation systems have still some limitations. Some variables and configurations that may have an impact on the behavior of these systems (e.g., network latency, data partitioning and query properties) are not sufficiently defined; this affects the results and repeatability of independent evaluation studies, and hence the insights that can be obtained from them. In this paper we evaluate FedBench, the most comprehensive testbed up to now, and empirically probe the need of considering additional dimensions and variables. The evaluation has been conducted on three SPARQL query federation systems, and the analysis of these results has allowed to uncover properties of these systems that would normally be hidden with the original testbeds.

We tackle the problem of improving the relevance of automatically selected tags in large-scale ontology-based information systems. Contrary to traditional settings where tags can be chosen arbitrarily, we focus on the problem of recommending tags (e.g., concepts) directly from a collaborative, user-driven ontology. We compare the effectiveness of a series of approaches to select the best tags ranging from traditional IR techniques such as TF/IDF weighting to novel techniques based on ontological distances and latent Dirichlet allocation. All our experiments are run against a real corpus of tags and documents extracted from the ScienceWise portal, which is connected to

ArXiv

.

org

and is currently used by growing number of researchers. The datasets for the experiments are made available online for reproducibility purposes.

In this paper we present the Quonto Inconsistent Data handler (QuID). QuID is a reasoner for OWL 2 QL that is based on the system Quonto and is able to deal with inconsistent ontologies. The central aspect of QuID is that it implements two different, orthogonal strategies for dealing with inconsistency: ABox repairing techniques, based on data manipulation, and consistent query answering techniques, based on query rewriting. Moreover, by exploiting the ability of Quonto to delegate the management of the ABox to a relational database system (DBMS), such techniques are potentially able to handle very large inconsistent ABoxes. For the above reasons, QuID allows for experimentally comparing the above two different strategies for inconsistency handling in the context of OWL 2 QL. We thus report on the experimental evaluation that we have conducted using QuID. Our results clearly point out that inconsistency-tolerance in OWL 2 QL ontologies is feasible in practical cases. Moreover, our evaluation singles out the different sources of complexity for the data manipulation technique and the query rewriting technique, and allows for identifying the conditions under which one method is more efficient than the other.

In recent years, strategies for Linked Data consumption have caught attention in Semantic Web research. For direct consumption by users, Linked Data mashups, interfaces, and visualizations have become a popular research area. Many approaches in this field aim to make Linked Data interaction more user friendly to improve its accessibility for non-technical users. A subtask for Linked Data interfaces is to present entities and their properties in a concise form. In general, these summaries take individual attributes and sometimes user contexts and preferences into account. But the objective evaluation of the quality of such summaries is an expensive task. In this paper we introduce a game-based approach aiming to establish a ground truth for the evaluation of entity summarization. We exemplify the applicability of the approach by evaluating two recent summarization approaches.

We present a question answering system architecture which processes natural language questions in a pipeline consisting of five steps: i) question parsing and query template generation, ii) lookup in an inverted index, iii) string similarity computation, iv) lookup in a lexical database in order to find synonyms, and v) semantic similarity computation. These steps are ordered with respect to their computational effort, following the idea of layered processing: questions are passed on along the pipeline only if they cannot be answered on the basis of earlier processing steps, thereby invoking computationally expensive operations only for complex queries that require them. In this paper we present an evaluation of the system on the dataset provided by the 2nd Open Challenge on Question Answering over Linked Data (QALD-2). The main, novel contribution is a systematic empirical investigation of the impact of the single processing components on the overall performance of question answering over linked data.

Since 2001, the semantic web community has been working hard towards creating standards which will increase the accessibility of available information on the web. Yahoo research recently reported that 30% of all HTML pages contain structured data such as microdata, RDFa, or microformat. Although multilinguality of the web is a hurdle in information access, the rapid growth of the semantic web enables us to retrieve fine grained information across the language barrier. In this thesis, firstly, we focus on developing a methodology to perform cross-lingual semantic search over structured data (knowledge base), by transforming natural language queries into SPARQL. Secondly, we focus on improving the semantic similarity and relatedness measures, to overcome the semantic gap between the vocabulary in the knowledge base and the terms appearing in the query. The preliminary results are evaluated against the QALD-2 test dataset, which achieved a F1 score of 0.46, an average precision of 0.44, and an average recall of 0.48.

Assessing the quality of data published on the Web has been identified as an essential step in selecting reliable information for use in tasks such as decision making. This paper discusses a quality assessment framework based on semantic web technologies and outlines a role for provenance in supporting and documenting such assessments.

Personalization techniques aim at helping people dealing with the ever growing amount of information by filtering it according to their interests. However, to avoid the information overload, such techniques often create an over-personalization effect,

i.e.

users are exposed

only

to the content systems assume they would like. To break this “personalization bubble” we introduce the notion of

serendipity

as a performance measure for recommendation algorithms. For this, we first identify aspects from the user perspective, which can determine level and type of serendipity desired by users. Then, we propose a user model that can facilitate such user requirements, and enables serendipitous recommendations. The use case for this work focuses on TV recommender systems, however the ultimate goal is to explore the transferability of this method to different domains. This paper covers the work done in the first eight months of research and describes the plan for the entire PhD trajectory.

Provenance is an increasingly important aspect of data management that is often underestimated and neglected by practitioners. In our work, we target the problem of reconstructing provenance of files in a shared folder setting, assuming that only standard filesystem metadata are available. We propose a content-based approach that is able to reconstruct provenance automatically, leveraging several similarity measures and edit distance algorithms, adapting and integrating them into a multi-signal pipeline. We discuss our research methodology and show some promising preliminary results.

Due to recent developments in reasoning algorithms of the various OWL profiles, the classification time for an ontology has come down drastically. For all of the popular reasoners, in order to process an ontology, an implicit assumption is that the ontology should fit in primary memory. The memory requirements for a reasoner are already quite high, and considering the ever increasing size of the data to be processed and the goal of making reasoning Web scale, this assumption becomes overly restrictive. In our work, we study several distributed classification approaches for the description logic EL+ (a fragment of OWL 2 EL profile). We present the lessons learned from each approach, our current results, and plans for future work.

With the Semantic Web scaling up, and more triple-stores with

update

facilities being available, the need for higher levels of simultaneous triple-stores with identical information becomes more and more urgent. However, where such Data Replication approaches are common in the database community, there is no comprehensive approach for data replication for the Semantic Web. In this research proposal, we will discuss the problem space and scenarios of data replication in the Semantic Web, and explain how we plan on dealing with this issue.

Due to the decentralized nature of the Semantic Web, the same real world entity may be described in various data sources and assigned syntactically distinct identifiers. In order to facilitate data utilization in the Semantic Web, without compromising the freedom of people to publish their data, one critical problem is to appropriately interlink such heterogeneous data. This interlinking process can also be referred to as

Entity Coreference

, i.e., finding which identifiers refer to the same real world entity. This proposal will investigate algorithms to solve this entity coreference problem in the Semantic Web in several aspects. The essence of entity coreference is to compute the similarity of instance pairs. Given the diversity of domains of existing datasets, it is important that an entity coreference algorithm be able to achieve good precision and recall across domains represented in various ways. Furthermore, in order to scale to large datasets, an algorithm should be able to intelligently select what information to utilize for comparison and determine whether to compare a pair of instances to reduce the overall complexity. Finally, appropriate evaluation strategies need to be chosen to verify the effectiveness of the algorithms.

Data streams are being continually generated in diverse application domains such as traffic monitoring, smart buildings, and so on. Stream Reasoning is the area that aims to combine reasoning techniques with data streams. In this paper, we present our approach to enable rule-based reasoning on semantic data streams using data flow networks in a distributed manner.

Designing domain ontologies from scratch is a time-consuming endeavor requiring a lot of close collaboration with domain experts. However, domain descriptions such as XML Schemas are often available in early stages of the ontology development process. For my dissertation, I propose a method to convert XML Schemas to OWL ontologies in an automatic way. The approach addresses the transformation of any XML Schema documents by using the XML Schema metamodel, which is completely represented by the XML Schema Metamodel Ontology. Automatically, all Schema declarations and definitions are converted to class axioms, which are intended to be enriched with additional domain-specific semantic information in form of domain ontologies.

In this paper, we present a doctoral thesis which introduces a new approach of time series enrichment with semantics. The paper shows the problem of assigning time series data to the right party of interest and why this problem could not be solved so far. We demonstrate a new way of processing semantic time series and the consequential ability of addressing users. The combination of time series processing and Semantic Web technologies leads us to a new powerful method of data processing and data generation, which offers completely new opportunities to the expert user.

In real world cases, building

reliable problem centric views

over Linked Data [1] is a challenging task. An ideal method should include a formal representation of the requirements of the needed dataset and a controlled process moving from the original sources to the outcome. We believe that a goal oriented approach, similar to the AI planning problem, could be successful in controlling the process of linked data fusion, as well as to formalize the relations between requirements, process and result.

Knowledge interaction in Web context is a challenging problem. For instance, it requires to deal with complex structures able to filter knowledge by drawing a meaningful context boundary around data. We assume that these complex structures can be formalized as Knowledge Patterns (KPs), aka frames. This Ph.D. work is aimed at developing methods for extracting KPs from the Web and at applying KPs to exploratory search tasks. We want to extract KPs by analyzing the structure of Web links from rich resources, such as Wikipedia.

Complex event processing is currently done primarily with proprietary definition languages. Future smart environments will require collaboration of multi-platform sensors operated by multiple parties. The goal of my research is to verify the applicability of standard-compliant SPARQL for complex event processing tasks. If successful, semantic web standards RDF, SPARQL and OWL with their established base of tools have many other benefits for event processing including support for interconnecting disjoint vocabularies, enriching event information with linked open data and reasoning over semantically annotated content. A software platform capable of continuous incremental evaluation of multiple parallel SPARQL queries is a key enabler of the approach.

A pair of RDF instances are said to corefer when they are intended to denote the same thing in the world, for example, when two nodes of type foaf:Person describe the same individual. This problem is central to integrating and inter-linking semi-structured datasets. We are developing an online, unsupervised coreference resolution framework for heterogeneous, semi-structured data. The online aspect requires us to process new instances as they appear and not as a batch. The instances are heterogeneous in that they may contain terms from different ontologies whose alignments are not known in advance. Our framework encompasses a two-phased clustering algorithm that is both flexible and distributable, a probabilistic multidimensional attribute model that will support robust schema mappings, and a consolidation algorithm that will be used to perform instance consolidation in order to improve accuracy rates over time by addressing data spareness.

We address the problem of developing a scaleable composition framework for Linked Data-based services, that retains the advantages of the loose coupling fostered by REST.