Scaling up the DBpedia extraction framework by Nilesh Chakraborty

The DBpedia extraction framework extracts different kinds of structured information from Wikipedia to generate various datasets. Performing a full extraction of Wikipedia dumps of all languages (or even just the mapping-based languages) takes a significant amount of time. The distributed extraction framework runs the extraction on top of Apache Spark so that users can leverage multi-core machines or a distributed cluster of commodity machines to perform faster extraction. For example, performing extraction of the 30-40 mapping based languages on a machine with a quad-core CPU and 16G RAM takes about 36 hours. Running the distributed framework in the same setting using three such worker nodes takes around 10 hours. It’s easy to achieve faster running times by adding more cores or more machines. Apart from the Spark-based extraction framework, we have also implemented a distributed wiki-dump downloader to download Wikipedia dumps for multiple languages, from multiple mirrors, on a cluster in parallel. This is still a work in progress, and in this talk I will discuss the methods and challenges involved in this project, and our immediate goals and timeline.

Automating RDF Dataset Transformation and Enrichment by Mohamed Sherif

With the adoption of RDF across several domains, come growing requirements pertaining to the completeness and quality of RDF datasets. Currently, this problem is most commonly addressed by manually devising means of enriching an input dataset. The few tools that aim at supporting this endeavour usually focus on supporting the manual definition of enrichment pipelines. In this talk, we present a supervised learning approach based on a refinement operator for enriching RDF datasets. We show how we can use exemplary descriptions of enriched resources to generate accurate enrichment pipelines. We evaluate our approach against eight manually defined enrichment pipelines and show that our approach can learn accurate pipelines even when provided with a small number of training examples.

Structured Machine Learning in Life Science (PhD progress report) by Patrick Westphal

The utilization of machine learning techniques to solve life science tasks has become widespread within the last years. Mainly working on unstructured data one question is whether such techniques could benefit from the provision of structured background knowledge. One prevalent way to express background knowledge in the life sciences is the Web Ontology Language (OWL). Accordingly there is a great variety of different domain ontologies covering anatomy, genetics, biological processes or chemistry that can be used to form structured machine learning approaches in the life science domain. The talk will give a brief overview of tasks and problems of structured machine learning in life science. Besides the special characteristics observed when applying the state-of-the-art concept learning approaches to life science tasks, a short description of the actual differences to concept learning setups in other domains is given. Further, some directions for machine learning based techniques are shown that could support concept learning in life science tasks.

About the AKSW Colloquium

This event is part of a series of events about Semantic Web technology. Please see http://wiki.aksw.org/Colloquium for further information about previous and future events. As always, Bachelor and Master students are able to get points for attendance and there is complimentary coffee and cake after the session.

This colloquium features two talks. First the Self-Service Semantic Suite (S4) platform is presented by Marin Dimitrov (Ontotext), followed up by Jörg Unbehauens report on Googles effort on using factual correctness as a ranking factor.

RDF database-as-a-service (DBaaS) via Self-Service Semantic Suite (S4) platform

In this talk Marin Dimitrov (Ontotext) will introduce the RDFdatabase-as-a-service (DBaaS) options for managing RDF data in the Cloud via the Self-Service Semantic Suite (S4) platform. With S4 developers and researchers can instantly get access to fully managed RDF DBaaS, without the need for hardware provisioning, maintenance and operations. Additionally, the S4 platform provides on-demand access to text analytics services for news, social media and life sciences, as well as access to knowledge graphs (DBpedia, Freebase and GeoNames).

The goal of the S4 platform is to make it easy for developers and researchers to develop smart/semantic applications, without the need to spend time and effort on infrastructure provisioning and maintenance. Marin will also provide examples of EC funded research projects – DaPaaS, ProDataMarket and KConnect — that plan to utilise the S4 platform for semantic data management

More information on S4 will be available in [1][2] and [3]

[1] Marin Dimitrov, Alex Simov and Yavor Petkov.  On-demand Text Analytics and Metadata Management with S4. In: proceedings of Workshop on Emerging Software as a Service and Analytics (ESaaSA 2015) at the 5th International Conference on Cloud Computing and Services Science (CLOSER 2015), Lisbon, Portugal.

[2] Marin Dimitrov, Alex Simov and Yavor Petkov. Text Analytics and Linked Data Management As-a-Service with S4. In: proceedings of 3rd International Workshop on Semantic Web Enterprise Adoption and Best Practice (WaSABi 2015) part of the Extended Semantic Web Conference (ESWC 2015), May 31st 2015, Portoroz, Slovenia

[3] Marin Dimitrov, Alex Simov and Yavor Petkov. Low-cost Open Data As-a-Service in the Cloud. In: proceedings of 2nd Semantic Web Enterprise Developers Workshop (SemDev 2015) part of the Extended Semantic Web Conference (ESWC 2015), May 31st 2015, Portoroz, Slovenia

Report on: “Knowledge-Based Trust: Estimating the Trustworthiness of Web Sources”

by Xin Luna Dong, Evgeniy Gabrilovich, Kevin Murphy, Van Dang, Wilko Horn, Camillo Lugaresi, Shaohua Sun, Wei Zhang

Link to the paper

Presentation by Jörg Unbehauen

Abstract:

“The quality of web sources has been traditionally evaluated using exogenous signals such as the hyperlink structure of the graph. We propose a new approach that relies on endogenous signals, namely, the correctness of factual information provided by the source. A source that has few false facts is considered to be trustworthy. The facts are automatically extracted from each source by information extraction methods commonly used to construct knowledge bases. We propose a way to distinguish errors made in the extraction process from factual errors in the web source per se, by using joint inference in a novel multi-layer probabilistic model. We call the trustworthiness score we computed Knowledge-Based Trust (KBT). On synthetic data, we show that our method can reliably compute the true trustworthiness levels of the sources. We then apply it to a database of 2.8B facts extracted from the web, and thereby estimate the trustworthiness of 119M webpages. Manual evaluation of a subset of the results confirms the effectiveness of the method. “

Linked Data Quality Assessment and its Application to Societal Progress Measurement

Abstract: In recent years, the Linked Data (LD) paradigm has emerged as a simple mechanism for employing the Web as a medium for data and knowledge integration where both documents and data are linked. Many different communities on the Internet such as geographic, media, life sciences and government have already adopted these LD principles. In all these use cases utilizing LD, one crippling problem is the underlying data quality. Incomplete, inconsistent or inaccurate data affects the end results gravely, thus making them unreliable. Data quality is commonly conceived as fitness for use, be it for a certain application or use case.  A key challenge is to assess the quality of datasets published on the Web and make this quality information explicit. Assessing data quality is particularly a challenge in LD as the underlying data stems from a set of multiple, autonomous and evolving data sources. Moreover, the dynamic nature of LD makes assessing the quality crucial to measure the accuracy of representing the real-world data. In this thesis, we first unify 18 data quality dimensions and provide a total of 69 metrics for assessment of LD. Next, three different methodologies for linked data quality assessment are evaluated namely (i) user-driven; (ii) crowdsourcing and (iii) semi-atuomated use case driven. Finally, we take into account a domain-specific use case that consumes LD and leverages on data quality. We show the advantages of this semi-automated assessment over the other types of quality assessment methodologies discussed earlier. The Observatory aims at evaluating the impact of research development on the economic and healthcare performance of each country per year. We illustrate the usefulness of LD in this use case and the importance of quality assessment for any data analysis.

Join us!

• Thursday, 9 April at 2pm, Room P702