The LODeXporter: Flexible Generation of Linked Open Data Triples from NLP Frameworks for Automatic Knowledge Base Construction

The LODeXporter: Flexible Generation of Linked Open Data Triples from NLP Frameworks for Automatic Knowledge Base Construction Ren´e Witte and Bahar Sateli Semantic Software Lab Concordia University, Montr´eal, QC, Canada http://www.semanticsoftware.info Abstract We present LODeXporter, a novel approach for exporting Natural Language Processing (NLP) results to a graph-based knowledge base, following Linked Open Data (LOD) principles. The rules for transforming NLP entities into Resource Description Framework (RDF) triples are described in a custom mapping language, which is defined in RDF Schema (RDFS) itself, providing a separation of concerns between NLP pipeline engineering and knowledge base engineering. LODeXporter is available as an open source component for the GATE (General Architecture for Text Engineering) framework. Keywords: Linked Open Data, Automatic Knowledge Base Construction, Semantic Web 1. Introduction Knowledge bases in standardized graph-based formats have become increasingly important in modern applications, such as personal intelligent assistants. Due to the large amount of formalized knowledge required, automatic knowledge base (KB) construction is a crucial step that typically spans multiple information sources (such as structured databases, sensors, and documents). A significant amount of knowl- edge resides in natural language text and can be automat- ically extracted through existing or custom-made Natural Language Processing (NLP) components and pipelines. Yet, so far no generic solution existed for exporting the results of an NLP pipeline into a knowledge base. Our contribution is a novel technique for converting the re- sults of an NLP pipeline into an LOD-compliant KB, based on the Resource Description Format (RDF) (Cyganiak et al., 2014) and RDF Schema (RDFS) (Brickley and Guha, 2014) W3C standards. The concrete mapping of NLP re- sults to entities in the knowledge base is defined through a mapping language, which is itself written in RDFS. This pro- vides high flexibility, as the same NLP pipeline can drive the generation of triples in different KBs with different vocabu- laries. It also relieves the NLP engineer from dealing with the technical details of generating correct Uniform Resource Identifiers (URIs), thereby providing an agile solution for bringing NLP results onto the Linked Open Data (LOD) cloud (Heath and Bizer, 2011). LODeXporter has been implemented based on the Gen- eral Architecture for Text Engineering (GATE) framework (Cunningham et al., 2013) and is available as open source (LGPLv3) on GitHub.1 2. Foundations In this section, we provide brief background information on the foundations of our approach. Linked Data. Linked Open Data (LOD) (Heath and Bizer, 2011; Wood et al., 2014) is the practice of using semantic 1LODeXporter on GitHub, https://github.com/ SemanticSoftwareLab/TextMining-LODeXporter web technologies, like RDF, to publish structured, open datasets that can be interlinked and shared automatically between machines. All data within the LOD cloud have unique URIs. When dereferenced, the URIs provide useful, machine-readable information about that entity, as well as links to other related entities on the web of data. Automatic Knowledge Base Construction. As semantic knowledge bases, triplestores are especially well suited for connecting previously isolated knowledge sources (so-called “information silos”). Towards this end, various solutions have been developed for converting …

Content truncated to 500 words