The Integration of Crowdsourcing into Knowledge Engineering Workflows -- Invited Talk

Beschreibung

The advent of the Web has significantly changed the context of systems that rely on formally specified knowledge, which led to opening up knowledge creation processes and tools to wider groups of contributors. Further broadening of this process allows large populations of non-experts to create knowledge through the use of crowdsourcing techniques such as games or mechanised labour platforms. Crowdsourcing techniques can provide effective means for solving a variety of ontology engineering (and other knowledge-rich) problems. Yet, crowdsourcing is mainly used as external support to ontology engineering, without being integrated into the work of ontology engineers. In this presentation I discuss the experiences made from the uComp Protege plugin, which facilitates the integration of typical crowdsourcing tasks into ontology engineering from within the Protege ontology editor. An evaluation of the plugin in a typical ontology engineering scenario demostrates significant benefits, such as lowering the overall task costs by 40% to 83% depending on the crowdsourcing settings [1]. The presentation will include important crowdsourcing aspects such as task batching, result aggregation and how to improve result quality. Furthermore, it will incorporate an introduction to the ontology learning system used in the experiments [2]. In a quick second part of the talk I will quickly address the work I started together with a student to study the use of Google's word2vec [3] for learning domain knowledge, and the ideas to apply word2vec for relation type detection for unlabelled relations between terms.

Zeitraum	1 Juni 2016
Ereignistitel	Workshop on Mathematic Models and Information Technology
Veranstaltungstyp	Keine Angaben