This paper describes the second edition of the shared task on Taxonomy Extraction Evaluation organised as part of SemEval 2016. This task aims to extract hypernym-hyponym relations between a given list of domain-specific terms and then to construct a domain taxonomy based on them. TExEval-2 introduced a multilingual setting for this task, covering four different languages including English, Dutch, Italian and French from domains as diverse as environment, food and science. A total of 62 runs submitted by 5 different teams were evaluated using structural measures, by comparison with gold standard taxonomies and by manual quality assessment of novel relations.
This paper describes the second edition of the shared task on Taxonomy Extraction Evaluation organised as part of SemEval 2016. This task aims to extract hypernym-hyponym relations between a given list of domain-specific terms and then to construct a domain taxonomy based on them. TExEval-2 introduced a multilingual setting for this task, covering four different languages including English, Dutch, Italian and French from domains as diverse as environment, food and science. A total of 62 runs submitted by 5 different teams were evaluated using structural measures, by comparison with gold standard taxonomies and by manual quality assessment of novel relations.
ing are analysed and evaluated. Our results show that it is possible to construct expert profiles starting from automatically extracted expertise topics and that topiccentric approaches outperform state-of-the-art language modelling approaches for expert finding.
Expertise modeling has been the subject of extensive research in two main disciplines: Information Retrieval (IR) and Social Network Analysis (SNA). Both IR and SNA approaches build the expertise model through a document-centric approach providing a macro-perspective on the knowledge emerging from large corpus of static documents. With the emergence of the Web of Data there has been a significant shift from static to evolving documents, through micro-contributions. Thus, the existing macro-perspective is no longer sufficient to track the evolution of both knowledge and expertise. In this paper we present a comprehensive, domain-agnostic model for expertise profiling in the context of dynamic, living documents and evolving knowledge bases. We showcase its application in the biomedical domain and analyze its performance using two manually created datasets.
Traditionally, relevance assessments for expert search have been gathered through self-assessment or based on the opinions of coworkers. We introduce three benchmark datasets 1 for expert search that use conference workshops for relevance assessment. Our data sets cover entire research domains as opposed to single institutions. In addition, they provide a larger number of topic-person associations and allow a more objective and fine-grained evaluation of expertise than existing data sets do. We present and discuss baseline results for a language modelling and a topic-centric approach to expert search. We find that the topic-centric approach achieves the best results on domain-specific datasets.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.