Marten Postma

VUA-background : When to Use Background Information to Perform Word Sense Disambiguation

We present in this paper our submission to task 13 of SemEval2015, which makes use of background information and external resources (DBpedia and Wikipedia) to automatically disambiguate texts. Our approach follows two routes for disambiguation: one route is proposed by a state‐of‐the‐art WSD system, and the other one by the predominant sense information extracted in an unsupervised way from an automatically built background corpus. We reached 4th position in terms of F1-score in task number 13 of SemEval2015: “Multilingual All-Words Sense Disambiguation and Entity Linking” (Moro and Navigli, 2015). All the software and code created for this approach are publicly available on GitHub 1 .

Moving away from semantic overfitting in disambiguation datasets

Entities and events in the world have no frequency, but our communication about them and the expressions we use to refer to them do have a strong frequency profile. Language expressions and their meanings follow a Zipfian distribution, featuring a small amount of very frequent observations and a very long tail of low frequent observations. Since our NLP datasets sample texts but do not sample the world, they are no exception to Zipf’s law. This causes a lack of representativeness in our NLP tasks, leading to models that can capture the head phenomena in language, but fail when dealing with the long tail. We therefore propose a referential challenge for semantic NLP that reflects a higher degree of ambiguity and variance and captures a large range of small real-world phenomena. To perform well, systems would have to show deep understanding on the linguistic tail.