Refine
Year of publication
- 2019 (2) (remove)
Document Type
Language
- English (2)
Has Fulltext
- yes (2)
Is part of the Bibliography
- no (2)
Keywords
- Bildmaterial (1)
- Concreteness (1)
- Distributional Semantics (1)
- Focus Group (1)
- Gruppeninterview (1)
- Imagery (1)
- Images (1)
- Konkretum <Linguistik> (1)
- Lexical Semantics (1)
- Open Repositories (1)
Institute
The NOA project collects and stores images from open access publications and makes them findable and reusable. During the project a focus group workshop was held to determine whether the development is addressing researchers’ needs. This took place before the second half of the project so that the results could be considered for further development since addressing users’ needs is a big part of the project. The focus was to find out what content and functionality they expect from image repositories.
In a first step, participants were asked to fill out a survey about their images use. Secondly, they tested different use cases on the live system. The first finding is that users have a need for finding scholarly images but it is not a routine task and they often do not know any image repositories. This is another reason for repositories to become more open and reach users by integrating with other content providers. The second finding is that users paid attention to image licenses but struggled to find and interpret them while also being unsure how to cite images. In general, there is a high demand for reusing scholarly images but the existing infrastructure has room to improve.
Concreteness of words has been studied extensively in psycholinguistic literature. A number of datasets have been created with average values for perceived concreteness of words. We show that we can train a regression model on these data, using word embeddings and morphological features, that can predict these concreteness values with high accuracy. We evaluate the model on 7 publicly available datasets. Only for a few small subsets of these datasets prediction of concreteness values are found in the literature. Our results clearly outperform the reported results for these datasets.