Abstract

Background: Abstract sentence classification modelling has the potential to advance literature discovery capability for the array of academic literature information systems, however, no artefact exists that categorises known models and identifies their key characteristics. Aims: To systematically categorise known abstract sentence classification models and make this knowledge readily available to future researchers and professionals concerned with abstract sentence classification model development and deployment. Method: An information systems taxonomy development methodology was adopted after a literature review to categorise 23 abstract sentence classification models identified from the literature. Corresponding dimensions and characteristics were derived from this process with the resulting taxonomy presented. Results: Abstract sentence classification modelling has evolved significantly with state-of-the-art models now leveraging neural networks to achieve high-performance sentence classification. The resulting taxonomy provides a novel means to observe the development of this research field and enables us to consider how such models can be further improved or deployed in real-world applications.

Share

COinS