Hélène de Ribaupierre

Aleksandra Edwards, Asahi Ushio, Jose Camacho-Collados et al.

Data augmentation techniques are widely used for enhancing the performance of machine learning models by tackling class imbalance issues and data sparsity. State-of-the-art generative language models have been shown to provide significant gains across different NLP tasks. However, their applicability to data augmentation for text classification tasks in few-shot settings have not been fully explored, especially for specialised domains. In this paper, we leverage GPT-2 (Radford A et al, 2019) for generating artificial training instances in order to improve classification performance. Our aim is to analyse the impact the selection process of seed training examples have over the quality of GPT-generated samples and consequently the classifier performance. We perform experiments with several seed selection strategies that, among others, exploit class hierarchical structures and domain expert selection. Our results show that fine-tuning GPT-2 in a handful of label instances leads to consistent classification improvements and outperform competitive baselines. Finally, we show that guiding this process through domain expert selection can lead to further improvements, which opens up interesting research avenues for combining generative models and active learning.

Anelia Kurteva, Hélène De Ribaupierre

The semantic linked data model is at the core of the Web due to its ability to model real world entities, connect them via relationships and provide context, which could help to transform data into information and information into knowledge. Linked Data, in the form of ontologies and knowledge graphs could be stored locally or could be made available to everyone online. For example, the DBpedia knowledge base, which provides global and unified access to knowledge graphs is open access. However, both access and usage of Linked Data require individuals to have expert knowledge in the field of the Semantic Web. Many of the existing solutions that are powered by Linked Data are developed for specific use cases such as building and exploring ontologies visually and are aimed at researchers with knowledge of semantic technology. The solutions that are aimed at non-experts are generic and, in most cases, information visualisation is not available. Instead, information is presented in textual format, which does not ease cognitive processes such as comprehension and could lead to problems such as information overload. In this paper, we present a web application with a user interface (UI), which combines features from applications for both experts and non-experts. The UI allows individuals with no previous knowledge of the Semantic Web to query the DBpedia knowledge base for definitions of a specific word and to view a graphical visualisation of the query results (the search keyword itself and concepts related to it).

Aleksandra Edwards, David Rogers, Jose Camacho-Collados et al.

The task of text and sentence classification is associated with the need for large amounts of labelled training data. The acquisition of high volumes of labelled datasets can be expensive or unfeasible, especially for highly-specialised domains for which documents are hard to obtain. Research on the application of supervised classification based on small amounts of training data is limited. In this paper, we address the combination of state-of-the-art deep learning and classification methods and provide an insight into what combination of methods fit the needs of small, domain-specific, and terminologically-rich corpora. We focus on a real-world scenario related to a collection of safeguarding reports comprising learning experiences and reflections on tackling serious incidents involving children and vulnerable adults. The relatively small volume of available reports and their use of highly domain-specific terminology makes the application of automated approaches difficult. We focus on the problem of automatically identifying the main themes in a safeguarding report using supervised classification approaches. Our results show the potential of deep learning models to simulate subject-expert behaviour even for complex tasks with limited labelled data.