This paper describes the development of a system for SemEval-2023 Shared Task 11 on Learning with Disagreements (Le-Wi-Di). Labelled data plays a vital role in the development of machine learning systems. The human-annotated labels are usually considered the truth for training or validation. To obtain truth labels, a traditional way is to hire domain experts to perform an expensive annotation process. Crowd-sourcing labelling is comparably cheap, whereas it raises a question on the reliability of annotators. A common strategy in a mixed-annotator dataset with various sets of annotators for each instance is to aggregate the labels among multiple groups of annotators to obtain the truth labels. However, these annotators might not reach an agreement, and there is no guarantee of the reliability of these labels either. With further problems caused by human label variation, subjective tasks usually suffer from the different opinions provided by the annotators. In this paper, we propose two simple heuristic functions to compute the annotator ranking scores, namely AnnoHard and AnnoSoft, based on the hard labels (i.e., aggregative labels) and soft labels (i.e., cross-entropy values). By introducing these scores, we adjust the weights of the training instances to improve the learning with disagreements among the annotators.

We model source-selection in multi-source Unsupervised Domain Adaptation (UDA) as an attention-learning problem, where we learn attention over the sources per given target instance. We first independently learn source-specific classification models, and a relatedness map between sources and target domains using pseudo-labelled target domain instances. Next, we learn domain-attention scores over the sources for aggregating the predictions of the source-specific models. Experimental results on two cross-domain sentiment classification datasets show that the proposed method reports consistently good performance across domains, and at times outperforming more complex prior proposals. Moreover, the computed domain-attention scores enable us to find explanations for the predictions made by the proposed method.

Lack of labelled data in the target domain for training is a common problem in domain adaptation. To overcome this problem, we propose a novel unsupervised domain adaptation method that combines projection and self-training based approaches. Using the labelled data from the source domain, we first learn a projection that maximises the distance among the nearest neighbours with opposite labels in the source domain. Next, we project the source domain labelled data using the learnt projection and train a classifier for the target class prediction. We then use the trained classifier to predict pseudo labels for the target domain unlabelled data. Finally, we learn a projection for the target domain as we did for the source domain using the pseudo-labelled target domain data, where we maximise the distance between nearest neighbours having opposite pseudo labels. Experiments on a standard benchmark dataset for domain adaptation show that the proposed method consistently outperforms numerous baselines and returns competitive results comparable to that of SOTA including self-training, tri-training, and neural adaptations.

Feature sparseness is a problem common to cross-domain and short-text classification tasks. To overcome this feature sparseness problem, we propose a novel method based on graph decomposition to find candidate features for expanding feature vectors. Specifically, we first create a feature-relatedness graph, which is subsequently decomposed into core-periphery (CP) pairs and use the peripheries as the expansion candidates of the cores. We expand both training and test instances using the computed related features and use them to train a text classifier. We observe that prioritising features that are common to both training and test instances as cores during the CP decomposition to further improve the accuracy of text classification. We evaluate the proposed CP-decomposition-based feature expansion method on benchmark datasets for cross-domain sentiment classification and short-text classification. Our experimental results show that the proposed method consistently outperforms all baselines on short-text classification tasks, and perform competitively with pivot-based cross-domain sentiment classification methods.