Continuous learning from free-text human feedback, such as error corrections, new knowledge, or alternative responses, is essential for today’s chatbots and virtual assistants to stay up-to-date, engaging, and socially acceptable. However, for research on methods for learning from such data, annotated data is scarce. To address this, we examine the error and user response types of six popular dialogue datasets from various types, including MultiWoZ, PersonaChat, Wizards-of-Wikipedia, and others, to assess their extendibility with the needed annotations. For this corpus study, we manually annotate a subset of each dataset with error and user response types using an improved version of the Integrated Error Taxonomy and a newly proposed user response type taxonomy. We provide the resulting dataset (EURTAD) to the community. Our findings provide new insights into dataset composition, including error types, user response types, and the relations between them.
Dialogue Systems are becoming ubiquitous in various forms and shapes - virtual assistants(Siri, Alexa, etc.), chat-bots, customer sup-port, chit-chat systems just to name a few. The advances in language models and their publication have democratised advanced NLP.However, data remains a crucial bottleneck. Our contribution to this essential pillar isMATILDA, to the best of our knowledge the first multi-annotator, multi-language dialogue annotation tool. MATILDA allows the creation of corpora, the management of users, the annotation of dialogues, the quick adaptation of the user interface to any language and the resolution of inter-annotator disagreement. We evaluate the tool on ease of use, annotation speed and interannotation resolution for both experts and novices and conclude that this tool not only supports the full pipeline for dialogue annotation, but also allows non-technical people to easily use it. We are completely open-sourcing the tool at
https://github.com/wluper/matilda and provide a tutorial video1.
Following the major success of neural language models (LMs) such as BERT or GPT-2 on a variety of language understanding tasks, recent work focused on injecting (structured) knowledge from external resources into these models. While on the one hand, joint pre-training (i.e., training from scratch, adding objectives based on external knowledge to the primary LM objective) may be prohibitively computationally expensive, post-hoc fine-tuning on external knowledge, on the other hand, may lead to the catastrophic forgetting of distributional knowledge. In this work, we investigate models for complementing the distributional knowledge of BERT with conceptual knowledge from ConceptNet and its corresponding Open Mind Common Sense (OMCS) corpus, respectively, using adapter training. While overall results on the GLUE benchmark paint an inconclusive picture, a deeper analysis reveals that our adapter-based models substantially outperform BERT (up to 15-20 performance points) on inference tasks that require the type of conceptual knowledge explicitly present in ConceptNet and OMCS. We also open source all our experiments and relevant code under:
https://github.com/wluper/retrograph.
Classification tasks are usually analysed and improved through new model architectures or hyperparameter optimisation but the underlying properties of datasets are discovered on an ad-hoc basis as errors occur. However, understanding the properties of the data is crucial in perfecting models. In this paper we analyse exactly which characteristics of a dataset best determine how difficult that dataset is for the task of text classification. We then propose an intuitive measure of difficulty for text classification datasets which is simple and fast to calculate. We empirically prove that this measure generalises to unseen data by comparing it to state-of-the-art datasets and results. This measure can be used to analyse the precise source of errors in a dataset and allows fast estimation of how difficult a dataset is to learn. We searched for this measure by training 12 classical and neural network based models on 78 real-world datasets, then use a genetic algorithm to discover the best measure of difficulty. Our difficulty-calculating code and datasets are publicly available.