Active learning is an iterative labeling process that is used to obtain a small labeled subset, despite the absence of labeled data, thereby enabling to train a model for supervised tasks such as text classification.While active learning has made considerable progress in recent years due to improvements provided by pretrained language models, there is untapped potential in the often neglected unlabeled portion of the data, although it is available in considerably larger quantities than the usually small set of labeled data.In this work, we investigate how self-training, a semi-supervised approach that uses a model to obtain pseudo-labels for unlabeled data, can be used to improve the efficiency of active learning for text classification.Building on a comprehensive reproduction of four previous self-training approaches, some of which are evaluated for the first time in the context of active learning or natural language processing, we introduce HAST, a new and effective self-training strategy, which is evaluated on four text classification benchmarks.Our results show that it outperforms the reproduced self-training approaches and reaches classification results comparable to previous experiments for three out of four datasets, using as little as 25% of the data.The code is publicly available at https://github.com/chschroeder/self-training- for-sample-efficient-active-learning.

Christopher Schröder, Kim Bürgl, Yves Annanias, Andreas Niekler, Lydia Müller, Daniel Wiegreffe, Christian Bender, Christoph Mengs, Gerik Scheuermann, Gerhard Heyer. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

To ease the difficulty of argument stance classification, the task of same side stance classification (S3C) has been proposed.In contrast to actual stance classification, which requires a substantial amount of domain knowledge to identify whether an argument is in favor or against a certain issue, it is argued that, for S3C, only argument similarity within stances needs to be learned to successfully solve the task.We evaluate several transformer-based approaches on the dataset of the recent S3C shared task, followed by an in-depth evaluation and error analysis of our model and the task's hypothesis.We show that, although we achieve state-of-the-art results, our model fails to generalize both within as well as across topics and domains when adjusting the sampling strategy of the training and test set to a more adversarial scenario.Our evaluation shows that current state-of-the-art approaches cannot determine same side stance by considering only domain-independent linguistic similarity features, but appear to require domain knowledge and semantic inference, too.

Abstract Social media are an emerging new paradigm in interdisciplinary research in crisis informatics. They bring many opportunities as well as challenges to all fields of application and research involved in the project of using social media content for an improved disaster management. Using the Central European flooding 2013 as our case study, we optimize and apply methods from the field ofnatural language processingand unsupervised machine learning to investigate the thematic and temporal structure of German social media communication. By means of topic model analysis, we will investigate which kind of content was shared on social media during the event. On this basis, we will, furthermore, investigate the development of topics over time and apply temporal clustering techniques to automatically identify different characteristic phases of communication. From the results, we, first, want to reveal properties of social media content and show what potential social media have for improving disaster management in Germany. Second, we will be concerned with the methodological issue of finding and adapting natural language processing methods that are suitable for analysing social media data in order to obtain information relevant for disaster management. With respect to the first, application-oriented focal point, our study reveals high potential of social media content in the factual, organizational and psychological dimension of the disaster and during all stages of the disaster management life cycle. Interestingly, there appear to be systematic differences in thematic profile between the different platforms Facebook and Twitter and between different stages of the event. In context of our methodological investigation, we claim that if topic model analysis is combined with appropriate optimization techniques, it shows high applicability for thematic and temporal social media analysis in disaster management.

During the flood in 2013 in Germany and Austria, the engagement of volunteers was the highest ever known. Notably, these volunteers organized themselves mainly via social media and without being motivated or guided by professional management. The present paper wants to provide input and positive impulse for current discussions among the public authorities how to become more present in social networks and take benefit of their strength. By means of a corpus-based case study of German Facebook and Twitter messages during the flood in 2013, we show and analyze the real potential of social media for disaster management and reveal some of their communicative characteristics. At the same time, we discuss two of the main challenges, namely information overload and legal issues. Regarding the problem of information overload the paper shows by case of an example from state-of-the-art automatic language processing (topic model analysis), that today it is possible to establish the technical basis required to get efficient and flexible computer-based access to information in social media. With respect to the legal conditions of social media use in disaster management, paper is to give a concise overview of the current legal situation using Saxony as an example, to identify open problems and to present proposals for their potential solution.

We present a novel method to visualize multidimensional point clouds. While conventional visualization techniques, like scatterplot matrices or parallel coordinates, have issues with either overplotting of entities or handling many dimensions, we abstract the data using topological methods before presenting it. We assume the input points to be samples of a random variable with a high-dimensional probability distribution which we approximate using kernel density estimates on a suitably reconstructed mesh. From the resulting scalar field we extract the join tree and present it as a topological landscape, a visualization metaphor that utilizes the human capability of understanding natural terrains. In this landscape, dense clusters of points show up as hills. The nesting of hills indicates the nesting of clusters. We augment the landscape with the data points to allow selection and inspection of single points and point sets. We also present optimizations to make our algorithm applicable to large data sets and to allow interactive adaption of our visualization to the kernel window width used in the density estimation.

We present WCTAnalyze, a novel tool for accessing and visually analyzing huge collections of temporally indexed textual data. It is motivated by applications such as media resonance analysis, business intelligence etc. that need to perform high level analytics on top of linguistically and statistically processed unstructured textual data.