Semiology of Graphics: Diagrams Networks Maps.

Recent advances in language representation using neural networks have made it viable to transfer the learned internal states of large pretrained language models (LMs) to downstream natural language processing (NLP) tasks. This transfer learning approach improves the overall performance on many tasks and is highly beneficial when labeled data is scarce, making pretrained LMs valuable resources specially for languages with few annotated training examples. In this work, we train BERT (Bidirectional Encoder Representations from Transformers) models for Brazilian Portuguese, which we nickname BERTimbau. We evaluate our models on three downstream NLP tasks: sentence textual similarity, recognizing textual entailment, and named entity recognition. Our models improve the state-of-the-art in all of these tasks, outperforming Multilingual BERT and confirming the effectiveness of large pretrained LMs for Portuguese. We release our models to the community hoping to provide strong baselines for future NLP research: https://github.com/neuralmind-ai/portuguese-bert.

BERTimbau: Pretrained BERT Models for Brazilian Portuguese

Recent advances in language representation using neural networks have made it viable to transfer the learned internal states of a trained model to downstream natural language processing tasks, such as named entity recognition (NER) and question answering. It has been shown that the leverage of pre-trained language models improves the overall performance on many tasks and is highly beneficial when labeled data is scarce. In this work, we train Portuguese BERT models and employ a BERT-CRF architecture to the NER task on the Portuguese language, combining the transfer capabilities of BERT with the structured predictions of CRF. We explore feature-based and fine-tuning training strategies for the BERT model. Our fine-tuning approach obtains new state-of-the-art results on the HAREM I dataset, improving the F1-score by 1 point on the selective scenario (5 NE classes) and by 4 points on the total scenario (10 NE classes).

Portuguese Named Entity Recognition using BERT-CRF.

Immersive Analytics is a quickly evolving field that unites several areas such as visualisation, immersive environments, and human-computer interaction to support human data analysis with emerging technologies. This research has thrived over the past years with multiple workshops, seminars, and a growing body of publications, spanning several conferences. Given the rapid advancement of interaction technologies and novel application domains, this paper aims toward a broader research agenda to enable widespread adoption. We present 17 key research challenges developed over multiple sessions by a diverse group of 24 international experts, initiated from a virtual scientific workshop at ACM CHI 2020. These challenges aim to coordinate future work by providing a systematic roadmap of current directions and impending hurdles to facilitate productive and effective applications for Immersive Analytics.

/pdf/grand-challenges-in-immersive-analytics-1ay3f154h3.pdf

Grand Challenges in Immersive Analytics

A Space-Time Cube enables analysts to clearly observe spatio-temporal features in movement trajectory datasets in geovisualization. However, its general usability is impacted by a lack of depth cues, a reported steep learning curve, and the requirement for efficient 3D navigation. In this work, we investigate a Space-Time Cube in the Immersive Analytics domain. Based on a review of previous work and selecting an appropriate exploration metaphor, we built a prototype environment where the cube is coupled to a virtual representation of the analyst's real desk, and zooming and panning in space and time are intuitively controlled using mid-air gestures. We compared our immersive environment to a desktop-based implementation in a user study with 20 participants across 7 tasks of varying difficulty, which targeted different user interface features. To investigate how performance is affected in the presence of clutter, we explored two scenarios with different numbers of trajectories. While the quantitative performance was similar for the majority of tasks, large differences appear when we analyze the patterns of interaction and consider subjective metrics. The immersive version of the Space-Time Cube received higher usability scores, much higher user preference, and was rated to have a lower mental workload, without causing participants discomfort in 25-minute-long VR sessions.

Evaluating an Immersive Space-Time Cube Geovisualization for Intuitive Trajectory Data Exploration

In this work, we present the construction process of a large Web corpus for Brazilian Portuguese, aiming to achieve a size comparable to the state of the art in other languages. We also discuss our updated sentence-level approach for the strict removal of duplicated content. Following the pipeline methodology, more than 60 million pages were crawled and filtered, with 3.5 million being selected. The obtained multi-domain corpus, named brWaC, is composed by 2.7 billion tokens, and has been annotated with tagging and parsing information. The incidence of non-unique long sentences, an indication of replicated content, which reaches 9% in other Web corpora, was reduced to only 0.5%. Domain diversity was also maximized, with 120,000 different websites contributing content. We are making our new resource freely available for the research community, both for querying and downloading, in the expectation of aiding in new advances for the processing of Brazilian Portuguese.

The brWaC Corpus: A New Open Resource for Brazilian Portuguese

The use of novel displays and interaction resources to support immersive data visualization and improve analytical reasoning is a research trend in the information visualization community In this work, we evaluate the use of an HMD-based environment for the exploration of multidimensional data, represented in 3D scatterplots as a result of dimensionality reduction (DR) We present a new modeling for this problem, accounting for the two factors whose interplay determine the impact on the overall task performance: the difference in errors introduced by performing dimensionality reduction to 2D or 3D, and the difference in human perception errors under different visualization conditions This two-step framework offers a simple approach to estimate the benefits of using an immersive 3D setup for a particular dataset Here, the DR errors for a series of roll call voting datasets when using two or three dimensions are evaluated through an empirical task-based approach The perception error and overall task performance, on the other hand, are assessed through a comparative user study with 30 participants Results indicated that perception errors were low and similar in all approaches, resulting in overall performance benefits in both desktop and HMD-based 3D techniques The immersive condition, however, was found to require less effort to find information and less navigation, besides providing much larger subjective perception of accuracy and engagement

Immersive Visualization of Abstract Information: An Evaluation on Dimensionally-Reduced Data Scatterplots

The VirtualDesk metaphor is an opportunity for more comfortable and efficient immersive data exploration, using tangible interaction with the analyst's physical work desk and embodied manipulation of mid-air data representations. In this paper, we present an extended discussion of its underlying concepts, and review and compare two previous case studies where promising results were obtained in terms of user comfort, engagement, and usability. We also discuss findings of a novel study conducted with geovisualization experts, pointing directions for improvement and future research.

Comfortable Immersive Analytics With the VirtualDesk Metaphor

The availability of annotated corpora for research in the area of Readability Assessment is still very limited. On the other hand, the Web is increasingly being used by researchers as a source of written content to build very large and rich corpora, in the Web as Corpus (WaC) initiative. This paper proposes a framework for automatic generation of large corpora classified by readability. It adopts a supervised learning method to incorporate a readability filter based in features with low computational cost to a crawler, to collect texts targeted at a specific reading level. We evaluate this framework by comparing a readability-assessed web crawled corpus to a reference corpus (Both corpora are available in http://www.inf.ufrgs.br/pln/resource/CrawlingByReadabilityLevel.zip.). The results obtained indicate that these features are good at separating texts from level 1 (initial grades) from other levels. As a result of this work two Portuguese corpora were constructed: the Wikilivros Readability Corpus, classified by grade level, and a crawled WaC classified by readability level.

Jorge A. Wagner Filho

Papers

Evaluating an Immersive Space-Time Cube Geovisualization for Intuitive Trajectory Data Exploration

The brWaC Corpus: A New Open Resource for Brazilian Portuguese

Immersive Visualization of Abstract Information: An Evaluation on Dimensionally-Reduced Data Scatterplots

Comfortable Immersive Analytics With the VirtualDesk Metaphor

Crawling by Readability Level