Knowledge bases often consist of facts which are harvested from a variety of sources, many of which are noisy and some of which conflict, resulting in a level of uncertainty for each triple. Knowledge bases are also often incomplete, prompting the use of embedding methods to generalize from known facts, however, existing embedding methods only model triple-level uncertainty, and reasoning results lack global consistency. To address these shortcomings, we propose BEUrRE, a novel uncertain knowledge graph embedding method with calibrated probabilistic semantics. BEUrRE models each entity as a box (i.e. axis-aligned hyperrectangle) and relations between two entities as affine transforms on the head and tail entity boxes. The geometry of the boxes allows for efficient calculation of intersections and volumes, endowing the model with calibrated probabilistic semantics and facilitating the incorporation of relational constraints. Extensive experiments on two benchmark datasets show that BEUrRE consistently outperforms baselines on confidence prediction and fact ranking due to its probabilistic calibration and ability to capture high-order dependencies among facts.

Reasoning about tabular information presents unique challenges to modern NLP approaches which largely rely on pre-trained contextualized embeddings of text. In this paper, we study these challenges through the problem of tabular natural language inference. We propose easy and effective modifications to how information is presented to a model for this task. We show via systematic experiments that these strategies substantially improve tabular inference performance.

We take the first step towards multilingual style transfer by creating and releasing XFORMAL, a benchmark of multiple formal reformulations of informal text in Brazilian Portuguese, French, and Italian. Results on XFORMAL suggest that state-of-the-art style transfer approaches perform close to simple baselines, indicating that style transfer is even more challenging when moving multilingual.

Score-based generative models provide state-of-the-art quality for image and audio synthesis. Sampling from these models is performed iteratively, typically employing a discretized series of noise levels and a predefined scheme. In this note, we first overview three common sampling schemes for models trained with denoising score matching. Next, we focus on one of them, consistent annealed sampling, and study its hyper-parameter boundaries. We then highlight a possible formulation of such hyper-parameter that explicitly considers those boundaries and facilitates tuning when using few or a variable number of steps. Finally, we highlight some connections of the formulation with other sampling schemes.

This paper describes the acquisition, preprocessing, segmentation, and alignment of an Amharic-English parallel corpus. It will be useful for machine translation of an under-resourced language, Amharic. The corpus is larger than previously compiled corpora; it is released for research purposes. We trained neural machine translation and phrase-based statistical machine translation models using the corpus. In the automatic evaluation, neural machine translation models outperform phrase-based statistical machine translation models.

Social networks are widely used for information consumption and dissemination, especially during time-critical events such as natural disasters. Despite its significantly large volume, social media content is often too noisy for direct use in any application. Therefore, it is important to filter, categorize, and concisely summarize the available content to facilitate effective consumption and decision-making. To address such issues automatic classification systems have been developed using supervised modeling approaches, thanks to the earlier efforts on creating labeled datasets. However, existing datasets are limited in different aspects (e.g., size, contains duplicates) and less suitable to support more advanced and data-hungry deep learning models. In this paper, we present a new large-scale dataset with ~77K human-labeled tweets, sampled from a pool of ~24 million tweets across 19 disaster events that happened between 2016 and 2019. Moreover, we propose a data collection and sampling pipeline, which is important for social media data sampling for human annotation. We report multiclass classification results using classic and deep learning (fastText and transformer) based models to set the ground for future studies. The dataset and associated resources are publicly available. https://crisisnlp.qcri.org/humaid_dataset.html

The joy, anxiety, and anticipation of getting a COVID vaccine in America culminates, quite anticlimactically, with a piece of white cardstock. Some have already lost their vaccine cards or never got them to begin with. Others have their names misspelled and crossed out on it. Many are having trouble reconciling how something so simple--and easily forged--can carry such import and weight. The White House has recently clarified that there will be no federal vaccine passport.

Despite only being early in the year, significant events have already taken place in 2021. Mass vaccinations for Covid-19 have begun around the world, and new strains of the disease have surfaced in the United Kingdom, South Africa and Brazil. For companies, this news has had a direct impact on their ability to conduct business while further placing their pandemic response under the public microscope. How companies are being talked and written about is changing as the pandemic unfolds, and these nuances could reveal more than simply how effective an organization's marketing department is. What if shifts in sentiment could help traders make more informed financial decisions?

On many occasions, the greatest impediments to creating Artificial Intelligence solutions do not lie in the capacity of highly qualified teams, but in establishing an effective way of working between the different professional profiles involved in the life cycle of analytical models. This is one of the main tasks we are currently tackling at BBVA AI Factory. It is a task guided by three concepts: simplify, accelerate and reuse. My first direct contact with the AI Factory was in April 2020, in the middle of lockdown. I found myself with a team of data scientists who were extremely competent in creating AI models, but who needed to continue to push for common working guidelines in order to deal with the complexity – both organisational and technical – that exists in the Engineering domain.

Researchers at the University of Barcelona have developed an open access, deep learning-based web app that will enable the detection and quantification of floating plastics in the sea with a reliability of over 80%. Floating sea macro-litter is a threat to the conservation of marine ecosystems worldwide. According to UNESCO, plastic debris causes the deaths of more than a million seabirds every year, as well as more than 100,000 marine mammals. Eroded fragments, known as micro-plastics, are now prevalent across the food chain. The largest density of floating litter is found in the great ocean gyres (systems of circular currents) with litter being caught and spun in these vast cycles.