Information Extraction (IE) from text refers to the task of extracting structured knowledge from unstructured text. The task typically consists of a series of sub-tasks such as Named Entity Recognition and Relation Extraction. Sourcing entity and relation type specific training data is a major bottleneck in the above sub-tasks.In this work we present a slot filling approach to the task of biomedical IE, effectively replacing the need for entity and relation-specific training data, allowing to deal with zero-shot settings. We follow the recently proposed paradigm of coupling a Tranformer-based bi-encoder, Dense Passage Retrieval, with a Transformer-based reader model to extract relations from biomedical text. We assemble a biomedical slot filling dataset for both retrieval and reading comprehension and conduct a series of experiments demonstrating that our approach outperforms a number of simpler baselines. We also evaluate our approach end-to-end for standard as well as zero-shot settings. Our work provides a fresh perspective on how to solve biomedical IE tasks, in the absence of relevant training data. Our code, models and pretrained data are available at https://github.com/healx/biomed-slot-filling.
Representation learning for text via pretraining a language model on a large corpus has become a standard starting point for building NLP systems. This approach stands in contrast to autoencoders, also trained on raw text, but with the objective of learning to encode each input as a vector that allows full reconstruction. Autoencoders are attractive because of their latent space structure and generative properties. We therefore explore the construction of a sentence-level autoencoder from a pretrained, frozen transformer language model. We adapt the masked language modeling objective as a generative, denoising one, while only training a sentence bottleneck and a single-layer modified transformer decoder. We demonstrate that the sentence representations discovered by our model achieve better quality than previous methods that extract representations from pretrained transformers on text similarity tasks, style transfer (an example of controlled generation), and single-sentence classification tasks in the GLUE benchmark, while using fewer parameters than large pretrained models.
Research on crude oil price forecasting has attracted tremendous attention from scholars and policymakers due to its significant effect on the global economy. Besides supply and demand, crude oil prices are largely influenced by various factors, such as economic development, financial markets, conflicts, wars, and political events. Most previous research treats crude oil price forecasting as a time series or econometric variable prediction problem. Although recently there have been researches considering the effects of real-time news events, most of these works mainly use raw news headlines or topic models to extract text features without profoundly exploring the event information. In this study, a novel crude oil price forecasting framework, AGESL, is proposed to deal with this problem. In our approach, an open domain event extraction algorithm is utilized to extract underlying related events, and a text sentiment analysis algorithm is used to extract sentiment from massive news. Then a deep neural network integrating the news event features, sentimental features, and historical price features is built to predict future crude oil prices. Empirical experiments are performed on West Texas Intermediate (WTI) crude oil price data, and the results show that our approach obtains superior performance compared with several benchmark methods.
Digital humanities is an important subject because it enables developments in history, literature, and films. In this paper, we perform an empirical study of a Chinese historical text, Records of the Three Kingdoms (\textit{Records}), and a historical novel of the same story, Romance of the Three Kingdoms (\textit{Romance}). We employ natural language processing techniques to extract characters and their relationships. Then, we characterize the social networks and sentiments of the main characters in the historical text and the historical novel. We find that the social network in \textit{Romance} is more complex and dynamic than that of \textit{Records}, and the influence of the main characters differs. These findings shed light on the different styles of storytelling in the two literary genres and how the historical novel complicates the social networks of characters to enrich the literariness of the story.
The segmentation of complex images into semantic regions has seen a growing interest these last years with the advent of Deep Learning. Until recently, most existing methods for Historical Document Analysis focused on the visual appearance of documents, ignoring the rich information that textual content can offer. However, the segmentation of complex documents into semantic regions is sometimes impossible relying only on visual features and recent models embed both visual and textual information. In this paper, we focus on the use of both visual and textual information for segmenting historical registers into structured and meaningful units such as acts. An act is a text recording containing valuable knowledge such as demographic information (baptism, marriage or death) or royal decisions (donation or pardon). We propose a simple pipeline to enrich document images with the position of text lines containing key-phrases and show that running a standard image-based layout analysis system on these images can lead to significant gains. Our experiments show that the detection of acts increases from 38 % of mAP to 74 % when adding textual information, in real use-case conditions where text lines positions and content are extracted with an automatic recognition system.
Texts like news, encyclopedias, and some social media strive for objectivity. Yet bias in the form of inappropriate subjectivity - introducing attitudes via framing, presupposing truth, and casting doubt - remains ubiquitous. This kind of bias erodes our collective trust and fuels social conflict. To address this issue, we introduce a novel testbed for natural language generation: automatically bringing inappropriately subjective text into a neutral point of view ("neutralizing" biased text). We also offer the first parallel corpus of biased language. The corpus contains 180,000 sentence pairs and originates from Wikipedia edits that removed various framings, presuppositions, and attitudes from biased sentences. Last, we propose two strong encoder-decoder baselines for the task. A straightforward yet opaque CONCURRENT system uses a BERT encoder to identify subjective words as part of the generation process. An interpretable and controllable MODULAR algorithm separates these steps, using (1) a BERT-based classifier to identify problematic words and (2) a novel join embedding through which the classifier can edit the hidden states of the encoder. Large-scale human evaluation across four domains (encyclopedias, news headlines, books, and political speeches) suggests that these algorithms are a first step towards the automatic identification and reduction of bias.
Generating explanations for neural networks has become crucial for their applications in real-world with respect to reliability and trustworthiness. In natural language processing, existing methods usually provide important features which are words or phrases selected from an input text as an explanation, but ignore the interactions between them. It poses challenges for humans to interpret an explanation and connect it to model prediction. In this work, we build hierarchical explanations by detecting feature interactions. Such explanations visualize how words and phrases are combined at different levels of the hierarchy, which can help users understand the decision-making of black-box models. The proposed method is evaluated with three neural text classifiers (LSTM, CNN, and BERT) on two benchmark datasets, via both automatic and human evaluations. Experiments show the effectiveness of the proposed method in providing explanations that are both faithful to models and interpretable to humans.
The demand of competent robot assisted surgeons is progressively expanding, because robot-assisted surgery has become progressively more popular due to its clinical advantages. To meet this demand and provide a better surgical education for surgeon, we develop a novel robotic surgery education system by integrating artificial intelligence surgical module and augmented reality visualization. The artificial intelligence incorporates reinforcement leaning to learn from expert demonstration and then generate 3D guidance trajectory, providing surgical context awareness of the complete surgical procedure. The trajectory information is further visualized in stereo viewer in the dVRK along with other information such as text hint, where the user can perceive the 3D guidance and learn the procedure. The proposed system is evaluated through a preliminary experiment on surgical education task peg-transfer, which proves its feasibility and potential as the next generation of robot-assisted surgery education solution.
Recently, scene text recognition methods based on deep learning have sprung up in computer vision area. The existing methods achieved great performances, but the recognition of irregular text is still challenging due to the various shapes and distorted patterns. Consider that at the time of reading words in the real world, normally we will not rectify it in our mind but adjust our focus and visual fields. Similarly, through utilizing deformable convolutional layers whose geometric structures are adjustable, we present an enhanced recognition network without the steps of rectification to deal with irregular text in this work. A number of experiments have been applied, where the results on public benchmarks demonstrate the effectiveness of our proposed components and shows that our method has reached satisfactory performances. The code will be publicly available at https://github.com/Alpaca07/dtr soon.
In this paper, we formulate a novel task to synthesize speech in sync with a silent pre-recorded video, denoted as automatic voice over (AVO). Unlike traditional speech synthesis, AVO seeks to generate not only human-sounding speech, but also perfect lip-speech synchronization. A natural solution to AVO is to condition the speech rendering on the temporal progression of lip sequence in the video. We propose a novel text-to-speech model that is conditioned on visual input, named VisualTTS, for accurate lip-speech synchronization. The proposed VisualTTS adopts two novel mechanisms that are 1) textual-visual attention, and 2) visual fusion strategy during acoustic decoding, which both contribute to forming accurate alignment between the input text content and lip motion in input lip sequence. Experimental results show that VisualTTS achieves accurate lip-speech synchronization and outperforms all baseline systems.