MiDTD: A Simple and Effective Distillation Framework for Distantly Supervised Relation Extraction

Relation extraction (RE), an important information extraction task, faced the great challenge brought by limited annotation data. To this end, distant supervision was proposed to automatically label RE data, and thus largely increased the number of annotated instances. Unfortunately, lots of noise relation annotations brought by automatic labeling become a new obstacle. Some recent studies have shown that the teacher-student framework of knowledge distillation can alleviate the interference of noise relation annotations via label softening. Nevertheless, we find that they still suffer from two problems: propagation of inaccurate dark knowledge and constraint of a unified distillation temperature . In this article, we propose a simple and effective Multi-instance Dynamic Temperature Distillation (MiDTD) framework, which is model-agnostic and mainly involves two modules: multi-instance target fusion (MiTF) and dynamic temperature regulation (DTR). MiTF combines the teacher’s predictions for multiple sentences with the same entity pair to amend the inaccurate dark knowledge in each student’s target. DTR allocates alterable distillation temperatures to different training instances to enable the softness of most student’s targets to be regulated to a moderate range. In experiments, we construct three concrete MiDTD instantiations with BERT, PCNN, and BiLSTM-based RE models, and the distilled students significantly outperform their teachers and the state-of-the-art (SOTA) methods.

Arabic open information extraction system using dependency parsing

<p>Arabic is a Semitic language and one of the most natural languages distinguished by the richness in morphological enunciation and derivation. This special and complex nature makes extracting information from the Arabic language difficult and always needs improvement. Open information extraction systems (OIE) have been emerged and used in different languages, especially in English. However, it has almost not been used for the Arabic language. Accordingly, this paper aims to introduce an OIE system that extracts the relation tuple from Arabic web text, exploiting Arabic dependency parsing and thinking carefully about all possible text relations. Based on clause types' propositions as extractable relations and constituents' grammatical functions, the identities of corresponding clause types are established. The proposed system named Arabic open information extraction(AOIE) can extract highly scalable Arabic text relations while being domain independent. Implementing the proposed system handles the problem using supervised strategies while the system relies on unsupervised extraction strategies. Also, the system has been implemented in several domains to avoid information extraction in a specific field. The results prove that the system achieves high efficiency in extracting clauses from large amounts of text.</p>

The role of software in science: a knowledge graph-based analysis of software mentions in PubMed Central

Science across all disciplines has become increasingly data-driven, leading to additional needs with respect to software for collecting, processing and analysing data. Thus, transparency about software used as part of the scientific process is crucial to understand provenance of individual research data and insights, is a prerequisite for reproducibility and can enable macro-analysis of the evolution of scientific methods over time. However, missing rigor in software citation practices renders the automated detection and disambiguation of software mentions a challenging problem. In this work, we provide a large-scale analysis of software usage and citation practices facilitated through an unprecedented knowledge graph of software mentions and affiliated metadata generated through supervised information extraction models trained on a unique gold standard corpus and applied to more than 3 million scientific articles. Our information extraction approach distinguishes different types of software and mentions, disambiguates mentions and outperforms the state-of-the-art significantly, leading to the most comprehensive corpus of 11.8 M software mentions that are described through a knowledge graph consisting of more than 300 M triples. Our analysis provides insights into the evolution of software usage and citation patterns across various fields, ranks of journals, and impact of publications. Whereas, to the best of our knowledge, this is the most comprehensive analysis of software use and citation at the time, all data and models are shared publicly to facilitate further research into scientific use and citation of software.

DeHyFoNet: Deformable Hybrid Network for Formula Detection in Scanned Document Images

This work presents an approach for detecting mathematical formulas in scanned document images. The proposed approach is end-to-end trainable. Since many OCR engines cannot reliably work with the formulas, it is essential to isolate them to obtain the clean text for information extraction from the document. Our proposed pipeline comprises a hybrid task cascade network with deformable convolutions and a Resnext101 backbone. Both of these modifications help in better detection. We evaluate the proposed approaches on the ICDAR-2017 POD and Marmot datasets and achieve an overall accuracy of 96% for the ICDAR-2017 POD dataset. We achieve an overall reduction of error of 13%. Furthermore, the results on Marmot datasets are improved for the isolated and embedded formulas. We achieved an accuracy of 98.78% for the isolated formula and 90.21% overall accuracy for embedded formulas. Consequently, it results in an error reduction rate of 43% for isolated and 17.9% for embedded formulas.

Large Scale System for Social Media Data Warehousing

Social media data become an integral part in the business data and should be integrated into the decisional process for better decision making based on information which reflects better the true situation of business in any field. However, social media data are unstructured and generated in very high frequency which exceeds the capacity of the data warehouse. In this work, we propose to extend the data warehousing process with a staging area which heart is a large scale system implementing an information extraction process using Storm and Hadoop frameworks to better manage their volume and frequency. Concerning structured information extraction, mainly events, we combine a set of techniques from NLP, linguistic rules and machine learning to succeed the task. Finally, we propose the adequate data warehouse conceptual model for events modeling and integration with enterprise data warehouse using an intermediate table called Bridge table. For application and experiments, we focus on drug abuse events extraction from Twitter data and their modeling into the Event Data Warehouse.