Multi-Graph Cooperative Learning Towards Distant Supervised Relation Extraction

The Graph Convolutional Network (GCN) is a universal relation extraction method that can predict relations of entity pairs by capturing sentences’ syntactic features. However, existing GCN methods often use dependency parsing to generate graph matrices and learn syntactic features. The quality of the dependency parsing will directly affect the accuracy of the graph matrix and change the whole GCN’s performance. Because of the influence of noisy words and sentence length in the distant supervised dataset, using dependency parsing on sentences causes errors and leads to unreliable information. Therefore, it is difficult to obtain credible graph matrices and relational features for some special sentences. In this article, we present a Multi-Graph Cooperative Learning model (MGCL), which focuses on extracting the reliable syntactic features of relations by different graphs and harnessing them to improve the representations of sentences. We conduct experiments on a widely used real-world dataset, and the experimental results show that our model achieves the state-of-the-art performance of relation extraction.

Impact of Sex on COVID-19 Media Exposure, Anxiety, Perceived Risk, and Severity in Middle-Aged and Older Adults

Objectives To examine associations between COVID-19 media exposure and anxiety/perceived risk/severity and investigate their dependency on sex in middle-aged/older adults. Methods Adults aged 50+ years completed online surveys: Coronavirus Anxiety Scale, COVID-19 media exposure, COVID-19 media dependency for health information, and COVID-19 perceived risk and severity. Multiple regressions examined independent and interactive (with sex) associations between COVID-19 media exposure/dependency and COVID-19 anxiety/perceived risk and severity. Analyses controlled for age, education, race, total medical conditions, and COVID-19 status. Results Higher COVID-19 media exposure was associated with higher COVID-19 anxiety among men (not women) and higher perceived risk/severity in both sexes. Higher COVID-19 media dependency was associated with higher COVID-19 anxiety and perceived risk/severity in both sexes. Conclusion In middle-aged/older adults, the use/dependency of media for COVID-19 information may be linked to negative psychological health and increased COVID-19 perceived risk and severity. Men may be at increased risk of anxiety related to media exposure.

Improving Cross-Domain Performance for Relation Extraction via Dependency Prediction and Information Flow Control

Relation Extraction (RE) is one of the fundamental tasks in Information Extraction and Natural Language Processing. Dependency trees have been shown to be a very useful source of information for this task. The current deep learning models for relation extraction has mainly exploited this dependency information by guiding their computation along the structures of the dependency trees. One potential problem with this approach is it might prevent the models from capturing important context information beyond syntactic structures and cause the poor cross-domain generalization. This paper introduces a novel method to use dependency trees in RE for deep learning models that jointly predicts dependency and semantics relations. We also propose a new mechanism to control the information flow in the model based on the input entity mentions. Our extensive experiments on benchmark datasets show that the proposed model outperforms the existing methods for RE significantly.

Very Early Initiation Reduces Benefits of Poststroke Rehabilitation Despite Increased Corticospinal Projections

Background. Although the effect of rehabilitation is influenced by aspects of the training protocol, such as initiation time and intensity of training, it is unclear whether training protocol modifications affect the corticospinal projections. Objective. The present study was designed to investigate how modification of initiation time (time-dependency) and affected forelimb use (use-dependency) influence the effects of rehabilitation on functional recovery and corticospinal projections. Methods. The time-dependency of rehabilitation was investigated in rats forced to use their impaired forelimb immediately, at 1 day, and 4 days after photothrombotic stroke. The use-dependency of rehabilitation was investigated by comparing rats with affected forelimb immobilization (forced nonuse), unaffected forelimb immobilization (forced use), and a combination of forced use and skilled forelimb training beginning at 4 days after stroke. Results. Although forced use beginning 1 day or 4 days after stroke caused significant functional improvement, immediate forced limb use caused no functional improvement. On the other hand, a combination of forced use and skilled forelimb training boosted functional recovery in multiple tasks compared to simple forced use treatment. Histological examination showed that no treatment caused brain damage. However, a retrograde tracer study revealed that immediate forced use and combination training, including forced use and skilled forelimb training, increased corticospinal projections from the contralesional and ipsilesional motor cortex, respectively. Conclusions. These results indicate that although both very early initiation time and enhanced skilled forelimb use increased corticospinal projections, premature initiation time hampers the functional improvement induced by poststroke rehabilitation.

Risk Factors for Fracture in Alcohol-dependent, Pre-menopausal Females

Aims: Alcoholism is known to be associated with increased risk of fracture. This study aimed to study bone turnover following alcohol detoxification and to investigate lifestyle factors for low bone density that might coexist with alcohol dependency, which might be amenable to modification. Method: Pre-menopausal female participants were recruited from an alcohol-use dependency unit to a cross-sectional study. A lifestyle questionnaire, including alcohol history, smoking, physical activity, dietary calcium intake, falls, and fracture history was completed. Quantitative heel ultrasonography was performed and broadband ultrasound attenuation (BUA), speed of sound (SOS), t score, and z score were recorded. Blood was taken for bone-turnover markers at baseline and day 5 following admission for alcohol withdrawal.Results: The mean age (SD) of alcohol dependent participants was 41.6 (8.3) years, with participants reporting high levels of current cigarette smoking, physical inactivity, and falls. BUA, SOS, t scores, and z scores were lower than the age-matched reference range in alcohol-dependent participants. Levels of type 1 procollagen (P1NP) increased significantly after five days (p < .001). Conclusions: Alcohol-dependent, pre-menopausal individuals have multiple risk factors for fracture, beyond alcohol excess. These should be addressed and targeted as modification may reduce fracture risk, especially given the apparent recovery of bone turnover on the withdrawal of alcohol.

SOIL CARBON MAPPING IN LOW RELIEF AREAS WITH COMBINED LAND USE TYPES AND PERCENTAGES

Accurate mapping of soil carbon in low relief areas is of great challenge because of the defect of conventional “soil-landscape” model. Efforts have been made to integrate the land use information in the modelling and mapping of soil organic carbon (SOC), in which the spatial context was ignored. With 256 topsoil samples collected from Jianghan Plain, we aim to (i) explore the land-use dependency of SOC via one-way ANOVA; (ii) investigate the “spillover effect” of land use on SOC content; (iii) examine the feasibility of land use types and percentages (obtained with a 200-meter buffer) for soil mapping via regression Kriging (RK) models. Results showed that the SOC of paddy fields was higher than that of woodlands and irrigated lands. The land use type could explain 20.5&amp;thinsp;% variation of the SOC, and the value increased to 24.7&amp;thinsp;% when the land use percentages were considered. SOC was positively correlated with the percentage of water area and irrigation canals. Further research indicated that SOC of irrigated lands was significantly correlated with the percentage of water area and irrigation canals, while paddy fields and woodlands did not show similar trends. RK model that combined land use types and percentages outperformed the other models with the lowest values of RMSE_C (5.644&amp;thinsp;g/kg) and RMSE_P (6.229&amp;thinsp;g/kg), and the highest R²_C (0.193) and R²_P (0.197). In conclusions, land use types and percentages serve as efficient indicators for the SOC mapping in plain areas. Additionally, irrigation facilities contributed to the farmland SOC sequestration especially in irrigated lands.