Search

Relation extraction is one of the fundamental subtasks of the information extraction. The purpose is to determine the implicit relation between two entities in a sentence. Therefore, Convolutional Neural Networks and Feature Attention-based Prototypical Networks (CNN-Proto-FATT), a typical few-shot learning method, is proposed and achieve competitive performance. However, convolutional neural networks suffer from the insufficient instances of relation in real scenes, leading to undesirable results. To extract long-distance features more comprehensively, the pre-trained model Bidirectional Encoder Representation from Transformers (BERT) is incorporated into CNN-Proto-FATT. In this model, named Bidirectional Encoder Representation from Transformers and Feature Attention-based Prototypical Networks (BERT-Proto-FATT), the multi-head attention helps the network extract semantic features cross long- and short-distance to enhance the encoded representations. Experimental results indicate that BERT-Proto-FATT demonstrates significant improvements on the FewRel dataset.

Background Forage nutritive value plays an important role in livestock nutrition and maintaining sustainable grassland ecosystems, and grazing management can affect the quality of forage. In this study, we investigated the effects of different grazing intensities on the nutritive values of Leymus chinensis (Trin.) Tzvelev, Artemisia spp. and Carex duriuscula C. A. Mey in the steppes of China during the growing seasons from 2011 to 2013. Five grazing management treatments were implemented: (1) rest grazing in spring, heavy grazing in summer and moderate grazing in autumn (RHM), (2) rest grazing in spring, moderate grazing in summer and heavy grazing in autumn (RMH), (3) heavy grazing though all seasons (HHH), (4) heavy grazing in spring and summer and moderate grazing in autumn (HHM) and (5) continuous moderate grazing in all seasons (MMM). Results There were significant effects of year, season, treatment, and year × season and year × treatment interactions only on the crude protein of L. chinensis (P 0.05) for ether extract, neutral detergent fiber, acid detergent fiber and Ca concentration for any of the grasses in spring and summer from 2011 to 2013 under the different grazing management treatments. Conclusions The nutritive value of L. chinensis was more responsive to grazing disturbance than Artemisia spp. and C. duriuscula, and heavy grazing maintained a relatively high crude protein content in all species. Seasonal and interannual seasonal differences in grazing management combinations were two of the most important factors in determining the variability of forage nutritive value, including crude protein, ether extract, neutral detergent fiber, acid detergent fiber and calcium, for L. chinensis, Artemisia spp. and C. duriuscula. We suggest that moderate grazing should be adopted to ensure the quality and yield of forage and promote the sustainable development of animal husbandry.

This paper proposes a lightweight mechanism for transparent migration of mobile agents. It introduces the continuation mechanism of the functional programming language into mobile agents, and implements the capture and reestablishment of agent's execution state by bytecode transformation mechanism. We have implemented a prototype, called JMobile, which is composed of three modules: continuation module, bytecode transformation module, and agent migration module based on Java socket. We have also measured growth in bytecode size and execution efficiency, and obtained relatively better results compared to other implementations.