Добірка наукової літератури з теми "KGEval"

Knowledge graph conflict resolution is a method to solve the knowledge conflict problem in constructing knowledge graphs. The existing methods ignore the time attributes of facts and the dynamic changes of the relationships between entities in knowledge graphs, which is liable to cause high error rates in dynamic knowledge graph construction. In this article, we propose a knowledge graph conflict resolution method, knowledge graph evolution algorithm based on deep learning (Kgedl), which can resolve facts confliction with high precision by combing time attributes, semantic embedding representations, and graph structure features. Kgedl first trains the semantic embedding vector through the relationships between entities. Then, the path embedding vector is trained from the graph structures of knowledge graphs, and the time attributes of entities are combined with the semantic and path embedding vectors. Finally, Kgedl uses a recurrent neural network to make the inconsistent facts appear in the dynamic evolution of the knowledge graph consistent. A large number of experiments on real datasets show that Kgedl outperforms the state-of-the-art methods. Especially, Kgedl achieves 23% higher performance than the classical method numerical Probabilistic Soft Logic (nPSL).in the metric HITS@10. Also, extensive experiments verified that our proposal possess better robustness by adding noise data.

The task of multi-hop question generation (QG) seeks to generate questions that require a complex reasoning process that spans multiple sentences and answers. Beyond the conventional challenges of what to ask and how to ask, multi-hop QG necessitates sophisticated reasoning from dispersed evidence across multiple sentences. To address these challenges, a knowledge graph-enhanced language model (KGEL) has been developed to imitate human reasoning for multi-hop questions.The initial step in KGEL involves encoding the input sentence with a pre-trained GPT-2 language model to obtain a comprehensive semantic context representation. Next, a knowledge graph is constructed using the entities identified within the context. The critical information in the graph that is related to the answer is then utilized to update the context representations through an answer-aware graph attention network (GAT). Finally, the multi-head attention generation module (MHAG) is performed over the updated latent representations of the context to generate coherent questions. Human evaluations demonstrate that KGEL generates more logical and fluent multi-hop questions compared to GPT-2. Furthermore, KGEL outperforms five prominent baselines in automatic evaluations, with a BLEU-4 score that is 27% higher than that of GPT-2.

En el municipio de San Vicente de Chucurí, Santander, se produce aproximadamente el 11,8% del cacao de Colombia. Asociado con la producción de los granos de cacao se producen de 74-86% de residuos durante el proceso de beneficio, lo que representaría entre 4715–5480 toneladas al año. Esta situación genera problemas ambientales y fitosanitarios que se deben atender en el futuro inmediato. Es así que en este trabajo se estudia el aprovechamiento de los residuos de cacao (cáscara) del material clon CCN-51 para la extracción de polifenoles y fibra total. Desde el punto de vista metodológico, se desarrollaron dos etapas; en la primera se caracterizaron las cáscaras de cacao evaluando el contenido de polifenoles totales (método de Folin-Ciocalteu) y fibra dietaria total (método enzimático-gravimétrico); en lo que respecta a la segunda etapa, se realizó la simulación (software SuperPro Designer ®v.9.0 académica) de una planta extractora usando como información de entrada al simulador los resultados obtenidos en la primera etapa. La simulación se desarrolló tomando como base de cálculo el 32,6% de la producción del municipio y los resultados fueron: el contenido de polifenoles totales y fibra dietaria total para el clon CCN-51 fueron respectivamente 61 mgEAG/g (expresados en peso seco) y 56,80%. En lo que respecta a la simulación, se propuso un diagrama de proceso en que se incluyeron las etapas de preparación, extracción, concentración de polifenoles y fibra total. Con esta configuración fue posible obtener 8,6 kgEAG/h de polifenoles totales y 80,3 kg/h de fibra total presentes en la torta de cáscara de cacao.Palabras clave: cáscara de cacao, fibra dietaria total, polifenoles, simulación.AbstractThe municipality of San Vicente de Chucurí, department of Santander, produces approximately 11,8% of Colombia cocoa. In the production of cocoa beans, specifically during the beneficiation process, between 74-86% of waste are generated, which represents between 4715-5480 tons per year. This situation generates environmental and phytosanitary issues that should be addressed in the near future. Thus, in this paper it was evaluated the use of cocoa waste (husk) of CCN51 material for the extraction of polyphenols and total dietary fiber. Methodologically, two stages were developed. In thefirst stage, the cocoa husks were characterized by determining the total polyphenol content (Folin-Ciocalteu method) and total dietary fiber (enzyme-gravimetric method). With respect to second stage, it was performed the simulation (SuperPro Designer® software, academic version 9.0) of an extraction plant, using as input data of the simulator the values of obtained experimentally in the first stage. The simulation was developed on the basis of calculating the 32.6% of production in the municipality of San Vicente and the results obtained were as follows: the content of total polyphenols and total dietary for clone CCN-51 were respectively 61 mgEGA/gdm y 56,80%. With respect to the simulation, a process diagram was proposed in which the stages of preparation, extraction and concentration of polyphenols and total dietary fiber were included. With this configuration, it was possible to obtain 8,6 kgEGA/h total polyphenols and 80,3 kg/h total dietary fiber present in the cocoa husk mass.Keywords: Cocoa Husk, Total dietary fiber, Polyphenol, Simulation

Entity linking involves mapping ambiguous mentions in documents to the correct entities in a given knowledge base. Most of the current methods are a combination of local and global models. The local model uses the local context information around the entity mention to independently resolve the ambiguity of each entity mention. The global model encourages thematic consistency across the target entities of all mentions in the document. However, the known global models calculate the correlation between entities from a semantic perspective, ignoring the correlation information between entities in nature. In this paper, we introduce knowledge graphs to enrich the correlation information between entities and propose an entity linking model that introduces the structural information of the knowledge graph (KGEL). The model can fully consider the relations between entities. To prove the importance of the knowledge graph structure, extensive experiments are conducted on multiple public datasets. Results illustrate that our model outperforms the baseline and achieves superior performance.

Abstract Background Knowledges graphs (KGs) serve as a convenient framework for structuring knowledge. A number of computational methods have been developed to generate KGs from biomedical literature and use them for downstream tasks such as link prediction and question answering. However, there is a lack of computational tools or web frameworks to support the exploration and visualization of the KG themselves, which would facilitate interactive knowledge discovery and formulation of novel biological hypotheses. Method We developed a web framework for Knowledge Graph Exploration and Visualization (KGEV), to construct and visualize KGs in five stages: triple extraction, triple filtration, metadata preparation, knowledge integration, and graph database preparation. The application has convenient user interface tools, such as node and edge search and filtering, data source filtering, neighborhood retrieval, and shortest path calculation, that work by querying a backend graph database. Unlike other KGs, our framework allows fast retrieval of relevant texts supporting the relationships in the KG, thus allowing human reviewers to judge the reliability of the knowledge extracted. Results We demonstrated a case study of using the KGEV framework to perform research on COVID-19. The COVID-19 pandemic resulted in an explosion of relevant literature, making it challenging to make full use of the vast and heterogenous sources of information. We generated a COVID-19 KG with heterogenous information, including literature information from the CORD-19 dataset, as well as other existing knowledge from eight data sources. We showed the utility of KGEV in three intuitive case studies to explore and query knowledge on COVID-19. A demo of this web application can be accessed at http://covid19nlp.wglab.org. Finally, we also demonstrated a turn-key adaption of the KGEV framework to study clinical phenotypic presentation of human diseases by Human Phenotype Ontology (HPO), illustrating the versatility of the framework. Conclusion In an era of literature explosion, the KGEV framework can be applied to many emerging diseases to support structured navigation of the vast amount of newly published biomedical literature and other existing biological knowledge in various databases. It can be also used as a general-purpose tool to explore and query gene-phenotype-disease-drug relationships interactively.