Search

The Sinian gas reservoir in the Sichuan Basin has the potential for natural gas exploration and development. Production prediction and risk quantification are important in planning of natural gas resources. Ultimate recoverable reserves (URRs) of Sinian gas reservoir are estimated. Hubbert and Gauss models are used to predict the growth trend of production in the gas reservoir. Based on the prediction results, the Monte Carlo simulation is used to calculate the probability of production realization. The evaluation matrix of risk level is established by using indices of production realization probability and dispersion degree for assessing the risk level of natural gas production. The results show that: (1) compared to the Hubbert model, the production prediction results of the Gauss model have higher accuracy. The Sinian gas reservoir will reach peak production of $$\left( {140 - 285} \right) \times 10^{8} {\text{m}}^{3} /{\text{a}}$$ 140 - 285 × 10 8 m 3 / a in 2036 and will have stable production from 2032 to 2040. By the end of the stable production stage, the URR exploitation degree is about 60% and (2) the Monte Carlo method can be used to obtain the production realization probability for each year. The risk level evaluation matrix can be established by taking the probability of realization and the dispersion degree as evaluation indices, which can provide the systematization of the risk levels. The study can help to better understand the guiding significance for the natural gas exploration and development.

Quantifying natural gas production risk can help guide natural gas exploration and development in Carboniferous gas reservoirs. In this study, the Monte Carlo probability method is used to obtain the probability distribution and growth curve of each production risk factor and production in a Carboniferous gas reservoir in eastern Sichuan. In addition, the fuzzy comprehensive evaluation method is used to conduct the sensitivity analysis of the risk factors, and the natural gas production and realization probability under different risk factors are obtained. The research results show that: (1) the risk factor–production growth curve and probability distribution are calculated by the Monte Carlo probability method. The average annual production under the stable production stage under different realization probabilities is obtained. The maximum probability range of annual production is $$\left( {43.43 - 126.35} \right) \times 10^{8} {\text{m}}^{3} /{\text{year}}$$ 43.43 - 126.35 × 10 8 m 3 / year , and the probability range is 14.59–92.88%. (2) The risk factor sensitivity analysis is significantly affected by the probability interval. In the entire probability interval, the more sensitive risk factors are the average production of the kilometer-deep well (D) and the production rate in the stable production stage (A). During the exploration and development of natural gas, these two risk factors can be adjusted to increase production.

Open-domain Question Answering models that directly leverage question-answer (QA) pairs, such as closed-book QA (CBQA) models and QA-pair retrievers, show promise in terms of speed and memory compared with conventional models which retrieve and read from text corpora. QA-pair retrievers also offer interpretable answers, a high degree of control, and are trivial to update at test time with new knowledge. However, these models fall short of the accuracy of retrieve-and-read systems, as substantially less knowledge is covered by the available QA-pairs relative to text corpora like Wikipedia. To facilitate improved QA-pair models, we introduce Probably Asked Questions (PAQ), a very large resource of 65M automatically generated QA-pairs. We introduce a new QA-pair retriever, RePAQ, to complement PAQ. We find that PAQ preempts and caches test questions, enabling RePAQ to match the accuracy of recent retrieve-and-read models, whilst being significantly faster. Using PAQ, we train CBQA models which outperform comparable baselines by 5%, but trail RePAQ by over 15%, indicating the effectiveness of explicit retrieval. RePAQ can be configured for size (under 500MB) or speed (over 1K questions per second) while retaining high accuracy. Lastly, we demonstrate RePAQ’s strength at selective QA, abstaining from answering when it is likely to be incorrect. This enables RePAQ to “back-off” to a more expensive state-of-the-art model, leading to a combined system which is both more accurate and 2x faster than the state-of-the-art model alone.

Achieving large Fermi energy splitting under illumination is critical for improving the performance of photovoltaic devices. In this work, we conducted the energy band structure modulation of MoS2 thin films via MoOx doping scheme to realize its photovoltaic operation on n-type c-Si by magnetron sputtering. It is found that, by capping a MoOx overlayer, the MoS2 electrons density decreased and the Fermi level shifted ~0.4 eV towards valence band, consequently the MoS2 conductive type changing. With this doped MoS2 layer, the built-in electric field of fabricated MoS2/n-Si heterojunction devices greatly improved and exceeded 400 mV owing to the larger interface energy-level differences. Meanwhile, the defects recombination losses reduced benefiting from the holes-selective behavior of O-deficiencies in MoOx, so that the photo-generated carriers extraction process were promoted. As a result, the doped MoS2/n-Si devices showed excellent rectification behavior with rectifying ratio to 105 and ideality of 1.12. The TRPL spectrums corroborated the enhanced carrier transport mechanism at MoS2/n-Si interface by MoOx modulation doping effect. Correspondingly, the significant improvement of doped MoS2/n-Si solar cells performance (Voc of 289 mV, FF of 60.72% and Jsc of 31.25 mA/cm2) was achieved, exhibiting an optimized conversion efficiency of 5.47%. The UPS analysis revealed that the doping mechanism of MoOx/MoS2 stacks lies in energy band overlap and electrons contact transfer between these two materials. This observation of MoOx modulation doping effect on large-area sputtering MoS2 could help in building a better interface carrier transport in photovoltaic and optoelectronic applications of 2D materials.

This paper presents a multi-timescale receding horizon framework for the load forecast of large power customers. The future load pattern of individual users could be very difficult to predict because of its chronological and high volatile properties. Also, the sampling of nonaggregated load data may suffer from severe information missing issues. To address these challenges, we first develop an online singular value thresholding (SVT) algorithm, which utilizes the approximate low-rank property of load data matrices to efficiently recover the missing information. Then, a combinatorial deep learning method is developed, which applies the multi-layer perception (MLP) neural network and the long short-term memory (LSTM) neural network with gated recurrent unit (GRU) to deal with the short-term and ultra-short-term load forecast, respectively. Specifically, an early stopping strategy is designed and implemented to avoid the over-fitting of model training. Moreover, the receding time window is imposed to dynamically update the data recovery and load forecast outcomes, which supports the online computing on a Spark platform. Numerical experiments on real-world load data from North China confirms the effectiveness of the proposed methodology, which can support more complex applications in embedded systems and cyber physical systems.

Failing to consider the strong correlations between weights and topological properties in capacity-weighted networks renders test results on the scale-free property unreliable. According to the preferential attachment mechanism, existing high-degree nodes normally attract new nodes. However, in capacity-weighted networks, the weights of existing edges increase as the network grows. We propose an optimized simplification method and apply it to international trade networks. Our study covers more than 1200 product categories annually from 1995 to 2018. We find that, on average, 38%, 38% and 69% of product networks in export, import and total trade are scale-free. Furthermore, the scale-free characteristics differ depending on the technology. Counter to expectations, the exports of high-technology products are distributed worldwide rather than concentrated in a few developed countries. Our research extends the scale-free exploration of capacity-weighted networks and demonstrates that choosing appropriate filtering methods can clarify the properties of complex networks.

During the development of complex gas reservoirs, the risk decision-making problem often emerges. Thus, the study on risk assessment is an important tool used to identify potential hazards and create appropriate avoidance measures accordingly. Based on the analysis of seven types of risk factors in gas reservoir development planning, this paper aims to clarify the logical relationship between the risk factors in the strategic planning of natural gas development. The comprehensive research on target risks in the gas reservoir development planning based on stochastic simulation was carried out. The “probability curve scanning method” was used to evaluate objective risk factors, while the decision-making risk factors were evaluated using the “probability curve displacement method”. According to the realization probability and dispersion degree of the planned target combined with the risk grade evaluation matrix, the planning target evaluation risk grade was implemented. Moreover, the planning unit risk grade evaluation was obtained at different stages. Regarding the specific production capacity conditions in gas wells (horizontal and vertical wells) and gas reservoir water invasion – the probability method with Monte Carlo stochastic simulation was used to calculate the production and water invasion volumes. The established decision-making risk technology for gas reservoir development, along with the associated supporting procedures can be used to evaluate the risks of reservoir development planning, production, and water invasion.

In this paper, we aim to improve abstractive dialogue summarization quality and, at the same time, enable granularity control. Our model has two primary components and stages: 1) a two-stage generation strategy that generates a preliminary summary sketch serving as the basis for the final summary. This summary sketch provides a weakly supervised signal in the form of pseudo-labeled interrogative pronoun categories and key phrases extracted using a constituency parser. 2) A simple strategy to control the granularity of the final summary, in that our model can automatically determine or control the number of generated summary sentences for a given dialogue by predicting and highlighting different text spans from the source text. Our model achieves state-of-the-art performance on the largest dialogue summarization corpus SAMSum, with as high as 50.79 in ROUGE-L score. In addition, we conduct a case study and show competitive human evaluation results and controllability to human-annotated summaries., Comment: ACL-Findings 2021. Code is released at https://github.com/salesforce/ConvSumm

After nearly 60 years of development, many old gas fields in the Sichuan Basin have come to middle–late development stages with low pressure and low yield, and some are even on the verge of abandonment, but there are plenty remaining gas resources still undeveloped. Analysis shows that gas fields which have the conditions for the secondary development are faced with many difficulties. For example, it is difficult to produce low permeable reserves and to unset the hydraulic seal which is formed by active formation water. In this paper, therefore, the technical route and selection conditions of old gas fields for the secondary development were comprehensively elaborated with its definition as the beginning. Firstly, geological model forward modeling and production performance inversion characteristic curve diagnosis are performed by using the pressure normalization curve and the identification and quantitative description method for multiple sets of storage–seepage body of complex karst fracture–cavity systems is put forward, after the multiple storage–seepage body mode of fracture–cavity systems is established. Combined with the new occurrence mode of gas and water in U-shape pipes, a new calculation technology for natural gas reserves of multiple fracture–cavity systems with strong water invasion is developed. Secondly, a numerical model of pore–cavity–fracture triple media is built, and simulation and result evaluation technology for the production pattern of “drainage by horizontal wells + gas production by vertical wells” in bottom-water fracture and cavity gas reservoirs with strong water invasion is developed. Thirdly, the geological model of gas reservoirs is reconstructed with the support of the integration technologies which are formed based on fine gas reservoir description. Low permeable reserves of gas reservoirs are evaluated based on each classification. The effective producing ratio is increased further by using the technologies of well pattern optimization, horizontal-well geosteering and staged acid fracturing. And fourthly, overall simulation, optimization and prediction technology for regional pipeline net-works is developed by building a multi-node multi-link gas transmission pipeline network model. Application shows that this technology plays an important role in productivity construction, recovery factor improvement, production decline delay and production stabilization of old gas fields.

We demonstrate the integration of the Archives Unleashed Toolkit, a scalable platform for exploring web archives, with Google's Tensor-Flow deep learning toolkit to provide scholars with content-based image analysis capabilities. By applying pretrained deep neural networks for object detection, we are able to extract images of common objects from a 4TB web archive of GeoCities, which we then compile into browsable collages. This case study illustrates the types of interesting analyses enabled by combining big data and deep learning capabilities.

A core problem of information retrieval (IR) is relevance matching, which is to rank documents by relevance to a user’s query. On the other hand, many NLP problems, such as question answering and paraphrase identification, can be considered variants of semantic matching, which is to measure the semantic distance between two pieces of short texts. While at a high level both relevance and semantic matching require modeling textual similarity, many existing techniques for one cannot be easily adapted to the other. To bridge this gap, we propose a novel model, HCAN (Hybrid Co-Attention Network), that comprises (1) a hybrid encoder module that includes ConvNet-based and LSTM-based encoders, (2) a relevance matching module that measures soft term matches with importance weighting at multiple granularities, and (3) a semantic matching module with co-attention mechanisms that capture context-aware semantic relatedness. Evaluations on multiple IR and NLP benchmarks demonstrate state-of-the-art effectiveness compared to approaches that do not exploit pretraining on external data. Extensive ablation studies suggest that relevance and semantic matching signals are complementary across many problem settings, regardless of the choice of underlying encoders.

The microstructure of MoS2 films is crucial important for its photoelectrical properties, which has significant impact on its application in the field of photovoltaic devices. In this paper, a set of MoS2 thin films with varied thickness was prepared by magnetron sputtering. The influence of the MoS2 microstructure on its optical and electrical properties and, especially, the MoS2/p-Si solar cells photovoltaic performance was investigated in terms of thin films growth. The results reveal that, as the thickness increases, the microstructure of the MoS2 films evolves from amorphous to microcrystalline with the in-plane E2g1 and out-of-plane A1g Raman modes enhancing and the parallel (002) growth orientation increasing. The films electrical conductivity significantly improves owing to the enhanced MoS2 crystallinity. When the thickness is 10 nm, the small ordered MoS2 crystal clusters gradient distribute from the surface to the bulk in the films, contributing to the stress release in the amorphous precursor. Simultaneously, the optical band gap of the MoS2 films shows a maximum of 1.3 eV, due to the less band tail states in the films at the initial phase transition. The optimized optical-electrical properties of the MoS2 films lead to the reducing defects recombination and the enhanced carrier transportation at MoS2/p-Si interface, which promotes the rectification behavior of the pn-junction up to 102. Besides, the significant improvement of the photo-response in MoS2/p-Si heterojunction is obtained, especially in short wavelength range. Correspondingly, the photovoltaic characteristics (Voc, Jsc and FF) of the MoS2/p-Si solar cells increase, and thus an enhancement of the cells conversion efficiency is achieved with 10-nm MoS2 layer.

We propose a Multi-task learning approach for Abstractive Text Summarization (MATS), motivated by the fact that humans have no difficulty performing such task because they have the capabilities of multiple domains. Specifically, MATS consists of three components: (i) a text categorization model that learns rich category-specific text representations using a bi-LSTM encoder; (ii) a syntax labeling model that learns to improve the syntax-aware LSTM decoder; and (iii) an abstractive text summarization model that shares its encoder and decoder with the text categorization and the syntax labeling tasks, respectively. In particular, the abstractive text summarization model enjoys significant benefit from the additional text categorization and syntax knowledge. Our experimental results show that MATS outperforms the competitors.1

As an indigenous concept of China, guanxi has gained broad acceptance by the west without further translation or interpretation during the past 30 years, which is a fascinating topic to investigate. In order to discover the dissemination mechanism of the concept of guanxi, this study employs co-citation analysis and citation analysis to visualize the intellectual base and the evolution of guanxi studies; also it pinpoints the most productive authors, institutions and journals. The findings may provide some guidance for the scientific knowledge dissemination.

Influenza viruses have been isolated from dogs in China; however, the extent of influenza infection among dogs is not yet clear. Here, we examined the seroprevalence of avian-origin canine H3N2, pandemic H1N1/09 and human seasonal H3N2 influenza viruses in pet dogs in China during January 2012 to June 2013. The seropositivity rate of canine H3N2, H1N1/09 and human H3N2 were 3.5%, 1.5%, and 1.2%, respectively. Dogs aged 2–5 years were most commonly seropositive to canine H3N2 virus. It is worth noting that two serum samples were positive against both canine H3N2 and H1N1/09 viruses, suggesting the possibility of coinfection with both viruses. Our findings emphasize the necessity for continued surveillance of influenza viruses in dogs in China.

Spammer detection on social network is a challenging problem. The rigid anti-spam rules have resulted in emergence of "smart" spammers. They resemble legitimate users who are difficult to identify. In this paper, we present a novel spammer classification approach based on Latent Dirichlet Allocation(LDA), a topic model. Our approach extracts both the local and the global information of topic distribution patterns, which capture the essence of spamming. Tested on one benchmark dataset and one self-collected dataset, our proposed method outperforms other state-of-the-art methods in terms of averaged F1-score., NAACL-HLT 2016, Student Research Workshop

Testing-as-a-service (TaaS) is a new model to provide testing capabilities to end users. Users save the cost of complicated maintenance and upgrade effort, and service providers can upgrade their services without impact on the end-users. Due to uneven volumes of concurrent requests, it is important to address the elasticity of TaaS platform in a cloud environment. Scheduling and dispatching algorithms are developed to improve the utilization of computing resources. We develop a prototype of TaaS over cloud, and evaluate the scalability of the platform by increasing the test task load, analyze the distribution of computing time on test task scheduling and test task processing over the cloud, and examine the performance of proposed algorithms by comparing others

This paper describes a concept to support scalable multicast communications for small audio/video conferencing groups on the Internet. The solution presented in this paper is based on extensions of IPv6 and the session description protocol (SDP). A goal of the concept called multicast for small conferences (MSC) is the smooth deployment in the Internet.