Self-charging zinc batteries that combine energy harvesting technology with are candidates for reliable self-charging power systems. However, the lack of rational materials design results in unsatisfactory performance. Here, a covalent organic framework containing pyrene-4,5,9,10-tetraone groups (COF-PTO) is reported as cathode material aqueous batteries. The ordered channel structure COF-PTO provides excellent capacity retention 98% after 18 000 cycles at 10 A g

Answer selection for question answering is a challenging task, since it requires effective capture of the complex semantic relations between questions and answers. Previous remarkable approaches mainly adopt general Compare-Aggregate framework that performs word-level comparison aggregation. In this paper, unlike previous models which utilize traditional attention mechanism to generate corresponding vector before comparison, we propose novel named Dynamic-Clip Attention directly integrated...

Feature Selection (FS) is one of the most important issues in Text Categorization (TC). Empirical studies show that Information Gain (IG) an effective method FS. However, as traditional IG gives little attention to term frequency and takes into account situation does not appear, effect ideal. In this paper, we put forward improved information gain-based feature selection using balance factor(IGTB) for statistical machine learning-based text categorization. Our strives precisely pick out key...

Collaborative filtering is a very important technology in e-commerce. Unfortunately, with the increase of users and commodities, user rating data extremely sparse, which leads to low efficient collaborative recommendation system. To address these issues, an optimized algorithm based on item proposed. While calculating similarity two items, we obtain ratio who rated both items those each them. The taken into account this method. experimental results show that proposed can improve quality filtering.

Chinese grammatical error correction (CGEC) is practically useful for learners of as a second language, but it rather challenging task due to the complex and flexible nature language so that existing methods English cannot be directly applied. In this paper, we introduce convolutional sequence model into CGEC first time, since many errors are concentrated between three four words neural network can better capture local context. A convolution-based obtain representations context by fixed size...

Due to the explosive growth of Web pages, centralized crawlers are no longer sufficient run on efficiently. There many distributed in wide use; however, none them is suitable for template-customized vertical crawling. In this paper, we present a crawler which specially used crawling Internet forums. The Client-Server architecture system and function every module described detail can be extended other fields easily. A crawling-period based distribution strategy also proposed, with manager...

This paper presents a systematic solution of the information retrieval in online Chinese resume. resume's contents have several expression and structure resume is complex. So this applies rule-based statistical algorithm to extract information. We get high accuracy experiment on 1500 resumes, which are offered by www.chinahr.com.

Semi-structured Chinese document analysis is the most difficult task for complex structure and semantics. According to generic characteristics of semi-structured specific resume document, paper researched on block based pattern matching, multi-level information identification feedback control algorithms was also prompted. Based research, parser system implemented ChinaHR, which biggest recruitment Website. It can read, analysis, retrieval store automatically. all kinds experiments results,...

目前的关联规则挖掘算法主要依靠基于支持度的剪切策略来减小组合搜索空间.如果挖掘潜在的令人感兴趣的低支持度模式,这种策略并非有效.为此,提出一种新的关联模式——可信关联规则(credible association rule,简称CAR),规则中每个项目的支持度处于同一数量级,规则的置信度直接反映其可信程度,从而可以不必再考虑传统的支持度.同时,提出MaxCliqueMining算法,该算法采用邻接矩阵产生2-项可信集,进而利用极大团思想产生所有可信关联规则.提出并证明了几个相关命题以说明这种规则的特点及算法的可行性和有效性.在告警数据集及Pumsb数据集上的实验表明,该算法挖掘CAR具有较高的效率和准确性.;Existing association-rule mining algorithms mainly rely on the support-based pruning strategy to prune its combinatorial search space. This is not quite effective in process of potentially...

Chinese named entity recognition is one of the most important tasks in NLP. The paper mainly describes our work on NER tasks. built up a system under framework conditional random fields (CRFs) model. With an improved tag set gets F-value 93.49 using SIGHAN2007 MSRA corpus.

Text classification is a foundational task in natural language processing (NLP). Traditional methods rely heavily on human-designed features, while deep learning models based neural networks can automatically capture contextual information. We explore and introduce various network architectures to extract information key components texts. An extensive set of experiments comparisons accuracy, speed, memory-consumption are conducted. Methods the proposed won first place Zhihu Machine Learning...

In this paper, we present a special crawler for Internet forums. Different from general and focused crawler, it can get structured information directly the most valuable Web resources by utilizing least system resources, filter useless to maximum extent finally supply users with high-precision information. This adopts template-based processing method which is use regular expressions extract The URL queue initialized URLs set in seeds file are extracted pages added into during crawling...

Microblogging is one of the most popular on-line social services by which people can share information and communicate with others. In paper, it proved that in a microblogging's network, there exists small-world, scale-free high clustering coefficient characteristics, are three main properties complex network. It means network Besides, verified number microblogs frequency posting follow power-law distributions, an evidence shows human dynamics non-Poissonian. The related geographical...

Keywords extraction is the process of choosing several words from a document to express its main idea. help people understand an article quickly and clearly. In recent years, more researchers pay attention research since important role in text clustering, classification, automatic abstracting, retrieval. This paper proposes algorithm called EC-DC extract keywords based on centrality measures complex network. A mapped network with vertices relations between edges. Then, importance evaluated...

Unknown word recognition is a key issue in Chinese information processing. The traditional algorithms of unknown can be broadly classified into two types: the rule-based methods and statistical methods. However, these have some limitations identifying words which are created on Internet. Internet no obvious rules composed common words, so them; while also them for they use mutual information. Therefore, this paper proposes an algorithm recognition, based bigram model uses method mining...

This article puts forward a new multi-criteria ranking method using Analytic Hierarchy Process (AHP) analysis to evaluate key nodes in complex networks. As is well known, most of existing methods only consider one factor (e.g. betweenness, degree), but not the integration multiple factors evaluating important nodes, so those each have limited application range. In addition, because different are contribution on importance, then how determine ratio problem. So this study, AHP used get best...

We introduce a novel approach that is used to convert images into the corresponding language descriptions. This method follows most popular encoder-decoder architecture. The encoder uses recently proposed densely convolutional neural network (DenseNet) extract feature maps. Meanwhile, decoder long short time memory (LSTM) parse maps predict next word of descriptions by taking effective combination with embedding current input "visual attention switch". Finally, we compare performance model...

A Bayesian approach termed the BAyesian Least Squares Optimization with Nonnegative L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> -norm constraint (BALSON) is proposed. The error distribution of data fitting described by Gaussian likelihood. parameter assumed to be a Dirichlet distribution. With Bayes rule, searching for optimal parameters equivalent finding mode posterior In order explicitly characterize nonnegative parameters, we...

Rich information is contributed to blogs by millions of users all around the world with development blogsphere. However, few work has been done on study blog extraction so far. Unlike traditional template-dependent wrapper, not only articles but also blogroll extracted template-independent wrapper in this paper. In our method, formalized as a machine learning problem and learned using labeled pages from single site. Testing are obtained 10 popular Chinese sites. And experimental results 300...

Credible association rule(CAR) is a new type of pattern in which items are highly affiliated with each other. The presence an item one transaction strongly implies the every other same CAR. And maximal CAR whose superset isn't CAR, so specifies more compact representation group CARs. In this paper, we introduce some measures for all represent affinity A mining method based on clique also presented to mine experimental results demonstrate that effective than stand methods rules.