一种新的词语权重计算方法

词语权重计算是自然语言处理中的重要问题,是文档语义表示的重要实现手段.词语的权重主要由两部分决定,一部分是词语在文档中的重要性度量,另一部分是基于统计方法所得的词语在区分不同文档能力上的度量即词语全局权重.本文就针对有相对稳定性的词语全局权重计算方法分析比较了现有的几种词语全局权重计算方法,并在此基础上提出了一种新的组合词语全局权重计算方法,实验证明本文提出的新方法能够有效地提高了系统的性能.     

全信息理论:定义与测度

迄今,人们所熟悉的信息理论是 C.E.Shannon 在1948年建立的概率统计型的信息理论,即 Shannon 信息论.本文试图提出一种新的信息理论,即全信息理论,以适应新的发展的需要.文中将阐明全信息理论产生的背景、全信息的基本概念、全信息的度量方法、以及全信息理论应用的前景.     

关于磨细粉煤灰混凝土早期强度的研究

为解决在粉煤灰掺入混凝土及砂浆时其早期强度偏低的问题，采用了掺化学外加剂的方法，以激发粉煤灰的潜在活性，达到解决早期强度偏低的缺陷。掺入早强磨细粉煤灰25％～30％取代同等量水泥，不仅降低工程造价，而且为利用工业废渣开辟了新的途径。     

一起基于油色谱分析的变压器潜伏性故障分析

针对某500kV变压器油色谱数据异常情况,采用改良的特征气体法、三比值法进行分析判断,结合局部放电试验进行验证分析,发现了一起变压器绝缘件受潮的潜伏性故障,通过后期返厂解体检查,查明了故障点,分析了故障原因,并提出了相关建议。     

Analysis of Technology Potential of a Metal-based Latent Heat Storage for Backup Process Steam Supply

An analysis of technology potential of a metal-based high temperature latent heat storage for the application of process steam backup for industrial parks in Germany was carried out. To investigate the integration of the storage into a process steam network, dynamic process simulations were performed. Furthermore, the analysis includes ecological and economical calculations for various fuel types of industrial power plants, plant sizes and auxiliary fuels. Basis of all calculations is an extensive research conducted on the industrial parks in Germany with a dedicated power plant and process steam utilization.     

Supporting exact indexing of arbitrarily rotated shapes and periodic time series under Euclidean and warping distance measures

Shape matching and indexing is important topic in its own right, and is a fundamental subroutine in most shape data mining algorithms. Given the ubiquity of shape, shape matching is an important problem with applications in domains as diverse as biometrics, industry, medicine, zoology and anthropology. The distance/similarity measure for used for shape matching must be invariant to many distortions, including scale, offset, noise, articulation, partial occlusion, etc. Most of these distortions are relatively easy to handle, either in the representation of the data or in the similarity measure used. However, rotation invariance is noted in the literature as being an especially difficult challenge. Current approaches typically try to achieve rotation invariance in the representation of the data, at the expense of discrimination ability, or in the distance measure, at the expense of efficiency. In this work, we show that we can take the slow but accurate approaches and dramatically speed them up. On real world problems our technique can take current approaches and make them four orders of magnitude faster without false dismissals. Moreover, our technique can be used with any of the dozens of existing shape representations and with all the most popular distance measures including Euclidean distance, dynamic time warping and Longest Common Subsequence. We further show that our indexing technique can be used to index star light curves, an important type of astronomical data, without modification. Reproducible Research Statement: All datasets and images used in this work are freely available at .     

Paillier''s Trapdoor Function Hides up to O ( n ) Bits

At EuroCrypt '99 Paillier proposed a new encryption scheme based on higher residuosity classes. The new scheme was proven to be one-way under the assumption that computing N -residuosity classes in Z _N2 ^* is hard. Similarly the scheme can be proven to be semantically secure under a much stronger decisional assumption: given w ∈ Z _N2 ^* it is impossible to decide if w is an N -residue or not. In this paper we examine the bit security of Paillier's scheme. We prove that if computing residuosity classes is hard, then given a random w it is impossible to predict the least significant bit of its class significantly better than at random. This immediately yields a way to obtain semantic security without relying on the decisional assumption (at the cost of several invocations of Paillier's original function). In order to improve efficiency we then turn to the problem of simultaneous security of many bits. We prove that Paillier's scheme hides n-b (up to O(n) ) bits if one assumes that computing the class c of a random w remains hard even when we are told that c<2 ^b . We thoroughly examine the security of this stronger version of the intractability of the class problem. An important theoretical implication of our result is the construction of the first trapdoor function that hides super-logarithmically (up to O(n) ) many bits. We generalize our techniques to provide sufficient conditions for a trapdoor function to have this property.     

Knowledge-based learning using Conceptual Transformers

The notion of concept based on the semantics of objects is defined and illustrated. An underlying thread connecting a subset of concepts is identified. This class of concepts, called the Conceptual Transformer is defined and illustrated with real-world examples. This class finds a natural application in any area where objects can be characterized by functionality. Some interesting application areas are knowledge classification, manufacturing automation, and pattern synthesis. The salient features of this class are elaborated and a knowledge structure for representing concepts is proposed. The effect of these transformers on knowledge-directed classification, which results in the formation of virtual clusters, is examined in detail. We make use of examples from real life to bring out the efficacy of the proposed transformerbased, concept-directed classification.     

可移植的汉语人—机接口系统的研究与实现

可移植汉语人一机接口是我国界当前研究的热点之一。本文通过分析汉语的特点, 对在微机上建造中小型应用系统的可移植汉语人一机接口, 提出了一种切实可行的实现方法, 文中还介绍了反用这一方法实现的实验系统的具体结构。