基于融合特征的虚假评论检测方法

针对现有虚假评论检测方法未充分利用用户历史行为中蕴含的动态信息,首先利用时序分析模型从这些动态信息中挖掘能够刻画用户行为的动态特征;其次,融合这些动态特征与用户层面静态特征发现可疑用户,并将用户可疑概率传播至用户所发表评论得到评论可疑概率;最后,融合评论可疑概率与评论层面静态特征形成融合特征,使用PU-Learning分类策略实现虚假评论的检测。真实数据集上的实验表明,本文方法的性能优于现有方法。     

嵌入式视觉的测距系统设计

行车操作人员在实际操作过程中,会出现对与之相邻障碍物之间的距离太近而不能及时规避情况,导致碰撞事故的发生.针对这一问题,笔者构建了以ARM11微处理器为核心,以USB摄像头为图像采集设备的测距系统,在基于机器视觉和图像处理的基础上,将单目视觉和激光技术相结合,考虑对应点匹配和单个特征点提取误差对测量结果影响,将激光点作为特征点的单目视觉测距的方法.实验结果表明,该系统能有效的测量与障碍物的距离,测量准确,可靠性高且满足实时性要求.     

Identification of intra-individual variation in intracranial arterial flow by MRI and the effect on computed hemodynamic descriptors

Magnetic Resonance Materials in Physics, Biology and Medicine - To determine the intra-individual flow variation in serially acquired studies, and the influence of this variation on subsequent...     

Evaluation on wear behavior of Cr-Ag-N and Cr-W-N PVD nanocomposite coatings using two different types of tribometer

Practice has proven that CrN performs better than TiN under some specific tribological applications. Nevertheless, the relatively soft nature of CrN still remains a problem. This paper reports experimental results on increases in hardness of sputtered PVD CrN coatings by means of additions of varying content of Ag or W. The resulting Cr-Ag-N and Cr-W-N coatings, grown on JIS SKH51 steel substrates, were characterized using SEM, EDS, WDS, XRD, micro-indentation hardness testing and scratch adhesion tests. Moreover, wear behavior was studied on two types of tribometer, employing different contact regimes — a ‘cylinder-on-disk’ line-contact reciprocating-sliding regime and a ‘ball-on-disk’ point-contact rotating-sliding regime.The experimental results can be summarized as follows: The hardness of Cr-W-N coatings increased with increasing W content; reversely, that of Cr-Ag-N coatings decreased with increasing Ag content. The additions of Ag and W resulted in a formation of secondary phases, elemental Ag and WN, respectively, uniformly embedded in the CrN coatings. With the two different types of tribometer, the observed trends for wear behavior, wear and friction coefficient, were nearly identical, indicating that both Ag and W additions to CrN coatings were beneficial. However, the Cr-W-N coatings were significantly more wear resistant than the Cr-Ag-N coatings. With the addition of W at 6.8 at.%, the largest wear improvement of 73%-85% was achieved.     

REACTIVE DIFFUSION BETWEEN Si AND NbC AT 1300℃

ＲＥＡＣＴＩＶＥＤＩＦＦＵＳＩＯＮＢＥＴＷＥＥＮＳｉＡＮＤＮｂＣＡＴ１３００℃Ｃ．Ｒ．Ｋａｏ（ＤｅｐａｒｔｍｅｎｔｏｆＣｈｅｍｉｃａｌＥｎｇｉｎｅｅｒｉｎｇＮａｔｉｏｎａｌＣｅｎｔｒａｌＵｎｉｖｅｒｓｉｔｙＣｈｕｎｇＬｉ，Ｔａｉｗａｎ）Ｊ．Ｗｏｏｄｆ...     

Phase equilibria of the cu-in system II: Thermodynamic assessment and calculation of phase diagram

The phases in the Cu-In binary were modelled thermodynamically using the Redlich-Kister expression for the Gibbs energies of the solution phases, the Wagner-Schottky model for those of the η (η)’)-Cu₂ln phase (taking η and η)’ to be a single phase), and assuming line compound behavior for the other intermetallic phases. The model parameters were obtained using primarily the thermodynamic data, as well as the phase equilibrium data. The thermodynamic values for the various phases calculated from the models are in reasonable agreement with the experimentally determined thermodynamic data that are available in the literature. The entropies of melting for the intermetallic phases obtained from the models are in accord with the values calculated from the empirical formulas suggested by Kubaschewski. The calculated phase diagram is also in reasonable agreement with the experimentally determined diagram, with the calculated temperatures for all the invariant equilibria within 1°C of the experimental values. The discrepancies between the calculated and experimental phase boundaries at the invariant temperatures are less than 1 at.% except those involving βCu₄Inn and γCu₇ln₃. These two phases were taken to be line compounds in the present study, although experimentally they exist over appreciable ranges of homogeneity. Current address: Dept. of Chemical Engineering, National Tsing Hua University, Taiwan.     

Enhancing the accuracy of WLAN-based location determination systems using predicted orientation information

Indoor location determination has emerged as a significant research topic due to the wide-spread deployment of wireless local area networks (WLANs) and the demand for context-aware services inside buildings. However, prediction accuracy remains a primary issue surrounding the practicality of WLAN-based location determination systems. This study proposes a novel scheme that utilizes mobile user orientation information to improve prediction accuracy. Theoretically, if the precise orientation of a user can be identified, then the location determination system can predict that user’s location with a high degree of accuracy by using the training data of this specific-orientation. In reality, a mobile user’s orientation can be estimated only by comparing variations in received signal strength; and nevertheless the predicted orientation may be incorrect. Incorrect orientation information causes the accuracy of the entire system to decrease. Therefore, this study presents an accumulated orientation strength algorithm which can utilize uncertain estimated orientation information to improve prediction accuracy. Implementation of this system is based on the Bayesian model, and the experimental results indeed show the effectiveness of our proposed approach.     

Effects of insulin therapy on non-insulin-dependent diabetics with secondary oral hypoglycemic agent failure

A recent trend among physicians is the categorisation of lung scans as normal [excludes pulmonary embolism (PE)], high probability (confirms PE) and non-diagnostic (no judgement on PE risk). The low probability scan is therefore being eliminated as a functional category. This occasional survey contends that such an approach is misguided. Correction of the original PIOPED data with certain assumptions provides a more reproducible, albeit restricted, low probability scan category which excludes PE in 97% of cases in the low pre-test clinical category. Patients with a low probability scan with risk factors for PE (i.e. medium clinical risk) will require further investigation. More important, the very low probability scan category excludes PE in 98% of patients with low and more than 92% of patients with medium pre-test clinical likelihood. The demise of "low probability" is premature.