Entity

Time filter

Source Type


Chen D.,Chinese Academy of Agricultural Sciences | Wang F.,Chinese Academy of Agricultural Sciences | Qian Y.,Chinese Academy of Agricultural Sciences | Luo C.,Chinese Academy of Agricultural Sciences | And 7 more authors.
Acta Tabacaria Sinica | Year: 2014

Flue-cured tobacco variety Zhongyan204 was developed by crossing female parent tobacco strain 88 which is resistant to multiple diseases and male parent varietyK326 which has moderate resistance to tobacco bacterial wilt but susceptible to brown spot, viral disease and weather fleck. Results showed that the new variety had immunity to TMV and strong tolerance to CMV, PVY. It was resistant to black shank, weather fleck, and sub infection to Angular spot. Cured leaf was rich in oil and loose in structure and displayed well-distributed orange color. Chemical composition was favorable and well proportioned to meet cigarette production requirements. Compared with NC89, the new variety was better in major economic characters. In conclusion, Zhongyan204 was a new hybrids featuring good productivity, quality, economic performance and strong resistance to diseases. Source


Ren Z.,Shandong Agricultural University | Li X.,Shandong Tobacco Research Institute | Xue X.,Shandong Agricultural University | Li C.,Shandong Agricultural University
Acta Tabacaria Sinica | Year: 2015

Multivariate ordinal Logit regression model based on survey data of 542 tobacco family farms in Weifang was employed to analyze factors influencing development of tobacco family farm in terms of economic benefit, social benefit and ecological benefit. Results showed that: management capability, fund adequacy, external support and cooperative service ability were among the most important factors. It is suggested that training for managerial staff need to be strengthened so as to improve basic operating conditions, reduce cost, enhance efficiency, and beef up external support to the development of tobacco family farms. ©, 2015, State Tobacco Monopoly Bureau and China Tobacco Society. All right reserved. Source


Song N.,Shandong Tobacco Research Institute | Hou J.,JiNan Shandong Tobacco Company Ltd | Liu P.,Shandong Tobacco Research Institute | Han H.,Shandong University of Finance and Economics | And 2 more authors.
Proceedings - 2015 International Conference on Intelligent Transportation, Big Data and Smart City, ICITBS 2015 | Year: 2015

This paper aims to solve the intelligent and personalized tobacco brand recommendation problem, which greatly affects the sales performance of tobacco enterprises. Firstly, we discuss how to mine the internal correlations between different users to compute user similarity. Particularly, we estimate user similarity by constructing user feature vectors using Cosine distance. Secondly, a novel intelligent and personalized tobacco brand recommendation algorithm is given, and the top ranked tobacco brands are output as the tobacco brand recommendation results. Finally, experiments test the effectiveness of the proposed algorithm by two main aspects, and positive results are achieved. © 2016 IEEE. Source


Gong H.,Ocean University of China | Xu X.,Ocean University of China | Song N.,Shandong Tobacco Research Institute | Liu P.,Shandong Tobacco Research Institute
Journal of Computational Information Systems | Year: 2014

A new novel projection algorithm (PCP) is proposed to discriminate the subtle difference of high dimensional data in similar data. The identification analysis of similar spectra is done by the proposed method based on principal component analysis (PCA) and Parallel coordinates. Through data feature extracting by PCA and data dimension filtering by parallel coordinates method, high-dimensional information of the similar samples will be compressed in the low-dimensional space and analyzed with vivid and clear expression. The results of K-Nearest Neighbor classification after the similar spectrum analysis of samples show that the proposed method not only has an equivalently high classification performance to subtle difference of multiple samples and keeps a good visualization in high dimensional data, but also gives the higher accuracy in the classification than the conventional PCA method. © 2014 Binary Information Press. Source


Cheng D.,Ocean University of China | Ding X.,Ocean University of China | Song N.,Shandong Tobacco Research Institute | Liu P.,Shandong Tobacco Research Institute
WIT Transactions on Information and Communication Technologies | Year: 2014

As multi-view face detection is an important problem in computer vision research field, in this paper, we propose a novel multi-view face detection based on a modified multi-class SVM, which can solve the limitations in standard SVM. Firstly, the structure of the SVM classifiers for multi-view face detection is present, of which the face detection problem is divided into three categories (“Front view”, “Left view”, and “Right view”). Secondly, the left projecting vector and the right projecting vector are defined, and the the constraint conditions of multi-class are illustrated as well. Furthermore, the multi-class classification problem is converted to finding the best solution from all feasible solutions. Thirdly, visual features of images are extracted to construct the image vector space, and then the testing images can be classified by the multi-class SVM to make multi-view human face detection. Finally, to make performance evaluation for the proposed algorithm, experiments are conducted using some standard human face datasets, and the conclusions can be drawn that the proposed is quite effective. © 2013 WIT Press. Source

Discover hidden collaborations