Table of Content

    20 March 2018
    Volume 53 Issue 3
    User influence analysis of social media with temporal characteristics
    LIAO Xiang-wen, ZHANG Ling-ying, WEI Jing-jing, GUI Lin, CHENG Xue-qi, CHEN Guo-long
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  1-12.  doi:10.6040/j.issn.1671-9352.0.2017.371
    Abstract ( 598 )   PDF (2074KB) ( 324 )   Save
    References | Related Articles | Metrics
    Since both the temporal characteristics and online learning are not fully considered in exsiting tensor influence models, a novel method with temporal characteristics is proposed in this paper. This method constructs tensor with users opinion, activity and network centrality information. Then, a factorizes tensor with stochastic gradient descent algorithm which is constrained by temporal characteristics matrix is deployed in our model. Base on these two steps above, this method calculates user influence by combining different slices of tensor in the end. The advantages of this method are that it can decompose tensor efficiently and satisfy the need of online learning. Experimental results show that the average accuracy of the proposed method is 2% to 6% better than the baseline method such as TwitterRank, OOLAM and constrained nonnegative tensor factorization method. Besides, the running time of the proposed method is only 30% to 50% of constrained nonnegative tensor factorization method.
    Deep representative learning based sentiment analysis in the cross-lingual environment
    YU Chuan-ming, FENG Bo-lin, TIAN Xin, AN Lu
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  13-23.  doi:10.6040/j.issn.1671-9352.0.2017.064
    Abstract ( 793 )   PDF (1135KB) ( 345 )   Save
    References | Related Articles | Metrics
    Transfer learning focuses on solving the problem that it is difficult for supervised learning to obtain good classification results with the small training sets. Compared with the traditional supervised methods, it does not require the training and testing sets follow the same or similar data distributions. The model can be trained on the language which has rich data and labeling resources(source language), and the source and target language documents can be projected into the same feature space. In this way the large data from the source domain or language can be leveraged to solve the low performance problem in the target language. The reviews of three product categories, i.e. books, DVD and music, from Amazon, which are written in Chinese, English and Japanese, are collected as the experimental data. A novel model, i.e. the Cross Lingual Deep Representation Learning(CLDRL)is proposed and empirical study is conducted upon the experimental data. From the experimental results, it shows that the best performance of CLDRL achieves 78.59%, which prove the effectiveness of the proposed model.
    Community detection algorithm based on effective resistance of network
    ZHANG Jun, LI Jing-fei, ZHANG Rui, RUAN Xing-mao, ZHANG Shuo
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  24-29.  doi:10.6040/j.issn.1671-9352.1.2017.041
    Abstract ( 486 )   PDF (875KB) ( 228 )   Save
    References | Related Articles | Metrics
    Community detection has a number of important applications in many areas. Inspired by the principle of effective resistance in traditional electrical circuit, we propose a novel community detection algorithm which can discover communities by iteratively calling the edge-cutting selection model based on the total effective resistance of a network. In each iteration, the edge-cutting selection model cuts an appropriate number of edges in order to maximize the total effective resistance of the updated network. Theoretical analysis shows that our algorithm has relatively low algorithm complexity degree. Extensive empirical experiments have been conducted on simulated and real complex networks, the results show that the proposed algorithm has good performance.
    Fusion of pointwise and deep learning methods for passage ranking
    PANG Bo, LIU Yuan-chao
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  30-35.  doi:10.6040/j.issn.1671-9352.1.2017.012
    Abstract ( 922 )   PDF (914KB) ( 254 )   Save
    References | Related Articles | Metrics
    Intelligent question answering is an important way to make information acquisition more intelligent and convenient. Intelligent Q&A oriented passage ranking is very important for accurately grasping the user's query intention, improving the user experience and the accuracy of feedback. We use deep learning techniques to capture semantic information about query and passages, and build the mapping model to the tag. Then the training model is used to predict the correlation between new query and the passage. Finally, we use the predicted correlation index of the passages and the query to sort the multiple answers of the same question. The experimental results show that our method can reach 3.979 on DCG@3 and 5.396 for DCG@5.
    Text feature extraction method for sentiment analysis based on order-preserving submatrix and frequent sequential pattern mining
    CHEN Xin, XUE Yun, LU Xin, LI Wan-li, ZHAO Hong-ya, HU Xiao-hui
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  36-45.  doi:10.6040/j.issn.1671-9352.1.2017.093
    Abstract ( 790 )   PDF (676KB) ( 222 )   Save
    References | Related Articles | Metrics
    Feature extraction is one of the key steps in text sentiment analysis, which is also the main factor that affects the result. According to the variant expression of online review, the synonyms TF-IDF(term frequency-inverse document frequency)weight vector is obtained based on the semantic similarity. Then in view of the different length of online review, the local patterns of the feature vectors are identified with OPSM(order-preserving submatrix)biclustering algorithm. We improve PrefixSpan algorithm to detect the frequent classification phrase feature, which contain word order information. Furthermore some important factors, such as the separation of word, are also employed to improve the discriminative ability of sentiment orientation. Finally, the proposed method is applied to the sentiment analysis task experiment of the product reviews, and the results show that the text feature extraction has a better performance.
    Lexical and semantic relevance matching based neural document ranking
    ZHANFG Fang-fang, CAO Xing-chao
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  46-53.  doi:10.6040/j.issn.1671-9352.1.2017.001
    Abstract ( 820 )   PDF (758KB) ( 319 )   Save
    References | Related Articles | Metrics
    A deep neural network based on lexical correlation matching and semantic correlation matching is proposed, which can be used to calculate the matching score of a query and a document in the information retrieval task. The lexical relevance matching model is based upon the word co-occurrence matrix of a query and a document, which takes the word matching information into consideration, so as to consider the position information of the matching word. The semantic relevance matching model is grounded in pre-trained word vector, then the convolution network extracts the semantic matching information between a query and different positions of the documents, where the final matching score is the superposition of the two sub-models. Model parameters are updated in the training process by maximizing the fractional difference between positive and negative samples. Experimental results indicate that the NDCG@3 and NDCG@5 of the model can attain to 0.790 4 and 0.818 3 respectively on the validation set. which significantly outperforms the baselines, verifying the importance of word and semantic matching for information retrieval.
    Efficient multiple sets intersection using SIMD instructions
    SONG Xing-shen, YANG Yue-xiang, JIANG Yu
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  54-62.  doi:10.6040/j.issn.1671-9352.1.2017.040
    Abstract ( 500 )   PDF (3869KB) ( 190 )   Save
    References | Related Articles | Metrics
    Conjunctive Boolean query is one fundamental operation for document retrieval and widely used in many information systems and databases. In its most basic and popular form, a conjunctive query can be seen as the intersection problem of multiple sets of sorted integers, and how to improve its efficiency is becoming one important research highlight. Based on the traditional intersection algorithms, this paper proposes two optimizations on the essential searching algorithms using SIMD instructions. The optimized search algorithms are able to be adopted into various multiple sets intersection methods while improving intersection efficiency. Experiments show that the optimized algorithms performs much better than the traditional ones, even outperform the recent SIMD intersection algorithms,and the improvement is up to 37.3% at most.
    Steganalysis method based on shallow convolution neural network
    LIU Ming-ming, ZHANG Min-qing, LIU Jia, GAO Pei-xian
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  63-70.  doi:10.6040/j.issn.1671-9352.2.2017.294
    Abstract ( 752 )   PDF (2768KB) ( 264 )   Save
    References | Related Articles | Metrics
    In order to improve the detection rate of steganalysis, a method of image steganalysis based on shallow convolution neural network is proposed. Compared with the deep convolution neural network, the shallow convolution neural network can improve the convergence speed of the neural network and reduce the loss of the steganography feature by reducing the convolution layer and disabling the pool layer. At the same time, the generalization performance of the steganalysis network is improved by using batch normalization functions and using a single fully connected layer. The experimental results show that the detection accuracy can reach 96% and 81.7% respectively when the embedding rate is 0.4 bpp and 0.1 bpp. And the method is still maintain a better detection performance in the case of carrier source and embedding rate mismatch.
    Metric model for cloud computing security risk assessment
    RUAN Shu-hua, WENG Jun-hao, MAO Hui, CHEN Xue-lian
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  71-76.  doi:10.6040/j.issn.1671-9352.2.2017.380
    Abstract ( 1148 )   PDF (723KB) ( 293 )   Save
    References | Related Articles | Metrics
    From three aspects related to cloud computing of policies, management and technologies, an indicator system of cloud security risk assessment is established for the security risk assessment problem in cloud computing environment. A metric model of security risk in cloud computing environment is established by fusing Delphi method, fuzzy analytical hierarchy process and fuzzy comprehensive evaluation method. Measurement results of risk instances show that the metric model could provide effective quantitative evaluation for the security risk assessment in cloud computing environment.
    Finite element simulation of relaxation properties on lumbarintervertebal disc under compression
    LUAN Yi-chao, YANG Xiu-ping, ZHANG Jing-jing, LIU Qing, ZHANG Chun-qiu
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  77-81.  doi:10.6040/j.issn.1671-9352.0.2017.264
    Abstract ( 564 )   PDF (2865KB) ( 193 )   Save
    References | Related Articles | Metrics
    In order to study the relaxation properties of intervertebal disc, a poroelastic finite element model of the human spine L3/L4 segments was developed by ABAQUS with considering the nonlinear porous elastic properties, and the relaxation properties with different permeability and under different strain conditions were calculated. The results show that nonlinear permeability rises the stiffness of the segments and increases the stress, the stress relaxation curve presents the index under compression strain. And the stress increases with the rise of strain. The pore pressure and the effective stress of the annulus fibrosus(AF)are higher than those of the nucleus pulposus. The pore pressure and the effective stress of the posterior AF are greater than those of the anterior AF, so the lumbar disc herniation occurs mostly on the posterior AF.
    Silk fibroin-type Ⅱ collagen cartilage scaffold fabricated by 3D printing technology
    YUAN Qing-xian, GAO Li-lan, LI Rui-xin, LIU Ying-jie, LIN Xiang-long, ZHANG Xi-zheng
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  82-87.  doi:10.6040/j.issn.1671-9352.0.2017.501
    Abstract ( 770 )   PDF (4225KB) ( 276 )   Save
    References | Related Articles | Metrics
    The macroscopic structure of cartilage scaffold was designed by using Solidworks, and the silk fibroin-type Ⅱ collagen cartilage scaffold was prepared by 3D printing technique and freeze-drying technique. The density, porosity and elastic modulus of the scaffolds were tested by experiments. The proliferation of the cells was analyzed by MTT assay, HE staining and scanning electron microscopy. The results show that the silk fibroin-type II collagen scaffold is dependent on the strain rate. The elastic modulus of scaffold increases with the increase of strain rate. The density and porosity of scaffold were(0.086 6±0.008 4)g/cm3 and(89.3±3.26)%, respectively. The cell growth and proliferation were accelerated after 7 days of inoculation. By analyzing the results of HE staining, it is found that the cells grow most in the surface area and there are the least cells in the deep region. The microscopic images by Scanning electron microscopy(SEM)reveal that the diameter of scaffold is regular and the permeability is better. The cells are mostly distributed on the surface of the spine.
    Catalytic amination of sym-triazinetriol with diethylenetriamine
    MENG Li, WANG Qian, CHAI Shu, ZHU Wei-qun
    JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(3):  88-94.  doi:10.6040/j.issn.1671-9352.0.2017.377
    Abstract ( 632 )   PDF (1249KB) ( 205 )   Save
    References | Related Articles | Metrics
    As a carbon resource, it has become one of the hot spots at present through the chemical utilization of CO2 to achieve emission reduction purposes. Sym-triazinetriol is kind of CO2 sequestration products, and it has broad prospects to develop its downstream products. In this paper, synthesis of 6-divinyl three amine-1,3,5-triazine base-2,4(1H,3H)-diketone by diethylenetriamine(DETA)and Sym-triazinetriol(CA)was studied. The effects of molar ratio, reaction time, reaction temperature and catalyst on the yields were investigated by single-factor method. Then, the product was characterized by IR, 1H NMR and 13C NMR. The optimum conditions were as follows: the molar ratio of CA to DETA was 1∶4, the reaction temperature was 210 ℃, the reaction time was 4 h, the catalyst was 5A molecular sieve. The yield of product can reach 70.9%.