Table of Content

  

20 September 2018
Volume 53 Issue 9

Anomaly detection model of host group based on graph-evolution events

YE Xiao-ming, CHEN Xing-shu, YANG Li, WANG Wen-xian, ZHU Yi, SHAO Guo-lin, LIANG Gang

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  1-11.  doi:10.6040/j.issn.1671-9352.2.2017.169

Abstract ( 2305 )   PDF (5937KB) ( 1444 )   Save

References | Related Articles | Metrics

Aiming at the communication behavior based on service aggregation and the new collaborative attack mode that is typical of distributed attack in the network environment, the anomaly detection model of host group based on graph-evolution events is proposed. It analyzes the potential socialization of actors, the clustering of host clusters and the dynamics of their group behavior. The model has the characteristics of no parameters and extensible data magnitude. The dynamic evolution events and detection algorithms are defined and proposed to detect abnormal host groups. The model is implemented and deployed on Spark, and the data from the actual computer and network environment is analyzed and verified. The experimental results show that this model can effectively describe group behavior, expose important graph-evolution events, and locate the host group with abnormal occurrence accurately. The detection rate of group members is 95.09%.

Differential privacy partitioning algorithm based on adaptive density grids

YAN Yan, HAO Xiao-hong

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  12-22.  doi:10.6040/j.issn.1671-9352.0.2017.418

Abstract ( 2326 )   PDF (1990KB) ( 643 )   Save

References | Related Articles | Metrics

In order to balance the influence of noise error and uniform hypothesis error for the two-dimensional partitioning publishing, a new hierarchical differential privacy partitioning algorithm DP-ADG is proposed. Firstly, the position space is clustered to form the density adaptive grids in the first layer. Then in the second layer, different partitioning methods are adopted for different density blocks. The noise error introduced by a large number of null nodes is avoided while reducing the uniform hypothesis error. While using the hierarchical partitioning strategy, different Laplace noise of different privacy budgets is added to the results of two phases according to the sequential composition of differential privacy, in order to realize the overall ε differential privacy protection for the publishing data. Experimental results show that the algorithm has good effect on improving the accuracy of range counting query, saving unnecessary spatial decomposition process, as well as improving the efficiency of the algorithm.

Dynamic discovery of authors research interest based on the combined topic evolutional model

YU Chuan-ming, ZUO Yu-heng, GUO Ya-jing, AN Lu

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  23-34.  doi:10.6040/j.issn.1671-9352.1.2017.044

Abstract ( 2448 )   PDF (4162KB) ( 859 )   Save

References | Related Articles | Metrics

We propose a new combined topic model, i.e. author topic time-latent dirichlet allocation(ATT-LDA)with author ranking(AR), for the of dynamic discovery of researchers' interest, which is based on the academic literature in the financial field. Through the proposed model, we can easily acquire the probability distribution of the authors' interest, as well as the probability distribution of topics on deferent words. The influence of the ranking in the co-author list are fully taken into consideration. The empirical study shows that the proposed method can effectively reveal the dynamic change of interest of the authors in the financial field.

Reader emotion classification with news and comments

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  35-39.  doi:10.6040/j.issn.1671-9352.1.2017.003

Abstract ( 2253 )   PDF (573KB) ( 672 )   Save

References | Related Articles | Metrics

The news and comments are important resources to classify the reader emotion. However, previous studies only used news texts or mixed two types of texts as a general feature, which did not make the best use of the differences and connections between different textual features. Based on it, the paper proposed a new approach named dual-channel LSTM, which treated two types of texts as different features. First, the approach learned a LSTM representation with a LSTM recurrent neural network. Then, it proposed a joint learning method to learn the relationship between the features. Empirical studies demonstrate the effectiveness of the proposed approach to reader emotion classification.

Extraction of Chinese multiword expressions based on Web text

GONG Shuang-shuang, CHEN Yu-feng, XU Jin-an, ZHANG Yu-jie

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  40-48.  doi:10.6040/j.issn.1671-9352.1.2017.060

Abstract ( 2493 )   PDF (645KB) ( 689 )   Save

References | Related Articles | Metrics

A Multiword Expression is a kind of fixed and semi-fixed collocation in natural language, especially in network text, MWEs appear frequently, which brings a great challenge to the subsequent segmentation and text comprehension. Therefore, we propose a double-layer extraction strategy to achieve the recognition of MWEs in this paper. In the first layer, we use the LRE+EMI algorithm to achieve the initial extraction of MWEs; In the second layer, we use SVM classifier and construct the characteristics of context and word vector to classify the MWEs and non-MWEs, in order to further filter the MWEs candidate list on the basis of the MWEs candidate list got from the first layer. After the experiment, the F value of MWEs reached 84.92% in the first layer and the F value of MWEs reached 89.58% in the second layer, which have greatly improved performance compared with the baseline system. The experimental result shows that the double-layer extraction strategy can availably extract MWEs, and can effectively improve the segmentation results.

Design and implementation of topic detection in Russian news based on ontology

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  49-54.  doi:10.6040/j.issn.1671-9352.0.2017.650

Abstract ( 2317 )   PDF (661KB) ( 814 )   Save

References | Related Articles | Metrics

Aiming at the problem of topic detection in Russian news, using automatic morphological analysis and named entity recognition as the auxiliary means, a method for describing Russian news elements and calculating their similarities based on ontology was designed. The Single-pass algorithm was used to carry out text clustering experiments for topic detection. By comparing the results of vector space model(VSM)model and ontology model, it is proved that the latter has relatively high accuracy and validity.

Design and approximation of SISO three layers feedforward neural network based on Bernstein polynomials

XIAO Wei-ming, WANG Gui-jun

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  55-61.  doi:10.6040/j.issn.1671-9352.0.2017.606

Abstract ( 2240 )   PDF (627KB) ( 659 )   Save

References | Related Articles | Metrics

A single input single output(SISO)three layers feedforward neural network was designed by using the difference value between adjacent equidistant subdivision points of unary Bernstein polynomial with a Sigmodial transfer function, and a method of selecting the connection weights and thresholds was given. In addition, according to the approximation theorem for unary Bernstein polynomial, we proved that SISO three layers feedforward neural network could also approximate a continuous function. The analytical expression of the neural network was obtained by an example.

Quantum secret sharing scheme realizing all hyperstar quantum access structure

JIAO Hong-ru, QIN Jing

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  62-68.  doi:10.6040/j.issn.1671-9352.0.2017.569

Abstract ( 1933 )   PDF (423KB) ( 474 )   Save

References | Related Articles | Metrics

To realise more quantum access structure, we put forward a new quantum secret sharing scheme by sharing non-objected secret sharing states to some authorised subsets in the hyperstar quantum access structure. Compared with other similar sechems, all partners in our scheme obtain quantum share and our scheme possess unconditional security.

Equilibrium decisions of a two-layer supply chain network considering retailers horizontal fairness

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  69-82.  doi:10.6040/j.issn.1671-9352.0.2018.033

Abstract ( 2183 )   PDF (904KB) ( 821 )   Save

References | Related Articles | Metrics

We extended the vertical fairness research on a supply chain with one manufacturer and one retailer, and considered the horizontal fairness preference in a two-layer supply chain network consisted by an upper tier for a single manufacturer and a lower tier for multiple retailers and demand markets. The retailer chooses his decision based on the horizontal fairness preference comparing the profit of other retailers. Considering retailers horizontal fairness preference, the Nash game model of the retailers and the Stackelberg-Nash game model of the two-layer supply chain network were constructed. The optimal decisions of a two-layer supply chain network were obtained by the penalty function algorithm. The numerical results show that the horizontal fairness preference among retailers lead to different changes in the decisions of the supply chain members, and the maximum profit of manufacturer and the maximum utility of retailers are lower than those of neutral retailers. Finally, some advices are provided to the supply chain members to deal with the horizontal fairness preference.

Research on supply chain decision making of equitable retailers with fair sensitivities

ZHANG Ke-yong, LI Jiang-xin, YAO Jian-ming, LI Chun-xia

JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE). 2018, 53(9):  83-94.  doi:10.6040/j.issn.1671-9352.0.2018.032

Abstract ( 2357 )   PDF (1177KB) ( 624 )   Save

References | Related Articles | Metrics

In order to study the influence of heterogeneity of subject of fair preference on the decision of supply chain, fair sensitivity was introduced to the decision model of fair preference decision-maker. We build a secondary supply chain consisting of a manufacturer, two competing retailers, one of which has a fair preference and fair sensitivity. Based on the social comparison theory, a supply chain decision model with horizontal equity preference and vertical equity preference was established, and the influence of fair sensitivity coefficient on decision variables and utility of decision makers was obtained. Finally, numerical examples were used to verify the conclusion. The results show that the existence of fair preference and fair sensitivity does not increase the utility value of retailer itself, and it increases the competition between the two retailers under the horizontal equity preference. In the adverse unfair situation, the utility of the retailer increases with the increase of the fair sensitivity coefficient, and decreases with the increase of the fair sensitivity coefficient in the favorable unfair situation. Therefore, in the case of unfair adverse circumstances, the retailer should remain humble inferiority, and should not overestimate his own abilities and status. But in the case of favorable unfair circumstances, the retailer should be confident. Only in this way can he keep the utility maximization.