Search

PurposeSocial networking platforms are increasingly using the Follower Link Prediction tool in an effort to expand the number of their users. It facilitates the discovery of previously unidentified individuals and can be employed to determine the relationships among the nodes in a social network. On the other hand, social site firms use follower–followee link prediction (FFLP) to increase their user base. FFLP can help identify unfamiliar people and determine node-to-node links in a social network. Choosing the appropriate person to follow becomes crucial as the number of users increases. A hybrid model employing the Ensemble Learning algorithm for FFLP (HMELA) is proposed to advise the formation of new follower links in large networks.Design/methodology/approachHMELA includes fundamental classification techniques for treating link prediction as a binary classification problem. The data sets are represented using a variety of machine-learning-friendly hybrid graph features. The HMELA is evaluated using six real-world social network data sets.FindingsThe first set of experiments used exploratory data analysis on a di-graph to produce a balanced matrix. The second set of experiments compared the benchmark and hybrid features on data sets. This was followed by using benchmark classifiers and ensemble learning methods. The experiments show that the proposed (HMELA) method predicts missing links better than other methods.Practical implicationsA hybrid suggested model for link prediction is proposed in this paper. The suggested HMELA model makes use of AUC scores to predict new future links. The proposed approach facilitates comprehension and insight into the domain of link prediction. This work is almost entirely aimed at academics, practitioners, and those involved in the field of social networks, etc. Also, the model is quite effective in the field of product recommendation and in recommending a new friend and user on social networks.Originality/valueThe outcome on six benchmark data sets revealed that when the HMELA strategy had been applied to all of the selected data sets, the area under the curve (AUC) scores were greater than when individual techniques were applied to the same data sets. Using the HMELA technique, the maximum AUC score in the Facebook data set has been increased by 10.3 per cent from 0.8449 to 0.9479. There has also been an 8.53 per cent increase in the accuracy of the Net Science, Karate Club and USAir databases. As a result, the HMELA strategy outperforms every other strategy tested in the study.

Active learning approach is well known method for labeling huge un-annotated dataset requiring minimal effort and is conducted in a cost efficient way. This approach selects and adds most informative instances to the training set iteratively such that the performance of learner improves with each iteration. Named entity recognition (NER) is a key task for information extraction in which entities present in sequences are labeled with correct class. The traditional query sampling strategies for the active learning only considers the final probability value of the model to select the most informative instances. In this paper, we have proposed a new active learning algorithm based on the hybrid query sampling strategy which also considers the sentence similarity along with the final probability value of the model and compared them with four other well known pool based uncertainty query sampling strategies based active learning approaches for named entity recognition (NER) i.e. least confident sampling, margin of confidence sampling, ratio of confidence sampling and entropy query sampling strategies. The experiments have been performed over three different biomedical NER datasets of different domains and a Spanish language NER dataset. We found that all the above approaches are able to reach to the performance of supervised learning based approach with much less annotated data requirement for training in comparison to that of supervised approach. The proposed active learning algorithm performs well and further reduces the annotation cost in comparison to the other sampling strategies based active algorithm in most of the cases.

Modern society is prominently dependent on information and communication technology over the last few decades has carried it with more vulnerability to an enormous variety of cyber-attacks. One of ...

Purpose The purpose of this paper is to design a sustainable supply chain network (SCN) for omnichannel environment in order to provide better service to customers through flexible distribution. Thus, there is a need to incorporate multiple-channel distribution in the network design of supply chains (SCs). Design/methodology/approach A multiple-channel distribution supply chain network (MCDSCN) has been proposed under omnichannel environment. This proposed model integrates online giants with local retailer’s distribution network in an uncertain environment with sustainability. To incorporate sustainability, an objective function is added to reduce carbon content along with other objectives of minimization of SC cost. The model turns out to be mixed-integer linear programming model which is coded in GAMS and solved using CPLEX solver. Findings The proposed MCDSCN model is compared with conventional SCN. Furthermore, it was found that the proposed MCDSCN model has achieved significant saving in SC cost and is also more sustainable than conventional SCN. The proposed model also enables online giants to integrate their distribution network with local retailer’s distribution network. Practical implications Through proposed model, customers are free to access product and services as per their choice of channels which increases their convenience, reach and satisfaction. Originality/value The proposed MCDSCN model is a novel approach to design flexible distribution systems. This would significantly help organizations to design their distribution network more effectively to meet global competition.

The Internet of Things (IoT) is getting the reputation as one of the most optimistic networking paradigms that is reducing the gap between the cyber world and physical world. Most of the participating nodes in IoT network are sensors, which are limited in terms of resource such as energy, computation power, memory and so on. In IoT network, nodes communicate with each other via wireless medium, which makes the IoT network vulnerable to many security threats including eavesdropping. The IoT network is deployed in a large area and work on 24/7 hours, so an energy efficient scheme is one of the major issues in IoT. To achieve secure and energy efficient network, this article contributes: (1) A hierarchical topology for IoT network deployment; (2) symmetric matrix-based pair-wise key generation for secure communication; (3) A secure and energy efficient secure routing algorithm for the proposed model.

With the arrival of the internet of things and advancement in every front of technology, the field of the smart transportation system is becoming prevailing among the researchers. To avoid the problems in existing transportation systems and to provide better services and information to its users, smart transportation system can be used. This paper incorporates the wireless sensor network (WSN) technology with the smart transportation system. WSNs play a vital role in monitoring and gathering information from the environment. To provide better coverage and connectivity with least message overhead, this paper uses a combination of k-means and harmony search algorithm. The proposed approach is compared on the basis of parameters like message overhead, coverage, throughput and end-to-end delay. The simulation results show that the proposed work is efficient, provides good coverage and reduces message overhead.

The primary motive of this survey is to impart complete analysis of social connectivity issues using concepts of big data analysis. Issues related to social networking include community detection, cluster analysis, personalized search, anomaly detection, searching cut space, friendship selection, load balancing, structural balancing, etc. These issues can be improvised using machine learning techniques including particle swarm optimization, evolutionary algorithms, genetic algorithm, properties of graphs such as embeddedness and triadic closure. These properties and algorithms enhance the behavior of social networking issues when applied to datasets of various social networking platforms.

Different machine learning techniques require large amount of labeled training data to work well. It is very costly and difficult to label the unlabelled data. The active learning approach can be used in this case as it selects the most appropriate data from the unlabeled data pool. In this work, we have applied active learning using CRF classifier over biomedical disease names and adverse effect corpus using margin sampling strategy. Experiment have been carried for selecting the most uncertain samples for labeling by defining different threshold values for the margin used to measure the uncertainty. The active learning approach performs well with much less data requirement and its performance increases with the increase in threshold values of the margin.

In this paper, we propose a hierarchical framework for mobile ad hoc IoT applications such as animal monitoring system, which supports the scalable network and uniform energy consumptions across all the nodes to prolong the network lifetime. The proposed framework uses the clustering algorithm to divide the network into equal zones. Each node belongs to some particular zone. Nodes are mobile (based on Random Waypoint Model) and able to move from one zone to another zone in the entire network. Some nodes select as zone head, zone coordinator, and relay node. Node forwards the data packets to the base station via a nearest relay node. Relay node forwards the packets to the zone head, and zone head transmits these packets to base stations via upper layer zone coordinator. The comparison has been made of the proposed framework with static model, LEACH and EESAA routing protocols. It identified that the proposed technique performs well in terms of energy consumption and network lifetime.

In this paper, we propose a network framework for dynamic IoT applications such as smart healthcare system, animal monitoring system, which supports the core IoT feature: scalability. The proposed framework uses hierarchical topology for node deployment. To support uniform energy consumptions, the network is divided into equal size zone so that network load can be distributed equally and network lifetime can be increased. To make this framework more realistic as animal monitoring and smart healthcare system, Random Waypoint Model is used for node mobility under scheme 1 and scheme 2. In scheme 1, nodes can move in entire network and in scheme 2 nodes can move in their respective zones only. To avoid long distance communication, some nodes are promoted as relay node, zone head and zone coordinator. Sensor nodes begin the communication via nearest relay node. Relay node forward the packets to the zone head and zone head transmits these packets to base stations via upper layer zone coordinator. In the RN selection, we consider the ratio of node to zone head (ZH) distance and their residual energy. The node with shortest ratio is selected as RN. The comparison has been made of the proposed framework with static model, mobile under scheme 1 and 2.

The field of wireless sensor networks has witnessed an increasing traction among researchers in the erstwhile few decades. Since in large networks, sensors are not uniformly distributed; thus, a problem arises in low- and medium-density part of that large network which has a high probability of the packets getting stuck at a specific node due to the problem of local minima. Thus, this paper tries to solve the problem of unsuccessful routing in low–medium-density networks by implementing updated form of azimuthal routing protocol which removes the overhead of keeping the routing details and also ensures that the packets reach the sink and is not affected by the presence of holes which is normally the case in lower density network. This protocol has been compared with greedy protocol which shows that it exhibits excellent performance with around 98.78% of average successful ratio compared to 45.38% in case of greedy.

The Boxplot is one of exploratory tools of data mining. It was originated by Tukey in 1970 [1]. The Boxplot uses five-points to display information which includes the median, the quartiles and the first and the last points. To improve on the information conveyed by existing Boxplots it is proposed to add the clustering methods during the computation of Boxplots. The clustering methods used are K-Means and DBSCAN. The results are plotted using all the above three techniques of constructing the Boxplots. The results obtained from the proposed methods of construction of Boxplots show that they are different from the original Boxplots. The results are compared with the traditional technique of generating boxplots and it is demonstrated that cluster driven boxplot generation is more effective for data analysis in real life scenario.

Particle swarm optimization (PSO) is a stochastic swarm-based algorithm inspired by the intelligent collective behavior of some animals. There are very few parameters to adjust in PSO which makes PSO easy to implement. One of the important parameter is inertia weight (ω) which balances the exploration and exploitation properties of PSO in a search space. In this paper, a new variation of PSO has been proposed, which utilizes a novel adaptive inertia weight strategy based on the binomial probability distribution for global optimization. This new technique improves final accuracy and the convergence speed of PSO with better performance. This new strategy has been tested against a set of ten benchmark functions and compared with four other PSO variants. The result shows that this new strategy is better and very competitive in most of the cases than other PSO variants.

The Boxplot uses five point summary of data to display information. K-Means and DBSCAN clustering methods are used to improve the Boxplots. Both the above methods have parameters which affect the clusters formed by the algorithms. The results obtained by varying these parameters i.e. $K$ in K-Means and Eps and MinPts in DBSCAN clustering methods are studied in this paper. The Boxplots obtained by varying the above mentioned parameters are plotted and the effects of varying the parameters have been studied. A comparative figure of traditional Boxplot and Boxplots generated using K-Means and DBSCAN clustering has also been plotted. The comparisons and differences have been discussed between the above mentioned Boxplots.

This paper proposes a technique for improving the convergence speed and the final accuracy of the Particle Swarm Optimization (PSO) by introducing a new adaptive inertia weight strategy based on Euclidean distance. This change does not inflict any major modifications to the basic algorithm. The proposed technique has shown significantly better performance as compared to other PSO variants on a test suite of ten optimization test functions evaluated on following performance metrics: time to locate the solution, scalability, quality of the final solution, and frequency of hitting the optima.

With the popularity of the Internet of things (IoT), an enormous number of sensors and smart object, which is heterogeneous in nature deployed into a network, send/receive data through the internet on the web. It is tedious to secure the communicated data and the device identity otherwise it cannot be adopted by their natives. But, existing security techniques that are very popular on the Internet are too complex to integrate on small, constrained objects. While initiating security through some protocol and techniques, the life of battery operated sensor can be reduced, so there is need to provide a proper balance between security and energy. This paper presents a survey of existing security protocols applied on IoT, their comparison and possible solution for various attacks.