Link Prediction Approach Research Articles

MODEL: Motif-Based Deep Feature Learning for Link Prediction

Link prediction plays an important role in network analysis and applications. Recently, approaches for link prediction have evolved from traditional similarity-based algorithms into embedding-based algorithms. However, most existing approaches fail to exploit the fact that real-world networks are different from random networks. In particular, real-world networks are known to contain motifs, natural network building blocks reflecting the underlying network-generating processes. In this article, we propose a novel embedding algorithm that incorporates network motifs to capture higher order structures in the network. To evaluate its effectiveness for link prediction, experiments were conducted on three types of networks: social networks, biological networks, and academic networks. The results demonstrate that our algorithm outperforms both the traditional similarity-based algorithms (by 20%) and the state-of-the-art embedding-based algorithms (by 19%).

A Novel Link Prediction Method for Opportunistic Networks Based on Random Walk and a Deep Belief Network

Link prediction is to estimate the possibility of future links among nodes by utilizing known information such as network topology and node attributes. According to the characteristics of opportunistic networks (topological time-variation, node mobility and intermittent connections), this paper proposes a novel link prediction approach (IRWR-DBN) for opportunistic networks that is based on random walk and a deep belief network. First, we reconstruct the Markov probability transition matrix and define a similarity index- improved random walk with restart (IRWR). Second, we divide the opportunistic network into network snapshots. Then, the similarity matrix of each snapshot is calculated by using the IRWR index to construct a sample set. Finally, a predictive model is constructed based on a deep belief network which extracts the time-domain characteristics in the process of dynamic evolution of the opportunistic network. The experimental results on the ITC and MIT Reality datasets show that compared with methods, such as the similarity-based index (CN, AA, Katz, RA, RWR), convolutional neural network, and recurrent neural network, the proposed method is more accurate and stable.

DTIP: A Comparative Analytical Framework for Chemogenomic Drugtarget Interactions Prediction.

Prediction of drug-target interactions is an essential step in drug discovery. Given drug-target interactions network, the objective of this task is to predict probable missing edges from known interactions. Computationally predicting drug-target interactions is an appropriate alternative for the time-consuming and costly experimental process of drug-target interaction prediction. A large number of computational methods for solving this problem have been proposed in recent years. In recent years, several review articles have been published in the field of drug-target interactions prediction. Compared to other review articles, this paper includes a qualitative analysis in the form of a framework, a drug-target interactions prediction (DTIP) framework. The framework consists of three sections. Initially, a classification has been presented for drug-target interactions prediction methods based on the link prediction approaches used in these methods. Secondly, general evaluation criteria have been introduced for analyzing approaches. Finally, a qualitative comparison is made between each approach in terms of their advantages and disadvantages. By providing a new classification of the drug-target interactions prediction approaches and comparing them with the proposed evaluation criteria, this framework provides a convenient and efficient way to select and compare the methods. Moreover, using the framework, we can improve these techniques further. This paper provides a study to select, compare, and improve chemogenomic drugtarget interactions prediction methods. To this aim, an analytical framework is presented.

DeepLink: A novel link prediction framework based on deep learning

Recently, link prediction has attracted more attention from various disciplines such as computer science, bioinformatics and economics. In link prediction, numerous information such as network topology, profile information and user-generated contents are considered to discover missing links between nodes. Whereas numerous previous researches had focused on the structural features of the networks for link prediction, recent studies have shown more interest in profile and content information, too. So, some of these researches combine structural and content information. However, some issues such as scalability and feature engineering need to be investigated to solve a few remaining problems. Moreover, most of the previous researches are presented only for undirected and unweighted networks. In this article, a novel link prediction framework named ‘DeepLink’ is presented, which is based on deep learning techniques. While deep learning has the advantage of extracting automatically the best features for link prediction, many other link prediction algorithms need manual feature engineering. Moreover, in the proposed framework, both structural and content information are employed. The framework is capable of using different structural feature vectors that are prepared by various link prediction methods. It learns all proximity orders that are presented on a network during the structural feature learning. We have evaluated the effectiveness of DeepLink on two real social network datasets, Telegram and irBlogs. On both datasets, the proposed framework outperforms several other structural and hybrid approaches for link prediction.

A Similarity-Inclusive Link Prediction Based Recommender System Approach

Despite being a challenging research field with many unresolved problems, recommender systems are getting more popular in recent years. These systems rely on the personal preferences of users on items given in the form of ratings and return the preferable items based on choices of like-minded users. In this study, a graph-based recommender system using link prediction techniques incorporating similarity metrics is proposed. A graph-based recommender system that has ratings of users on items can be represented as a bipartite graph, where vertices correspond to users and items and edges to ratings. Recommendation generation in a bipartite graph is a link prediction problem. In current literature, modified link prediction approaches are used to distinguish between fundamental relational dualities of like vs. dislike and similar vs. dissimilar. However, the similarity relationship between users/items is mostly disregarded in the complex domain. The proposed model utilizes user-user and item-item cosine similarity value with the relational dualities in order to improve coverage and hits rate of the system by carefully incorporating similarities. On the standard MovieLens Hetrec and MovieLens datasets, the proposed similarity-inclusive link prediction method performed empirically well compared to other methods operating in the complex domain. The experimental results show that the proposed recommender system can be a plausible alternative to overcome the deficiencies in recommender systems.

A hotel recommendation system based on customer location: a link prediction approach

Hotel recommendation is one of the most used application areas in recommendation systems. So far, many hotel recommendation systems have been proposed. Most of these systems are based collaborative filtering, content-based filtering, and association rule methods and employ the features of hotel, the ratings given by user, online reviews and comments in social network about the related hotel as data. However, due to the difficulty of processing the data used, the performance rates and speeds of these methods are relatively slow. As a solution to these problems in this paper, we propose a novel hotel recommendation system based on link prediction method. For this purpose, a customer-hotel bipartite network was first constructed and the relationship information in this network was used as data. Then, a supervised link prediction method that consider customers’ location was presented. To the best of our knowledge, this is the first study that recommends hotel by using link prediction method. The experimental results conducted on data crawled from TripAdvisor.com demonstrate that the proposed method captures an accuracy of 89.5% and outperforms the other recent related algorithms.

Interpolative self-training approach for link prediction

Interpolative self-training approach for link prediction

Link prediction in paper citation network to construct paper correlation graph

Nowadays, recommender system has become one of the main tools to search for users’ interested papers. Since one paper often contains only a part of keywords that a user is interested in, recommender system returns a set of papers that satisfy the user’s need of keywords. Besides, to satisfy the users’ requirements of further research on a certain domain, the recommended papers must be correlated. However, each paper of an existing paper citation network hardly has cited relationships with others, so the correlated links among papers are very sparse. In addition, while a mass of research approaches have been put forward in terms of link prediction to address the network sparsity problems, these approaches have no relationship with the effect of self-citations and the potential correlations among papers (i.e., these correlated relationships are not included in the paper citation network as their published time is close). Therefore, we propose a link prediction approach that combines time, keywords, and authors’ information and optimizes the existing paper citation network. Finally, a number of experiments are performed on the real-world Hep-Th datasets. The experimental results demonstrate the feasibility of our proposal and achieve good performance.

A node representation learning approach for link prediction in social networks using game theory and K-core decomposition

The role of social networks in people’s daily life is undeniable. Link prediction is one of the most important tasks of complex network analysis. Predicting links is currently becoming a concerned topic in social network analysis. Although many link prediction methods have been proposed in the recent years, most of the existing link prediction methods have unsatisfactory performance because of high time complexity, network size, sparsity, and similarity measures between node pairs for processing topological information. In this paper, we proposed a method for node representation learning and also a new embedding technique is used for generating latent features. We also proposed an improved version of the weighted random walk based on game theoretical technique and k-core decomposition. Node representations are generated via skipgram method. Although most of the link prediction methods have high time complexity, since our method uses Stochastic Gradient Descent for the optimization process, it has linear time complexity with respect to the number of vertices. This causes our algorithm to be scalable to large networks. In addition to that, sparsity is a huge challenge in complex networks and we cannot infer enough information from the structure of the network to make predictions. By learning a low-dimensional representation that captures the network structure, classification of nodes and edges can be done more easily. The performance of the proposed method was evaluated with some benchmark heuristic scores and state-of-the-art techniques on link prediction in several real-world networks. The experimental results show that the proposed method obtains higher accuracy in comparison with considered methods and measures. However, the time complexity is not improved effectively.

A Link Prediction Approach in Temporal Networks Based on Game Theory

A Link Prediction Approach in Temporal Networks Based on Game Theory

ELPKG: A High-Accuracy Link Prediction Approach for Knowledge Graph Completion

Link prediction in knowledge graph is the task of utilizing the existing relations to infer new relations so as to build a more complete knowledge graph. The inferred new relations plus original knowledge graph is the symmetry of completion knowledge graph. Previous research on link predication only focuses on path or semantic-based features, which can hardly have a full insight of features between entities and may result in a certain ratio of false inference results. To improve the accuracy of link predication, we propose a novel approach named Entity Link Prediction for Knowledge Graph (ELPKG), which can achieve a high accuracy on large-scale knowledge graphs while keeping desirable efficiency. ELPKG first combines path and semantic-based features together to represent the relationships between entities. Then it adopts a probabilistic soft logic-based reasoning method that effectively solves the problem of non-deterministic knowledge reasoning. Finally, the relation between entities is completed based on the entity link prediction algorithm. Extensive experiments on real dataset show that ELPKG outperforms baseline methods on hits@1, hits@10, and MRR.

Time-aware link prediction based on strengthened projection in bipartite networks

The traditional projection models in the bipartite networks involve many node pairs that consist of weak relationships. These node pairs lead to poor quality predictions as well as high computation time. In this paper, to cope with these problems, we firstly propose a novel projection model, which is called “Strengthened Projection Model”. Then, to predict the potential links in the future, we present a new link prediction approach based on the proposed projection model. Thanks to the proposed model, the computation time is shortened, and the high probability predictions are extracted. The majority of the previous works conducted in this area have used the classical proximity measure algorithms that only take into account the current network structure, regardless of when events occur in the network evolution. To overcome these limited methods, in this paper, we also propose a novel proximity measure algorithm that considers the bipartite network evolution. To the best of our knowledge, this is the first attempt that takes into account the time-awareness in bipartite networks. To evaluate the performance of our proposed approach, we conducted experiments on the academic information network. To construct this bipartite network, we collected data from IEEE Xplore. The experimental results show that the success of the proposed method is promising.

Who will trade bauxite with whom? Finding potential links through link prediction

In order to find potential links in bauxite international trade and help bauxite trading countries find new partners as well as diversify trading partners. A link prediction approach was used to find potential bauxite trade links from the perspective of the topological relationship of trade networks in various countries and trade rules. The results show that over the next five to six years, Denmark and the United States, China and Hungary; Guinea and the Netherlands, Hungary and the United States; and Australia and the United Kingdom are most likely to establish bauxite trade relations. However, Brazil and Turkey, Guyana and Hungary are the least likely to trade bauxite. The results of this study show that we can understand the potential changing trends of the international bauxite trade and provide relevant government policy recommendations for it.

Link prediction in dynamic networks based on the attraction force between nodes

As an important technology of social network analysis, link prediction is widely applied in computer science and many other fields. Link prediction can be used to detect missing links or predict whether two unconnected nodes will connect in the future. Various link prediction approaches have been proposed based on similarity metrics or learning in recent years; however, most failed to consider the direct changes during network development, and hence they are not applied to dynamic networks whose structures change continuously over time. In this paper, a novel approach for link prediction in dynamic networks based on the attraction force between nodes (DLPA) is proposed for detecting missing links and for predicting whether potential links will become real links in the future. First, a level is assigned to each node, which is used to represent the influence strength of the node compared to its neighbours in the initial network snapshot. The level must be updated with changes in the nodes. Then, the connection probability of each potential link is calculated based on the levels of the corresponding nodes and the attraction force between them. Thus, missing links can be detected and potential links can be predicted. In addition, the connection probabilities of potential links calculated via the proposed approach can vary with the evolution of the network. Experiments on static and dynamic real-world networks are conducted to evaluate the performance of the proposed approach, and the results demonstrate that the proposed approach outperforms several baseline algorithms in terms of prediction accuracy.

A pattern based supervised link prediction in directed complex networks

Link prediction is one of the most interesting tasks in complex network analysis. Numerous recently published link prediction methods have focused on utilizing network models close to real networks to improve performance of link prediction. Directed, temporal, weighted and heterogeneous network models are some examples of the favored network models. Most published link prediction metrics cannot take into account the effect of links directions on link formation. In this study, we propose a pattern based supervised link prediction approach to improve link prediction accuracy of Triad Closeness (TC) metric in directed complex networks. The proposed pattern based link prediction metric is compared with TC metric and the state-of-the-art link prediction metrics to evaluate the effectiveness of the proposed metric. Experimental results in two citation networks show that the proposed metric improves remarkably link prediction accuracy of TC metric and obtains the highest link prediction performance compared to the state-of-the-art link prediction metrics.

Enhanced recommender system using predictive network approach

Recommender systems have a unique role in on-line trading companies due to building relationships among users and items to reduce big information load. There exist several successful algorithms in the recommender systems like collaborative filtering (CF), although most of them suffer from the sparsity problem. Here, we propose a novel integrated recommendation approach based on the tools of network science to mitigate the sparsity problem. The link prediction approach is used to extract hidden structures among users, and diffusion of information is applied to enhance the rating matrix in our proposed framework. Not only, the sparsity problem is alleviated through a more efficient way, but the proposed approach also can be applied in a hybrid way with the well-known algorithms. The proposed approach is examined on several datasets via standard evaluation criteria. The experimental results show that the proposed approach outperforms the earlier methods.

Link Prediction Approach for Opportunistic Networks Based on Recurrent Neural Network

The target of link prediction is used to estimate the possibility of future links among nodes through known network structures and nodes information. According to the time-varying characteristics of the opportunistic network, the historical information of node pairs has a significant influence on the future connection state. We propose a novel link prediction approach which is based on the recurrent neural network link prediction (RNN-LP) framework. With the help of time series method, we define the vector that is made up of the node information and historical connection information of the node pairs, in which a sequence vector is constructed. Benefiting from RNN in sequence modeling, the time domain characteristics were extracted in the process of the dynamic evolution of the opportunistic network. Hence, the future link prediction becomes significantly better. By utilizing iMote traces Cambridge and MIT reality datasets, experimental results are obtained to reveal that RNN-LP method gives better accuracy and stability than the prediction techniques of the common neighbor, Adamic-Adar, resource allocation, local path, and Katz.

Variational Random Function Model for Network Modeling.

Link prediction is a fundamental problem in network modeling. A family of link prediction approaches is to treat network data as an exchangeable array whose entries can be explained by random functions (e.g., block models and Gaussian processes) over latent node factors. Despite their powerful ability in modeling missing links, these models tend to have a large computational complexity and thus are hard to deal with large networks. To address this problem, we develop a novel variational random function model by defining latent Gaussian processes on exchangeable arrays. This model not only inherits the ability of Gaussian process to describe the nonlinear interactions between nodes, but also enjoys significant reduction on computational complexity. To further make the model scalable to large network data, we develop an efficient key-value-free strategy under the map-reduce framework to tremendously reduce the inference time. Experimental results on large network data have demonstrated both the efficacy and efficiency of the proposed method over state-of-the-arts methods in network modeling.

Evaluating reliability of question-disease relations in online health forms: A link prediction approach

The Internet has become an indispensable part of human life in today. People can now easily find answers to questions they are curious about via the internet. The short, effortless and free way that the Internet provides is extremely attractive for people to have an idea in subjects they wonder related to their health. There are many online health forums where people can ask questions answered by health professionals. Every day, people ask thousands of questions on these sites and get answers about which diseases their complaints may be related to. The frequent use of online forum sites by people has led to the selection of these forums as data source for this study, and analysis of reliability. Firstly, in this study, link prediction in bipartite social networks, where intensive works have been done and it is applied on many areas nowadays, is tried to be carried out on question-disease bipartite network constructed with data obtained from analysis of online health forums whose use rate increase substantially. For this purpose, a novel link prediction method called as intensive link prediction is proposed, and prediction success of this method is compared with five of similarity-based link prediction methods. Better results have been obtained with the proposed method than the other methods. Then, the accuracy of the answers given to the users on online health forums which received intense interest are tested. The reliability of online health forums is measured by the accuracy analysis performed.

Predicting potential drug-drug interactions on topological and semantic similarity features using statistical learning.

Drug-drug interaction (DDI) is a change in the effect of a drug when patient takes another drug. Characterizing DDIs is extremely important to avoid potential adverse drug reactions. We represent DDIs as a complex network in which nodes refer to drugs and links refer to their potential interactions. Recently, the problem of link prediction has attracted much consideration in scientific community. We represent the process of link prediction as a binary classification task on networks of potential DDIs. We use link prediction techniques for predicting unknown interactions between drugs in five arbitrary chosen large-scale DDI databases, namely DrugBank, KEGG, NDF-RT, SemMedDB, and Twosides. We estimated the performance of link prediction using a series of experiments on DDI networks. We performed link prediction using unsupervised and supervised approach including classification tree, k-nearest neighbors, support vector machine, random forest, and gradient boosting machine classifiers based on topological and semantic similarity features. Supervised approach clearly outperforms unsupervised approach. The Twosides network gained the best prediction performance regarding the area under the precision-recall curve (0.93 for both random forests and gradient boosting machine). The applied methodology can be used as a tool to help researchers to identify potential DDIs. The supervised link prediction approach proved to be promising for potential DDIs prediction and may facilitate the identification of potential DDIs in clinical research.