Frequent Subgraph Mining Research Articles

A qualitative survey on frequent subgraph mining

AbstractData mining is a popular research area that has been studied by many researchers and focuses on finding unforeseen and important information in large databases. One of the popular data structures used to represent large heterogeneous data in the field of data mining is graphs. So, graph mining is one of the most popular subdivisions of data mining. Subgraphs that are more frequently encountered than the user-defined threshold in a database are called frequent subgraphs. Frequent subgraphs in a database can give important information about this database. Using this information, data can be classified, clustered and indexed. The purpose of this survey is to examine frequent subgraph mining algorithms (i) in terms of frequent subgraph discovery process phases such as candidate generation and frequency calculation, (ii) categorize the algorithms according to their general attributes such as input type, dynamicity of graphs, result type, algorithmic approach they are based on, algorithmic design and graph representation as well as (iii) to discuss the performance of algorithms in comparison to each other and the challenges faced by the algorithms recently.

Big graph classification frameworks based on Extreme Learning Machine

Graph data analysis is a hot topic in recent research area. Graph classification is one of the most important graph data analysis problems, which choose the most probable class labels of graphs using models based on the training dataset. It has wildly applications in protein group identification, chemical compounds classification and so on. Many existing research of graph learning suffer from high computation cost as large scale graph data are dramatically increased. In order to realize big graph classification with real-time learning ability and good scalability, efficient feature extraction approaches and ELM variants are utilized in this paper. To be specific, we present three frameworks of big graph classification based on ELMs: (1) a framework with a compression-based frequent subgraph mining method to reduce graph size; (2) an incremental framework to handle dynamic graphs; (3) a distributed framework with distributed ELMs to provide good scalability and easy implementation on cloud platforms. Extensive experiments are conducted on clusters with large real-world graph datasets. The experimental results demonstrate that our frameworks are efficient in big graph classification applications, and well suitable for dynamic networks. The results also validate that ELM and its variants have good classification performance on large-scale graphs.

Predicting ICU readmission using grouped physiological and medication trends.

Patients who are readmitted to an intensive care unit (ICU) usually have a high risk of mortality and an increased length of stay. ICU readmission risk prediction may help physicians to re-evaluate the patient's physical conditions before patients are discharged and avoid preventable readmissions. ICU readmission prediction models are often built based on physiological variables. Intuitively, snapshot measurements, especially the last measurements, are effective predictors that are widely used by researchers. However, methods that only use snapshot measurements neglect predictive information contained in the trends of physiological and medication variables. Mean, maximum or minimum values take multiple time points into account and capture their summary statistics, however, these statistics are not able to catch the detailed picture of temporal trends. In this study, we find strong predictors with ability of capturing detailed temporal trends of variables for 30-day readmission risk and build prediction models with high accuracy. We study physiological measurements and medications from the Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC-II) clinical dataset. Time series of each variable are converted into trend graphs with nodes being discretized measurements of each variable. Then we extract important temporal trends by applying frequent subgraph mining on the trend graphs. The frequency of a subgraph is a good cue to find important temporal trends since similar patients often share similar trends regarding their pathophysiological evolution under medical interventions. Important temporal trends are then grouped automatically by non-negative matrix factorization. The grouped trends could be considered as an approximate representation of patients' pathophysiological states and medication profiles. We train a logistic regression model to predict 30-day ICU readmission risk based on snapshot measurements, grouped physiological trends and medication trends. Our dataset consists of 1170 patients who are alive 30 days after discharge from ICU and have at least 12 h of data. In the dataset, 860 patients were not readmitted and 310 were readmitted, within 30 days after discharge. Our model outperforms all comparison models, and shows an improvement in the area under the receiver operating characteristic curve (AUC) of almost 4% from the best comparison model. Grouped physiological and medication trends carry predictive information for ICU readmission risk. In order to build predictive models with higher accuracy, we should add grouped physiological and medication trends as complementary features to snapshot measurements.

Two bi-objective hybrid approaches for the frequent subgraph mining problem

This paper proposes two hybrid bi-objective evolutionary algorithms to solve the frequent subgraph mining problem. In this contribution, we propose to improve the search ability of the stochastic local search (SLS) and the variable neighborhood search (VNS) algorithms by adding genetic operators (crossover and mutation) and Pareto dominance concept. A mined subgraph is defined by a bi-objective function which uses two parameters, support and size. We combine GA and SLS in a hybrid method denoted GASLS and GA with VNS in a hybrid method denoted GAVNS to solve the considered problem. The two proposed methods are implemented and evaluated on two synthetic and five real-world datasets of various sizes and their performance were compared against a single-objective stochastic local search algorithm and the well-known NSGA-II algorithm. The proposed methods are able to discover efficiently diversified subgraphs in the search space by exploring new solutions. The numerical results show that in general the GASLS method provides competitive results and finds high quality solutions compared to the other considered algorithms.

A Survey of Frequent Subgraph Mining Algorithms

Graphs are broadly used data structure. Graphs are very useful in representing/analyzing and processing real world data. Evolving graphs are graphs which are frequently changing in nature. There is either increase or decrease in their size i.e. change in number of edges or/and vertices. Mining is the process done for knowledge discovery in graphs. Detecting specific patterns with their number of repetition more than a predefined threshold in graph is known as frequent subgraph mining or FSM. Real Timed data representing graphs are high volumetric or of very large in size, handling such graphs require processing them with special mechanisms and algorithms. Our review paper discovers present FSM techniques and tries to give their comparative study.

CmFSM: a scalable CPU-MIC coordinated drug-finding tool by frequent subgraph mining

BackgroundFrequent subgraphs mining is a significant problem in many practical domains. The solution of this kind of problem can particularly used in some large-scale drug molecular or biological libraries to help us find drugs or core biological structures rapidly and predict toxicity of some unknown compounds. The main challenge is its efficiency, as (i) it is computationally intensive to test for graph isomorphisms, and (ii) the graph collection to be mined and mining results can be very large. Existing solutions often require days to derive mining results from biological networks even with relative low support threshold. Also, the whole mining results always cannot be stored in single node memory.ResultsIn this paper, we implement a parallel acceleration tool for classical frequent subgraph mining algorithm called cmFSM. The core idea is to employ parallel techniques to parallelize extension tasks, so as to reduce computation time. On the other hand, we employ multi-node strategy to solve the problem of memory constraints. The parallel optimization of cmFSM is carried out on three different levels, including the fine-grained OpenMP parallelization on single node, multi-node multi-process parallel acceleration and CPU-MIC collaborated parallel optimization.ConclusionsEvaluation results show that cmFSM clearly outperforms the existing state-of-the-art miners even if we only hold a few parallel computing resources. It means that cmFSM provides a practical solution to frequent subgraph mining problem with huge number of mining results. Specifically, our solution is up to one order of magnitude faster than the best CPU-based approach on single node and presents a promising scalability of massive mining tasks in multi-node scenario. More source code are available at:Source Code: https://github.com/ysycloud/cmFSM.

Ap-FSM: A parallel algorithm for approximate frequent subgraph mining using Pregel

Large graphs are scale-free, ubiquitous having irregular relationships and non-trivial topology. Frequent subgraph mining is a popular method for knowledge extraction from graphs. Most of the existing frequent subgraph mining algorithms are centralized algorithms that cannot handle a single large graph efficiently and incur high communication cost. However, to make the task of subgraph mining less expensive computationally, approximate subgraph mining can be applied which will capture similar structure subgraphs as of exact subgraph mining. In this paper, we propose an approximate subgraph mining algorithm named Ap-FSM implemented on distributed graph environment Pregel. The working of Ap-FSM is divided into three phases. The first phase selects the representative graph from the original graph while preserving the original graph properties. The second phase efficiently performs subgraph extension. Phase 3 introduces a novel two-step optimization for performing subgraph pruning. Analyzing such large graph data will be beneficial from the perspective of expert and intelligent systems, as discovered patterns can be used for knowledge discovery and decision making. To evaluate the performance of Ap-FSM, experiments are performed over three real life datasets having up to billion edges. The results show that the proposed Ap-FSM significantly outperforms the state-of-art frequent subgraph mining algorithms and overcome the challenges of performing frequent subgraph mining on a massive large graph. It is also shown that Ap-FSM achieves high scalability and speedup in distributed graph environment and is highly accurate in finding frequent subgraphs from a single large graph.

Mining frequent subgraphs in multigraphs

For more than a decade, extracting frequent patterns from single large graphs has been one of the research focuses. However, in this era of data eruption, rich and complex data is being generated at an unprecedented rate. This complex data can be represented as a multigraph structure - a generic and rich graph representation. In this paper, we propose a novel frequent subgraph mining approach MuGraM that can be applied to multigraphs. MuGraM is a generic frequent subgraph mining algorithm that discovers frequent multigraph patterns. MuGraM efficiently performs the task of subgraph matching, which is crucial for support measure, and further leverages several optimization techniques for swift discovery of frequent subgraphs. Our experiments reveal two things: MuGraM discovers multigraph patterns, where other existing approaches are unable to do so; MuGraM, when applied to simple graphs, outperforms the state of the art approaches by at least one order of magnitude.

A Parallel Approach for Frequent Subgraph Mining in a Single Large Graph Using Spark

Frequent subgraph mining (FSM) plays an important role in graph mining, attracting a great deal of attention in many areas, such as bioinformatics, web data mining and social networks. In this paper, we propose SSiGraM (Spark based Single Graph Mining), a Spark based parallel frequent subgraph mining algorithm in a single large graph. Aiming to approach the two computational challenges of FSM, we conduct the subgraph extension and support evaluation parallel across all the distributed cluster worker nodes. In addition, we also employ a heuristic search strategy and three novel optimizations: load balancing, pre-search pruning and top-down pruning in the support evaluation process, which significantly improve the performance. Extensive experiments with four different real-world datasets demonstrate that the proposed algorithm outperforms the existing GraMi (Graph Mining) algorithm by an order of magnitude for all datasets and can work with a lower support threshold.

A Two-Phase Algorithm for Differentially Private Frequent Subgraph Mining.

Mining frequent subgraphs from a collection of input graphs is an important task for exploratory data analysis on graph data. However, if the input graphs contain sensitive information, releasing discovered frequent subgraphs may pose considerable threats to individual privacy. In this paper, we study the problem of frequent subgraph mining (FSM) under the rigorous differential privacy model. We present a two-phase differentially private FSM algorithm, which is referred to as DFG. In DFG, frequent subgraphs are privately identified in the first phase, and the noisy support of each identified frequent subgraph is calculated in the second phase. In particular, to privately identity frequent subgraphs, we propose a frequent subgraph identification approach, which can improve the accuracy of discovered frequent subgraphs through candidate pruning. Moreover, to compute the noisy support of each identified frequent subgraph, we devise a lattice-based noisy support computation approach, which leverages the inclusion relations between the discovered frequent subgraphs to improve the accuracy of the noisy supports. Through formal privacy analysis, we prove that DFG satisfies ϵ-differential privacy. Extensive experimental results on real datasets show that DFG can privately find frequent subgraphs while achieving high data utility.

Optimized and Frequent Subgraphs: How Are They Related?

Frequent subgraph mining (FSM) is one of the most challenging tasks in graph mining. FSM consists of applying the data mining algorithms to extract interesting, unexpected, and useful graph patterns from the graphs. It also aspires to offer a richer apprehension of the given graph data. FSM has been applied to many domains, such as graphical data management and knowledge discovery, social network analysis, Bioinformatics, and security. In this context, a large number of techniques have been suggested to deal with the graph data, with the objective to extract the frequently occurring graph patterns. Such patterns are called frequent subgraph patterns (FSPs). FSPs are extracted using the traditional support threshold parameter. However, there exists no specialized scheme to decide the discovered FSPs are optimized as well. Thus, the aim of this paper is to suggest an optimization strategy to uncover the association between the frequent and the optimized subgraph patterns. For exploring the existence of the potential association between the FSPs and the optimized subgraph, a Particle Swarm Optimization technique is suggested. This relationship will be very handy to reduce the FSPs, by choosing those FSPs which were also discovered as optimized FSPs. Different experiments are performed using benchmark graph data sets to validate the existence of the aforementioned relationship between the optimized and the frequent FSPs.

A New Framework of Frequent Uncertain Subgraph Mining

Nowadays, frequent graph mining approaches are very popular in several applications and big data domains. However, some problems are still not solved and need more attention from researchers such as inexact subgraph matching when similarity among subgraphs is not totally proved. Furthermore, many research works dealt with structural graph matching, but a little attention was paid to semantic graph matching when graph vertices and/or edges are attributed and typed. Therefore, this paper proposes a new possibilistic flexible graph mining approach to discover similar subgraphs, by applying possibilistic similarity rather than using hard structural exact similarity. This novel approach is hybrid and uses two similarity measures. An approximate structural similarity function of graph edit distance function and a semantic vertices similarity function based on possibilistic information affinity function. Experimental evaluation was conducted on real datasets borrowed from various domains and tests have shown a highly superiority of our hybrid frequent subgraph mining approach by achieving a good time processing performance and simultaneously discovering more similar subgraphs.

Incremental Frequent Subgraph Mining on Large Evolving Graphs

Frequent subgraph mining is a core graph operation used in many domains, such as graph data management and knowledge exploration, bioinformatics, and security. Most existing techniques target static graphs. However, modern applications, such as social networks, utilize large evolving graphs. Mining these graphs using existing techniques is infeasible, due to the high computational cost. In this paper, we propose IncGM+ , a fast incremental approach for continuous frequent subgraph mining on a single large evolving graph. We adapt the notion of “fringe” to the graph context, that is the set of subgraphs on the border between frequent and infrequent subgraphs. IncGM+ maintains fringe subgraphs and exploits them to prune the search space. To boost the efficiency, we propose an efficient index structure to maintain selected embeddings with minimal memory overhead. These embeddings are utilized to avoid redundant expensive subgraph isomorphism operations. Moreover, the proposed system supports batch updates. Using large real-world graphs, we experimentally verify that IncGM+ outperforms existing methods by up to three orders of magnitude, scales to much larger graphs and consumes less memory.

Resling: a scalable and generic framework to mine top-k representative subgraph patterns

Mining subgraph patterns is an active area of research due to its wide-ranging applications. Examples include frequent subgraph mining, discriminative subgraph mining, statistically significant subgraphs. Existing research has primarily focused on mining all subgraph patterns in the database. However, due to the exponential subgraph search space, the number of patterns mined, typically, is too large for any human-mediated analysis. Consequently, deriving insights from the mined patterns is hard for domain scientists. In addition, subgraph pattern mining is posed in multiple forms: the function that models if a subgraph is a pattern varies based on the application and the database could be over multiple graphs or a single, large graph. In this paper, we ask the following question: Given a subgraph importance function and a budget k, which are the k subgraph patterns that best represent all other patterns of interest? We show that the problem is NP-hard, and propose a generic framework called Resling that adapts to arbitrary subgraph importance functions and generalizable to both transactional graph databases as well as single, large graphs. Resling derives its power by structuring the search space in the form of an edit map, where each subgraph is a node, and two subgraphs are connected if they have an edit distance of one. We rank nodes in the edit map through two random walk based algorithms: vertex-reinforced random walks ( Resling -VR) and negative-reinforced random walks( Resling -NR). Experiments show that Resling-VR is up to 20 times more representative of the pattern space and two orders of magnitude faster than the state-of-the-art techniques. Resling-NR further improves the running time while maintaining comparable or better performance in representative power.

Interval graph mining

Frequent subgraph mining is a difficult data mining problem aiming to find the exact set of frequent subgraphs into a database of graphs. Current subgraph mining approaches make use of the canonical encoding which is one of the key operations. It is well known that canonical encodings have an exponential time complexity. Consequently, mining all frequent patterns for large and dense graphs is computationally expensive. In this paper, we propose an interval approach to handle canonicity, leading to two encodings, lower and upper encodings, with a polynomial time complexity, allowing to tightly enclose the exact set of frequent subgraphs. These two encodings lead to an interval graph mining algorithm where two minings are launched in parallel, a lower mining (resp. upper mining) using the lower (resp. upper) encoding. The interval graph mining approach has been implemented within the state of the art Gaston miner. Experiments performed on synthetic and real graph databases coming from stock market and biological datasets show that our interval graph mining is effective on dense graphs.

Mining frequent subgraphs from tremendous amount of small graphs using MapReduce

Frequent subgraph mining from a tremendous amount of small graphs is a primitive operation for many data mining applications. Existing approaches mainly focus on centralized systems and suffer from the scalability issue. Consider the increasing volume of graph data and mining frequent subgraphs is a memory-intensive task, it is difficult to tackle this problem on a centralized machine efficiently. In this paper, we therefore propose an efficient and scalable solution, called MRFSE, using MapReduce. MRFSE adopts the breadth-first search strategy to iteratively extract frequent subgraphs, i.e., all frequent subgraphs with $$i+1$$ edges are generated based on frequent subgraphs with i edges at the ith iteration. In our design, existing frequent subgraph mining techniques in centralized systems can be easily extended and integrated. More importantly, new frequent subgraphs are generated without performing any isomorphism test which is costly and imperative in existing frequent subgraph mining techniques. Besides, various optimization techniques are proposed to further reduce the communication and I/O cost. Extensive experiments conducted on our in-house clusters demonstrate the superiority of our proposed solution in terms of both scalability and efficiency.

Discovery of Functional Motifs from the Interface Region of Oligomeric Proteins Using Frequent Subgraph Mining.

Modeling the interface region of a protein complex paves the way for understanding its dynamics and functionalities. Existing works model the interface region of a complex by using different approaches, such as, the residue composition at the interface region, the geometry of the interface residues, or the structural alignment of interface regions. These approaches are useful for ranking a set of docked conformation or for building scoring function for protein-protein docking, but they do not provide a generic and scalable technique for the extraction of interface patterns leading to functional motif discovery. In this work, we model the interface region of a protein complex by graphs and extract interface patterns of the given complex in the form of frequent subgraphs. To achieve this, we develop a scalable algorithm for frequent subgraph mining. We show that a systematic review of the mined subgraphs provides an effective method for the discovery of functional motifs that exist along the interface region of a given protein complex. In our experiments, we use three PDB protein structure datasets. The first two datasets are composed of PDB structures from different conformations of two dimeric protein complexes: HIV-1 protease (329 structures), and triosephosphate isomerase (TIM) (86 structures). The third dataset is a collection of different enzyme structures protein structures from the six top-level enzyme classes, namely: Oxydoreductase, Transferase, Hydrolase, Lyase, Isomerase, and Ligase. We show that for the first two datasets, our method captures the locking mechanism at the dimeric interface by taking into account the spatial positioning of the interfacial residues through graphs. Indeed, our frequent subgraph mining based approach discovers the patterns representing the dimerization lock which is formed at the base of the structure in 323 of the 329 HIV-1 protease structures. Similarly, for 86 TIM structures, our approach discovers the dimerization lock formation in 50 structures. For the enzyme structures, we show that we are able to capture the functional motifs (active sites) that are specific to each of the six top-level classes of enzymes through frequent subgraphs.

Mechanism of Cerebralcare Granule® for Improving Cognitive Function in Resting-State Brain Functional Networks of Sub-healthy Subjects.

Cerebralcare Granule® (CG), a Chinese herbal medicine, has been used to ameliorate cognitive impairment induced by ischemia or mental disorders. The ability of CG to improve health status and cognitive function has drawn researchers' attention, but the relevant brain circuits that underlie the ameliorative effects of CG remain unclear. The present study aimed to explore the underlying neurobiological mechanisms of CG in ameliorating cognitive function in sub-healthy subjects using resting-state functional magnetic resonance imaging (fMRI). Thirty sub-healthy participants were instructed to take one 2.5-g package of CG three times a day for 3 months. Clinical cognitive functions were assessed with the Chinese Revised Wechsler Adult Intelligence Scale (WAIS-RC) and Wechsler Memory Scale (WMS), and fMRI scans were performed at baseline and the end of intervention. Functional brain network data were analyzed by conventional network metrics (CNM) and frequent subgraph mining (FSM). Then 21 other sub-healthy participants were enrolled as a blank control group of cognitive functional. We found that administrating CG can improve the full scale of intelligence quotient (FIQ) and Memory Quotient (MQ) scores. At the same time, following CG treatment, in CG group, the topological properties of functional brain networks were altered in various frontal, temporal, occipital cortex regions, and several subcortical brain regions, including essential components of the executive attention network, the salience network, and the sensory-motor network. The nodes involved in the FSM results were largely consistent with the CNM findings, and the changes in nodal metrics correlated with improved cognitive function. These findings indicate that CG can improve sub-healthy subjects' cognitive function through altering brain functional networks. These results provide a foundation for future studies of the potential physiological mechanism of CG.

A novel rule generator for intrusion detection based on frequent subgraph mining

espanolEl desarrollo actual de las tecnologias ha impulsado el uso de servicios de telecomunicaciones. Este hecho implica un aumento en los volumenes de datos generados en las empresas de telecomunicaciones. Dichos datos requieren ser procesados con el fin de detectar potenciales intrusos o fraudes. Las tecnicas de evaluacion regla son ampliamente utilizadas en estos contextos de aplicacion por su alta efectividad sobre ataques conocidos. La incorporacion de un generador automatico de reglas en dichas tecnicas, permite obtener reglas a partir de grandes volumenes de datos, lo cual contribuye al trabajo de los analistas de informacion e incrementa la precision durante la deteccion de intrusos. En este trabajo, se presenta un metodo de generacion automatica de reglas, que comprende una estrategia basada en el procesamiento de patrones extraidos de un conjunto de entrenamiento y la conformacion de reglas de clasificacion. Finalmente, nuestra propuesta es evaluada y comparada con respecto a otros clasificadores, alcanzando buenos resultados. EnglishThe current development of technologies has boosted the use of telecommunication services. This fact brings an increase in the volumes of data generated in telecommunication companies. Such data need to be processed in order to detect potential intruders or fraud. The rule evaluation techniques are widely used in these application contexts due to their high effectiveness over known attacks. The incorporation of an automatic rule generator allows it to obtain rules in large volumes of data, for assisting information analysts; thus, the accuracy of intrusion detection is increased. In this paper, an automatic rule generation method is presented, including a strategy based on processing the patterns extracted from a training set and building classification rules. Finally, our proposal is evaluated and compared regarding other classifiers, achieving good results.

Extension of Canonical Adjacency Matrices for Frequent Approximate Subgraph Mining on Multi-Graph Collections

Into the data mining field, frequent approximate subgraph (FAS) mining has become an important technique with a broad spectrum of real-life applications. This fact is because several real-life phenomena can be modeled by graphs. In the literature, several algorithms have been reported for mining frequent approximate patterns on simple-graph collections; however, there are applications where more complex data structures, as multi-graphs, are needed for modeling the problem. But to the best of our knowledge, there is no FAS mining algorithm designed for dealing with multi-graphs. Therefore, in this paper, a canonical form (CF) for simple-graphs is extended to allow representing multi-graphs and a state-of-the-art algorithm for FAS mining is also extended for processing multi-graph collections by using the extended CF. Our experiments over different synthetic and real-world multi-graph collections show that the proposed algorithm has a good performance in terms of runtime and scalability. Additionally, we show the usefulness of the patterns computed by our algorithm in an image classification problem where images are represented as multi-graphs.