Graph pattern matching is fundamental to social network analysis. Its effectiveness for identifying social communities and social positions, making recommendations and so on has been repeatedly demonstrated. However, the social network analysis raises new challenges to graph pattern matching. As real-life social graphs are typically large, it is often prohibitively expensive to conduct graph pattern matching over such large graphs, e.g., NP-complete for subgraph isomorphism, cubic time for bounded simulation, and quadratic time for simulation. These hinder the applicability of graph pattern matching on social network analysis. In response to these challenges, the thesis presents a series of effective techniques for querying large, dynamic, and distributively stored social networks. First of all, we propose a notion of query preserving graph compression, to compress large social graphs relative to a class Q of queries. We then develop both batch and incremental compression strategies for two commonly used pattern queries. Via both theoretical analysis and experimental studies, we show that (1) using compressed graphs Gr benefits graph pattern matching dramatically; and (2) the computation of Gr as well as its maintenance can be processed efficiently. Secondly, we investigate the distributed graph pattern matching problem, and explore parallel computation for graph pattern matching. We show that our techniques possess following performance guarantees: (1) each site is visited only once; (2) the total network traffic is independent of the size of G; and (3) the response time is decided by the size of largest fragment of G rather than the size of entire G. Furthermore, we show how these distributed algorithms can be implemented in the MapReduce framework. Thirdly, we study the problem of answering graph pattern matching using views since view based techniques have proven an effective technique for speeding up query evaluation. We propose a notion of pattern containment to characterise graph pattern matching using views, and introduce efficient algorithms to answer graph pattern matching using views. Moreover, we identify three problems related to graph pattern containment, and provide efficient algorithms for containment checking (approximation when the problem is intractable). Fourthly, we revise graph pattern matching by supporting a designated output node, which we treat as “query focus”. We then introduce algorithms for computing the top-k relevant matches w.r.t. the output node for both acyclic and cyclic pattern graphs, respectively, with early termination property. Furthermore, we investigate the diversified top-k matching problem, and develop an approximation algorithm with performance guarantee and a heuristic algorithm with early termination property. Finally, we introduce an expert search system, called ExpFinder, for large and dynamic social networks. ExpFinder identifies top-k experts in social networks by graph pattern matching, and copes with the sheer size of real-life social networks by integrating incremental graph pattern matching, query preserving compression and top-k matching computation. In particular, we also introduce bounded (resp. unbounded) incremental algorithms to maintain the weighted landmark vectors which are used for incremental maintenance for cached results.

The accumulation of high-throughput genomic and proteomic data allows for the reconstruction of the increasingly large and complex metabolic networks. In order to analyze the accumulated data and reconstructed networks, it is critical to identify network patterns and evolutionary relations between metabolic networks. But even finding similar networks becomes computationally challenging. The dissertation addresses these challenges with discrete optimization and the corresponding algorithmic techniques. Based on the property of the gene duplication and function sharing in biological network,we have formulated the network alignment problem which asks the optimal vertex-to-vertex mapping allowing path contraction, vertex deletion, and vertex insertions. We have proposed the first polynomial time algorithm for aligning an acyclic metabolic pattern pathway with an arbitrary metabolic network. We also have proposed a polynomial-time algorithm for patterns with small treewidth and implemented it for series-parallel patterns which are commonly found among metabolic networks. We have developed the metabolic network alignment tool for free public use. We have performed pairwise mapping of all pathways among five organisms and found a set of statistically significant pathway similarities. We also have applied the network alignment to identifying inconsistency, inferring missing enzymes, and finding potential candidates.

Sequential patterns mining is an important pattern discovery technique used to identify frequently observed sequential occurrence of items across ordered transactions over time. It has been intensively studied and there exists a great diversity of algorithms. However, there is a major problem associated with the conventional sequential patterns mining in that patterns derived are often large and not very easy to understand or use. In addition, more complex relations among events are often hidden behind sequences. A novel model for sequential patterns called Sequential Patterns Graph (SPG) is proposed. The construction algorithm of SPG is presented with experimental results to substantiate the concept. The thesis then sets out to define some new structural patterns such as concurrent branch patterns, exclusive patterns and iterative patterns which are generally hidden behind sequential patterns. Finally, an integrative framework, named Post Sequential Patterns Mining (PSPM), which is based on sequential patterns mining, is also proposed for the discovery and visualisation of structural patterns. This thesis is intended to prove that discrete sequential patterns derived from traditional sequential patterns mining can be modelled graphically using SPG. It is concluded from experiments and theoretical studies that SPG is not only a minimal representation of sequential patterns mining, but it also represents the interrelation among patterns and establishes further the foundation for mining structural knowledge (i.e. concurrent branch patterns, exclusive patterns and iterative patterns). from experiments conducted on both synthetic and real datasets, it is shown that Concurrent Branch Patterns (CBP) mining is an effective and efficient mining algorithm suitable for concurrent branch patterns.

Pattern matching approaches in reverse engineering aim to incorporate domain knowledge and system documentation in the software architecture extraction process, hence provide a user/tool collaborative environment for architectural design recovery. This thesis presents a model and an environment for recovering the high level design of legacy software systems based on user defined architectural patterns and graph matching techniques. In the proposed model, a high-level view of a software system in terms of the system components and their interactions is represented as a query, using a description language. A query is mapped onto a pattern-graph, where a module and its interactions with other modules are represented as a group of graph-nodes and a group of graph-edges, respectively. Interaction constraints can be modeled by the description language as a part of the query. Such a pattern-graph is applied against an entity-relation graph that represents the information extracted from the source code of the software system. An approximate graph matching process performs a series of graph edit operations (i. e. , node/edge insertion/deletion) on the pattern-graph and uses a ranking mechanism based on data mining association to obtain a sub-optimal solution. The obtained solution corresponds to an extracted architecture that complies with the given query. An interactive prototype toolkit implemented as part of this thesis provides an environment for architecture recovery in two levels. First the system is decomposed into a number of subsystems of files. Second each subsystem can be decomposed into a number of modules of functions, datatypes, and variables.

Pattern matching approaches in reverse engineering aim to incorporate domain knowledge and system documentation in the software architecture extraction process, hence provide a user/tool collaborative environment for architectural design recovery. This thesis presents a model and an environment for recovering the high level design of legacy software systems based on user defined architectural patterns and graph matching techniques. In the proposed model, a high-level view of a software system in terms of the system components and their interactions is represented as a query, using a description language. A query is mapped onto a pattern-graph, where a module and its interactions with other modules are represented as a group of graph-nodes and a group of graph-edges, respectively. Interaction constraints can be modeled by the description language as a part of the query. Such a pattern-graph is applied against an entity-relation graph that represents the information extracted from the source code of the software system. An approximate graph matching process performs a series of graph edit operations (i. e. , node/edge insertion/deletion) on the pattern-graph and uses a ranking mechanism based on data mining association to obtain a sub-optimal solution. The obtained solution corresponds to an extracted architecture that complies with the given query. An interactive prototype toolkit implemented as part of this thesis provides an environment for architecture recovery in two levels. First the system is decomposed into a number of subsystems of files. Second each subsystem can be decomposed into a number of modules of functions, datatypes, and variables.

Ces dernières années ont connu un regain d'intérêt pour l'utilisation des graphes comme moyen fiable de représentation et de modélisation des données, et ce, dans divers domaines de l'informatique. En particulier, pour les grandes masses de données, les graphes apparaissent comme une alternative prometteuse aux bases de données relationnelles. Plus particulièrement, le recherche de sous-graphes s'avère être une tâche cruciale pour explorer ces grands jeux de données. Dans cette thèse, nous étudions deux problématiques principales. Dans un premier temps, nous abordons le problème de la détection de motifs dans les grands graphes. Ce problème vise à rechercher les k-meilleures correspondances (top-k) d'un graphe motif dans un graphe de données. Pour cette problématique, nous introduisons un nouveau modèle de détection de motifs de graphe nommé la Simulation Relaxée de Graphe (RGS), qui permet d’identifier des correspondances de graphes avec un certain écart (structurel) et ainsi éviter le problème de réponse vide. Ensuite, nous formalisons et étudions le problème de la recherche des k-meilleures réponses suivant deux critères, la pertinence (la meilleure similarité entre le motif et les réponses) et la diversité (la dissimilarité entre les réponses). Nous considérons également le problème des k-meilleures correspondances diversifiées et nous proposons une fonction de diversification pour équilibrer la pertinence et la diversité. En outre, nous développons des algorithmes efficaces basés sur des stratégies d’optimisation en respectant le modèle proposé. Notre approche est efficiente en terme de temps d’exécution et flexible en terme d'applicabilité. L’analyse de la complexité des algorithmes et les expérimentations menées sur des jeux de données réelles montrent l’efficacité des approches proposées. Dans un second temps, nous abordons le problème de recherche agrégative dans des documents XML. Pour un arbre requête, l'objectif est de trouver des motifs correspondants dans un ou plusieurs documents XML et de les agréger dans un seul agrégat. Dans un premier temps nous présentons la motivation derrière ce paradigme de recherche agrégative et nous expliquons les gains potentiels par rapport aux méthodes classiques de requêtage. Ensuite nous proposons une nouvelle approche qui a pour but de construire, dans la mesure du possible, une réponse cohérente et plus complète en agrégeant plusieurs résultats provenant de plusieurs sources de données. Les expérimentations réalisées sur plusieurs ensembles de données réelles montrent l’efficacité de cette approche en termes de pertinence et de qualité de résultat. / Recent years have witnessed a growing renewed interest in the use of graphs as a reliable means for representing and modeling data. Thereby, graphs enable to ensure efficiency in various fields of computer science, especially for massive data where graphs arise as a promising alternative to relational databases for big data modeling. In this regard, querying data graph proves to be a crucial task to explore the knowledge in these datasets. In this dissertation, we investigate two main problems. In the first part we address the problem of detecting patterns in larger graphs, called the top-k graph pattern matching problem. We introduce a new graph pattern matching model named Relaxed Graph Simulation (RGS), to identify significant matches and to avoid the empty-set answer problem. We formalize and study the top-k matching problem based on two classes of functions, relevance and diversity, for ranking the matches according to the RGS model. We also consider the diversified top-k matching problem, and we propose a diversification function to balance relevance and diversity. Moreover, we provide efficient algorithms based on optimization strategies to compute the top-k and the diversified top-k matches according to the proposed model. The proposed approach is optimal in terms of search time and flexible in terms of applicability. The analyze of the time complexity of the proposed algorithms and the extensive experiments on real-life datasets demonstrate both the effectiveness and the efficiency of these approaches. In the second part, we tackle the problem of graph querying using aggregated search paradigm. We consider this problem for particular types of graphs that are trees, and we deal with the query processing in XML documents. Firstly, we give the motivation behind the use of such a paradigm, and we explain the potential benefits compared to traditional querying approaches. Furthermore, we propose a new method for aggregated tree search, based on approximate tree matching algorithm on several tree fragments, that aims to build, the extent possible, a coherent and complete answer by combining several results. The proposed solutions are shown to be efficient in terms of relevance and quality on different real-life datasets