Nurcan Yuruk

SCAN: a structural clustering algorithm for networks

Network clustering (or graph partitioning) is an important task for the discovery of underlying structures in networks. Many algorithms find clusters by maximizing the number of intra-cluster edges. While such algorithms find useful and interesting structures, they tend to fail to identify and isolate two kinds of vertices that play special roles - vertices that bridge clusters (hubs) and vertices that are marginally connected to clusters (outliers). Identifying hubs is useful for applications such as viral marketing and epidemiology since hubs are responsible for spreading ideas or disease. In contrast, outliers have little or no influence, and may be isolated as noise in the data. In this paper, we proposed a novel algorithm called SCAN (Structural Clustering Algorithm for Networks), which detects clusters, hubs and outliers in networks. It clusters vertices based on a structural similarity measure. The algorithm is fast and efficient, visiting each vertex only once. An empirical evaluation of the method using both synthetic and real datasets demonstrates superior performance over other methods such as the modularity-based algorithms.

A structural approach for finding functional modules from large biological networks

BackgroundBiological systems can be modeled as complex network systems with many interactions between the components. These interactions give rise to the function and behavior of that system. For example, the protein-protein interaction network is the physical basis of multiple cellular functions. One goal of emerging systems biology is to analyze very large complex biological networks such as protein-protein interaction networks, metabolic networks, and regulatory networks to identify functional modules and assign functions to certain components of the system. Network modules do not occur by chance, so identification of modules is likely to capture the biologically meaningful interactions in large-scale PPI data. Unfortunately, existing computer-based clustering methods developed to find those modules are either not so accurate or too slow.ResultsWe devised a new methodology called SCAN (Structural Clustering Algorithm for Networks) that can efficiently find clusters or functional modules in complex biological networks as well as hubs and outliers. More specifically, we demonstrated that we can find functional modules in complex networks and classify nodes into various roles based on their structures. In this study, we showed the effectiveness of our methodology using the budding yeast (Saccharomyces cerevisiae) protein-protein interaction network. To validate our clustering results, we compared our clusters with the known functions of each protein. Our predicted functional modules achieved very high purity comparing with state-of-the-art approaches. Additionally the theoretical and empirical analysis demonstrated a linear running-time of the algorithm, which is the fastest approach for networks.ConclusionWe compare our algorithm with well-known modularity based clustering algorithm CNM. We successfully detect functional groups that are annotated with putative GO terms. Top-10 clusters with minimum p-value theoretically prove that newly proposed algorithm partitions network more accurately then CNM. Furthermore, manual interpretations of functional groups found by SCAN show superior performance over CNM.

AHSCAN: Agglomerative Hierarchical Structural Clustering Algorithm for Networks

Many systems in sciences, engineering and nature can be modeled as networks. Examples include the internet, WWW and social networks. Finding hidden structures is important for making sense of complex networked data. In this paper we present a new network clustering method that can find clusters in an agglomerative fashion using structural similarity of vertices in the given network. Experiments conducted on real datasets demonstrate promising performance of the new method.

A Divisive Hierarchical Structural Clustering Algorithm for Networks

Many systems in sciences, engineering and nature can be modeled as networks. Examples are internet, metabolic networks and social networks. Network clustering algorithms aimed to find hidden structures from networks are important to make sense of complex networked data. In this paper we present a new clustering method for networks. The proposed algorithm can find hierarchical structure of clusters without requiring any input parameters. The experiments using real data demonstrate an outstanding performance of the new method.

Mining concept associations for knowledge discovery in large textual databases

In this paper, we describe a new approach for mining concept associations from large text collections. The concepts are short sequences of words that occur frequently together across the text collections. It is these concepts that convey most of the meaning in any language. Our goal is to extract interesting associations among concepts that co-occur within the text collections. Interesting association between the concepts is mined using association rule mining algorithm. Finally we construct directed graph from current rules. The experimental result shows that our approach can efficiently find interesting concept associations in large text collections.

Finding Functional Modules

Biological systems can be modeled as complex network systems with many interactions between the components. These interactions give rise to the function and behavior of that system. For example, the protein–protein interaction (PPI) network is the physical basis of multiple cellular functions. One goal of emerging systems biology is to analyze very large complex biological networks such as protein–protein interaction networks, metabolic networks, and regulatory networks to identify functional modules and assign functions to certain components of the system. Network modules do not occur by chance, so identification of modules is likely to capture the biologically meaningful interactions in large-scale PPI data. Unfortunately, existing computer-based clustering methods developed to find those modules are either not so accurate or are too slow. We devised a new methodology called SCAN (Structural Clustering Algorithm for Networks) that can efficiently find clusters or functional modules in complex biological networks as well as hubs and outliers. More specifically, we demonstrated that we can find functional modules in complex networks and classify nodes into various roles based on their structures. In this chapter, we show the effectiveness of our methodology using the budding yeast (Saccharomyces cerevisiae) PPI network. To validate our clustering results, we compared our clusters with the known functions of each protein. Our predicted functional modules achieved very high purity comparing with state-of-the-art approaches. Additionally the theoretical and empirical analysis demonstrated a linear running time of the algorithm, which is the fastest approach for networks. We compare our algorithm with well-known modularity-based clustering algorithm CNM. We successfully detect functional groups that are annotated with putative Gene Ontology (GO) terms. Top-10 clusters with minimum p-value theoretically prove that newly proposed algorithm partitions network more accurately then CNM. Furthermore, manual interpretations of functional groups found by SCAN show superior performance over CNM.

Analyzing Organizational Structures Using Social Network Analysis

Technological changes have aided modern companies to gather enormous amounts of data electronically. The availability of electronic data has exploded within the past decade as communication technologies and storage capacities have grown tremendously. The need to analyze this collected data for creating business intelligence and value continues to grow rapidly as more and more apparently unbiased information can be extracted from these data sets. In this paper we focus in particular, on email corpuses, from which a great deal of information can be discerned about organization structure and their unique cultures. We hypothesize that a broad based analysis of information exchanges (ex. emails) among a company’s employees could give us deep information about their respective roles within the organization, thereby revealing hidden organizational structures that hold immense intrinsic value. Enron email corpus is used as a case study to predict the unknown status of Enron employees and identify homogeneous groups of employees and hierarchy among them within Enron organization. We achieve this by using classification and cluster techniques. As a part of this work, we have also developed a web-based graphical user interface to work with feature extraction and composition.

Knowledge Discovery in Textual Databases: A Concept-Association Mining Approach

The number of scientific publications is exploding as online digital libraries and the World Wide Web grow. MEDLINE, the premier bibliographic database of the National Library of Medicine (NLM) , contains about 18 million records from more than 7,300 different publications dating from 1965; it is growing by about 400,000 citations each year. The explosive growth of information in textual documents creates great need for techniques for knowledge discovery from text collections.

On Structural Analysis of Large Networks

Containing much valuable information, networks such as the World Wide Web, social networks and metabolic networks draw increasingly attention in scientific communities. Network clustering (or graph partitioning) is the discovery of underlying clusters of related vertices in networks. But beyond organizing vertices into clusters of peers is the question of what role each vertex play in the network. This paper presents some new ways of uncovering underlying structures, including the roles that vertices play in the network. Identifying vertex roles is useful for applications such as viral marketing and epidemiology. For example, hubs are responsible for spreading ideas or disease. We applied our algorithm to analyze some real networks. The results demonstrate a superior performance over other methods such as modularity-based algorithms.

Supporting keyword queries on structured databases with limited search interfaces

Many Web sources provide forms to allow users to query their hidden data. For instance, online stores such as Amazon.com have search interfaces, using which users can query information about books by providing conditions on attributes of title, author, and publisher. We propose a novel system framework that supports keyword queries on structured data behind such limited search forms. It provides user-friendly query interfaces for users to type in IR-style keyword queries to find relevant records. We study research challenges in the framework and conduct extensive experiments on real datasets to show the practicality of our framework and evaluate different algorithms.