Search

We propose a definition for frequent approximate patterns in order to model important subgraphs in a graph database with incomplete or inaccurate information. By our definition, frequent approximate patterns possess three main properties: possible absence of exact match, maximal representation, and the Apriori Property. Since approximation increases the number of frequent patterns, we present a novel randomized algorithm (called RAM) using feature retrieval. A large number of real and synthetic data sets are used to demonstrate the effectiveness and efficiency of the frequent approximate graph pattern model and the RAM algorithm. [ABSTRACT FROM AUTHOR]

Most of today's structured data is stored in relational data- bases. Such a database consists of multiple relations that are linked together conceptually via entity-relationship links in the design of relational database schemas. Multi-relational classification can be widely used in many disciplines including financial decision making and medical research. However, most classification approaches only work on single "flat" data relations. It is usually difficult to convert multiple relations into a single flat relation without either introducing huge "universal relation" or losing essential information. Previous works using Inductive Logic Programming approaches (recently also known as Relational Mining) have proven effective with high accuracy in multi-relational classification. Unfortunately, they fail to achieve high scalability w.r.t. the number of relations in databases because they repeatedly join different relations to search for good literals. In this paper we propose CrossMine, an efficient and scalable approach for multi-relational classification. CrossMine employs tuple ID propagation, a novel method for virtually joining relations, which enables flexible and efficient search among multiple relations. CrossMine also uses aggregated information to provide essential statistics for classification. A selective sampling method is used to achieve high scalability w.r.t. the number of tuples in the databases. Our comprehensive experiments on both real and synthetic databases demonstrate the high scalability and accuracy of CrossMine. [ABSTRACT FROM AUTHOR]

Mobile objects have become ubiquitous in our everyday lives, ranging from cellular phones to sensors, therefore, analyzing and mining mobile data becomes an interesting problem with great practical importance. For instance, by finding trajectory patterns of the mobile clients, the mobile communication network can allocate resources more efficiently. However, due to the limited power of the mobile devices, we are only able to obtain the imprecise location of a mobile object at a given time. Sequential patterns are a popular data mining model. By applying the sequential pattern model on the set of imprecise trajectories of the mobile objects, we may uncover important information or further our understanding of the inherent characteristics of the mobile objects, e.g., constructing a classifier based on the discovered patterns or using the patterns to improve the accuracy of location prediction. Since the input data is highly imprecise, it may not be possible to directly apply any existing sequential pattern discovery algorithm to the problem in this paper. Thus, we propose the model of the trajectory patterns and a novel measure to represent the expected occurrences of a pattern in a set of imprecise trajectories. The concept of pattern groups is introduced to present the trajectory patterns in a concise manner. Since the Apriori property no longer holds on the trajectory patterns, a new min-max property is identified and a novel TrajPattern algorithm is devised based on the newly discovered property. Last but not least, we apply the TrajPattern algorithm on a wide range of real and synthetic data sets to demonstrate the usefulness, efficiency, and scalability of this approach. [ABSTRACT FROM AUTHOR]

Efficiently finding the most relevant publications in large corpus is an important research topic in information retrieval. The number of biological literatures grows exponentially in various publication databases. The objective of this paper is to quickly identify useful publications from a large number of biological documents. In this paper, we introduce a new iterative search paradigm that integrates biomedical background knowledge in organizing the results returned by search engines and utilizes user feedbacks in pruning irrelevant documents by document classification. A new term weighting strategy based on Gene Ontology is proposed to represent biomedical literatures. A prototype text retrieval system is built on this iterative search approach. Experimental results on MEDLINE abstracts and different keyword inputs show that the system can filter a large number of irrelevant documents in a reasonable time while keeping most of the useful documents. The results also show that the system is robust against different inputs and parameter settings. [ABSTRACT FROM AUTHOR]

Microarrays are one of the latest breakthroughs in experimental molecular biology. Thousands of different research groups generate tens of thousands of microarray gene expression profiles based on different tissues, species, and conditions. Combining such vast amount of microarray data sets is an important and yet challenging problem. In this paper, we introduce a "correlation signature" method that allows the coherent interpretation and integration of microarray data across disparate sources. The proposed algorithm first builds, for each gene (row) in a table, a correlation signature that captures the system-wide dependencies existing between the gene and the other genes within the table, and then compares the signatures across the tables for further analysis. We validate our framework with an experimental study using real microarray data sets, the result of which suggests that such an approach can be a viable solution for the microarray data integration and analysis problems. [ABSTRACT FROM AUTHOR]