Search

Complex networks have been widely used to model complex systems composed of interacting entities. It is possible to recommend new relationships between entities according to the network topology and the entities' properties. The likelihood of forming a missing or potential relationship is often captured by similarity measures. As community clusters are also based on the entities' similarities, traditional link recommendation/prediction methods naturally tend to recommend links within a community. However, potential links valuable across communities are often overlooked and may cause the problem of information cocoons. We focus on link recommendation across communities based on homogeneous and heterogeneous information networks, which aims to improve the diversity of recommender systems. We propose to solve this problem using a novel similarity calculation and heterogeneous network embedding methods. Our comprehensive experiments on real-world datasets and synthetic datasets show that our methods strike a good balance between accuracy and efficiency, while generating valuable unconventional recommendations in practical application scenarios. [ABSTRACT FROM AUTHOR]

Currently, graphs serve as fundamental data structures for many applications, such as road networks, social and communication networks, and web requests. In many applications, graph edges stream in and users are only interested in the recent data. In data exploration, the storage and processing of such massive amounts of graph stream data has become a significant problem. As the categorical attributes of vertices and edges are often referred to as labels, we propose a labeled graph sketch that stores real-time graph structural information using only sublinear space and that supports graph queries of diverse types. This sketch also works for sliding-window queries. We conduct extensive experiments on real-world datasets in six different domains and compare the results with a state-of-the-art method to show the accuracy, efficiency, and practicability of our proposed approach. [ABSTRACT FROM AUTHOR]

Abstract Many big data applications today require querying highly dynamic and large-scale data streams to find the top- k most frequent items in the most recent window of a specified size at a specific time. This is a challenging problem. We propose a novel approach called Floating Top-k. Our algorithm does not need to explicitly maintain any item counts over time or deal with count updates upon item entry and expiration. Succinctly , we use only a small-size data structure to retrieve the top- k items dynamically in a window of any specified size within an upper bound. We prove that the space and time costs of Floating Top- k grow only logarithmically with the window size rather than the linear growth of previous methods. Our comprehensive experiments using three real-world datasets show that Floating Top- k not only provides accuracy guarantees but also has a memory footprint two to three orders of magnitude smaller and is one to two orders of magnitude faster than previous approaches. Hence, Floating Top- k is both effective and scalable, significantly outperforming competing methods. In addition, we devise a concise and efficient solution called Progressive Trend Model to address a related problem of tracking the frequency of selected items, improving upon previous methods by 20 to 30 times in model conciseness while maintaining the same accuracy and efficiency. [ABSTRACT FROM AUTHOR]