Mining exceptional closed patterns in attributed graphs

Abstract

Geo-located social media provide a large amount of information describing urban areas based on user descriptions and comments. Such data makes possible to identify meaningful city neighborhoods on the basis of the footprints left by a large and diverse population that uses this type of media. In this paper, we present some methods to exhibit the predominant activities and their associated urban areas to automatically describe a whole city. Based on a suitably attributed graph model, our approach identifies neighborhoods with homogeneous and exceptional characteristics. We introduce the novel problem of exceptional subgraph mining in attributed graphs and propose a complete algorithm that takes benefits from closure operators, new upper bounds and pruning properties. We also define an approach to sample the space of closed exceptional subgraphs within a given time-budget. Experiments performed on 10 real datasets are reported and demonstrate the relevancy of both approaches, and also show their limits.

SIAS-miner: mining subjectively interesting attributed subgraphs

“…As stated in Section 6, there is no approach that supports the discovery of subjectively interesting attributed subgraphs in the literature. The closest method to SIAS-Miner-Enum is Cenergetics [2] that aims at discovering closed exceptional attributed subgraphs involving overrepresented and/or underrepresented attributes, and which mined the London graph used here in the experiments (and on similar graphs of other cities). It assesses exceptionality with the weighted relative accuracy (WRAcc) measure that accounts for margins but cannot account for other prior knowledge.…”

“…Few works in this direction exist. For example Atzmueller et al [1] study mining communities (densely connected subgraphs) that can also be described well in terms of attribute values, while Bendimerad et al [2] look for exceptional subgraphs that are connected. Various quality measures are used in the first work and the second relies on Weighted Relative Accuracy.…”

Exceptional Attributed Subgraph Mining to Understand the Olfactory Percept

“…Condition (3) assesses the exceptionality of the attributes, while condition (4) enforces the subgraph to be cohesive. Such patterns can be mined using the algorithm presented in [4], originally designed to exhibit the predominant activities and their associated urban areas in graphs that model urban areas. The algorithm, named Cenergetics, mines exceptional subgraph in attributed graphs.…”

“…As mentioned in the previous section, exceptional attributed subgraph definition can be extended to make possible the discovery of subgraphs whose attribute values are lower than what observed on the rest of the graphs. For mean and rank modelings, we just adapted the WRAcc measure to catch under-represented attributes [4]. In the pairwise inequality based modeling, we consider all pairs of attributes, so it is not necessary to consider under-represented attribute values.…”

“…To achieve this goal, we propose to model fMRI images as an attributed graph where the vertices are the voxels (brain unit), the edges encode the adjacency relationship, and vertex attributes stand for the hemodynamic response to an odor. We propose to analyze such a graph with Cenergetics [4] which makes possible to identify brain areas with exceptional hemodynamic response in some experimental settings.…”