Detecting Community Structure by Using a Constrained Label Propagation Algorithm

Differences in the distribution of triggers among resting state networks in patients with juvenile myoclonic epilepsy explained by network analysis

Background

Juvenile myoclonus epilepsy (JME) is an idiopathic generalized epilepsy syndrome. Functional connectivity studies based on graph theory have demonstrated changes in functional connectivity among different brain regions in patients with JME and healthy controls. However, previous studies have not been able to clarify why visual stimulation or increased cognitive load induces epilepsy symptoms in only some patients with JME.

Methods

This study constructed a small-world network for the visualization of functional connectivity of brain regions in patients with JME, based on system mapping. We used the node reduction method repeatedly to identify the core nodes of the resting brain network of patients with JME. Thereafter, a functional connectivity network of the core brain regions in patients with JME was established, and it was analyzed manually with white matter tracks restriction to explain the differences in symptom distribution in patients with JME.

Results

Patients with JME had 21 different functional connections in their resting state, and no significant differences in their distribution were noted. The thalamus, cerebellum, basal ganglia, supplementary motor area, visual cortex, and prefrontal lobe were the core brain regions that comprised the functional connectivity network in patients with JME during their resting state. The betweenness centrality of the prefrontal lobe and the visual cortex in the core functional connectivity network of patients with JME was lower than that of the other brain regions.

Conclusion

The functional connectivity and node importance of brain regions of patients with JME changed dynamically in the resting state. Abnormal discharges originating from the thalamus, cerebellum, basal ganglia, supplementary motor area, visual cortex, and prefrontal cortex are most likely to lead to seizures in patients with JME. Further, the low average value of betweenness centrality of the prefrontal and visual cortices explains why visual stimulation or increased cognitive load can induce epileptic symptoms in only some patients with JME.

Disambiguation of author entities in ADS using supervised learning and graph theory methods

Disambiguation of authors in digital libraries is essential for many tasks, including efficient bibliographical searches and scientometric analyses to the level of individuals. The question of how to link documents written by the same person has been given much attention by academic publishers and information retrieval researchers alike. Usual approaches rely on publications’ metadata such as affiliations, email addresses, co-authors, or scholarly topics. Lack of homogeneity in the structure of bibliographic collections and discipline-specific dissimilarities between them make the creation of general-purpose disambiguators arduous. We present an algorithm to disambiguate authorships in the Astrophysics Data System (ADS) following an established semi-supervised approach of training a classifier on authorship pairs and clustering the resulting graphs. Due to the lack of high-signal features such as email addresses and citations, we engineer additional content- and location-based features via text embeddings and named-entity recognition. We train various nonlinear tree-based classifiers and detect communities from the resulting weighted graphs through label propagation, a fast yet efficient algorithm that requires no tuning. The resulting procedure reaches reasonable complexity and offers possibilities for interpretation. We apply our method to the creation of author entities in a recent ADS snapshot. The algorithm is evaluated on 39 manually-labeled author blocks comprising 9545 authorships from 562 author profiles. Our best approach utilizes the Random Forest classifier and yields a micro- and macro-averaged BCubed $$\mathrm {F}_1$$ F 1 score of 0.95 and 0.87, respectively. We release our code and labeled data publicly to foster the development of further disambiguation procedures for ADS.

Modeling with Node Popularities for Autonomous Overlapping Community Detection

Overlapping community detection has triggered recent research in network analysis. One of the promising techniques for finding overlapping communities is the popular stochastic models, which, unfortunately, have some common drawbacks. They do not support an important observation that highly connected nodes are more likely to reside in the overlapping regions of communities in the network. These methods are in essence not truly unsupervised, since they require a threshold on probabilistic memberships to derive overlapping structures and need the number of communities to be specified a priori . We develop a new method to address these issues for overlapping community detection. We first present a stochastic model to accommodate the relative importance and the expected degree of every node in each community. We then infer every overlapping community by ranking the nodes according to their importance. Second, we determine the number of communities under the Bayesian framework. We evaluate our method and compare it with five state-of-the-art methods. The results demonstrate the superior performance of our method. We also apply this new method to two applications, showing its superb performance on practical problems.

Overcoming randomness does not rule out the importance of inherent randomness for functionality

Randomness is intrinsic to many natural processes. It is also clear that, under certain conditions, disorders are not associated with functionality. Several examples in which overcoming, suppressing, or combining both randomness and non-randomness is required are drawn from various fields. However, the need to suppress or overcome randomness does not negate its importance under certain conditions and its significance in valid processes and organ functions. Randomness should be acknowledged rather than ignored or suppressed; it can be viewed, at worst, as a disturbing disorder that may be treated to produce order, or, at best, as a 'beneficial disorder' that can be considered as a higher level of functionality.

Voting Simulation based Agglomerative Hierarchical Method for Network Community Detection

Community detection has been paid much attention in many fields in recent years, and a great deal of community-detection methods have been proposed. But the time consumption of some of them is heavy, limiting them from being applied to large-scale networks. On the contrary, there exist some lower-time-complexity methods. But most of them are non-deterministic, meaning that running the same method many times may yield different results from the same network, which reduces their practical utility greatly in real-world applications. To solve these problems, we propose a community-detection method in this paper, which takes both the quality of the results and the efficiency of the detecting procedure into account. Moreover, it is a deterministic method which can extract definite community structures from networks. The proposed method is inspired by the voting behaviours in election activities in the social society, in which we first simulate the voting procedure on the network. Every vertex votes for the nominated candidates following the proposed voting principles, densely connected groups of vertices can quickly reach a consensus on their candidates. At the end of this procedure, candidates and their own voters form a group of clusters. Then, we take the clusters as initial communities, and agglomerate some of them into larger ones with high efficiency to obtain the resulting community structures. We conducted extensive experiments on some artificial networks and real-world networks, the experimental results show that our proposed method can efficiently extract high-quality community structures from networks, and outperform the comparison algorithms significantly.

Micro-blog user community discovery using generalized SimRank edge weighting method

Community discovery is one of the most popular issues in analyzing and understanding a network. Previous research suggests that the discovery can be enhanced by assigning weights to the edges of the network. This paper proposes a novel edge weighting method, which balances both local and global weighting based on the idea of shared neighbor ranging between users and the interpersonal significance of the social network community. We assume that users belonging to the same community have similar relationship network structures. By controlling the measure of "neighborhood", this method can adequately adapt to real-world networks. Therefore, the famous similarity calculation method-SimRank-can be regarded as a special case of our method. According to the practical significance of social networks, we propose a new evaluation method that uses the communication rate to measure its divided demerit to better express users' interaction relations than the ordinary modularity Q. Furthermore, the fast Newman algorithm is extended to weighted networks. In addition, we use four real networks in the largest Chinese micro-blog website Sina. The results of experiments demonstrate that the proposed method easily meets the balancing requirements and is more robust to different kinds of networks. The experimental results also indicate that the proposed algorithm outperforms several conventional weighting methods.

A semi-synchronous label propagation algorithm with constraints for community detection in complex networks

Community structure is an important feature of a complex network, where detection of the community structure can shed some light on the properties of such a complex network. Amongst the proposed community detection methods, the label propagation algorithm (LPA) emerges as an effective detection method due to its time efficiency. Despite this advantage in computational time, the performance of LPA is affected by randomness in the algorithm. A modified LPA, called CLPA-GNR, was proposed recently and it succeeded in handling the randomness issues in the LPA. However, it did not remove the tendency for trivial detection in networks with a weak community structure. In this paper, an improved CLPA-GNR is therefore proposed. In the new algorithm, the unassigned and assigned nodes are updated synchronously while the assigned nodes are updated asynchronously. A similarity score, based on the Sørensen-Dice index, is implemented to detect the initial communities and for breaking ties during the propagation process. Constraints are utilised during the label propagation and community merging processes. The performance of the proposed algorithm is evaluated on various benchmark and real-world networks. We find that it is able to avoid trivial detection while showing substantial improvement in the quality of detection.