An Optimization-based Approach To Node Role Discovery in Networks: Approximating Equitable Partitions

Similar to community detection, partitioning the nodes of a network according to their structural roles aims to identify fundamental building blocks of a network. The found partitions can be used, e.g., to simplify descriptions of the network connectivity, to derive reduced order models for dynamical processes unfolding on processes, or as ingredients for various graph mining tasks. In this work, we offer a fresh look on the problem of role extraction and its differences to community detection and present a definition of node roles related to graph-isomorphism tests, the Weisfeiler-Leman algorithm and equitable partitions. We study two associated optimization problems (cost functions) grounded in ideas from graph isomorphism testing, and present theoretical guarantees associated to the solutions of these problems. Finally, we validate our approach via a novel"role-infused partition benchmark", a network model from which we can sample networks in which nodes are endowed with different roles in a stochastic way.

Paper

References (52)

Scroll for more · 38 remaining

Similar papers

Peer review

Reviewer rRxj5/10 · confidence 3/52023-07-06

Summary

The paper offers a new perspective on the problem of role extraction while defining node roles based on the ideas of equitable partitions and and graph-isomorphism tests, the Weisfeiler-Leman algorithm. The paper studies two associated optimization problems (cost functions) inspired by graph isomorphism testing and provides theoretical guarantees for their solutions. To validate the approach, a novel "role-infused partition benchmark" is introduced, enabling the generation of networks where nodes possess different roles in a stochastic manner.

Strengths

(+) The paper offers a new perspective on node role extraction, focusing on equitable partitions and providing a principled stochastic notion of node roles. This approach adds to the existing methods in the field and offers a new way to define and quantify node roles in complex networks. (+) The paper presents a quantitative definition of node roles based on equitable partitions, allowing for the numerical measurement of deviation from an exact equivalence. This provides a more nuanced understanding of node roles compared to traditional definitions. (+) The technical aspects of the paper are thorough and detailed. Also, the technical details aeem to be correct. (+) Numerical experiments show the effectiveness and superiority of the proposed method over a graph neural network, GIN, for three different tasks.

Weaknesses

(-) While the paper briefly mentions different categories of node role extraction approaches (graphlet-based, walk-based, and matrix-factorization-based), it does not provide a detailed comparison or analysis of how the proposed equitable partitions approach compares to these existing methods. A more rigorous comparative analysis, including performance metrics and evaluation on benchmark datasets, would strengthen the paper's contribution and demonstrate the advantages of the proposed approach. (-) The idea of Equitable partitions for node role discovery is interesting. However, the details regarding "why this approach" makes sense is missing. There can be other ways too? Why this method should work? At a conceptual level, a more rigorous explanation related to the definition of roles based on equitable partition is missing. I think this is crucially important. In other words, the paper focuses on providing details of "designing" and approach based on EP, while leaving details of "why this is appropriate" (-) A more thorough literature review might be needed. For instance, the following paper provides a nice simple algorithm for computing equitable partitions. (I am not sure it is better or worse; however it might be handy to have a look at it.) - Zhang et al., "Upper and Lower Bounds for Controllable Subspaces of Networks of Diffusively Coupled Agents," IEEE Transactions on Automatic control, 2014. Also, there is some recent work on counting subgraphs (that may define node roles based on the structural attributes of the graph), for instance, - Hassan et al., "Computing Graph Descriptors on Edge Streams." ACM Transactions on Knowledge Discovery from Data, 2023. (-) The paper introduces a family of cost functions to assess the quality of a role partitioning. However, it does not thoroughly discuss the selection, design, or properties of these cost functions. Providing a more in-depth exploration and analysis of the different cost functions, their properties, and how they relate to the problem of node role extraction would enhance the technical understanding of the proposed approach. (-) It is unclear how the ground truth for the sampled adjacency matrices is computed in Experiment 1 of Section 5. Moreover, GIN is a relatively old graph neural network. There are recent methods that show better results on several downstream tasks and could have been considered for baseline comparisons. Some of these works include the work of Bouritsas et al, and Ramp\'a\v{s}ek et al, below. - Bouritsas et al., "Improving graph neural network expressivity via subgraph isomorphism counting," IEEE Transactions on Pattern Analysis and Machine Intelligence. 2022 Feb 24;45(1):657-68. - Ramp\'a\v{s}ek, et al., "Recipe for a general, powerful, scalable graph transformer," Advances in Neural Information Processing Systems. 2022.

Questions

(-) It is mentioned that Equitable partitions imply an upper bound on the expressivity of Graph Neural Networks while citing references [26] and [43]. There is nothing of that sort in these papers. Could you point out where such results are mentioned in the said references. (-) As coarsest equitable partition is an important concept for the paper, an example to illustrate this concept or a proper reference to another paper is vital and missing in the draft. (-) As mentioned in the "weakness", can you explain more rigorously the motivation for using equitable partitions for node role discovery. (-) In Theorem 3, what is the interpretation of the parameter $\delta$ and what is the computational complexity to calculate it? (A minor comment) There are minor grammatical errors, e.g., - "Thus, from any node $v$ within in the same class ..." (in is not needed). - "This also allows for Gradient Descent approaches like e.g. GNNs." (like and e.g., are used simultaneously).

Rating

5: Borderline accept: Technically solid paper where reasons to accept outweigh reasons to reject, e.g., limited evaluation. Please use sparingly.

Confidence

3: You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.

Soundness

3 good

Presentation

2 fair

Contribution

3 good

Limitations

The limitations and possibly approaches to tackle these limitations are presented in the paper.

Reviewer nMtV5/10 · confidence 3/52023-07-06

Summary

This paper presents a novel perspective on the problem of node role extraction in complex networks, highlighting its distinctions from community detection. The authors propose a definition of node roles and introduce two optimization problems based on graph-isomorphism tests, the Weisfeiler-Leman algorithm, and equitable partitions. Theoretical guarantees are provided, and the approach is validated using a newly introduced "role-infused partition benchmark" that allows for stochastic assignment of different roles to nodes in sampled networks. The findings contribute to network analysis, graph mining, and the development of reduced order models for dynamical processes on networks.

Strengths

- The research topic, role discovery in graphs, is very important and is worth studying, especially compared to the community detection problem, role discovery is also important but lacks attention. - The design of the model based on equitable partitions is technically sound and theoretically guaranteed. - Experimental results on several datasets from different perspectives show the effectiveness of the proposed method.

Weaknesses

- Baseline selection. Some relevant and/or important baselines have not been compared. - Downstream tasks. The proposed method is not tested on widely used downstream tasks to show its effectiveness. - Dataset. There are some more widely used datasets for role discovery that are not used in this paper.

Questions

Most of my questions are related to the experiments: - There are some relevant baselines that are not compared, for example, MMSB [1] and RIDRs [2]. The first one you can also discuss the relation with your proposed method, e.g., if it is possible to extend the proposed method to soft role assignment like MMSB. The second method is also based on equitable partitions. - There are some commonly used datasets that can be used for evaluating role discovery, for example, these datasets used in [3]. - In previous role discovery studies, different downstream tasks have been selected to show the effectiveness of the discovered roles, e.g., node classification and link prediction. For example, these tasks in RolX [4] and struc2vec [3]. Also, there is no evaluation of the discovered roles. For example, in RolX, the quantitative evaluation using clustering metric has been used. - Efficiency issue. Although the proposed method is technically sound, a question is how efficient the proposed method is, e.g., how large a graph this method can handle. [1] Airoldi E M, Blei D, Fienberg S, et al. Mixed membership stochastic blockmodels[J]. Advances in neural information processing systems, 2008, 21. [2] Gupte P V, Ravindran B, Parthasarathy S. Role discovery in graphs using global features: Algorithms, applications and a novel evaluation strategy[C]//2017 IEEE 33rd International Conference on Data Engineering (ICDE). IEEE, 2017: 771-782. [3] Henderson K, Gallagher B, Eliassi-Rad T, et al. Rolx: structural role extraction & mining in large graphs[C]//Proceedings of the 18th ACM SIGKDD international conference on Knowledge discovery and data mining. 2012: 1231-1239. [4] Ribeiro L F R, Saverese P H P, Figueiredo D R. struc2vec: Learning node representations from structural identity[C]//Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining. 2017: 385-394. ============================================== After the rebuttal, I increased my overall rating.

Rating

5: Borderline accept: Technically solid paper where reasons to accept outweigh reasons to reject, e.g., limited evaluation. Please use sparingly.

Confidence

3: You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.

Soundness

2 fair

Presentation

3 good

Contribution

3 good

Limitations

No potential negative societal impact.

Reviewer WiNY6/10 · confidence 4/52023-07-06

Summary

The paper considers a relaxed definition of the coarsest equitable partition (CEP), which equals the final partition of Weisfeiler-Leman or the color refinement algorithm in the original version. The authors allow to specify the number of cells of the partition and derive a related optimization problem. From this, an algorithm for role discovery is derived, outperforming techniques such as role2vec on real-world graphs.

Strengths

* The concept of structural roles is closely related to the Weisfeiler-Leman algorithm and fundamental for graph learning. * The relaxation of the final WL partition with a fixed number of cells is novel. * The RIP model for graph generation is innovative.

Weaknesses

* Algorithm 1 is not analyzed sufficiently. It would be interesting to have approximation guarantees. Moreover, it would be nice to investigate the case where $k$ equals the number of cells in the cEP. I believe that the current version would not guarantee that the cEP is found since the cluster method obtains only $X$ as an argument neglecting $H$, which should be required to guarantee this. Is this correct? * Several other methods compute relaxed WL partitions without allowing to specify $k$ and are discussed in section 2.1. These could also be used in the experimental comparison to assess the advantages of the proposed method.

Questions

* Could you explain whether Algorithm 1 can recover the cEP for a suitable choice of the parameter $k$?

Rating

6: Weak Accept: Technically solid, moderate-to-high impact paper, with no major concerns with respect to evaluation, resources, reproducibility, ethical considerations.

Confidence

4: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.

Soundness

3 good

Presentation

3 good

Contribution

2 fair

Limitations

Limitations are sufficiently discussed.

Reviewer ZAin6/10 · confidence 3/52023-07-07

Summary

The authors propose the notion of equitable partitions from graph isomorphism literature in order to partition the nodes of a network according to their structural roles. They study two optimization problems for approximately recovering such equitable partitions. Analogous to the SBM model, the authors devise the RIP (role-infused partition) stochastic model and validate their findings on this model.

Strengths

The paper is very well-written. The proposed cost functions for computing approximate equitable partitions are very well-motivated and natural. Approximate versions of the Weisfeiler-Leman algorithm is an important question which deserves much study. The proposed RIP model is a very satisfcatory benchmark for role discovery.

Weaknesses

The experimental work is inadequate in order to understand the impact of the proposed formalism to real-world data (Section 5, Experiment 3). The considered datasets are too few, specialized and small-sized in order to deduce that approximate equitable partitions (and approximate Weisfeiler-Leman) is at the core of role discovery.

Questions

1) On the considered molecular datasets, does the approximate number of equitable partitions differ (the parameter "k" \in {2,20}) differ significantly from the actual number of equitable partitions, on an average? Since the average number of nodes is 30, and since I would expect most graphs to have discrete equitable partitions (i.e. one node per class), it will be interesting to know if the optimal values of k were indeed small (say 2-6).

Rating

6: Weak Accept: Technically solid, moderate-to-high impact paper, with no major concerns with respect to evaluation, resources, reproducibility, ethical considerations.

Confidence

3: You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.

Soundness

3 good

Presentation

4 excellent

Contribution

3 good

Limitations

Yes.

Reviewer nMtV2023-08-15

Thanks for the responses

Dear authors, Thanks for your efforts on the rebuttal. These answers addressed my concerns. I thereby increase my ratings.

Authorsrebuttal2023-08-17

Thanks for your response

Dear Reviewer nMtV, Thanks for your response! Cheers, the authors

Reviewer WiNY2023-08-16

Thanks for answering my question. The link to the adapted WL in [1] was very helpful. Please consider addressing this question also in the paper. I have increased my score accordingly.

Authorsrebuttal2023-08-17

Thanks for your response

Dear reviewer WiNY, Thanks for your response! If space permits, we will add this perspective on our algorithm to the final version of our paper. Best, the authors

Reviewer ZAin2023-08-19

Thanks for your response. After going through the rebuttal and other responses, I have decided to maintain my original score.

Program Chairsdecision2023-09-21

Decision

Accept (poster)

© 2026 NYSGPT2525 LLC