Performance of community detection algorithms supported by node embeddings
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Abstract The grouping of nodes into subsets that are relatively densely interconnected and separable from the rest of the network is a property often displayed in many complex real-world networks; this feature is known as a community structure. There is a growing demand for algorithms that can find partitions that resemble the community structure of a given network as closely as possible. However, most popular algorithms for community detection in graphs have one serious drawback, namely, they are heuristic-based and in many cases are unable to find a near-optimal solution. Moreover, their results are volatile, impacting the replicability of their results. In this paper, we investigate if the performance of greedy algorithms might be improved by initialising such algorithms with some carefully chosen partition of nodes, namely a partition obtained by embedding the nodes into real numbers space and then running a clustering algorithm on this latent representation. We believe that embedding will filter unwanted noise while retaining the proximity of nodes belonging to the same community or will learn more complex and elusive relations between nodes. Then, clustering algorithms run on this embedding will create a stable partitioning that will reduce the uncertainty in the initial phases of the community detection algorithms. The experiments show that the proposed procedure significantly improves the results over baseline community detection algorithms, namely Louvain and Leiden. It also reduces the inherent volatility of such algorithms. The impact depends on the given graph’s properties, especially the strength of the community structure and degree distribution. The largest boost in performance is given in the cases when networks are ‘noisier’, that is, when the community structure is less pronounced and there are many connections between communities. Furthermore, the design and parametrization of the procedure depend on the network’s topology, not on the size of the network itself.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it