PaperlayerView exact revision on arXiv

Source-linked AI summary

Andrzej Pekalski networks of scientific interests with internal degrees of freedom through self-citation analysis

M. Ausloos, R. Lambiotte, A. Scharnhorst, I. Hellsten

arXiv:0710.1800v1physics.soc-phphysics.hist-ph

TL;DR

The paper asks whether APE’s self-citations can reveal changing research interests and collaboration structure, rather than being treated as bibliometric noise. It analyzes APE’s self-citation network with co-authorship and topic information using Optimal Percolation Method, finding separated clusters and temporal patterns associated with his scientific trajectory.

  • Problem

    The paper asks whether self-citation networks can reveal a scientist’s research topics, collaborations, and field mobility.

    Source: §I · pp. 1–2

  • Method

    The method analyzes self-citations with co-authorship and article-topic information, using Optimal Percolation Method to identify network substructures.

    Source: §II · p. 2

  • Results

    APE’s self-citation network forms three separated clusters, revealing patterns in research interests and collaboration over time.

    Source: A. Basic Statistics and Clusters · p. 3, §IV · p. 5

  • Takeaways & Limitations

    Self-citation patterns can provide information about an author’s scientific interests, field mobility, and collaboration networks.

    Source: §I · p. 1, §IV · p. 5

Abstract

from arXiv · show

Old and recent theoretical works by Andrzej Pekalski (APE) are recalled as possible sources of interest for describing network formation and clustering in complex (scientific) communities, through self-organisation and percolation processes. Emphasis is placed on APE self-citation network over four decades. The method is that used for detecting scientists field mobility by focusing on author's self-citation, co-authorships and article topics networks as in [1,2]. It is shown that APE's self-citation patterns reveal important information on APE interest for research topics over time as well as APE engagement on different scientific topics and in different networks of collaboration. Its interesting complexity results from "degrees of freedom" and external fields leading to so called internal shock resistance. It is found that APE network of scientific interests belongs to independent clusters and occurs through rare or drastic events as in irreversible "preferential attachment processes", similar to those found in usual mechanics and thermodynamics phase transitions.

I. INTRODUCTION

The paper treats APE’s self-citations as informative evidence of changing research interests, collaborations, and field mobility rather than noise. It asks how self-citation structures relate to topics, co-authorship, and career evolution.

I. INTRODUCTION

  • Self-citations are analyzed as information about APE’s research development, rather than discarded as noise or treated only as bias.The study explicitly avoids evaluating work quality, external citations, or impact factors.Source: §I · pp. 1–2
  • APE’s publication history spans interconnected work on statistical mechanics, diffusion, biophysics, population dynamics, and socio-economic physics.The recalled trajectory includes random spin systems, magnetic lattice gases, oxygen diffusion, predator-prey models, population evolution, wealth dynamics, and prison riots.Source: §I · p. 1
  • The study uses self-citations together with co-authorships and article keywords to trace scientists’ mobility across research fields.The approach draws on a de-clustering method rooted in percolation theory.Source: §I · p. 2
  • The research asks which structures APE’s self-citation network contains, whether they correspond to research topics, and whether they reveal field mobility.It also asks whether changing co-authorships track changes in self-citation networks and scientific focus.Source: §I · p. 2
  • The paper situates APE’s record alongside studies of other scientists, including Werner Ebeling, while focusing specifically on APE’s scientific journey.Both scientists are described as having led theoretical-physics schools and changing collaborator networks, but their publication records differ.Source: §I · p. 2

II. DATA AND METHOD

The method constructs APE’s self-citation network from Web of Science records and detects article subgroups using Optimal Percolation Method. The analysis combines network structure, clustering, co-authorship, and temporal development.

II. DATA AND METHOD

  • The dataset can omit relevant papers because they were not indexed in Web of Science or because author names were misspelled or spelled differently.The authors note that missing reference lists and name variation can produce disagreement between publication records.Source: §II · p. 2
  • The empirical basis is 95 Web of Science articles, excluding non-ISI journal papers, unpublished work, books, and book chapters.Self-citations are papers by APE that cite other papers co-authored by APE, without normalization by publication or reference counts.Source: §II · p. 2
  • OPM differs from link-removal percolation methods because it operates on unweighted networks and removes nodes rather than links.PIBM instead applies to weighted correlation matrices and removes links according to their values.Source: §II · p. 3
  • Optimal Percolation Method identifies network substructures by removing k nodes that minimize connectivity, where c = N(N − 1)/2 for a connected network.The removed nodes define intersections between subgroups, which are then displayed in different tones.Source: §II · pp. 2–3
  • The analysis defines clusters as groups of articles more strongly linked to one another than to the rest of the network.This clustering is interpreted as indicating that articles within a subgroup share something in common.Source: §II · p. 3
  • The workflow examines publication frequency, the overall self-citation network, OPM-derived structures, co-authorships, and article types by year and detected topic.These five stages are intended to support a combined view of network organization and scientific development.Source: §II · p. 3

A. Basic Statistics and Clusters

APE’s publication and self-citation records reveal three largely distinct research-field clusters, with limited overlap among collaborators and keywords. The clusters correspond to different theoretical topics and provide a basis for tracing his scientific journey.

Basic publication statistics

  • APE published continuously over four decades, averaging about 3 papers per year with yearly output ranging from 0 to 9 and accelerating after 1980.Productivity also increased toward the end of the century, although the supplied figures provide the record counts rather than additional yearly values.Source: A. Basic Statistics and Clusters · p. 3, Figure 2 · p. 3

Co-authorship structure

  • The collaborator network comprises 11 co-authors overall, while 20 single-authored papers remain persistent over time as the co-author network grows.The collaborator list includes local students, senior colleagues, and two members of the Liège group.Source: A. Basic Statistics and Clusters · pp. 3–4

Self-citation clusters

  • The self-citation network divides into three subgroups containing 9, 7, and 32 nodes, with 18, 19, and 68 citation links, respectively.These clusters combine articles from different years and should not be interpreted as evolutionary trees.Source: A. Basic Statistics and Clusters · p. 3

Research-field interpretation

  • Only a few keywords overlap among clusters, supporting their use for demarcating research fields and examining APE’s creativity and adaptation.The conclusion is based on keyword comparisons within the self-citing articles.Source: A. Basic Statistics and Clusters · p. 4
  • The three clusters correspond to random or disordered spin models, magnetic lattice-gas physics, and diffusion, adaptation, self-organisation, and predator-prey research.The magnetic lattice-gas work introduces an additional degree of freedom and external-field coupling, while the third cluster links several connected questions.Source: A. Basic Statistics and Clusters · p. 4

Co-authorship structure

  • Few co-authors occur across multiple clusters, indicating loosely similar co-authorship lists rather than a strong field-specific dependence of collaborator choice.Only a few co-authors occur in three clusters and very few in two; the table reports co-authors appearing at least twice within a cluster.Source: A. Basic Statistics and Clusters · p. 4, Table I · p. 4

B. Time dependence

APE’s research fields occupy concentrated and largely non-overlapping periods, with transitions between them appearing sharp and irreversible. The record moves from random spin systems through magnetic lattice-gas work toward self-organisation, irreversible processes, and nonlinear dynamics.

Temporal structure

  • APE’s activities in different research fields are concentrated in distinct periods with barely any temporal overlap, resembling phase-transition-like changes.The temporal visualization assigns each self-citing article to one of the three percolation-derived structures.Source: B. Time dependence · p. 4

Research-field sequence

  • The career progresses from random spin systems in the 1970s–1980s to magnetic lattice-gas research around 1985–1995, followed by self-organisation and nonlinear-dynamics topics.Later topics extend toward biophysics, economy, and social dynamics.Source: B. Time dependence · p. 4

Irreversibility

  • Transitions between subfields are sharp and irreversible: APE does not return to a subfield after inactivity and remains active in one for long periods.The paper characterizes this pattern as high internal resistance to external fields or shocks.Source: B. Time dependence · pp. 4–5

Productivity over time

  • Productivity increases over time: the first ten ISI-indexed self-citing articles took ten years, whereas eighteen BSE articles took six years and the last 24 took about 12 years.The paper relates this pattern to prior work reporting increasing scientific productivity during careers, up to declining years.Source: B. Time dependence · p. 5

IV. DISCUSSION

The paper argues that self-citations, combined with co-authorships, can trace APE’s field mobility and reveal clustered research interests rather than mere citation noise. APE’s network comprises disconnected clusters and appears to develop through rare or drastic preferential-attachment-like events.

  • Combining self-citations with co-authorships provides a potential tool for tracing scientists’ field mobility.This approach treats self-citations as information about research development rather than automatically excluding them as noise or bias.Source: §IV · p. 5
  • APE’s self-citation network contains three well-separated clusters comprising 48 linked articles, while 47 of 95 indexed articles remain disconnected.The clusters contain 9, 7, and 32 articles, respectively; disconnected articles were excluded from the network-structure analysis.Source: §IV · p. 5
  • APE’s scientific interests appear to form independent clusters through rare or drastic preferential-attachment-like events involving coauthor groups.The paper relates this complexity to degrees of freedom coupled to external fields and possible internal shock resistance.Source: §IV · p. 5
Loading 0710.1800v1…