Egocentric Network Research Articles

Hyperauthored papers disproportionately amplify important egocentric network metrics

Abstract Hyperauthorship, a phenomenon whereby there are a disproportionately large number of authors on a single paper, is increasingly common in several scientific disciplines, but with unknown consequences for network metrics used to study scientific collaboration. The validity of coauthorship as a proxy for scientific collaboration is affected by this. Using bibliometric data from publications in the field of genomics, we examine the impact of hyperauthorship on metrics of scientific collaboration, and propose a method to determine a suitable cutoff threshold for hyperauthored papers and compare coauthorship networks with and without hyperauthored works. Our analysis reveals that including hyperauthored papers dramatically impacts the structural positioning of central authors and the topological characteristics of the network, while producing small influences on whole-network cohesion measures. We present two solutions to minimize the impact of hyperauthorship: using a mathematically grounded and reproducible calculation of threshold cutoff to exclude hyperauthored papers or fractional counting to weight network results. Our findings affirm the structural influences of hyperauthored papers and suggest that scholars should be mindful when using coauthorship networks to study scientific collaboration.

Multi-class Bitcoin mixing service identification based on graph classification

Due to its anonymity and decentralization, Bitcoin has long been a haven for various illegal activities. Cyber-criminals generally legalize illicit funds by Bitcoin mixing services. Therefore, it is critical to investigate the mixing services in cryptocurrency anti-money laundering. Existing studies treat different mixing services as a class of suspicious Bitcoin entities. Furthermore, they are limited by relying on expert experience or needing to deal with large-scale networks. So far, multi-class mixing service identification has not been explored yet. It is challenging since mixing services share a similar procedure, presenting no sharp distinctions. However, mixing service identification facilitates the healthy development of Bitcoin, supports financial forensics for cryptocurrency regulation and legislation, and provides technical means for fine-grained blockchain supervision. This paper aims to achieve multi-class Bitcoin Mixing Service Identification with a Graph Classification (BMSI-GC) model. First, BMSI-GC constructs 2-hop ego networks (2-egonets) of mixing services based on their historical transactions. Second, it applies graph2vec, a graph classification model mainly used to calculate the similarity between graphs, to automatically extract address features from the constructed 2-egonets. Finally, it trains a multilayer perceptron classifier to perform classification based on the extracted features. BMSI-GC is flexible without handling the full-size network and handcrafting address features. Moreover, the differences in transaction patterns of mixing services reflected in the 2-egonets provide adequate information for identification. Our experimental study demonstrates that BMSI-GC performs excellently in multi-class Bitcoin mixing service identification, achieving an average identification F1-score of 95.08%.

Resilience or Risk? Evaluating Three Pathways Linking Hispanic Immigrant Networks and Health.

There are competing perspectives on the impact of Hispanic immigrants' social networks on health; the Hispanic health paradox views networks as sources of resilience, whereas the tenuous ties perspective views networks as sources of risk. In this study, I explore the effect of networks on health by examining three network pathways: social capital, social bonding, and network stress. Using egocentric social network data from the VidaSana Study, a survey of 547 Hispanic immigrants in Indiana, I investigate how each network pathway is associated with physical health, mental health, and health care utilization. Results show that networks with greater capital, namely, more network health knowledge, promote physical health and health care access, whereas social bonding, operationalized as close and dense networks, benefits mental health and health care utilization. Network stress contributes to worse mental health yet improved health care access. Implications for social networks and health research among the Hispanics and more broadly are discussed.

How the size and structure of egocentric networks change during a life transition

While social networks are typically relatively stable in size over time, major changes in life circumstances can result in opportunities to acquire new friends. How young adults manage their relationships with their wider network of friends and family during such transitions is, however, not well understood. Using a prospective longitudinal design, we investigate changes in the size and composition of complete egocentric networks of two cohorts of young adults moving away from home to college. We show that, although networks grow rapidly due to an influx of new friends made at college, the social overload that would result is partially mitigated through the progressive loss of pre-transition friendships (but not family relationships). In addition, most of the new relationships are placed in the outermost, emotionally less close network layers that are less costly to maintain. In contrast, the more intimate inner layers of the network remain stable in size, with efforts being made to conserve these relationships. The overriding importance of face-to-face interaction in creating and maintaining ties (compared to digital media) results in the emotional quality of a tie being traded off against the constraints imposed by physical distance. The most reliable predictor of the proportion of original members with whom relationships were maintained post-transition was pre-transition network size, with weaker effects due to geographical proximity and personal popularity in the new social context. These findings have implications for managing transitions to a new environment at any life stage.

Navigating Complex Relationships: Support Networks Among Women in Community Corrections

In this article, we explored the social networks of 20 women who were in the process, or had completed, a sentence at a residential community corrections facility. Drawing on egocentric social network data and in-depth qualitative interviews, we analyzed the kinds of support that were most and least available to women, who were members of their network and what type of support they provided, and the role of complex relationships and peers. We found that women often relied on complex relationships and strategically navigated these relationships to access support. Many women also expressed a desire to mend and heal damaged relationships, recognizing their significance in their support system. Although peers can play a critical role in providing support to system-involved women, community corrections must invest in facilitating supportive connections and relationships among women. Ultimately, this research offers insights that can guide community corrections organizations in building and repairing women’s networks.

Social support networks of incarcerated women with opioid use disorder: Differences associated with jail-based substance use treatment

IntroductionSocial support is a critical factor for women's engagement with substance use treatment and recovery, particularly for women with criminal-legal system involvement. However, less is known about the social support networks of incarcerated women, particularly as a function of service engagement. Thus, this paper aims to describe the structure, composition, and function of social support networks of women with opioid use disorder (OUD) incarcerated in jails; and compare network differences between women receiving jail-based substance use treatment and non-treatment participants. MethodsAs part of a larger clinical trial under the NIDA-funded Justice Community Opioid Innovation Network (JCOIN), staff conducted an egocentric social network inventory with women who were randomly selected from eight jails in Kentucky, screened for OUD, and consented (N = 445). Women were asked to name people (alters) who provided them with support in the past 90 days and respond to questions about alters who were named. Bivariate comparisons examined differences in social networks among women who were currently receiving jail-based treatment (29.9 %, n = 133) and those who were incarcerated, but not receiving treatment (70.1 %, n = 312). ResultsOn average, women's social support networks consisted of 2.4 alters (range 1–9). Compared to women with OUD who were not receiving treatment, those in treatment were significantly more likely to name at least one alter who was in recovery (57.9 % vs. 43.9 %, p = .007) and less likely to name a partner (21.8 % vs. 37.8 %, p = .001) or someone who had recently used opioids (9.8 % vs. 24.7 %, p < .001). On average, women in treatment also felt significantly closer to their alters (4.70 vs. 4.55 out of 5, p = .021) and rated alters as fulfilling more types of social support functions (5.54 vs. 5.18 out of 6, p < .001) and recovery support functions (2.83 vs. 2.70 out of 3, p = .016). ConclusionResults suggest that women participating in jail-based substance use treatment reported more positive social support network attributes, including network function and composition. Future research should assess longitudinal changes in networks and associated differences in recovery outcomes as women are released to the community.

Conceptualizing the social networks of children of parents with serious mental illness: a thematic analysis.

Social networks, defined as the set of active and significant ties surrounding an individual, influence the wellbeing of vulnerable children. The best evidenced mechanism through which this occurs is where networks act as a vehicle to access social support. Little is known about the content and function of social networks of children of parents with severe and enduring mental illness (COPMI). COPMI are a frequently under-identified vulnerable child population at risk of negative outcomes. This qualitative study investigates the structure, role and function of these children's networks. Researchers conducted 17 semi-structured egocentric social network interviews. Interviews incorporated personal network mapping as a data collection method. COPMI were recruited through third sector organizations and interviewed across three sites in England. Data was analyzed using an inductive thematic analysis. Five network features were identified (i) parents as primary providers of support (i) limited networks and diminished connections over time (iii) substitutable ties (formal and informal) (iv) peer connections as source of both support and strain (v) coping strategies: self-censorship, avoidance and animals. Children of parents with severe and enduring mental illness networks are structurally typical of vulnerable children in that they are limited, rely on parents as primary ties but allow for some substitution of support ties. COPMI-specific features included peer relationships at times as source of strain and network level coping strategies used to manage wellbeing, including pets. This latter reflects previous findings in vulnerable adult populations so far unevidenced in children. Little evidence as to the mechanistic effect at work within networks was collected. However, COPMI were clearly shown to be engaged in active management and strategising in network navigation approaches, indicating the need to engage with children in this capacity, rather than approaching them as passive recipients of support. As such, effective network level interventions for this group are likely to prioritize access to beneficial substitute ties when support is limited. Additionally, interventions that promote network navigation skills and help foster productive coping strategies can capitalize on the child's active management role within their network.

Network of Categories: A Method to Aggregate Egocentric Network Survey Data into a Whole Network Structure

We propose a new method for analyzing and visualizing information on a large collection of personal networks to uncover the socio-centric structure of relationships among aggregated actors that we clustered into categories. The network of categories identifies the links between groups of sampled ego actors sharing a given attribute (e.g., “being immigrant organizations”) and abstract attribute-based groups or types of alters (e.g., “any immigrant organization”). This method takes advantage of both whole networks and ego-centered networks analytical approaches since it gains insights into the context’s whole network structure while generalizing the results to a wider population. The method can be applied to data obtained through modified versions of instruments such as position or resource generators, which are easy to collect within standard large- N representative surveys. We present a concrete application of the method to the analysis of immigrant organizations' position in local governance networks in 40 French and German cities.

Syphilis Positivity Among Men Who Have Sex With Men With Direct, Indirect, and No Linkage to Female Sex Partners: Exploring the Potential for Sex Network Bridging in Baltimore City, MD.

Syphilis epidemics among women and men who have sex with men (MSM) may be connected, but these connections are poorly understood. Using egocentric network data from a US urban MSM cohort, we examined sociodemographics, behaviors, and syphilis positivity among MSM with (1) direct (MSM who report sex with women, MSMW); (2) indirect (MSM who only report male partners, some of whom are MSMW, MSMO/W); and (3) no (MSM who only report male partners and whose partners only have sex with men, MSMO/O) connection to women. Sexually active MSM aged 18 to 45 years were administered behavioral and network interviews (recall period: 3 months) and syphilis/HIV testing. Syphilis positivity was defined as rapid plasma reagin titer ≥1:8. Modified Poisson regression was used to test for differences across groups. Among 385 MSM, 14.5% were MSMW and 22.3% were MSMO/W. MSMW and MSMO/W were significantly more likely than MSMO/O to report sex behaviors associated with increased syphilis acquisition/transmission risk, including ≥2 sex partners (MSMW: adjusted prevalence ratio [aPR], 1.28 [0.98-1.68]; MSMO/W: aPR, 1.35 [1.09-1.69]) and concurrent sex partners (MSMW: aPR, 1.50 [1.17-1.92]; MSMO/W: aPR, 1.39 [1.11-1.74]), and for MSMW only, transactional sex (aPR, 2.07 [1.11-3.88]). Syphilis positivity was 16.4% and was lower among MSMW (9.4%) and MSMO/W (14.1%) than MSMO/O (18.5%), but differences were not significant. There may be considerable connectivity between MSM and female sex partners that could facilitate syphilis transmission, and behaviors that increase acquisition/transmission risk among MSMW and MSMO/W may be distinct from MSMO/O. Future work should focus on examining the context and temporal patterns of sex partnerships among MSMW and MSMO/W.

Impact of Personal Networks of Ethnic Minority Female Entrepreneurs on Growth Construct: Blend of Egocentric and Narrative Analysis

The paper aims to explore and explain the content and structure of personal networks of immigrant and British-born Pakistani female entrepreneurs for business growth. Pakistani female entrepreneurs represent a distinct segment of ethnic minority entrepreneurs as they represent an interesting blend of socio-economic, religious and socio-cultural traits. This research study underscores the interactional dimension of personal networks and its impact on the growth of their business. It is mainly the contents and interactions within a network that determines the quality of a network and its influence on relevant aspects of network members. In order to explore the meaning and perceptions attached to relationships and their use for entrepreneurial ventures’ growth, an egocentric network analysis is applied to narratives of Pakistani female entrepreneurs (the ego). Narratives were collected through in-depth interviews with ‘name generator’ questions. Egocentric network analysis was carried out using sociograms to explore the formation and nature of network ties. By generating an in depth understanding of the distinctive use of personal networks for growth of business by Pakistani female entrepreneurs the paper provides in depth knowledge and understanding of a particular ethnic minority group that will be of use to business owners (Pakistani female entrepreneurs), academic researchers and policy makers.

빅데이터를 활용한 유보통합에 대한 사회적 인식 분석

The purpose of this study is to find out the social discourse of ECEC integration using big data. Big data related to ECEC integration from 2013 to 2023 were collected using Textorm. The collected raw data were refined, and among them 50 keywords were selected as a result of frequency analysis. The results of the study are as follows. First, the social discourse of ECEC integration showed high frequency in the order of ‘teacher’, ‘child care center’, ’kindergarten’, ‘child care’, and ‘qualification’. As a result of cluster analysis, social discourse was formed centering on three core discourses such as teacher qualification acquisition, policy direction, and consultation subject. Second, major keywords such as ‘child care center’, and ‘kindergarten’ have a higher degree of centrality, while word such as ‘policy’, and ‘Korea National Open University’ have a higher closeness centrality. Third, the ego network analysis was conducted on the main keywords with higher centrality and frequency, such as ‘teacher’ and ‘qualification’. In addition, the ego network analysis was performed and visualized the main keywords and visualized. Through this study, the social discourse was investigated and implications for approach of ECEC integration were discussed.

Unraveling the Dynamics of Employee Engagement: A Study of Employees’ Information-Sharing Networks and Voice Behavior within Organizations

ABSTRACT This study explores employee engagement by examining the complex relationship between information-sharing networks, employee voice behavior, and employee engagement. By focusing on the often-neglected dynamics of coworker relationships, this study draws on a social capital perspective and uses egocentric network analysis to examine how various attributes of employees’ information sharing networks – namely network size, strength, diversity, and ratio of friends – affect employee outcomes. Through an online egocentric survey of 400 full-time employees, the findings revealed that employees with larger, more diverse networks, and frequent interactions with coworkers are more likely to engage in voice behavior. The results also underscored the importance of friendship ties among coworkers, highlighting that employees who perceive more of their coworkers as friends within these networks are more likely to engage in voice behavior. This voice behavior, in turn, promotes employee engagement. The findings establish the significance of employees’ information-sharing networks with coworkers to leaders and communication professionals in nurturing a culture of engagement.

Advancing the Measurement of Social Functioning in Schizophrenia: Applications of Egocentric Social Network Analysis.

People with schizophrenia face challenges with forming and maintaining social relationships, often resulting in poor social functioning. Commonly used measures of social functioning provide broad information relating to social relationships, but they do not adequately capture information regarding network structure and characteristics of network members. One method that can assess these more detailed aspects of social networks and provide a more comprehensive understanding of social functioning deficits is egocentric social network analysis (SNA). SNA is a scientific discipline that uses principles of network science and graph theory to analyze social relations quantitatively. Even though some types of SNA have been applied in prior schizophrenia studies, its application as a framework to measure social functioning has been extremely limited. Therefore, this article aims to formally introduce SNA and select quantitative SNA metrics, including measures of network composition, structure, homophily, and centrality, to schizophrenia researchers as novel ways of measuring components of social functioning. To demonstrate the application of SNA, we provide illustrative examples of the SNA metrics and graphical diagrams of social networks for two individuals with schizophrenia.

Efficiently Quantifying Egocentric Social Network Cannabis Use: Initial Psychometric Validation of the Brief Cannabis Social Density Assessment.

Social environment is a key determinant of substance use, but cannabis-related social network analysis is not common, in part due to the assessment burden of comprehensive egocentric social network analysis. The current pre-registered secondary analysis assessed the psychometric properties (i.e., convergent, criterion-related, incremental validity) of the Brief Cannabis Social Density Assessment (B-CaSDA) in a cross-sectional sample of adults who use cannabis (N = 310) using a survey-based design. The B-CaSDA assesses the quantity and frequency of cannabis use for the respondent's four closest (non-parent) relationships. Cannabis use severity was elevated for each additional person who used cannabis at all or daily in the individual's social network. B-CaSDA indices (i.e., frequency, quantity, total score) were positively correlated with cannabis consumption, cannabis use severity indicators, and established risk factors for harmful cannabis use. B-CaSDA indices also discriminated between those above and below a clinical cutoff on the Cannabis Use Disorder Identification Test - Revised (CUDIT-R). Finally, in omnibus models that included common risk factors for cannabis use severity, the B-CaSDA quantity index contributed additional variance when predicting CUDIT-R total score, and B-CaSDA frequency contributed additional variance in predicting the CUDIT-R quantity-frequency subscale. The results suggest that the B-CaSDA has the potential to expand social network research on cannabis use and misuse by increasing its assessment feasibility in diverse designs.

Homophily and social mixing in a small community: Implications for infectious disease transmission.

Community mixing patterns by sociodemographic traits can inform the risk of epidemic spread among groups, and the balance of in- and out-group mixing affects epidemic potential. Understanding mixing patterns can provide insight about potential transmission pathways throughout a community. We used a snowball sampling design to enroll people recently diagnosed with SARS-CoV-2 in an ethnically and racially diverse county and asked them to describe their close contacts and recruit some contacts to enroll in the study. We constructed egocentric networks of the participants and their contacts and assessed age-mixing, ethnic/racial homophily, and gender homophily. The total size of the egocentric networks was 2,544 people (n = 384 index cases + n = 2,160 recruited peers or other contacts). We observed high rates of in-group mixing among ethnic/racial groups compared to the ethnic/racial proportions of the background population. Black or African-American respondents interacted with a wider range of ages than other ethnic/racial groups, largely due to familial relationships. The egocentric networks of non-binary contacts had little age diversity. Black or African-American respondents in particular reported mixing with older or younger family members, which could increase the risk of transmission to vulnerable age groups. Understanding community mixing patterns can inform infectious disease risk, support analyses to predict epidemic size, or be used to design campaigns such as vaccination strategies so that community members who have vulnerable contacts are prioritized.

Web Application to Enable Online Social Interactions in a Parkinson Disease Risk Cohort: Feasibility Study and Social Network Analysis.

There is evidence that social interaction has an inverse association with the development of neurodegenerative diseases. PREDICT-Parkinson Disease (PREDICT-PD) is an online UK cohort study that stratifies participants for risk of future Parkinson disease (PD). This study aims to explore the methodological approach and feasibility of assessing the digital social characteristics of people at risk of developing PD and their social capital within the PREDICT-PD platform, making hypotheses about the relationship between web-based social engagement and potential predictive risk indicators of PD. A web-based application was built to enable social interaction through the PREDICT-PD portal. Feedback from existing members of the cohort was sought and informed the design of the pilot. Dedicated staff used weekly engagement activities, consisting of PD-related research, facts, and queries, to stimulate discussion. Data were collected by the hosting platform. We examined the pattern of connections generated over time through the cumulative number of posts and replies and ego networks using social network analysis. We used network metrics to describe the bonding, bridging, and linking of social capital among participants on the platform. Relevant demographic data and Parkinson risk scores (expressed as an odd 1:x) were analyzed using descriptive statistics. Regression analysis was conducted to estimate the relationship between risk scores (after log transformation) and network measures. Overall, 219 participants took part in a 4-month pilot forum embedded in the study website. In it, 200 people (n=80, 40% male and n=113, 57% female) connected in a large group, where most pairs of users could reach one another either directly or indirectly through other users. A total of 59% (20/34) of discussions were spontaneously started by participants. Participation was asynchronous, with some individuals acting as "brokers" between groups of discussions. As more participants joined the forum and connected to one another through online posts, distinct groups of connected users started to emerge. This pilot showed that a forum application within the cohort web platform was feasible and acceptable and fostered digital social interaction. Matching participants' web-based social engagement with previously collected data at individual level in the PREDICT-PD study was feasible, showing potential for future analyses correlating online network characteristics with the risk of PD over time, as well as testing digital social engagement as an intervention to modify the risk of developing neurodegenerative diseases. The results from the pilot suggest that an online forum can serve as an intervention to enhance social connectedness and investigate whether patterns of online engagement can impact the risk of developing PD through long-term follow-up. This highlights the potential of leveraging online platforms to study the role of social capital in moderating PD risk and underscores the feasibility of such approaches in future research or interventions.

Health-related social control among U.S. adults ages 30–80: Associations with alcohol use over four years

Health-related social control (HRSC) includes efforts to regulate or influence others' health behaviors and is an important way interpersonal relationships can affect individual-level health. This study used egocentric network data to describe the size and composition of HRSC networks, identify trajectories of HRSC receipt, and examine how HRSC is related to binge drinking and alcohol-related problems. Data come from a U.S. nationally representative sample of 1235 adults age 30 and older (baseline mean age = 52, 52% female, 64% White) who completed four annual surveys between 2019 and 2022. On average, 30% of adults' network members were HRSC agents who told or reminded them to do things to protect their health. At baseline, 50% of respondents identified a spouse/partner as a HRSC agent, 56% a relative, 46% a friend, and 12% someone else. Respondents' relationships with HRSC agents were generally strong, 93% of agents were described as people “whose opinion matters,” and only 10% were described as hassling or making life difficult for the respondent. Growth mixture modeling identified five trajectories of HRSC receipt over the four-year period: Stable High (36% of sample), Stable Moderate (47%), Stable Low (14%), Decreasing (2%), and Increasing (2%). Binge drinking was relatively consistent for the three Stable HRSC classes (ranging from 11% to 15% of individuals), decreased steadily for the Increasing HRSC class (32%–16%), and fluctuated for the Decreasing HRSC class (decreasing from 10% to 2%, then increasing to 8%). For alcohol problems, the Increasing HRSC class showed the largest increase (2%–21%) before dropping to near-baseline levels (4%), whereas the Decreasing HRSC class fluctuated during the first three waves followed by no individuals reporting alcohol problems at the last wave. Results highlight the importance of examining heterogeneity in adults’ HRSC experiences because of its implications for understanding social influences on health-related behaviors.

Value implications of followers in social marketplaces: insights into ego network structures

PurposeIn social marketplaces, follower ego networks are integral social capital assets for online sellers. While previous research has underscored the positive impact of the follower number on seller performance, little attention has been given to the structure of follower networks and their value implications. This research investigates two structural properties of follower networks—network centralization and density—and examines their main and contingent effects on sellers’ sales performance.Design/methodology/approachA 13-month panel dataset of 1,150 sellers in Etsy, a social marketplace for handmade and vintage products, was collected and analyzed. A fixed effects model was adopted to validate the hypotheses on the main effect of centralization and density, as well as the moderating effects of two store attributes: store age and product diversification.FindingsWe find that both network centralization and density negatively impact sellers’ sales performance, and these effects vary across store age and product diversification levels. Specifically, the negative effect of network centralization is less pronounced for older stores than young ones, whereas the negative effect of density is more severe for stores with high product diversification.Originality/valueThis research contributes to social commerce research by highlighting the significance of network structure, alongside network size, in assessing the value of followers and offers practical guidance for sellers in social marketplaces seeking to optimize their follower networks.

A parameterised model for link prediction using node centrality and similarity measure based on graph embedding

Link prediction is a crucial aspect of graph machine learning, with applications as diverse as disease prediction, social network recommendations, and drug discovery. It involves the prediction of potential new links between nodes within a network. Despite its importance, current models for link prediction exhibit notable limitations. Graph Convolutional Networks have shown high efficiency in link prediction across various datasets. However, they face significant challenges when applied to short-path networks and ego networks, resulting in poor performance. This issue represents a critical area of concern that our work seeks to address. This paper introduces the Node Centrality and Similarity-based Parameterised Model (NCSM), a novel method for link prediction tasks. NCSM uniquely integrates node centrality and similarity measures as edge features in a customised Graph Neural Network (GNN) layer, effectively leveraging the topological information of large networks. This model represents the first parameterised GNN-based link prediction model that considers topological information. The proposed model was evaluated on five benchmark graph datasets, each comprising thousands of nodes and edges. Experimental results highlight NCSM's superiority over existing state-of-the-art models like Graph Convolutional Networks and Variational Graph Autoencoder, as it outperforms them across various metrics and datasets. This exceptional performance can be attributed to NCSM's innovative integration of node centrality, similarity measures, and its efficient use of topological information.

Research on scientific knowledge evolution patterns based on ego-centered fine-granularity citation network

A deep comprehension of the development of scientific knowledge is necessary to properly appreciate technological advancement and the process of knowledge innovation. However, fine-grained citation networks are less frequently utilized to characterize the evolution of scientific knowledge. To bridge this gap, we first constructed the fine-grained citation networks based on the PubMed and Web of Science (WOS) databases, then extracted the ego-centered networks of scientific knowledge, and finally employed Exponential Random Graph Models (ERGMs) to analyze the factors influencing the formation of Ego-centered Fine-granularity Citation Networks (EFCNs), taking into account the endogenous network structure and exogenous knowledge attribute variables. The results reveal that both types of variables play pivotal roles in the evolution of scientific knowledge. Furthermore, we found that (1) the in-degree and the out-degree centrality have a positive effect on knowledge evolution, respectively, in the 98.5 % and 99.9 % sample networks, while the clustering coefficient only has a positive effect on the edge formation of the 6.8 % sample network at the 0.05 significance level. (2) The citation behavior and domain impact of authors positively influence the scientific knowledge evolution, respectively, in the 63.2 % and 78.3 % sample networks. (3) There is a tendency to form citation relationships between scientific knowledge units of similar age in 67.1 % of the sample networks. (4) There is a greater possibility of developing a citation relationship between scientific knowledge with the same journal impact rank and knowledge type. Our findings indicate that the evolution of scientific knowledge is influenced not only by the process of scholarly communication but also by self-organizing mechanisms at the fine granularity level.