Learning attribute and homophily measures through random walks
Abstract We investigate the statistical learning of nodal attribute functionals in homophily networks using random walks. Attributes can be discrete or continuous. A generalization of various existing canonical models, based on preferential attachment is studied (model class $$\mathscr {P}$$ P ), wh...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2023-06-01
|
| Series: | Applied Network Science |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s41109-023-00558-3 |
| _version_ | 1827910903601823744 |
|---|---|
| author | Nelson Antunes Sayan Banerjee Shankar Bhamidi Vladas Pipiras |
| author_facet | Nelson Antunes Sayan Banerjee Shankar Bhamidi Vladas Pipiras |
| author_sort | Nelson Antunes |
| collection | DOAJ |
| description | Abstract We investigate the statistical learning of nodal attribute functionals in homophily networks using random walks. Attributes can be discrete or continuous. A generalization of various existing canonical models, based on preferential attachment is studied (model class $$\mathscr {P}$$ P ), where new nodes form connections dependent on both their attribute values and popularity as measured by degree. An associated model class $$\mathscr {U}$$ U is described, which is amenable to theoretical analysis and gives access to asymptotics of a host of functionals of interest. Settings where asymptotics for model class $$\mathscr {U}$$ U transfer over to model class $$\mathscr {P}$$ P through the phenomenon of resolvability are analyzed. For the statistical learning, we consider several canonical attribute agnostic sampling schemes such as Metropolis-Hasting random walk, versions of node2vec (Grover and Leskovec, 2016) that incorporate both classical random walk and non-backtracking propensities and propose new variants which use attribute information in addition to topological information to explore the network. Estimators for learning the attribute distribution, degree distribution for an attribute type and homophily measures are proposed. The performance of such statistical learning framework is studied on both synthetic networks (model class $$\mathscr {P}$$ P ) and real world systems, and its dependence on the network topology, degree of homophily or absence thereof, (un)balanced attributes, is assessed. |
| first_indexed | 2024-03-13T01:56:06Z |
| format | Article |
| id | doaj.art-11a32c9656f24ad3a1d20dabe6294f69 |
| institution | Directory Open Access Journal |
| issn | 2364-8228 |
| language | English |
| last_indexed | 2024-03-13T01:56:06Z |
| publishDate | 2023-06-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | Applied Network Science |
| spelling | doaj.art-11a32c9656f24ad3a1d20dabe6294f692023-07-02T11:10:23ZengSpringerOpenApplied Network Science2364-82282023-06-018112710.1007/s41109-023-00558-3Learning attribute and homophily measures through random walksNelson Antunes0Sayan Banerjee1Shankar Bhamidi2Vladas Pipiras3Center for Computational and Stochastic Mathematics, University of LisbonDepartment of Statistics and Operations Research, University of North CarolinaDepartment of Statistics and Operations Research, University of North CarolinaDepartment of Statistics and Operations Research, University of North CarolinaAbstract We investigate the statistical learning of nodal attribute functionals in homophily networks using random walks. Attributes can be discrete or continuous. A generalization of various existing canonical models, based on preferential attachment is studied (model class $$\mathscr {P}$$ P ), where new nodes form connections dependent on both their attribute values and popularity as measured by degree. An associated model class $$\mathscr {U}$$ U is described, which is amenable to theoretical analysis and gives access to asymptotics of a host of functionals of interest. Settings where asymptotics for model class $$\mathscr {U}$$ U transfer over to model class $$\mathscr {P}$$ P through the phenomenon of resolvability are analyzed. For the statistical learning, we consider several canonical attribute agnostic sampling schemes such as Metropolis-Hasting random walk, versions of node2vec (Grover and Leskovec, 2016) that incorporate both classical random walk and non-backtracking propensities and propose new variants which use attribute information in addition to topological information to explore the network. Estimators for learning the attribute distribution, degree distribution for an attribute type and homophily measures are proposed. The performance of such statistical learning framework is studied on both synthetic networks (model class $$\mathscr {P}$$ P ) and real world systems, and its dependence on the network topology, degree of homophily or absence thereof, (un)balanced attributes, is assessed.https://doi.org/10.1007/s41109-023-00558-3Attributed networksHomophilyNetwork modelResolvabilityRandom walk samplingsDiscrete and continuous attributes |
| spellingShingle | Nelson Antunes Sayan Banerjee Shankar Bhamidi Vladas Pipiras Learning attribute and homophily measures through random walks Applied Network Science Attributed networks Homophily Network model Resolvability Random walk samplings Discrete and continuous attributes |
| title | Learning attribute and homophily measures through random walks |
| title_full | Learning attribute and homophily measures through random walks |
| title_fullStr | Learning attribute and homophily measures through random walks |
| title_full_unstemmed | Learning attribute and homophily measures through random walks |
| title_short | Learning attribute and homophily measures through random walks |
| title_sort | learning attribute and homophily measures through random walks |
| topic | Attributed networks Homophily Network model Resolvability Random walk samplings Discrete and continuous attributes |
| url | https://doi.org/10.1007/s41109-023-00558-3 |
| work_keys_str_mv | AT nelsonantunes learningattributeandhomophilymeasuresthroughrandomwalks AT sayanbanerjee learningattributeandhomophilymeasuresthroughrandomwalks AT shankarbhamidi learningattributeandhomophilymeasuresthroughrandomwalks AT vladaspipiras learningattributeandhomophilymeasuresthroughrandomwalks |