Inference of node attributes from social network assortativity
Social networks are known to be assortative with respect to many attributes, such as age, weight, wealth, level of education, ethnicity and gender: Similar people according to these attributes tend to be more connected. This can be explained by influences and homophily. Independently of its origin,...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Springer London
2021
|
| Online Access: | https://hdl.handle.net/1721.1/130066 |
| _version_ | 1826204276593000448 |
|---|---|
| author | Pentland, Alexander (Sandy) |
| author2 | Massachusetts Institute of Technology. Media Laboratory |
| author_facet | Massachusetts Institute of Technology. Media Laboratory Pentland, Alexander (Sandy) |
| author_sort | Pentland, Alexander (Sandy) |
| collection | MIT |
| description | Social networks are known to be assortative with respect to many attributes, such as age, weight, wealth, level of education, ethnicity and gender: Similar people according to these attributes tend to be more connected. This can be explained by influences and homophily. Independently of its origin, this assortativity gives us information about each node given its neighbors. Assortativity can thus be used to improve individual predictions in a broad range of situations, when data are missing or inaccurate. This paper presents a general framework based on probabilistic graphical models to exploit social network structures for improving individual predictions of node attributes. Using this framework, we quantify the assortativity range leading to an accuracy gain in several situations, with various individual prediction profiles. We finally show how specific characteristics of the network can enhance performances further. For instance, the gender assortativity in real-world mobile phone data drastically changes according to some communication attributes. In this case, using the network topology indeed improves local predictions of node labels and moreover enables inferring missing node labels based on a subset of known vertices. In both cases, the performances of the proposed method are statistically significantly superior to the ones achieved by state-of-the-art label propagation and feature extraction schemes in most settings. |
| first_indexed | 2024-09-23T12:51:44Z |
| format | Article |
| id | mit-1721.1/130066 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T12:51:44Z |
| publishDate | 2021 |
| publisher | Springer London |
| record_format | dspace |
| spelling | mit-1721.1/1300662024-06-25T17:50:11Z Inference of node attributes from social network assortativity Pentland, Alexander (Sandy) Massachusetts Institute of Technology. Media Laboratory Social networks are known to be assortative with respect to many attributes, such as age, weight, wealth, level of education, ethnicity and gender: Similar people according to these attributes tend to be more connected. This can be explained by influences and homophily. Independently of its origin, this assortativity gives us information about each node given its neighbors. Assortativity can thus be used to improve individual predictions in a broad range of situations, when data are missing or inaccurate. This paper presents a general framework based on probabilistic graphical models to exploit social network structures for improving individual predictions of node attributes. Using this framework, we quantify the assortativity range leading to an accuracy gain in several situations, with various individual prediction profiles. We finally show how specific characteristics of the network can enhance performances further. For instance, the gender assortativity in real-world mobile phone data drastically changes according to some communication attributes. In this case, using the network topology indeed improves local predictions of node labels and moreover enables inferring missing node labels based on a subset of known vertices. In both cases, the performances of the proposed method are statistically significantly superior to the ones achieved by state-of-the-art label propagation and feature extraction schemes in most settings. 2021-03-03T19:37:43Z 2021-03-03T19:37:43Z 2019-01-07 2020-12-04T04:24:12Z Article http://purl.org/eprint/type/JournalArticle 9781509066391 9781509066384 0941-0643 1433-3058 https://hdl.handle.net/1721.1/130066 Mulders, Dounia et al. “Inference of node attributes from social network assortativity.” Paper in the Neural computing & applications, WSOM 2017, Nancy, France, June 28-30, 2017, Springer London: 18023–18043 © 2019 The Author(s) en https://doi.org/10.1007/s00521-018-03967-z Neural computing & applications Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ Springer-Verlag London Ltd., part of Springer Nature application/pdf Springer London Springer London |
| spellingShingle | Pentland, Alexander (Sandy) Inference of node attributes from social network assortativity |
| title | Inference of node attributes from social network assortativity |
| title_full | Inference of node attributes from social network assortativity |
| title_fullStr | Inference of node attributes from social network assortativity |
| title_full_unstemmed | Inference of node attributes from social network assortativity |
| title_short | Inference of node attributes from social network assortativity |
| title_sort | inference of node attributes from social network assortativity |
| url | https://hdl.handle.net/1721.1/130066 |
| work_keys_str_mv | AT pentlandalexandersandy inferenceofnodeattributesfromsocialnetworkassortativity |