# University of Twente Student Theses

## How well can personal network features be retrieved?

Arends, B.
(2023)
*How well can personal network features be retrieved?*

PDF
1MB |

Abstract: | This paper investigates how good certain algorithms can retrieve node features of nodes in networks when only observing the network connections on random uniformly distributed graphs up to 300 nodes. Algorithms like node2vec use random walks to retrieve information from a network, after which it makes an embedding of the graph. We are interested in the statistics of the original graph and the embedding to see if the algorithm can really capture the node features of the graph well. We investigated the degree of the nodes, the amount of triangles, the clustering and the number of edges that are in the graph but not in the embedding and vice versa. We also looked if changing parameters to favor breadth first search or depth first search of the random walks changed the outcome. Changing certain parameters did not change the properties of the embedded graph, for example changing the parameters of node2vec that favor breadth first search or depth first search. One parameter that really changed the properties of the embedded graphs was changing the dimension of the the random geometric graph. We could conclude that changing the parameters that favor either depth first search or breadth first search did not change the statistics for the random graph and the embedding. Changing the dimension of the embedding also did not influence the statistics. However, changing the dimension of the random generated graph did significantly change the outcome. |

Item Type: | Essay (Bachelor) |

Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |

Subject: | 31 mathematics |

Programme: | Applied Mathematics BSc (56965) |

Link to this item: | https://purl.utwente.nl/essays/94349 |

Export this item as: | BibTeX EndNote HTML Citation Reference Manager |

Repository Staff Only: item control page