Enhancing Relationship Link Prediction With Hierarchical Feature Enhancement
In this study, we endeavor to assess the potential for establishing friendships between users, a challenge that intersects with the domains of friend recommendation and link prediction. Notably, prevailing models in friend recommendation predominantly consider only the attributes of users, whereas l...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10759655/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In this study, we endeavor to assess the potential for establishing friendships between users, a challenge that intersects with the domains of friend recommendation and link prediction. Notably, prevailing models in friend recommendation predominantly consider only the attributes of users, whereas link prediction models primarily emphasize network structure. In this paper, we propose a novel methodology by regarding users’ attributes as node-level features, extracting them from tweet text using GPT-style models, and considering the structure of the constructed network as network-level features. The network-level features are obtained through a soft-partitioning overlapping community detection algorithm, namely BIGCLAM(Cluster Affiliation Model for Big Networks). We then integrate the node-level and network-level features to mutually enhance their predictive capabilities. Through rigorous experimental validation across two distinct dataset types—a homogeneous network and a heterogeneous network—our results demonstrate that the fusion of node-level and network-level features significantly improves performance. |
|---|---|
| ISSN: | 2169-3536 |