Abstract:Name ambiguity is common in academic digital libraries, such as multiple authors having the same name. This creates challenges for academic data management and analysis, thus name disambiguation becomes necessary. The procedure of name disambiguation is to divide publications with the same name into different groups, each group belonging to a unique author. A large amount of attribute information in publications makes traditional methods fall into the quagmire of feature selection. These methods always select attributes artificially and equally, which usually causes a negative impact on accuracy. The proposed method is mainly based on representation learning for heterogeneous networks and clustering and exploits the self-attention technology to solve the problem. The presentation of publications is a synthesis of structural and semantic representations. The structural representation is obtained by meta-path-based sampling and a skip-gram-based embedding method, and meta-path level attention is introduced to automatically learn the weight of each feature. The semantic representation is generated using NLP tools. Our proposal performs better in terms of name disambiguation accuracy compared with baselines and the ablation experiments demonstrate the improvement by feature selection and the meta-path level attention in our method. The experimental results show the superiority of our new method for capturing the most attributes from publications and reducing the impact of redundant information.

Using Lexical and Thematic Knowledge for Name Disambiguation.

Multi-document Chinese Name Disambiguation Based on Latent Semantic Analysis

A Unified Semi-supervised Framework for Author Disambiguation in Academic Social Network.

Name Disambiguation By Collective Classification

Name Disambiguation Using Web Connection

A Unified Framework for Name Disambiguation

Scholar Name Disambiguation with Search-enhanced LLM Across Language

Author Name Disambiguation via Heterogeneous Network Embedding from Structural and Semantic Perspectives

Resolving Regular Polysemy in Named Entities

Two Supervised Learning Approaches For Name Disambiguation In Author Citations

A Constraint-Based Topic Modeling Approach for Name Disambiguation

ADANA: Active Name Disambiguation

On Graph-Based Name Disambiguation.

A Constraint-Based Probabilistic Framework for Name Disambiguation

Robust hybrid name disambiguation framework for large databases

Research on the Application of a Chinese Semantic Knowledge Base in Chinese Phrase Disambiguation

An Effective Approach for Automatic Author Name Disambiguation Based on Multiple Strategies

Social Network Based Cross-Document Personal Name Disambiguation

Name Disambiguation Using Atomic Clusters.

Exploiting external knowledge sources to improve kernel-based Word Sense Disambiguation

A Graph-Based Author Name Disambiguation Method and Analysis via Information Theory