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

Area of Science:

• Computer Science
• Information Science
• Data Science

Background:

• Name ambiguity is a significant challenge in information retrieval and academic data analysis.
• Existing author name disambiguation methods often rely solely on document attributes or co-author relationships, leading to limitations.
• Attribute-based methods lack flexibility, while graph-based methods neglect feature information.

Purpose of the Study:

• To propose a novel name disambiguation model that integrates both document attributes and relationships.
• To address the limitations of existing methods in academic author name disambiguation.
• To enhance the accuracy and interpretability of author name disambiguation.

Main Methods:

• Developed a representation learning-based model incorporating document attributes and co-author relationships.
• Utilized information theory to enhance model interpretability.
• Conducted experiments on a public real dataset to validate the model's effectiveness.

Main Results:

• The proposed model demonstrated superior performance compared to several state-of-the-art graph-based methods.
• Experimental results confirmed the effectiveness of integrating attributes and relationships for name disambiguation.
• The approach improved the interpretability of the disambiguation process.

Conclusions:

• The novel representation learning model effectively resolves author name ambiguity by integrating diverse information sources.
• This approach offers a more robust and interpretable solution for academic data analysis.
• The findings suggest potential improvements in document retrieval and information integration systems.