Author disambiguation using multi-aspect similarity indicators

Area of Science:

• Bibliometrics
• Scientometrics
• Information Science

Background:

• Accurate bibliometric analysis relies on precise author-to-work linkage.
• Existing methods struggle with null data fields and suboptimal precision-recall balance.

Purpose of the Study:

• To develop an advanced algorithm for author disambiguation.
• To enhance the precision and recall of bibliometric data linkage.

Main Methods:

• Developed a novel algorithm for author disambiguation.
• Implemented a dynamic similarity calculation using all data fields.
• Incorporated author contribution and publication age differences into similarity metrics.

Main Results:

• Achieved very high precision and recall in author disambiguation.
• Addressed issues of discarded records due to null data fields.
• Demonstrated significant improvements over previous and standalone techniques using a heterogeneous catalysis dataset.
• Reported significantly high average F-measure scores.

Conclusions:

• The developed algorithm offers a superior approach to author disambiguation in bibliometrics.
• The method effectively handles data limitations and enhances linkage accuracy.
• This advancement has substantial implications for the reliability of bibliometric analyses.