Network motifs for translator stylometry identification.

Area of Science:

• Computational Linguistics
• Stylometry
• Network Analysis

Background:

• Translator stylometry is an understudied field with applications in education, intellectual property, and forensic linguistics.
• Traditional lexical measures, such as vocabulary richness, have proven insufficient for identifying translator stylometry.
• Existing research highlights the need for novel approaches to analyze translator-specific linguistic patterns.

Purpose of the Study:

• To evaluate the effectiveness of existing lexical measures for translator stylometry.
• To develop and assess a new approach for identifying translator stylometry using network motifs.
• To demonstrate the utility of complex network analysis in solving translator stylometry problems.

Main Methods:

• A two-stage process was employed, initially evaluating traditional lexical measures.
• A novel approach utilizing network motifs (small sub-graphs) was designed to capture local network structures.
• Complex network analysis and network motif mining were used to engineer distinctive features.

Main Results:

• Traditional vocabulary richness measures were ineffective in identifying translator stylometry.
• The proposed network motif approach achieved an average accuracy of 83% in three-way classification.
• Augmenting classic lexical features with non-parametric scaling and network analysis proved effective.

Conclusions:

• Translator stylometry analysis requires advanced methods beyond traditional lexical features.
• Network motif analysis offers a powerful tool for uncovering distinctive translator patterns.
• This research provides a robust methodology for translator stylometry with implications for forensic linguistics and authorship attribution.