Organizing Books and Authors by Multilayer SOM

Area of Science:

• Computer Science
• Information Retrieval
• Machine Learning

Background:

• Organizing large collections of electronic books (e-books) and author data presents significant challenges.
• Existing methods struggle with the complexity of structured data representations for authors and their works.

Purpose of the Study:

• To introduce a novel framework for organizing e-books and authors using a multilayer self-organizing map (MLSOM).
• To develop an efficient system for managing structured author data and facilitate book and author recommendations.

Main Methods:

• Modeling authors using a rich tree-structured representation encompassing biography, books, pages, and paragraphs.
• Employing a multilayer self-organizing map (MLSOM) algorithm as a clustering technique for e-books and authors.
• Implementing a book and author recommender system based on the proposed MLSOM framework.

Main Results:

• The framework was evaluated on a large dataset of 3868 authors and 10500 e-books.
• MLSOM visualization revealed hidden relevance patterns within author clusters.
• The proposed method demonstrated superior performance compared to content-based models like LDA and PLSI.

Conclusions:

• The MLSOM-based framework provides an effective solution for organizing structured e-book and author data.
• The approach offers a promising method for enhancing book recommendation, author recommendation, and data visualization.
• This work advances the field of information retrieval and recommender systems for digital libraries.