GESIS Search study finds semantic similarity best for social science recommendations

A new study published on arXiv details the integration and evaluation of various recommendation systems within GESIS Search, a specialized search engine for social sciences. Researchers implemented and compared lexical similarity, transformer-based semantic similarity, and session-based recommendation algorithms. The findings indicate that users generally prefer recommendations based on semantic similarity over term-similarity and session-based methods, though performance varies across different categories within the search engine. AI

IMPACT Findings suggest semantic similarity is key for academic search recommendations, potentially guiding future development in scholarly information retrieval.

RANK_REASON Academic paper detailing research findings on recommendation systems. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.IR (Information Retrieval) →

• alphaXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• GESIS Search
• Gotit.pub
• Hugging Face
• Influence Flower
• Litmaps
• Mehmet Deniz Türkmen
• ScienceCast
• scite Smart Citations
• STELLA

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →