Global Academic Libraries User Reviews Dataset: Zero-Shot Classification and Fortified VADER Sentiment Analysis

This dataset is composed of 8542 textual user reviews from 93 academic libraries worldwide, collected from the Google Maps platform between May and June 2026 using the Outscraper tool. Spanning the years 2008 to 2026, with a significant concentration of user-generated content from 2022 to 2026, the corpus captures nearly two decades of evolving patron experiences. The dataset has been processed and annotated using a Transformer-Based Zero-Shot Classification architecture, which successfully maps unstructured, short-form feedback onto eleven literature-derived operational dimensions. In addition, it includes sentiment polarity scores generated via a VADER sentiment analysis pipeline. This sentiment data is enhanced by a fortification mechanism that cross-references raw compound textual scores with explicit numeric star ratings, effectively correcting for sarcasm and contextual discrepancies to provide a highly accurate reflection of user satisfaction. Note: This dataset has been developed as a part of the paper submitted for consideration for publication in Electronics under the Special Issue Advances in Web Data Management.