Paper Number
3252
Paper Type
Complete
Abstract
The performance of recommender systems is highly dependent on the quality of the underlying data. However, while several studies have investigated the impact of data quality dimensions such as completeness and accuracy, the impact of currency has remained unexplored. Existing research focused on the impact of data age reported conflicting evidence, indicating that data age alone cannot account for the actual currency of information contained in reviews. Against this background, we perform an experiment to investigate the impact of currency on the performance of recommender systems using ten real-world data sets containing customer reviews from Yelp. Our results demonstrate that the impact of currency is significant and even exceeds that of data age. Furthermore, our results indicate that considering currency at the aspect-level has the most significant impact on the performance of recommender systems. Equipped with these insights, future design-oriented work can further improve recommender systems with currency-aware recommendation algorithms.
Recommended Citation
Hägele, Lukas Jakob; Klier, Mathias; Obermeier, Andreas; and Sparn, Christian, "Oldie But Goodie – Currency Beats Data Age of Customer Reviews When It Comes to Recommender System Performance" (2024). ICIS 2024 Proceedings. 8.
https://aisel.aisnet.org/icis2024/data_soc/data_soc/8
Oldie But Goodie – Currency Beats Data Age of Customer Reviews When It Comes to Recommender System Performance
The performance of recommender systems is highly dependent on the quality of the underlying data. However, while several studies have investigated the impact of data quality dimensions such as completeness and accuracy, the impact of currency has remained unexplored. Existing research focused on the impact of data age reported conflicting evidence, indicating that data age alone cannot account for the actual currency of information contained in reviews. Against this background, we perform an experiment to investigate the impact of currency on the performance of recommender systems using ten real-world data sets containing customer reviews from Yelp. Our results demonstrate that the impact of currency is significant and even exceeds that of data age. Furthermore, our results indicate that considering currency at the aspect-level has the most significant impact on the performance of recommender systems. Equipped with these insights, future design-oriented work can further improve recommender systems with currency-aware recommendation algorithms.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
13-DataAnalytics