Treffer: A game-theory-based method for retraining-free semantic enhancement in classification correction.

In order to better explain the working mechanism of text classification models and enhance the interpretability of classification models, this study proposes an interpretable, retraining-free semantic enhancement method. By using the TokenShapley values proposed in the study to quantify the contribution of input text tokens, and by removing interfering information to help the model strengthen semantics, more accurate classification can be achieved. The study conducted comparative experiments on three representative Chinese datasets with some common classification models. The results show that the corrected and strengthened models have improved in various evaluation metrics, proving the effectiveness of the method, and making an important contribution to in-depth exploration of text interpretability research. [ABSTRACT FROM AUTHOR]

Copyright of Enterprise Information Systems is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)