Incremental and Data-Efficient Concept Formation to Support Masked Word Prediction
This work addresses efficient language model learning for masked word prediction, offering a novel incremental approach that is competitive with existing methods but incremental in nature.
The paper tackles the problem of masked word prediction by introducing Cobweb4L, an incremental and data-efficient method that builds on Cobweb to learn probabilistic concept hierarchies, achieving performance comparable to or better than Word2Vec and outperforming BERT with less training data.
This paper introduces Cobweb4L, a novel approach for efficient language model learning that supports masked word prediction. The approach builds on Cobweb, an incremental system that learns a hierarchy of probabilistic concepts. Each concept stores the frequencies of words that appear in instances tagged with that concept label. The system utilizes an attribute value representation to encode words and their surrounding context into instances. Cobweb4L uses the information theoretic variant of category utility and a new performance mechanism that leverages multiple concepts to generate predictions. We demonstrate that with these extensions it significantly outperforms prior Cobweb performance mechanisms that use only a single node to generate predictions. Further, we demonstrate that Cobweb4L learns rapidly and achieves performance comparable to and even superior to Word2Vec. Next, we show that Cobweb4L and Word2Vec outperform BERT in the same task with less training data. Finally, we discuss future work to make our conclusions more robust and inclusive.