Domain-Specific Word Embeddings with Structure Prediction
This addresses the need for domain-specific embeddings in fields like Digital Humanities, offering a more integrated approach compared to existing methods, though it is incremental in improving dynamic embedding techniques.
The paper tackled the problem of learning dynamic word embeddings across domains without requiring post-alignment, proposing a method that simultaneously provides general and domain-specific representations along with structure prediction. The result showed better performance than baselines in analogy tests and structure prediction on datasets like New York Times articles and Wikipedia.
Complementary to finding good general word embeddings, an important question for representation learning is to find dynamic word embeddings, e.g., across time or domain. Current methods do not offer a way to use or predict information on structure between sub-corpora, time or domain and dynamic embeddings can only be compared after post-alignment. We propose novel word embedding methods that provide general word representations for the whole corpus, domain-specific representations for each sub-corpus, sub-corpus structure, and embedding alignment simultaneously. We present an empirical evaluation on New York Times articles and two English Wikipedia datasets with articles on science and philosophy. Our method, called Word2Vec with Structure Prediction (W2VPred), provides better performance than baselines in terms of the general analogy tests, domain-specific analogy tests, and multiple specific word embedding evaluations as well as structure prediction performance when no structure is given a priori. As a use case in the field of Digital Humanities we demonstrate how to raise novel research questions for high literature from the German Text Archive.