Novel contexts, comprising a set of terms referring to one or more concepts, may often arise in complex querying scenarios such as in evidence-based medicine (EBM) involving biomedical literature. These may not explicitly refer to entities or canonical concept forms occurring in a fact-based knowledge source, e.g. the UMLS ontology. Moreover, hidden associations between related concepts meaningful in the current context, may not exist within a single document, but across documents in the collection. Predicting semantic concept tags of documents can therefore serve to associate documents related in unseen contexts, or categorize them, in information filtering or retrieval scenarios. Thus, inspired by the success of sequence-to-sequence neural models, we develop a novel sequence-to-set framework with attention, for learning document representations in a unique unsupervised setting, using no human-annotated document labels or external knowledge resources and only corpus-derived term statistics to drive the training, that can effect term transfer within a corpus for semantically tagging a large collection of documents. Our sequence-to-set modeling approach to predict semantic tags, gives to the best of our knowledge, the state-of-the-art for both, an unsupervised query expansion (QE) task for the TREC CDS 2016 challenge dataset when evaluated on an Okapi BM25–based document retrieval system; and also over the MLTM system baseline baseline (Soleimani and Miller, 2016), for both supervised and semi-supervised multi-label prediction tasks on the del.icio.us and Ohsumed datasets. We make our code and data publicly available.

Contextualized language modeling using deep Transformer networks has been applied to a variety of natural language processing tasks with remarkable success. However, we find that these models are not a panacea for a question-answering dialogue agent corpus task, which has hundreds of classes in a long-tailed frequency distribution, with only thousands of data points. Instead, we find substantial improvements in recall and accuracy on rare classes from a simple one-layer RNN with multi-headed self-attention and static word embeddings as inputs. While much research has used attention weights to illustrate what input is important for a task, the complexities of our dialogue corpus offer a unique opportunity to examine how the model represents what it attends to, and we offer a detailed analysis of how that contributes to improved performance on rare classes. A particularly interesting phenomenon we observe is that the model picks up implicit meanings by splitting different aspects of the semantics of a single word across multiple attention heads.