Clinical machine learning is increasingly multimodal, collected in both structured tabular formats and unstructured forms such as free text. We propose a novel task of exploring fairness on a multimodal clinical dataset, adopting equalized odds for the downstream medical prediction tasks. To this end, we investigate a modality-agnostic fairness algorithm - equalized odds post processing - and compare it to a text-specific fairness algorithm: debiased clinical word embeddings. Despite the fact that debiased word embeddings do not explicitly address equalized odds of protected groups, we show that a text-specific approach to fairness may simultaneously achieve a good balance of performance classical notions of fairness. Our work opens the door for future work at the critical intersection of clinical NLP and fairness.

Automatically generating question answer (QA) pairs from the rapidly growing coronavirus-related literature is of great value to the medical community. Creating high quality QA pairs would allow researchers to build models to address scientific queries for answers which are not readily available in support of the ongoing fight against the pandemic. QA pair generation is, however, a very tedious and time consuming task requiring domain expertise for annotation and evaluation. In this paper we present our contribution in addressing some of the challenges of building a QA system without gold data. We first present a method to create QA pairs from a large semi-structured dataset through the use of transformer and rule-based models. Next, we propose a means of engaging subject matter experts (SMEs) for annotating the QA pairs through the usage of a web application. Finally, we demonstrate some experiments showcasing the effectiveness of leveraging active learning in designing a high performing model with a substantially lower annotation effort from the domain experts.

In this paper we present a new method for intent recognition for complex dialog management in low resource situations. Complex dialog management is required because our target domain is real world mixed initiative food ordering between agents and their customers, where individual customer utterances may contain multiple intents and refer to food items with complex structure. For example, a customer might say “Can I get a deluxe burger with large fries and oh put extra mayo on the burger would you?” We approach this task as a multi-level sequence labeling problem, with the constraint of limited real training data. Both traditional methods like HMM, MEMM, or CRF and newer methods like DNN or BiLSTM use only homogeneous feature sets. Newer methods perform better but also require considerably more data. Previous research has done pseudo-data synthesis to obtain the required amounts of training data. We propose to use a k-NN learner with heterogeneous feature set. We used windowed word n-grams, POS tag n-grams and pre-trained word embeddings as features. For the experiments we perform a comparison between using pseudo-data and real world data. We also perform semi-supervised self-training to obtain additional labeled data, in order to better model real world scenarios. Instead of using massive pseudo-data, we show that with only less than 1% of the data size, we can achieve better result than any of the methods above by annotating real world data. We achieve labeled bracketed F-scores of 75.46, 52.84 and 49.66 for the three levels of sequence labeling where each level has a longer word span than its previous level. Overall we achieve 60.71F. In comparison, two previous systems, MEMM and DNN-ELMO, achieved 52.32 and 45.25 respectively.

With the resurgence of chat-based dialog systems in consumer and enterprise applications, there has been much success in developing data-driven and rule-based natural language models to understand human intent. Since these models require large amounts of data and in-domain knowledge, expanding an equivalent service into new markets is disrupted by language barriers that inhibit dialog automation. This paper presents a user study to evaluate the utility of out-of-the-box machine translation technology to (1) rapidly bootstrap multilingual spoken dialog systems and (2) enable existing human analysts to understand foreign language utterances. We additionally evaluate the utility of machine translation in human assisted environments, where a portion of the traffic is processed by analysts. In English→Spanish experiments, we observe a high potential for dialog automation, as well as the potential for human analysts to process foreign language utterances with high accuracy.

With the increasing number of communication platforms that offer variety of ways of connecting two interlocutors, there is a resurgence of chat-based dialog systems. These systems, typically known as chatbots have been successfully applied in a range of consumer and enterprise applications. A key technology in such chat-bots is robust natural language understanding (NLU) which can significantly influence and impact the efficacy of the conversation and ultimately the user-experience. While NLU is far from perfect, this paper illustrates the role of underspecification and its impact on successful dialog completion.

The accuracy of statistical parsing models can be improved with the use of lexical information. Statistical parsing using Lexicalized tree adjoining grammar (LTAG), a kind of lexicalized grammar, has remained relatively unexplored. We believe that is largely in part due to the absence of large corpora accurately bracketed in terms of a perspicuous yet broad coverage LTAG. Our work attempts to alleviate this difficulty. We extract different LTAGs from the Penn Treebank. We show that certain strategies yield an improved extracted LTAG in terms of compactness, broad coverage, and supertagging accuracy. Furthermore, we perform a preliminary investigation in smoothing these grammars by means of an external linguistic resource, namely, the tree families of an XTAG grammar, a hand built grammar of English.

Many traditional TAG parsers handle ambiguity by considering all of the possible choices as they unfold during parsing. In contrast , D-theory parsers cope with ambiguity by using underspecified descriptions of trees. This paper introduces a novel approach to parsing TAG, namely one that explores how D-theoretic notions may be applied to TAG parsing. Combining the D-theoretic approach to TAG parsing as we do here raises new issues and problems. D-theoretic underspecification is used as a novel approach in the context of TAG parsing for delaying attachment decisions. Conversely, the use of TAG reveals the need for additional types of underspecification that have not been considered so far in the D-theoretic framework. These include combining sets of trees into their underspecified equivalents as well as underspecifying combinations of trees. In this paper, we examine various issues that arise in this new approach to TAG parsing and present solutions to some of the problems. We also describe other issues which need to be resolved for this method of parsing to be implemented.