This work explores the effectiveness of employing Clinical BERT for Relation Extraction (RE) tasks in medical texts within an Active Learning (AL) framework. Our main objective is to optimize RE in medical texts through AL while examining the trade-offs between performance and computation time, comparing it with alternative methods like Random Forest and BiLSTM networks. Comparisons extend to feature engineering requirements, performance metrics, and considerations of annotation costs, including AL step times and annotation rates. The utilization of AL strategies aligns with our broader goal of enhancing the efficiency of relation classification models, particularly when dealing with the challenges of annotating complex medical texts in a Human-in-the-Loop (HITL) setting. The results indicate that uncertainty-based sampling achieves comparable performance with significantly fewer annotated samples across three categories of supervised learning methods, thereby reducing annotation costs for clinical and biomedical corpora. While Clinical BERT exhibits clear performance advantages across two different corpora, the trade-off involves longer computation times in interactive annotation processes. In real-world applications, where practical feasibility and timely results are crucial, optimizing this trade-off becomes imperative.

Information extraction from clinical text has the potential to facilitate clinical research and personalized clinical care, but annotating large amounts of data for each set of target tasks is prohibitive. We present a German medical Named Entity Recognition (NER) system capable of cross-domain knowledge transferring. The system builds on a pre-trained German language model and a token-level binary classifier, employing semantic types sourced from the Unified Medical Language System (UMLS) as entity labels to identify corresponding entity spans within the input text. To enhance the system’s performance and robustness, we pre-train it using a medical literature corpus that incorporates UMLS semantic term annotations. We evaluate the system’s effectiveness on two German annotated datasets obtained from different clinics in zero- and few-shot settings. The results show that our approach outperforms task-specific Condition Random Fields (CRF) classifiers in terms of accuracy. Our work contributes to developing robust and transparent German medical NER models that can support the extraction of information from various clinical texts.

Training a model with access to human explanations can improve data efficiency and model performance on in- and out-of-domain data. Adding to these empirical findings, similarity with the process of human learning makes learning from explanations a promising way to establish a fruitful human-machine interaction. Several methods have been proposed for improving natural language processing (NLP) models with human explanations, that rely on different explanation types and mechanism for integrating these explanations into the learning process. These methods are rarely compared with each other, making it hard for practitioners to choose the best combination of explanation type and integration mechanism for a specific use-case. In this paper, we give an overview of different methods for learning from human explanations, and discuss different factors that can inform the decision of which method to choose for a specific use-case.

We study the problem of incremental domain adaptation of a generic neural machine translation model with limited resources (e.g., budget and time) for human translations or model training. In this paper, we propose a novel query strategy for selecting “unlabeled” samples from a new domain based on sentence embeddings for Arabic. We accelerate the fine-tuning process of the generic model to the target domain. Specifically, our approach estimates the informativeness of instances from the target domain by comparing the distance of their sentence embeddings to embeddings from the generic domain. We perform machine translation experiments (Ar-to-En direction) for comparing a random sampling baseline with our new approach, similar to active learning, using two small update sets for simulating the work of human translators. For the prescribed setting we can save more than 50% of the annotation costs without loss in quality, demonstrating the effectiveness of our approach.

We present a multimodal dialogue system that allows doctors to interact with a medical decision support system in virtual reality (VR). We integrate an interactive visualization of patient records and radiology image data, as well as therapy predictions. Therapy predictions are computed in real-time using a deep learning model.

Over the last several years, speech-based question answering (QA) has become very popular in contrast to pure search engine based approaches on a desktop. Open-domain QA systems are now much more powerful and precise, and they can be used in speech applications. Speech-based question answering systems often rely on predefined grammars for speech understanding. In order to improve the coverage of such complex AI systems, we reused speech patterns used to generate textual entailment patterns. These can make multimodal question understanding more robust. We exemplify this in the context of a domain-specific dialogue scenario. As a result, written text input components (e.g., in a textual input field) can deal with more flexible input according to the derived textual entailment patterns. A multimodal QA dialogue spanning over several domains of interest, i.e., personal address book entries, questions about the music domain and politicians and other celebrities, demonstrates how the textual input mode can be used in a multimodal dialogue shell.

The objective of the Semiotic-based Ontology Evaluation Tool (S-OntoEval) is to evaluate and propose improvements to a given ontological model. The evaluation aims at assessing the quality of the ontology by drawing upon semiotic theory, taking several metrics into consideration for assessing the syntactic, semantic, and pragmatic aspects of ontology quality. We consider an ontology to be a semiotic object and we identify three main types of semiotic ontology evaluation levels: the structural level, assessing the ontology syntax and formal semantics; the functional level, assessing the ontology cognitive semantics and; the usability-related level, assessing the ontology pragmatics. The Ontology Evaluation Tool implements metrics for each semiotic ontology level: on the structural level by making use of reasoner such as the RACER System and Pellet to check the logical consistency of our ontological model (TBoxes and ABoxes) and graph-theory measures such as Depth; on the functional level by making use of a task-based evaluation approach which measures the quality of the ontology based on the adequacy of the ontological model for a specific task; and on the usability-profiling level by applying a quantitative analysis of the amount of annotation. Other metrics can be easily integrated and added to the respective evaluation level. In this work, the Ontology Evaluation Tool is used to test and evaluate the SWIntO Ontology of the SmartWeb project.

General purpose ontologies and domain ontologies make up the infrastructure of the Semantic Web, which allow for accurate data representations with relations, and data inferences. In our approach to multimodal dialogue systems providing question answering functionality (SMARTWEB), the ontological infrastructure is essential. We aim at an integrated approach in which all knowledge-aware system modules are based on interoperating ontologiesin a common data model. The discourse ontology is meant to provide the necessary dialogue- and HCI concepts. We present the ontological syntactic structure of multimodal question answering results as partof this discourse ontology which extends the W3C EMMA annotation framework and uses MPEG-7 annotations. In addition, we describe anextension to ontological result structures where automatic and context-based sorting mechanisms can be naturally incorporated.