Mapping free-text mentions in clinical notes to standardized terminologies such as SNOMED CT is essential for large-scale secondary use of electronic health records, but remains challenging due to linguistic variability, under-specified annotation guidelines, term ambiguity, and ontology scale. This work presents a two-stage entity linking pipeline that combines span detection with context-aware concept linking and evaluates it on the SNOMED CT Entity Linking Challenge dataset. Our work builds upon the SNOMED CT entity linking challenge (CITATION), resulting in a fully open-source system. To our knowledge, this is the first end-to-end open-source system for this task. For span detection, we compare multiple neural architectures together with dictionary-based matching. For concept linking, we adopt a context-aware bi-encoder, and construct a multi-source knowledge base enriched with context derived from the SNOMED CT ontology. Finally, we implement an agentic re-ranker and test the effectiveness of LLM-backed re-ranking with access to annotation guidelines. In contrast to findings from the original shared task submissions, we show that context is important for optimal performance, and that agentic re-ranking with a state-of-the-art LLM only marginally improves overall performance, suggesting that the current benchmark may be approaching its practical ceiling. This work provides the first fully open-source, reproducible system for SNOMED CT entity linking, offering a foundation for future research and practical deployment.

We introduce GLiREL, an efficient architecture and training paradigm for zero-shot relation classification. Identifying relationships between entities is a key task in information extraction pipelines. The zero-shot setting for relation extraction, where a taxonomy of relations is not pre-specified, has proven to be particularly challenging because of the computational complexity of inference, and because of the lack of labeled training data with sufficient coverage. Existing approaches rely upon distant supervision using auxiliary models to generate training data for unseen labels, upon very large general-purpose large language models (LLMs), or upon complex pipelines models with multiple inference stages. Inspired by the recent advancements in zero-shot named entity recognition, this paper introduces an approach to efficiently and accurately predict zero-shot relationship labels between multiple entities in a single forward pass. Experiments using the FewRel and WikiZSL benchmarks demonstrate that our approach achieves state-of-the-art results on the zero-shot relation classification task. In addition, we contribute a protocol for synthetically-generating datasets with diverse relation labels.

Previous work on automatic news timeline summarization (TLS) leaves an unclear picture about how this task can generally be approached and how well it is currently solved. This is mostly due to the focus on individual subtasks, such as date selection and date summarization, and to the previous lack of appropriate evaluation metrics for the full TLS task. In this paper, we compare different TLS strategies using appropriate evaluation frameworks, and propose a simple and effective combination of methods that improves over the stateof-the-art on all tested benchmarks. For a more robust evaluation, we also present a new TLS dataset, which is larger and spans longer time periods than previous datasets.

Multi-document summarization (MDS) aims to compress the content in large document collections into short summaries and has important applications in story clustering for newsfeeds, presentation of search results, and timeline generation. However, there is a lack of datasets that realistically address such use cases at a scale large enough for training supervised models for this task. This work presents a new dataset for MDS that is large both in the total number of document clusters and in the size of individual clusters. We build this dataset by leveraging the Wikipedia Current Events Portal (WCEP), which provides concise and neutral human-written summaries of news events, with links to external source articles. We also automatically extend these source articles by looking for related articles in the Common Crawl archive. We provide a quantitative analysis of the dataset and empirical results for several state-of-the-art MDS techniques.

We study several methods for full or partial sharing of the decoder parameters of multi-lingual NMT models. Using only the WMT 2019 shared task parallel datasets for training, we evaluate both fully supervised and zero-shot translation performance in 110 unique translation directions. We use additional test sets and re-purpose evaluation methods recently used for unsupervised MT in order to evaluate zero-shot translation performance for language pairs where no gold-standard parallel data is available. To our knowledge, this is the largest evaluation of multi-lingual translation yet conducted in terms of the total size of the training data we use, and in terms of the number of zero-shot translation pairs we evaluate. We conduct an in-depth evaluation of the translation performance of different models, highlighting the trade-offs between methods of sharing decoder parameters. We find that models which have task-specific decoder parameters outperform models where decoder parameters are fully shared across all tasks.

The centroid-based model for extractive document summarization is a simple and fast baseline that ranks sentences based on their similarity to a centroid vector. In this paper, we apply this ranking to possible summaries instead of sentences and use a simple greedy algorithm to find the best summary. Furthermore, we show possibilities to scale up to larger input document collections by selecting a small number of sentences from each document prior to constructing the summary. Experiments were done on the DUC2004 dataset for multi-document summarization. We observe a higher performance over the original model, on par with more complex state-of-the-art methods.