Abstract
Over these past five years, significant breakthroughs, led by Transformers and large language models, have been made in understanding natural language text. However, the ability to capture contextual nuances in longer texts is still an elusive goal, let alone the understanding of consistent fine-grained narrative structures in text. These unsolved challenges and the interest in the community are at the basis of the sixth edition of Text2Story workshop to be held in Dublin on April 2nd, 2023 in conjunction with the 45th European Conference on Information Retrieval (ECIR'23). In its sixth edition, we aim to bring to the forefront the challenges involved in understanding the structure of narratives and in incorporating their representation in well-established models, as well as in modern architectures (e.g., transformers) which are now common and form the backbone of almost every IR and NLP application. It is hoped that the workshop will provide a common forum to consolidate the multi-disciplinary efforts and foster discussions to identify the wide-ranging issues related to the narrative extraction and generation task. Text2Story includes sessions devoted to full research papers, work-in-progress, demos and dissemination papers, keynote talks and space for an informal discussion of the methods, of the challenges and of the future of this research area.

Abstract
The rise of social media has brought a great transformation to the way news are discovered and shared. Unlike traditional news sources, social media allows anyone to cover a story. Therefore, sometimes an event is already discussed by people before a journalist turns it into a news article. Twitter is a particularly appealing social network for discussing events, since its posts are very compact and, therefore, contain colloquial language and abbreviations. However, its large volume of tweets also makes it impossible for a user to keep up with an event. In this work, we present TweetStream2Story, a web app for extracting narratives from tweets posted in real time, about a topic of choice. This framework can be used to provide new information to journalists or be of interest to any user who wishes to stay up-to-date on a certain topic or ongoing event. As a contribution to the research community, we provide a live version of the demo, as well as its source code.

Abstract
The first edition of the Implicit Author Characterization from Texts for Search and Retrieval (IACT'23) aims at bringing to the forefront the challenges involved in identifying and extracting from texts implicit information about authors (e.g., human or AI) and using it in IR tasks. The IACT workshop provides a common forum to consolidate multi-disciplinary efforts and foster discussions to identify the wide-ranging issues related to the task of extracting implicit author-related information from the textual content, including novel tasks and datasets. We will also discuss the ethical implications of implicit information extraction. In addition, we announce a shared task focused on automatically determining the literary epochs of written books.

Abstract
IntroductionHereditary transthyretin amyloidosis (ATTRv amyloidosis) is a rare neurological hereditary disease clinically characterized as severe, progressive, and life-threatening while the age of onset represents the moment in time when the first symptoms are felt. In this study, we present and discuss our results on the study, development, and evaluation of an approach that allows for time-to-event prediction of the age of onset, while focusing on genealogical feature construction. Materials and methodsThis research was triggered by the need to answer the medical problem of when will an asymptomatic ATTRv patient show symptoms of the disease. To do so, we defined and studied the impact of 77 features (ranging from demographic and genealogical to familial disease history) we studied and compared a pool of prediction algorithms, namely, linear regression (LR), elastic net (EN), lasso (LA), ridge (RI), support vector machines (SV), decision tree (DT), random forest (RF), and XGboost (XG), both in a classification as well as a regression setting; we assembled a baseline (BL) which corresponds to the current medical knowledge of the disease; we studied the problem of predicting the age of onset of ATTRv patients; we assessed the viability of predicting age of onset on short term horizons, with a classification framing, on localized sets of patients (currently symptomatic and asymptomatic carriers, with and without genealogical information); and we compared the results with an out-of-bag evaluation set and assembled in a different time-frame than the original data in order to account for data leakage. ResultsCurrently, we observe that our approach outperforms the BL model, which follows a set of clinical heuristics and represents current medical practice. Overall, our results show the supremacy of SV and XG for both the prediction tasks although impacted by data characteristics, namely, the existence of missing values, complex data, and small-sized available inputs. DiscussionWith this study, we defined a predictive model approach capable to be well-understood by medical professionals, compared with the current practice, namely, the baseline approach (BL), and successfully showed the improvement achieved to the current medical knowledge.

Abstract
Textual health records of cancer patients are usually protracted and highly unstructured, making it very time-consuming for health professionals to get a complete overview of the patient's therapeutic course. As such limitations can lead to suboptimal and/or inefficient treatment procedures, healthcare providers would greatly benefit from a system that effectively summarizes the information of those records. With the advent of deep neural models, this objective has been partially attained for English clinical texts, however, the research community still lacks an effective solution for languages with limited resources. In this paper, we present the approach we developed to extract procedures, drugs, and diseases from oncology health records written in European Portuguese. This project was conducted in collaboration with the Portuguese Institute for Oncology which, besides holding over 10 years of duly protected medical records, also provided oncologist expertise throughout the development of the project. Since there is no annotated corpus for biomedical entity extraction in Portuguese, we also present the strategy we followed in annotating the corpus for the development of the models. The final models, which combined a neural architecture with entity linking, achieved..1 scores of 88.6, 95.0, and 55.8 per cent in the mention extraction of procedures, drugs, and diseases, respectively.

Abstract
Temporal information extraction (TIE) has attracted a great deal of interest over the last two decades. Such endeavors have led to the development of a significant number of datasets. Despite its benefits, having access to a large volume of corpora makes it difficult to benchmark TIE systems. On the one hand, different datasets have different annotation schemes, which hinders the comparison between competitors across different corpora. On the other hand, the fact that each corpus is disseminated in a different format requires a considerable engineering effort for a researcher/practitioner to develop parsers for all of them. These constraints force researchers to select a limited amount of datasets to evaluate their systems which consequently limits the comparability of the systems. Yet another obstacle to the comparability of TIE systems is the evaluation metric employed. While most research works adopt traditional metrics such as precision, recall, and..1, a few others prefer temporal awareness - a metric tailored to be more comprehensive on the evaluation of temporal systems. Although the reason for the absence of temporal awareness in the evaluation of most systems is not clear, one of the factors that certainly weighs on this decision is the need to implement the temporal closure algorithm, which is neither straightforward to implement nor easily available. All in all, these problems have limited the fair comparison between approaches and consequently, the development of TIE systems. To mitigate these problems, we have developed tieval, a Python library that provides a concise interface for importing different corpora and is equipped with domain-specific operations that facilitate system evaluation. In this paper, we present the first public release of tieval and highlight its most relevant features. The library is available as open source, under MIT License, at PyPI1 and GitHub(2).