Hasil untuk "History of Germany"

G. Lindzey

F. M. Dienerowitz, M. David

Abstract The question of how to deal with a pregnancy, whether desired or unwanted, is a complex biological, ethical, social, and medical issue going back for millennia. Every form of regulatory approach to this issue is culturally and temporally specific and is therefore subject to continuous change. Our look at its history and the medical, legal, and religious background begins in ancient times, progresses through history, and ends with a focus on the second half of the nineteenth century and especially the twentieth century in Germany. These ethical, moral, and medical questions are likely to have been discussed in a similar way in other parts of the Western world.

Elie Abdo, Joe Germany, Mohammad Khalil Hamdan et al.

We study ionic electrodiffusion modeled by the Nernst--Planck equations describing the evolution of $N$ ionic species in a three-dimensional incompressible fluid flowing through a porous medium. We address the long-time dynamics of the resulting system in the three-dimensional whole space $\mathbb{R}^3$. We prove that the $k$-th spatial derivatives of each ionic concentration decays to zero in $L^2$ with a sharp rate of order $t^{-\frac{3}{4}-\frac{k}{2}}$. Moreover, we investigate the behavior of the relative entropy associated with the model and show that it blows up in time with a sharp growth rate of order $\log t$.

Mohadeseh Ghavami pour sereshkeh , Amirreza Mahmoudi

With the development of technology, artificially intelligent beings have been used in almost all areas of our daily lives. It is expected that these beings, which are currently at the service of humans and facilitate their work, will reach human level in the future and be able to perform some professions. These developments also bring with them numerous legal and criminal questions. What is the legal status of artificially intelligent beings, who is criminally liable in the crimes that arise from their use, and what is their role in the criminal proceedings, are among the questions that need to be answered. The purpose of this study is to provide a general assessment of these questions by considering legal regulations. The study methodology is descriptive-analytical. In this study, we have reached the conclusion that AIs are in the position of “property”, cannot be held liable for crimes resulting from their use, and although they have an important contribution to the criminal justice process, they cannot replace the subjects of the proceedings (judge, prosecutor, lawyer). In this context, the current legal regulations are largely capable of resolving the problems that have arisen. However, if AIs are to reach the status of “humans” as fully independent and self-aware entities, fundamental changes will be needed in our legal system. 1. Introduction The term “Fourth Industrial Revolution,” which refers to the digitalization of the production chain, the coordination of machines, humans, infrastructure, and the development of intelligent systems, was first introduced in 2011 at an exhibition held in Hanover, Germany. With the Fourth Industrial Revolution, it is predicted that the use of industrial robots will increase rapidly and automation will completely replace production processes, digital technologies will have significant impacts on efficiency, productivity, income distribution, and the environment, and half of the world’s trade will be conducted using the interaction of intelligent entities. With significant advances in information technology, we are witnessing the increasing expansion of applications of artificial intelligence entities in various fields. Artificially intelligent entities are widely used in almost all areas, including military operations, the industrial sector, the service sector, insurance services, medicine, and law. Translation programs, speech recognition systems, facial recognition systems, unmanned aerial vehicles used in military operations, and software that makes legal decisions by examining litigation data are only a small part of the areas where artificial intelligence is actively used. The increasing use of artificial intelligence saves time, labor, and money. However, the increasing presence of artificial intelligence in our daily lives also brings with it numerous legal challenges. The evolution and progress in the field of law, which can be considered a set of rules regulating human behavior, occur much slower than technological advances. For this reason, it is difficult to constantly update legal regulations in line with technological advances; As a result, questions such as the definition of artificial intelligence, the legal status of artificial intelligence, and the responsibility for the results of decisions based on predictions and inferences of artificial intelligence still await answers. In the first part of this study, the concept of artificial intelligence, its types, and its historical developments are explained to provide a conceptual assessment of the subject. In the second part, the views existing in the legal doctrine on the legal status of artificial intelligence are examined, and in the third part, the assessment of artificial intelligence from the perspective of criminal law is analyzed. Finally, in the conclusion section, possible challenges in the criminal law status of artificial intelligence with the increasing use of artificial intelligent entities and related solutions and suggestions will be addressed. Definition and History of Artificial Intelligence In studies related to technological progress, the term artificial intelligence, which is called “Artificial Intelligence” in English, is frequently used. However, a single and universally accepted definition of artificial intelligence has not yet been provided. John McCarthy, who coined the term artificial intelligence, defines it as “the engineering and science of building intelligent machines.” Humans can use their intelligence to solve problems they encounter. Artificial intelligence is a human-made machine that, depending on its level of development, acquires the ability to solve problems and is able to imitate human behavior. The definition of artificial intelligence, which Sebastian Tran, director of the Stanford University Artificial Intelligence Laboratory, has called “understanding something complex and making appropriate decisions,” is now recognized as a comprehensive and accepted definition in this field. Some authors consider artificial intelligence as a machine and define it as “the ability of a machine to learn from experience, adapt to new stimuli, and perform tasks similar to humans.” In line with the explanations provided, artificial intelligence is defined as follows: "Artificial intelligence refers to the imitation of human mental processes such as perception, reasoning, learning, and problem-solving by machines." The goal of artificial intelligence technology, which is the simulation of intelligence similar to human intelligence, is to create a robotic system consisting of machines, computers, and software that acts similarly to human intelligence, understands and classifies events, learns, evaluates, makes decisions based on the results of the evaluation, and implements the decision made. Just as there is disagreement over the definition of artificial intelligence, there is no consensus on the definition of a robot. Neil M. Richards, a professor of law, defines a robot as: “Biologically non-living systems that are created to perform both physical and mental activities”; therefore, it has been emphasized that just as there is no single definition for artificial intelligence, there is no complete consensus among experts on its definition for a robot. This definition emphasizes two characteristics of robots: 1. Being biologically non-living; 2. The ability to perform physical and mental activities. 2. Methodology This research has been conducted using a descriptive-analytical method and its main goal is to examine the theoretical foundations, analyze the legal status, and identify challenges related to the criminal liability of artificial intelligence systems. Also, with a comparative look at the legal systems of other countries, an attempt is made to provide solutions to strengthen Iran's legal frameworks in this field. The required information has been collected through the library method and the sources used include specialized books, scientific articles, research reports, domestic and international laws, and official documents related to the research topic. In this research, data collected from various sources have been examined using comparative analysis, content analysis, and a critical approach. In the comparative analysis, the Iranian legal system has been compared with the laws and regulations of leading countries such as the European Union, the United States, and Japan to determine its strengths and weaknesses. Content analysis was used to examine and interpret the data from the perspective of legal arguments, and in a critical approach, the limitations and shortcomings of the Iranian legal system in facing the challenges of artificial intelligence were examined. This research was conducted in several stages: first, the background of the subject was studied and previous research and articles on the criminal liability of artificial intelligence were analyzed. Then, the existing legal situation was examined and domestic and international laws related to artificial intelligence were analyzed. Next, theoretical and practical challenges related to the criminal liability of artificial intelligence systems were identified, and finally, legal reforms and practical solutions appropriate to the conditions of Iran were proposed. This approach, which is a combination of comparative law methods and legal analysis, allows for the examination of laws and the presentation of reform proposals based on global experiences and helps the researcher to reach a comprehensive understanding of the current situation and provide practical solutions for amending laws and strengthening the Iranian legal system in this area. 3. Results and Discussion Artificial intelligence as a technological phenomenon has created numerous challenges in the field of criminal law. Studies show that in the current situation, these systems are considered more as “objects” than “persons” from a legal perspective. However, the expansion of the use of artificial intelligence and its possibility of achieving a level of self-awareness and decision-making autonomy requires a review of the fundamental concepts of criminal law. One of the important results of this research is that artificial intelligence systems in their current state lack will, awareness, and criminal intent and cannot be directly considered criminally liable. For this reason, in the event of a crime related to these systems, criminal liability is attributed to their users, programmers, or producers. This shows that current laws are mainly designed for the liability of human individuals and are inadequate in the face of the complexities of artificial intelligence systems. A comparative analysis of the legal systems of other countries showed that some countries, including the European Union, have addressed these challenges with the concept of “electronic personality.” In this approach, advanced AI systems are granted legal personality that allows them to be held accountable within a specific framework. However, this idea poses several philosophical and operational challenges and requires further studies and the development of detailed laws in Iranian criminal law. Another issue examined in this study is the role of AI in facilitating judicial processes. The findings show that AI can serve as an efficient tool in collecting evidence, analyzing data, and predicting the outcomes of lawsuits. However, there are concerns about algorithmic bias, privacy, and the potential for violating individuals’ fundamental rights, which require the development of detailed regulatory regulations to prevent the misuse of this technology. Another conclusion of the study is that with the increasing use of AI in self-driving cars and other complex devices, the likelihood of crimes such as accidents caused by defects in algorithms increases. In these cases, the current Iranian laws lack comprehensive provisions for determining criminal liability, and there is a need to more precisely define self-driving cars and determine the liability of individuals involved in the design, production, and use of these devices. Finally, the present study shows that the current Iranian laws, despite covering some aspects of the use of AI, are not sufficient to manage the emerging challenges of this technology. It is suggested that to address these shortcomings, a comprehensive legal framework be developed in which new concepts such as “electronic personality,” “producer liability,” and “joint liability” are considered. Also, the establishment of a specialized regulatory body to manage the legal challenges of AI seems necessary. These measures can help create a balance between exploiting the capabilities of AI and protecting fundamental human rights. 4. Conclusions and Future Research The advancement of AI technology and the expansion of its application have created complex issues in legal systems. The present study showed that AI in its current state lacks independent will and consciousness and cannot be recognized as criminally responsible. As a result, criminal liability for crimes related to it falls on users, producers, and programmers. However, if AI achieves the level of self-awareness, a review of legal concepts will be necessary. Current Iranian laws are mainly based on traditional approaches and are not sufficient to face the challenges of AI. In leading countries, concepts such as electronic personality have been proposed that could be the basis for AI liability in the future, but they need to be adapted to national legal structures. Also, AI has great potential to improve judicial processes, but risks such as algorithmic bias and privacy violations must also be managed. Future research could focus on exploring the possibility of granting legal personality to AI, developing regulations for self-driving cars, managing algorithmic biases, and conducting more comprehensive comparative studies to help develop more rigorous legal frameworks to manage the legal challenges posed by this technology

Yann Lelièvre, Quentin Jossart, Quentin Jossart et al.

Coastal ecosystems of sub-Antarctic islands are threatened by increasing climate-driven changes and direct anthropogenic pressures. Significant effects on marine communities are expected, but benthic ecosystems of these isolated islands remain largely under-explored. Effective preservation of these nearshore environments requires deeper ecological assessments and comprehensive biodiversity knowledge. In this regard, this study reports findings from a survey carried out in 2021 at two sites – Baie du Marin and Crique du Sphinx – located on the eastern coast of Ile de la Possession (sub-Antarctic Crozet archipelago, Southern Ocean). We investigated the composition and structure of nearshore benthic faunal communities using a quantitative fieldwork protocol and an integrative molecular- and morphology-based taxonomic approach. A total of 124 morphotypes were identified, including a high proportion (72%) of rare species. Both sites exhibited similar benthic invertebrate communities. Structurally complex habitats such as hard substrates or areas dominated by macroalgae exhibited higher species richness and diversity. The investigated benthic invertebrate communities are typical of the sub-Antarctic area but featured unique structures, including dense tube-dwelling polychaete colonies. This study will provide a baseline for future monitoring programs and for the preservation of sub-Antarctic coastal benthic ecosystems.

Jingjing Chen, Hongjie Fang, Chenxi Wang et al.

Many manipulation tasks require memory beyond the current observation, yet most visuomotor policies rely on the Markov assumption and thus struggle with repeated states or long-horizon dependencies. Existing methods attempt to extend observation horizons but remain insufficient for diverse memory requirements. To this end, we propose an object-centric history representation based on point tracking, which abstracts past observations into a compact and structured form that retains only essential task-relevant information. Tracked points are encoded and aggregated at the object level, yielding a compact history representation that can be seamlessly integrated into various visuomotor policies. Our design provides full history-awareness with high computational efficiency, leading to improved overall task performance and decision accuracy. Through extensive evaluations on diverse manipulation tasks, we show that our method addresses multiple facets of memory requirements - such as task stage identification, spatial memorization, and action counting, as well as longer-term demands like continuous and pre-loaded memory - and consistently outperforms both Markovian baselines and prior history-based approaches. Project website: http://tonyfang.net/history

Roland Baatz

This study examines the generalization performance and interpretability of machine learning (ML) models used for predicting crop yield and yield anomalies in Germany's NUTS-3 regions. Using a high-quality, long-term dataset, the study systematically compares the evaluation and temporal validation behavior of ensemble tree-based models (XGBoost, Random Forest) and deep learning approaches (LSTM, TCN). While all models perform well on spatially split, conventional test sets, their performance degrades substantially on temporally independent validation years, revealing persistent limitations in generalization. Notably, models with strong test-set accuracy, but weak temporal validation performance can still produce seemingly credible SHAP feature importance values. This exposes a critical vulnerability in post hoc explainability methods: interpretability may appear reliable even when the underlying model fails to generalize. These findings underscore the need for validation-aware interpretation of ML predictions in agricultural and environmental systems. Feature importance should not be accepted at face value unless models are explicitly shown to generalize to unseen temporal and spatial conditions. The study advocates for domain-aware validation, hybrid modeling strategies, and more rigorous scrutiny of explainability methods in data-driven agriculture. Ultimately, this work addresses a growing challenge in environmental data science: how can we evaluate generalization robustly enough to trust model explanations?

Pauline Corre

Rasa Čepaitienė

The concepts of “genocide” and “crimes against humanity”, which crystallised in Western legal and intellectual discourse in the 1960s, have been actively exploited in various domestic political struggles and international relations (the most prominent examples being the Holocaust and the Ukrainian Holodomor), and this has implications for the politics of history. The last 20 years have witnessed significant changes in the European Union’s historical politics, both in the EU’s enlargement to the East and in the EU’s own role in the world. Whereas the narrative that emerged in West Germany in the 1970s and 1980s and was consolidated in the EU in the 1980s and 1990s centred on the Holocaust as an event unique in world history in terms of its scale and consequences, with the wave of EU enlargement in 2004 it was challenged by the new EU members. The latter proposed to treat Nazi and Soviet totalitarianism equally, thus equating the Gulag with the Holocaust. This has made European historical politics more complicated and more susceptible to internal disagreements and misunderstandings between members. The aim of this article is to elaborate on the legal and political exploitation of the concept of ‘genocide’ in the intellectual discourse of the emerging European Community, and to identify the challenges posed to it by the new geopolitical circumstances, ranging from the collapse of the Soviet Union to the military invasion of Ukraine by Russia in 2022. Finally, the recent challenges posed by post-colonial and human rights discourses are discussed in more detail, disrupting the established politics of European history, even in Germany itself, which has been at the forefront of their formation and global implementation. The conclusion is that, despite attempts to find points of contact and compromise between these different models of memory, a consensus that satisfies everyone is unfortunately not currently possible, and is unlikely to ever be achieved.

O. Krause, O. Krause

<p>The British geographer Halford J. Mackinder developed two different concepts of a dystopian new global order. The first, developed in 1904 and known as the pivot area concept, was adopted by German geopolitician Karl Haushofer in the 1920s. The second, developed in 1919, was named the heartland theory and was adopted in Great Britain and the USA. Haushofer reversed the dystopian vision of the pivot area concept into a utopian concept for German world power. Due to Haushofer's adaptation, interest in Mackinder's theories rose in the USA in the 1940s. Within the process of adaptation in the USA, both concepts were intertwined, resulting in the perception of the two as a monolithic bloc. Through this multi-layered process of intercontinental reception and adaptation in Germany and the USA, the term “heartland” became a generic spatial denomination detached from the geographical region it originally prescribed, integrable with various geopolitical concepts as the centre of an imagined world order. The reduction of complexity of the theory through the translation of text into maps led to its popularization among the US public during the 1940s and 1950s. Mackinder himself laid out the flexibility of the theory's interpretive possibilities by reflexively revising the theory and adapting it to the history of events over the course of the first half of the 20th century itself. In consequence, the generic spatial denomination “heartland” and the associated adopted theory served as a geopolitical argument for the strategic narrative legitimizing US foreign policy in World War II and during the Cold War.</p>

Niels Taubert, Anne Hobert, Najko Jahn et al.

This study investigates the determinants for the uptake of institutional and subject repository Open Access (OA) in the university landscape of Germany and considers three factors: the disciplinary profile of universities, their OA infrastructures and services and large transformative agreements. The uptake of OA as well as the determinants are measured by combining several data sources (incl. Web of Science, Unpaywall, an authority file of standardised German affiliation information, the ISSN-Gold-OA 4.0 list, and lists of publications covered by transformative agreements). For universities OA infrastructures and services, a structured data collection was created by harvesting different sources of information and by manual online search. To determine the explanatory power of the different factors, a series of regression analyses was performed for different periods and for both institutional as well as subject repository OA. As a result of the regression analyses, the most determining factor for the explanation of differences in the uptake of both repository OA-types turned out to be the disciplinary profile, whereas all variables that capture local infrastructural support and services for OA turned out to be non-significant. The outcome of the regression analyses is contextualised by an interview study conducted with 20 OA officers of German universities. The contextualisation provides hints that the original function of institutional repositories, offering a channel for secondary publishing is vanishing, while a new function of aggregation of metadata and full texts is becoming of increasing importance.

W. Schütz

Pamela Akuku, Pamela Akuku, Palmira Saladié et al.

Palaeobiological and archeological excavations at the site of Ewass Oldupa, found in the western Plio-Pleistocene rift basin of Oldupai Gorge (also Olduvai Gorge), Tanzania, revealed rich fossiliferous levels and the earliest remains of human activity at Oldupai Gorge, dated to 2 million years ago. This paper provides zooarchaeological taxonomic, taphonomic, and behavioral analyses, applying several methods to explore the setting in which the assemblage was formed. We identified agency behind bone surface modifications, such as cut, tooth and percussion marks, and determined the frequency of carnivore tooth marks as well as their distribution on both discrete specimens and across species. In addition, our work revealed co-occurrence of modifications to include butchering marks and carnivore tooth marks. Ravaging levels were estimated as percentage. The faunal accumulation from Ewass Oldupa contains two cut marked specimens, together with low degrees of percussion and carnivore tooth marks, moderate ravaging, and diagenetic changes suggestive of water flow. Thus, multiple lines of evidence indicate a palimpsest accumulation. Taxonomic diversity is high, with up to 22 taxa representing diverse habitats, ranging from open grassland to wooded bushlands, as well as moist mosaics during Bed I. Overall, this archaeo-faunal assemblage speaks to increased behavioral versatility among Oldowan hominins and interactions with the carnivore guild.

Bhargav Kumar Kakumani, Suman Prabha Yadav

In this article, we consider a viscoelastic plate equation with past history, nonlinear damping, and logarithmic nonlinearity. We prove explicit and general decay rate results of the solution to the viscoelastic plate equation with past history. Convex properties, logarithmic inequalities, and generalised Young's inequality are mainly used to prove the decay estimate.

L. Szinicz

Rebecca M Puhl, Leah M Lessard, Mary S Himmelstein et al.

<h4>Background/objectives</h4>Considerable evidence from U.S. studies suggests that weight stigma is consequential for patient-provider interactions and healthcare for people with high body weight. Despite international calls for efforts to reduce weight stigma in the medical community, cross-country research is lacking in this field. This study provides the first multinational investigation of associations between weight stigma and healthcare experiences across six Western countries.<h4>Methods</h4>Participants were 13,996 adults residing in Australia, Canada, France, Germany, the UK, and the US who were actively enrolled in an internationally available behavioral weight management program. Participants completed identical online surveys in the dominant language for their country that assessed experienced weight stigma, internalized weight bias, and healthcare behaviors and experiences including perceived quality of care, avoidance or delay of seeking care, experiences with providers, and perceived weight stigma from doctors.<h4>Results</h4>Among participants who reported a history of weight stigma (56-61%), two-thirds of participants in each country reported experiencing weight stigma from doctors. Across all six countries, after accounting for demographics, BMI, and experienced stigma, participants with higher internalized weight bias reported greater healthcare avoidance, increased perceived judgment from doctors due to body weight, lower frequency of obtaining routine checkups, less frequent listening and respect from providers, and lower quality of healthcare. Additionally, experienced weight stigma (from any source) was indirectly associated with poorer healthcare experiences through weight bias internalization, consistently across the six countries.<h4>Conclusions</h4>Weight stigma in healthcare is prevalent among adults actively engaged in weight management across different Western countries, and internalized weight bias has negative implications for healthcare even after controlling for BMI. The similar findings across all six countries underscore the negative consequences of weight stigma on healthcare behaviors and experiences, and emphasize the need for collective international efforts to address this problem.

Jieling Xiao, Francesco Aletta, Antonella Radicchi et al.

The interrelationships between humans, smells and the built environment have been the focus of increasing numbers of research studies in the past ten years. This paper reviews these trends and identifies the challenges in smellscape research from three aspects: methodological approaches, artistic design interventions and museum practices, and odour policy making. In response to the gaps and challenges identified, three areas of future research have also been identified for this field: smell archives and databases, social justice within odour control and management, and research into advanced building materials.

Lutter JI, Jörres RA, Trudzinski FC et al.

Johanna I Lutter,1 Rudolf A J&ouml;rres,2 Franziska C Trudzinski,3 Peter Alter,4 Christina Kellerer,2,5 Henrik Watz,6 Tobias Welte,7 Robert Bals,8 Diego Kauffmann-Guerrero,9 J&uuml;rgen Behr,9 Rolf Holle,10 Claus F Vogelmeier,4 Kathrin Kahnert9 On behalf of the COSYCONET study group1Institute of Health Economics and Health Care Management, Helmholtz Zentrum M&uuml;nchen GmbH &ndash; German Research Center for Environmental Health, Comprehensive Pneumology Center Munich (CPC-M), Member of the German Center for Lung Research, Munich, Germany; 2Institute and Outpatient Clinic for Occupational, Social and Environmental Medicine, LMU Hospital, Comprehensive Pneumology Center Munich (CPC-M), Munich, Germany; 3Department of Pneumology and Critical Care Medicine, Thoraxklinik University of Heidelberg, Translational Lung Research Center Heidelberg (TLRC-H), Member of the German Center for Lung Research (DZL), Heidelberg, Germany; 4Department of Medicine, Pulmonary and Critical Care Medicine, University Medical Center Giessen and Marburg, Philipps-University Marburg (UMR), Germany, Member of the German Center for Lung Research (DZL), Marburg, Germany; 5School of Medicine, Institute of General Practice and Health Services Research, Technical University of Munich (TUM), Munich, Germany; 6Pulmonary Research Institute at LungenClinic Grosshansdorf, Airway Research Center North (ARCN), Member of the German Center for Lung Research (DZL), Grosshansdorf, Germany; 7Department of Pneumology, Hannover Medical School, Hannover, Germany; 8Department of Internal Medicine V &ndash; Pulmonology, Allergology, Respiratory Intensive Care Medicine, Saarland University Hospital, Homburg, Germany; 9Department of Medicine V, University of Munich (LMU), Comprehensive Pneumology Center, Member of the German Center for Lung Research (DZL), Munich, Germany; 10Institute for Medical Informatics, Biometry and Epidemiology, LMU Hospital, Munich, GermanyCorrespondence: Kathrin KahnertDepartment of Internal Medicine V, University of Munich (LMU), LMU Hospital, Ziemssenstra&szlig;e 1, Munich, 80336, GermanyEmail Kathrin.Kahnert@med.uni-muenchen.deBackground: In COPD patients of GOLD groups A and B, a high degree of treatment with inhaled corticosteroids (ICS) has been reported, which is regarded as overtreatment according to GOLD recommendations. We investigated which factors predict ICS use and which relationship it has to clinical and functional outcomes, or healthcare costs.Methods: We used pooled data from visits 1 and 3 of the COSYCONET cohort (n=2741, n=2053, interval 1.5 years) including patients categorized as GOLD grades 1&ndash; 4 and GOLD group A or B at both visits (n=1080). Comparisons were performed using ANOVA, and regression analyses using propensity matching and inverse probability weighting to adjust for differences between ICS groups. These were defined as having ICS at both visits (always) vs no ICS at both visits (never). Measures were divided into predictors of ICS treatment and outcomes.Results: Among 1080 patients, 608 patients were eligible for ICS groups (n=297 never, n=311 always). Prior to matching, patients with ICS showed significantly (p&lt; 0.05 each) impaired lung function, symptoms and exacerbation history. After matching, the outcomes generic quality of life and CO diffusing capacity were increased in ICS patients (p&lt; 0.05 each). Moreover, costs for respiratory medication, but not total health care costs, were significantly elevated in the ICS group by 780&euro; per year.Conclusion: ICS therapy in COPD GOLD A/B patients can have small positive and negative effects on clinical outcomes and health care costs, indicating that the clinical evaluation of ICS over-therapy in COPD requires a multi-dimensional approach.Keywords: COPD, inhaled corticosteroids, GOLD groups, overtreatment

Keping Bi, Pavel Metrikov, Chunyuan Li et al.

An effective email search engine can facilitate users' search tasks and improve their communication efficiency. Users could have varied preferences on various ranking signals of an email, such as relevance and recency based on their tasks at hand and even their jobs. Thus a uniform matching pattern is not optimal for all users. Instead, an effective email ranker should conduct personalized ranking by taking users' characteristics into account. Existing studies have explored user characteristics from various angles to make email search results personalized. However, little attention has been given to users' search history for characterizing users. Although users' historical behaviors have been shown to be beneficial as context in Web search, their effect in email search has not been studied and remains unknown. Given these observations, we propose to leverage user search history as query context to characterize users and build a context-aware ranking model for email search. In contrast to previous context-dependent ranking techniques that are based on raw texts, we use ranking features in the search history. This frees us from potential privacy leakage while giving a better generalization power to unseen users. Accordingly, we propose a context-dependent neural ranking model (CNRM) that encodes the ranking features in users' search history as query context and show that it can significantly outperform the baseline neural model without using the context. We also investigate the benefit of the query context vectors obtained from CNRM on the state-of-the-art learning-to-rank model LambdaMart by clustering the vectors and incorporating the cluster information. Experimental results show that significantly better results can be achieved on LambdaMart as well, indicating that the query clusters can characterize different users and effectively turn the ranking model personalized.

Guangzhao Cheng, Siliang Tang

In many real-world games, such as traders repeatedly bargaining with customers, it is very hard for a single AI trader to make good deals with various customers in a few turns, since customers may adopt different strategies even the strategies they choose are quite simple. In this paper, we model this problem as fast adaptive learning in the finitely repeated games. We believe that past game history plays a vital role in such a learning procedure, and therefore we propose a novel framework (named, F3) to fuse the past and current game history with an Opponent Action Estimator (OAE) module that uses past game history to estimate the opponent's future behaviors. The experiments show that the agent trained by F3 can quickly defeat opponents who adopt unknown new strategies. The F3 trained agent obtains more rewards in a fixed number of turns than the agents that are trained by deep reinforcement learning. Further studies show that the OAE module in F3 contains meta-knowledge that can even be transferred across different games.