Search

Question answering (QA) systems are among the most important and rapidly developing research topics in natural language processing (NLP). A reason, therefore, is that a QA system allows humans to interact more naturally with a machine, e.g., via a virtual assistant or search engine. In the last decades, many QA systems have been proposed to address the requirements of different question-answering tasks. Furthermore, many error scores have been introduced, e.g., based on n-gram matching, word embeddings, or contextual embeddings to measure the performance of a QA system. This survey attempts to provide a systematic overview of the general framework of QA, QA paradigms, benchmark datasets, and assessment techniques for a quantitative evaluation of QA systems. The latter is particularly important because not only is the construction of a QA system complex but also its evaluation. We hypothesize that a reason, therefore, is that the quantitative formalization of human judgment is an open problem.

Technological progress has led to powerful computers and communication technologies that penetrate nowadays all areas of science, industry and our private lives. As a consequence, all these areas are generating digital traces of data amounting to big data resources. This opens unprecedented opportunities but also challenges toward the analysis, management, interpretation and utilization of these data. Fortunately, recent breakthroughs in deep learning algorithms complement now machine learning and statistics methods for an efficient analysis of such data. Furthermore, advances in text mining and natural language processing, e.g., word-embedding methods, enable also the processing of large amounts of text data from diverse sources as governmental reports, blog entries in social media or clinical health records of patients. In this paper, we present a perspective on the role of artificial intelligence in these developments and discuss also potential problems we are facing in a digital society.

The number of scientific publications in the literature is steadily growing, containing our knowledge in the biomedical, health, and clinical sciences. Since there is currently no automatic archiving of the obtained results, much of this information remains buried in textual details not readily available for further usage or analysis. For this reason, natural language processing (NLP) and text mining methods are used for information extraction from such publications. In this paper, we review practices for Named Entity Recognition (NER) and Relation Detection (RD), allowing, e.g., to identify interactions between proteins and drugs or genes and diseases. This information can be integrated into networks to summarize large-scale details on a particular biomedical or clinical problem, which is then amenable for easy data management and further analysis. Furthermore, we survey novel deep learning methods that have recently been introduced for such tasks.

In this paper, we study several distance-based entropy measures on fullerene graphs. These include the topological information content of a graph I a ( G ) , a degree-based entropy measure, the eccentric-entropy I f &sigma, ( G ) , the Hosoya entropy H ( G ) and, finally, the radial centric information entropy H e c c . We compare these measures on two infinite classes of fullerene graphs denoted by A 12 n + 4 and B 12 n + 6 . We have chosen these measures as they are easily computable and capture meaningful graph properties. To demonstrate the utility of these measures, we investigate the Pearson correlation between them on the fullerene graphs.

In this paper, we describe {\sc quantitative graph theory} and argue it is a new graph-theoretical branch in network science, however, with significant different features compared to classical graph theory. The main goal of quantitative graph theory is the structural quantification of information contained in complex networks by employing a {\it measurement approach} based on numerical invariants and comparisons. Furthermore, the methods as well as the networks do not need to be deterministic but can be statistic. As such this complements the field of classical graph theory, which is descriptive and deterministic in nature. We provide examples of how quantitative graph theory can be used for novel applications in the context of the overarching concept network science.

Despite the development of numerous gene regulatory network (GRN) inference methods in the last years, their application, usage and the biological significance of the resulting GRN remains unclear for our general understanding of large-scale gene expression data in routine practice. In our study, we conduct a structural and a functional analysis of B-cell lymphoma GRNs that were inferred using 3 mutual information-based GRN inference methods: C3Net, BC3Net and Aracne. From a comparative analysis on the global level, we find that the inferred B-cell lymphoma GRNs show major differences. However, on the edge-level and the functional-level - that are more important for our biological understanding - the B-cell lymphoma GRNs were highly similar among each other. Also, the ranks of the degree centrality values and major hub genes in the inferred networks are highly conserved as well. Interestingly, the major hub genes of all GRNs are associated with the G-protein-coupled receptor pathway, cell-cell signaling and cell cycle. This implies that hub genes of the GRNs can be highly consistently inferred with C3Net, BC3Net and Aracne, representing prominent targets for signaling pathways. Finally, we describe the functional and structural relationship between C3Net, BC3Net and Aracne gene regulatory networks. Our study shows that these GRNs that are inferred from large-scale gene expression data are promising for the identification of novel candidate interactions and pathways that play a key role in the underlying mechanisms driving cancer hallmarks. Overall, our comparative analysis reveals that these GRNs inferred with considerably different inference methods contain large amounts of consistent, method independent, biological information.

This paper has been withdrawn., Comment: This paper has been withdrawn

In this paper, we investigate treatment cycles inferred from diabetes data by means of graph theory. We define the term treatment cycles graph-theoretically and perform a descriptive as well as quantitative analysis thereof. Also, we interpret our findings in terms of nursing and clinical management.

In this paper we present a method which assigns to each layer of a multilayer neural network, whose network dynamics is governed by a noisy winner-take-all mechanism, a neural temperature. This neural temperature is obtained by a least mean square fit of the probability distribution of the noisy winner-take-all mechanism to the distribution of a softmax mechanism, which has a well defined temperature as free parameter. We call this approximated temperature resulting from the optimization step the neural temperature. We apply this method to a multilayer neural network during learning the XOR-problem with a Hebb-like learning rule and show that after a transient the neural temperature decreases in each layer according to a power law. This indicates a self-organized annealing behavior induced by the learning rule itself instead of an external adjustment of a control parameter as in physically motivated optimization methods e.g. simulated annealing., Comment: 10 pages, 4 figures

Vignette of package GSAR. (PDF 732 kb)

The objective of this chapter is to take stock and to map the relevant issues related to data ownership. It covers the central aspects of a potential new property right regarding data: its rationale, characteristics, and implementation. Market failure can be a consequence of overly high transaction costs. According to literature, data ownership could correct this market failure by lowering transaction costs for contractual agreements involving data and by correcting a misallocation of costs and benefits, that is, the fact that data collectors are able to externalize costs while internalizing benefits. Data may be protected at three distinct levels: at the syntactic, semantic, and pragmatic level. According to World Economic Forum taxonomy, volunteered data is explicitly shared by individuals. If individuals share data that is not by them but rather about them, this data should be qualified as observed data. Finally, inferred data refers to different data types originating from various sources, used mostly for predictive purposes.

Statistical association between a single nucleotide polymorphism (SNP) genotype and a quantitative trait in genome-wide association studies is usually assessed using a linear regression model, or, in the case of non-normally distributed trait values, using the Kruskal-Wallis test. While linear regression models assume an additive mode of inheritance via equi-distant genotype scores, Kruskal-Wallis test merely tests global differences in trait values associated with the three genotype groups. Both approaches thus exhibit suboptimal power when the underlying inheritance mode is dominant or recessive. Furthermore, these tests do not perform well in the common situations when only a few trait values are available in a rare genotype category (disbalance), or when the values associated with the three genotype categories exhibit unequal variance (variance heterogeneity). We propose a maximum test based on Marcus-type multiple contrast test for relative effect sizes. This test allows model-specific testing of either dominant, additive or recessive mode of inheritance, and it is robust against variance heterogeneity. We show how to obtain mode-specific simultaneous confidence intervals for the relative effect sizes to aid in interpreting the biological relevance of the results. Further, we discuss the use of a related all-pairwise comparisons contrast test with range preserving confidence intervals as an alternative to Kruskal-Wallis heterogeneity test. We applied the proposed maximum test to the Bogalusa Heart Study dataset, and gained a remarkable increase in the power to detect association, particularly for rare genotypes. Our simulation study also demonstrated that the proposed non-parametric tests control family-wise error rate in the presence of non-normality and variance heterogeneity contrary to the standard parametric approaches. We provide a publicly available R library nparcomp that can be used to estimate simultaneous confidence intervals or compatible multiplicity-adjusted p-values associated with the proposed maximum test. DFG/Br 655/16-1 DFG/HO-1687/9