Agreement measures are useful tools to assess the effectiveness of classifiers, being either diagnostic exams or artificial classifiers such as neural networks. They are commonly used to compare the labelling of an investigated classifier and those of the gold standard. The resulting agreement values gauge the consistency of the classifications among the considered dataset: the greater the values, the higher the consistency. The best among two classifiers is the one having the greatest agreement value with the gold standard. While these metrics effectively rank classifiers based on their similarity to the gold standard, they lack intrinsic relevance measures. It isn't clear whether an agreement score like, for instance, 0.7 is statistically significant and how difficult it is to obtain. Various scales have been developed to add context and interpretability to raw agreement values, but these are often arbitrary and specific to certain metrics. More recently, σ-significativity was introduced as a way to assess the relevance of an agreement value by calculating its probability that two randomly selected classifiers would agree at least that much with each other. This approach avoids the arbitrariness that is implicit in the agreement scales, but does not consider the asymmetry between the considered classifiers when one of them is the gold standard. This work addresses this issue and introduces a new kind of σ-significativity that measures the relevance of an agreement value$c$as the probability that the agreement value between a fixed classifier - e.g., the gold standard - and a randomly chosen classifier is lower than$c$.

The Italian Conference on Computational Logic -CILC-is the annual conference organized by GULP (Group of researchers and Users of Logic Programming).Since the first event of the series, which took place in Genoa in 1986, the annual GULP conference represents the main opportunity for Italian users, researchers and developers working in the field of computational logic to meet and exchange ideas.Over the years the conference broadened its horizons from the specific field of logic programming to include declarative programming and applications in neighboring areas such as artificial intelligence and deductive databases.This special issue contains revised and extended versions of papers presented at the 34th Italian Conference on Computational Logic -CILC 2019-which was hosted by the University of Trieste, Italy, from June 19 to June 21, 2019.The authors of selected papers were invited to submit an improved, extended version to this special issue of Fundamenta Informaticae.Those papers went through a careful review by qualified international referees.The three papers in the special issue witness the multifaceted nature of CILC, covering important topics in formal verification, automated theorem proving, and knowledge representation.We would like to thank the Editorial Office of Fundamenta Informaticae, and in particular the Editor-in-Chief Damian Niwiński.Finally, we thank the authors of the papers

Agreement measures are useful metrics to both compare different evaluations of the same diagnostic outcomes and validate new rating systems or devices. While many of them have been proposed in the literature so far, Cohen's n is still the de facto standard in gauging the agreement.Information Agreement (IA) is a novel two-observers information-theoretic-based metric introduced to overcome all the limitations and alleged pitfalls of Cohen's n. It offers an operative meaning to the agreement since it measures - in both dichotomous and multi-value ordered-categorical cases-the information shared between two raters through the virtual diagnostic channel connecting them: the more information exchanged between the raters, the higher their agreement. Unfortunately, this measure is only able to deal with agreement matrices whose values are all strictly positive numbers.This work extends IA by admitting also 0 as a possible value for the entries of an agreement matrix. Moreover, a Python software library to compute the extended version of IA, together with some of the most used agreement measures, is presented and tested.

One application of the P300 brain electric signal is sentence spelling, which enables subjects who have lost control of their motor pathways to communicate by selecting characters in a matrix containing all alphabet symbols. This technology still suffers from both low communication/high error rates. A P300 speller, named PolyMorph, which jointly introduces the selection matrix polymorphism (reducing the matrix size by removing useless symbols) and sentence-based predictions (which forecast words on the basis of language statistics) is presented. This is accomplished by using a custom dynamic knowledge-base, tailored to the subject lexicon, and updated in real time with the selections of the subject. The effectiveness of the presented speller is measured in vivo and in silico. The results suggest that the use of PolyMorph increases the number of spelt characters per time unit and reduces the error rate.

Hybrid automata naturally represent systems that exhibit a mixed discrete-continuous behaviours. The undecidability of the reach ability problem over them constrains the chances of punctually investigating this kind of formalism. Established that this negative result and the presence of artifacts, which do not correspond to any observable phenomena, are mainly due to the density of the continuous domain, a class of finite precision semantics, named [epsilon]-semantics, has been proposed to analyze hybrid automata. This paper presents a Python package, pyHybrid Analysis, that both implements the [epsilon]-semantics framework and allows to analyze hybrid automata.

Many systems, both natural and artificial, exhibit a mixed discrete-continuous behavior that cannot be fully captured by either continuous nor discrete models: they evolve in accordance to continuous laws, but these laws are controlled by a finite set of modes. Hybrid automata were proposed to represent such kind of behaviors and they have been used to model numerous natural phenomena in the last decades. Unfortunately, the Model Checking problem over them was proved undecidable and, because of that, many techniques were suggested so far to both approximate the original models and reduce the analysis complexity. This paper surveys some of such techniques and reports some open questions.

Many software tools are currently available to solve the hard goal of assembling millions of fragments produced in sequencing projects. Such a variety includes packages for long and short reads, generated by classical and next-generation sequencing technologies. Often the result produced by different tools can diverge-sometime significantly-for many reasons: the underlying algorithm, the data structures employed, the heuristics implemented, default parameters, etc. On the ground of the above considerations, we were motivated in developing a methodology which may both guide in a comparison of different assembler's output and improve the overall quality of the genome assembly sequences,by merging the sequences produced by different assembly programs.