ESG (Environmental, Social, Governance)-thematics play an important role for investing as they can represent additional risk for a company and subsequently its stock price. These ESG-thematics can be extracted from alternative data sources such as news data. However, the extend to which an ESG-event influences a company can vary greatly depending on the ESG-aspect of the event and the current expectations of the affected company. In this paper, we present an approach where we leverage Dynamic Time Warping (DTW) in order to identify periods in which certain ESG-aspects are closely aligned with the stockprice movement of a company. We demonstrate this approach using more than 50,000 news articles mined for 42 german companies between January 2023 and September 2024, (i) answering the question of which aspects are important for a company in what time-frames and (ii) showing that this approach increases the performance of an ESG-sensitive portfolio by a large amount.

Fairness metrics are used to assess discrimination and bias in decision-making processes across various domains, including machine learning models and human decision-makers in real-world applications. This involves calculating the disparities between probabilistic outcomes among social groups, such as acceptance rates between male and female applicants. However, traditional fairness metrics do not account for the uncertainty in these processes and lack of comparability when two decision-makers exhibit the same disparity. Using Bayesian statistics, we quantify the uncertainty of the disparity to enhance discrimination assessments. We represent each decision-maker, whether a machine learning model or a human, by its disparity and the corresponding uncertainty in that disparity. We define preferences over decision-makers and utilize brute-force to choose the optimal decision-maker according to a utility function that ranks decision-makers based on these preferences. The decision-maker with the highest utility score can be interpreted as the one for whom we are most certain that it is fair.

Abstract In modern data analysis, time is often considered just another feature. Yet time has a special role that is regularly overlooked. Procedures are usually only designed for time-independent data and are therefore often unsuitable for the temporal aspect of the data. This is especially the case for clustering algorithms. Although there are a few evolutionary approaches for time-dependent data, the evaluation of these and therefore the selection is difficult for the user. In this paper, we present a general evaluation measure that examines clusterings with respect to their temporal stability and thus provides information about the achieved quality. For this purpose, we examine the temporal stability of time series with respect to their cluster neighbors, the temporal stability of clusters with respect to their composition, and finally conclude on the temporal stability of the entire clustering. We summarise these components in a parameter-free toolkit that we call Cluster Over-Time Stability Evaluation (CLOSE). In addition to that we present a fuzzy variant which we call FCSETS (Fuzzy Clustering Stability Evaluation of Time Series). These toolkits enable a number of advanced applications. One of these is parameter selection for any type of clustering algorithm. We demonstrate parameter selection as an example and evaluate results of classical clustering algorithms against a well-known evolutionary clustering algorithm. We then introduce a method for outlier detection in time series data based on CLOSE. We demonstrate the practicality of our approaches on three real world data sets and one generated data set.

Corpora of argumentative discourse are commonly analyzed in terms of argumentative units, consisting of claims and premises.Both argument detection and classification are complex discourse processing tasks.Our paper introduces a semantic classification of arguments that can help to facilitate argument detection.We report on our experiences with corpus annotations using a function-based classification of arguments and a procedure for operationalizing the scheme by using semantic templates.

Alpha matting aims to estimate the translucency of an object in a given image. The resulting alpha matte describes pixel-wise to what amount foreground and background colors contribute to the color of the composite image. While most methods in literature focus on estimating the alpha matte, the process of estimating the foreground colors given the input image and its alpha matte is often neglected, although foreground estimation is an essential part of many image editing workflows. In this work, we propose a novel method for foreground estimation given the alpha matte. We demonstrate that our fast multi-level approach yields results that are comparable with the state-of-the-art while outperforming those methods in computational runtime and memory usage.

As the availability and the inter-connectivity of RDF datasets grow, so does the necessity to understand the structure of the data. Understanding the topology of RDF graphs can guide and inform the development of, e.g. synthetic dataset generators, sampling methods, index structures, or query optimizers. In this work, we propose two resources: (i) a software framework (Resource URL of the framework: https://doi.org/10.5281/zenodo.2109469 ) able to acquire, prepare, and perform a graph-based analysis on the topology of large RDF graphs, and (ii) results on a graph-based analysis of 280 datasets (Resource URL of the datasets: https://doi.org/10.5281/zenodo.1214433 ) from the LOD Cloud with values for 28 graph measures computed with the framework. We present a preliminary analysis based on the proposed resources and point out implications for synthetic dataset generators. Finally, we identify a set of measures, that can be used to characterize graphs in the Semantic Web.

Learning is becoming increasingly digital, which leads to an increasing amount of data originating from educational environments. Research fields, such as educational data mining are investigating algorithms that can use this data to better understand students and the settings which they learn in. In recent years, this has repeatedly led to problems between companies that want to analyze the data, and students, parents, and schools, which do not agree with a non-transparent use of their data. In this work, we propose an approach that can lead to transparency, and thus confidence, in the use of educational data mining. Every student should decide for himself which of his data may be passed on to third parties and be used by them. In the form of opt-in checklists, the features or feature groups are provided for selection. Since not every student will allow everything, datasets with missing values are created. This requires algorithms or strategies to deal with such data. We simulate missing values on a student dataset and evaluate an approach to dealing with missing data for several prediction tasks of different difficulty levels. Depending on the amount of missing data in a dataset, and the predictive task, this approach provides useful results.

Modern e-learning systems offer students many interactive learning elements such as videos or quiz questions. For the optimization of the teaching, the learning elements themselves and the prediction of student performance, it is important for educators to figure out how student interactions can be tracked and how the tracked data can be used. Modern learning courses can consist of different Software-as-a-Service products by third-party providers. These learning elements and the associated learning environments can be both heterogeneous and distributed. This requires advanced tracking methods. For these reasons, a tracking architecture that is capable of tracking and integrating the data of different learning environments is needed. Based on the xAPI specification, we provide such an architecture in this paper.

Automatic keyphrase extraction aims at extracting a compact representation of a single document which can be used for various applications such as indexing, classification or summarization. Existing methods for keyphrase extraction usually define the set of phrases of a document as a crisp set and by scoring the phrases, they select the keyphrases of the document. In this work we define the set of phrases inside a document to be a fuzzy set, and based on the membership values of the phrases, we select the ones with higher membership values as the keyphrases of the document. Moreover we propose a novel evaluation method inspired by the Turing test which can be used for extractive summarization tasks.

A very valuable piece of information in newspaper articles is the tonality of extracted statements.For the analysis of tonality of newspaper articles either a big human effort is needed, when it is carried out by media analysts, or an automated approach which has to be as accurate as possible for a Media Response Analysis (MRA).To this end, we will compare several state-of-the-art approaches for Opinion Mining in newspaper articles in this paper.Furthermore, we will introduce a new technique to extract entropy-based word connections which identifies the word combinations which create a tonality.In the evaluation, we use two different corpora consisting of news articles, by which we show that the new approach achieves better results than the four state-of-the-art methods.

Several approaches for computing semantic similarity and relatedness measures between terms have been developed. This paper proposes a new semantic similarity measure between two nodes concentrating on nouns as well as their hypernym/hyponym relationships based on the structure of Wordnet. In particular, the similarity of two given nouns depends not only on their positions but also on their relevancy connections in the hierarchy. We evaluate our measure and the other ones on dataset of Miller-Charles and then compute the correlation coefficients to the human judgments. The experimental results show that our method outperforms edge-counting methods.

This paper addresses the issue of analyzing time series with temporal association rule mining techniques. Since originally association rule mining was developed for the analysis of transactional data, as it occurs for instance in market basket analysis, algorithms and time series have to be adapted in order to apply these techniques gainfully to the analysis of time series in general. Continuous time series of different origins can be discretized in order to mine several temporal association rules, what reveals interesting coherences in one and between pairs of time series. Depending on the domain, the knowledge about these coherences can be used for several purposes, e.g. for the prediction of future values of time series. We present a short review on different standard and temporal association rule mining approaches and on approaches that apply association rule mining to time series analysis. In addition to that, we explain in detail how some of the most interesting kinds of temporal association rules can be mined from continuous time series and present an prototype implementation. We demonstrate and evaluate our implementation on two large datasets containing river level measurement and stock data.

This paper presents an application of data mining to the medical domain sleep research, i.e. an approach for automatic sleep stage scoring and apnea-hypopnea detection. By several combined techniques (Fourier and wavelet transform, DDTW and waveform recognition), our approach extracts meaningful features (frequencies and special patterns) from EEG, ECG, EOG and EMG data, on which a decision trees classifier is built for classifying epochs into their sleep stages (according to the rules by Rechtschaffen and Kales) and annotating occurrences of apnea-hypopnea (total or partial cessation of respiration). After that, case-based reasoning is applied to improve quality. We evaluated our approach on 3 large public databases from PhysioBank, which showed an overall accuracy of 95.2% for sleep stage scoring and 94.5% for classifying apneic/non-apneic minutes.

In this work, we describe an approach which aims to make typed texts comparable with temporal data mining methods. This proposal was made in earlier work [11], but to our knowledge no significant research on this subject has been done yet. The basic idea is to derive artificial time series from texts by counting the occurrences of relevant keywords in a sliding window applied to them, and these time series can be compared with techniques of time series analysis. In this particular case the Dynamic Time Warping distance [3] was used. By extensive testing adequate parameters for time series calculation were derived, and we show that this approach might aid in the recognition of similar texts since the observed distances between similar documents are significantly lower than those between unrelated texts. Our idea might also be especially suitable for comparison in different languages since only the keyword translations must be known.

This paper considers the problem of integrating heterogeneous semi-structured data sources with the purpose of estimating integration quality (IQ). Integration of such data sources leads to results with unpredictable trustworthiness and none of the existing methods is capable of accounting for the uncertainty which is accumulated over all of the integration steps and which affects integration quality. To compute the uncertainties we suggest using a well-established statistical method Latent Class Analysis (LCA). This method allows to analyze the influence of the latent factors associated with the real-world entities on the set of data. We show on examples how the proposed approach can be used for evaluating and improving IQ giving an important tool to the users concerned with the data's trustworthiness.

S. Conrad, W. Hasselbring, A. James, D. Kambur, R.-D. Kutsche, P. Thiran; Report on the EFIS 2001 Workshop, The Computer Journal, Volume 45, Issue 2, 1 January

Abstract The paper presents concepts and ideas underlying an approach for consistency management in object‐oriented (OO) databases. In this approach constraints are considered as first class citizens and stored in a meta‐database called constraints catalog. When an object is created constraints of this object are retrieved from the constraints catalog and relationships between these constraints and the object are established. The structure of constraints has several features that enhance consistency management in OO database management systems which do not exist in conventional approaches in a satisfactory way. This includes: monitoring object consistency at different levels of update granularity, integrity independence, and efficiency of constraints maintenance; controlling inconsistent objects; enabling and disabling constraints, globally to all objects or locally to individual objects; and declaring constraints on individual objects. All these features are provided by means of basic notations of OO data models. Copyright © 2001 John Wiley & Sons, Ltd.

Techniques for the development of reliable information systems on the basis of their formal specification are the main concern in the project. Our work focuses on the specification language TROLL light which allows one to describe the part of the world to be modeled as a community of concurrently existing and communicating objects. Our specification language comes with an integrated, open development environment. The task of this environment is to give support for the creation of correct information systems. Two important ingredients of the environment are the animator and the proof support system.>

Within the compound project KorSo our team is concerned with the research on techniques and methods for the development of reliable information systems on the basis of formal specifications. Our work focuses on the specification language TROLL light which allows to describe the part of the world which is to be modeled as a community of concurrently existing and communicating objects by determining their structure as well as their behavior. Moreover we develop and implement a computer aided specification environment for TROLL light which permits a prototyping animation as well as the proof of properties of specifications. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.