• Last recent works on real-time flood forecasting in urban drainage systems are reviewed. • A bibliometric and in-depth critical review is conducted. • All points are classified in data collection and preparation, model development and performance assessment. • Real-time data requirements and developed real-time urban flood forecasting models are discussed. There has been a strong tendency in recent decades to develop real-time urban flood prediction models for early warning to the public due to a large number of worldwide urban flood occurrences and their disastrous consequences. While a significant breakthrough has been made so far, there are still some potential knowledge gaps that need further investigation. This paper presents a comprehensive review of the current state-of-the-art and future trends of real-time modelling of flood forecasting in urban drainage systems. Findings showed that the combination of various real-time sources of rainfall measurement and the inclusion of other real-time data such as soil moisture, wind flow patterns, evaporation, fluvial flow and infiltration should be more investigated in real-time flood forecasting models. Additionally, artificial intelligence is also present in most of the new RTFF models in UDS and consequently further developments of this technique are expected to appear in future works. 

A Critical Review of Real-Time Modelling of Flood Forecasting in Urban Drainage Systems

Data mining is the discovery of meaningful information or unrevealed patterns in data. Traditional data‐mining approaches, using statistical calculations, machine learning, artificial intelligence, and database technology, cannot interpret data on a conceptual or semantic level and fail to reveal the meanings within the data. This results in a user not being analyzed and determines its signification and implications. Several semantic data‐mining approaches have been proposed in the past decade that overcome these limitations by using a domain ontology as background knowledge to enable and enhance data‐mining performance. The main contributions of this literature survey include organizing the surveyed articles in a new way that provides ease of understanding for interested researchers, and the provision of a critical analysis and summary of the surveyed articles, identifying the contribution of these papers to the field, and the limitations of the analysis methods and approaches discussed in this corpus, with the intention of informing researchers in this growing field in their innovative approaches to new research. Finally, we identify the future trends and challenges in this study track that will be of concern to future researchers, such as dynamic knowledge‐based methods or big‐data tool collaboration. This survey article provides a comprehensive overview of the literature on domain ontologies as used in the various semantic data‐mining tasks, such as preprocessing, modeling, and postprocessing. We investigated the role of semantic data mining in the field of data science and the processes and methods of applying semantic data mining to a data resource description framework.

Semantic data mining in the information age: A systematic review

Virtual sensor-based imputed graph attention network for anomaly detection of equipment with incomplete data

Missing values are a common phenomenon in all areas of applied research. While various imputation methods are available for metrically scaled variables, methods for categorical data are scarce. An imputation method that has been shown to work well for high dimensional metrically scaled variables is the imputation by nearest neighbor methods. In this paper, we extend the weighted nearest neighbors approach to impute missing values in categorical variables. The proposed method, called $\mathtt{wNNSel_{cat}}$, explicitly uses the information on association among attributes. The performance of different imputation methods is compared in terms of the proportion of falsely imputed values. Simulation results show that the weighting of attributes yields smaller imputation errors than existing approaches. A variety of real data sets is used to support the results obtained by simulations.

Nearest Neighbor Imputation for Categorical Data by Weighting of Attributes

Most data-warehouse construction processes are performed manually by experts, which is laborious, time-consuming, and prone to error. Furthermore, special knowledge is required to design complex multidimensional models, such as a star schema. This predicament has motivated computer scientists to propose automation techniques to generate such models. For this reason, we present a new strategy that incorporates knowledge-based models into a framework, named the Semantic-based Star-schema Designer, that assists the automation of star schema construction. Our models provide reasoning capabilities needed by star schema designs, including those that can disambiguate heterogeneous terms, detect appropriate data types and attribute sizes, and organize data hierarchies to support online analytical processes. We also propose strategies to overcome the uncertainty arising when attribute names are not available in the data source. The names of unknown attributes are thus predicted using an arithmetic coding technique to infer column names. Our system also generates star schema from semi-structured data (e.g., comma-separated-value files and spreadsheets), which do not provide primary keys, foreign keys, or relationship cardinalities between tables. Our framework facilitates star schema construction and their relationship information without human intervention using homegrown algorithms. Experiments demonstrate that our technique predicts column names and data types that enable the effective generation of star schema better than baseline approaches.

Intelligent approach to automated star-schema construction using a knowledge base

Missing data are a major problem that affects data analysis techniques for forecasting. Traditional methods suffer from poor performance in predicting missing values using simple techniques, e.g., mean and mode. In this paper, we present and discuss a novel method of imputing missing values semantically with the use of an ontology model. We make three new contributions to the field: first, an improvement in the efficiency of predicting missing data utilizing Particle Swarm Optimization (PSO), which is applied to the numerical data cleansing problem, with the performance of PSO being enhanced using K-means to help determine the fitness value. Second, the incorporation of an ontology with PSO for the purpose of narrowing the search space, to make PSO provide greater accuracy in predicting numerical missing values while quickly converging on the answer. Third, the facilitation of a framework to substitute nominal data that are lost from the dataset using the relationships of concepts and a reasoning mechanism concerning the knowledge-based model. The experimental results indicated that the proposed method could estimate missing data more efficiently and with less chance of error than conventional methods, as measured by the root mean square error.

N. Kamkhad

Papers

Toward semantic data imputation for a dengue dataset