Inna Novalija

Bio: Inna Novalija is an academic researcher from Jožef Stefan Institute. The author has contributed to research in topics: Ontology (information science) & Computer science. The author has an hindex of 3, co-authored 16 publications receiving 53 citations.

OntoPlus : text-driven ontology extension using ontology content, structure and co-occurrence information

Abstract: This paper addresses the process of semi-automatic text-driven ontology extension using ontology content, structure and co-occurrence information. A novel OntoPlus methodology is proposed for semi-automatic ontology extension based on text mining methods. It allows for the effective extension of the large ontologies, providing a ranked list of potentially relevant concepts and relationships given a new concept (e.g., glossary term) to be inserted in the ontology. A number of experiments are conducted, evaluating measures for ranking correspondence between existing ontology concepts and new domain concepts suggested for the ontology extension. Measures for ranking are based on incorporating ontology content, structure and co-occurrence information. The experiments are performed using a well known Cyc ontology and textual material from two domains – finances and, fisheries & aquaculture. Our experiments show that the best results are achieved by combining content, structure and co-occurrence information. Furthermore, ontology content and structure seem to be more important than co-occurrence for our data in the financial domain. At the same time, ontology content and co-occurrence seem to have higher importance for our fisheries & aquaculture domain.

Human-Centric Artificial Intelligence Architecture for Industry 5.0 Applications

Abstract: Human-centricity is the core value behind the evolution of manufacturing towards Industry 5.0. Nevertheless, there is a lack of architecture that considers safety, trustworthiness, and human-centricity at its core. Therefore, we propose an architecture that integrates Artificial Intelligence (Active Learning, Forecasting, Explainable Artificial Intelligence), simulated reality, decision-making, and users' feedback, focusing on synergies between humans and machines. Furthermore, we align the proposed architecture with the Big Data Value Association Reference Architecture Model. Finally, we validate it on three use cases from real-world case studies.

XAI-KG: Knowledge Graph to Support XAI and Decision-Making in Manufacturing

Abstract: The increasing adoption of artificial intelligence requires accurate forecasts and means to understand the reasoning of artificial intelligence models behind such a forecast. Explainable Artificial Intelligence (XAI) aims to provide cues for why a model issued a certain prediction. Such cues are of utmost importance to decision-making since they provide insights on the features that influenced most certain forecasts and let the user decide if the forecast can be trusted. Though many techniques were developed to explain black-box models, little research was done on assessing the quality of those explanations and their influence on decision-making. We propose an ontology and knowledge graph to support collecting feedback regarding forecasts, forecast explanations, recommended decision-making options, and user actions. This way, we provide means to improve forecasting models, explanations, and recommendations of decision-making options. We tailor the knowledge graph for the domain of demand forecasting and validate it on real-world data.

Ontology Extension Towards Analysis of Business News

Abstract: This paper addresses the process of the ontology extension for a selected domain of interest which is defined by keywords and a glossary of relevant terms with descriptions. A new methodology for semiautomatic ontology extension, aggregating the elements of text mining and user-dialog approaches for ontology extension, is proposed and evaluated. We conduct a set of ranking, tagging and illustrative question answering experiments using Cyc ontology and business news collection. We evaluate the importance of using the textual content and structure of the ontology concept in the process of ontology extension. The experiments show that the best results are obtained with giving more to weight to ontology concept content and less weight to ontology concept structure.

Towards Active Learning Based Smart Assistant for Manufacturing

Abstract: Smart assistants in manufacturing can guide and aid on decision-making while also provide means to collect additional insights and information available to the users. A general approach for building a smart assistant that provides users with machine learning forecasts and a sequence of decision-making options is presented in this work. The system provides means for knowledge acquisition by gathering data from users. To minimize interactions and friction with users, we envision active learning can be used to get data labels for most data instances expected to be most informative. The system is demonstrated on a demand forecasting use case in manufacturing. The methodology can be extended to several use cases in manufacturing.

Using a contextual entropy model to expand emotion words and their intensity for the sentiment classification of stock market news

Abstract: Sentiment classification of stock market news involves identifying positive and negative news articles, and is an emerging technique for making stock trend predictions which can facilitate investor decision making. In this paper, we propose the presence and intensity of emotion words as features to classify the sentiment of stock market news articles. To identify such words and their intensity, a contextual entropy model is developed to expand a set of seed words generated from a small corpus of stock market news articles with sentiment annotation. The contextual entropy model measures the similarity between two words by comparing their contextual distributions using an entropy measure, allowing for the discovery of words similar to the seed words. Experimental results show that the proposed method can discover more useful emotion words and their corresponding intensity, thus improving classification performance. Performance was further improved by the incorporation of intensity into the classification, and the proposed method outperformed the previously-proposed pointwise mutual information (PMI)-based expansion methods.

From Artificial Intelligence to Explainable Artificial Intelligence in Industry 4.0: A Survey on What, How, and Where

Abstract: Nowadays, Industry 4.0 can be considered a reality, a paradigm integrating modern technologies and innovations. Artificial intelligence (AI) can be considered the leading component of the industrial transformation enabling intelligent machines to execute tasks autonomously such as self-monitoring, interpretation, diagnosis, and analysis. AI-based methodologies (especially machine learning and deep learning support manufacturers and industries in predicting their maintenance needs and reducing downtime. Explainable artificial intelligence (XAI) studies and designs approaches, algorithms and tools producing human-understandable explanations of AI-based systems information and decisions. This article presents a comprehensive survey of AI and XAI-based methods adopted in the Industry 4.0 scenario. First, we briefly discuss different technologies enabling Industry 4.0. Then, we present an in-depth investigation of the main methods used in the literature: we also provide the details of what, how, why, and where these methods have been applied for Industry 4.0. Furthermore, we illustrate the opportunities and challenges that elicit future research directions toward responsible or human-centric AI and XAI systems, essential for adopting high-stakes industry applications.

From Artificial Intelligence to Explainable Artificial Intelligence in Industry 4.0: A Survey on What, How, and Where

Abstract: Nowadays, Industry 4.0 can be considered a reality, a paradigm integrating modern technologies and innovations. Artificial intelligence (AI) can be considered the leading component of the industrial transformation enabling intelligent machines to execute tasks autonomously such as self-monitoring, interpretation, diagnosis, and analysis. AI-based methodologies (especially machine learning and deep learning support manufacturers and industries in predicting their maintenance needs and reducing downtime. Explainable artificial intelligence (XAI) studies and designs approaches, algorithms and tools producing human-understandable explanations of AI-based systems information and decisions. This article presents a comprehensive survey of AI and XAI-based methods adopted in the Industry 4.0 scenario. First, we briefly discuss different technologies enabling Industry 4.0. Then, we present an in-depth investigation of the main methods used in the literature: we also provide the details of what, how, why, and where these methods have been applied for Industry 4.0. Furthermore, we illustrate the opportunities and challenges that elicit future research directions toward responsible or human-centric AI and XAI systems, essential for adopting high-stakes industry applications.

CFinder: An intelligent key concept finder from text for ontology development

Abstract: Key concept extraction is a major step for ontology learning that aims to build an ontology by identifying relevant domain concepts and their semantic relationships from a text corpus. The success of ontology development using key concept extraction strongly relies on the degree of relevance of the key concepts identified. If the identified key concepts are not closely relevant to the domain, the constructed ontology will not be able to correctly and fully represent the domain knowledge. In this paper, we propose a novel method, named CFinder, for key concept extraction. Given a text corpus in the target domain, CFinder first extracts noun phrases using their linguistic patterns based on Part-Of-Speech (POS) tags as candidates for key concepts. To calculate the weights (or importance) of these candidates within the domain, CFinder combines their statistical knowledge and domain-specific knowledge indicating their relative importance within the domain. The calculated weights are further enhanced by considering an inner structural pattern of the candidates. The effectiveness of CFinder is evaluated with a recently developed ontology for the domain of 'emergency management for mass gatherings' against the state-of-the-art methods for key concept extraction including-Text2Onto, KP-Miner and Moki. The comparative evaluation results show that CFinder statistically significantly outperforms all the three methods in terms of F-measure and average precision.