Supporting consumers in making autonomous food choices that are sustainable and nutritionally complete is an increasingly complex task, that must take into account several needs to foster eating habits of health-conscious consumers, while reducing food waste and environmental impact. While generative AI and Large Language Models (LLMs) show promising results in this domain due to their natural language processing capabilities, they suffer from critical limitations, including hallucinations, knowledge gaps, and a lack of handling factual information. To mitigate such limitations, Retrieval Augmented Generation (RAG), which retrieves relevant information from external sources to enhance the capabilities of LLMs, has shown effectiveness in many domains. However, existing RAG approaches typically operate on unstructured text that lacks sophisticated symbolic representations of complex domain knowledge. This work proposes an ontology-enhanced conversational food advisory system, that integrates a modular ontology, named FoCOSA (Food Consumer-Oriented Sustainability-Aware), within several key tasks of a RAG-based system, enhancing LLM reasoning with domain knowledge, while the LLM enhances the interpretation of user requests, thus improving retrieval effectiveness and interaction fluidity. Experimental evaluations demonstrate the efficacy of the approach, and the study concludes with guidelines for selecting appropriate settings for food recommendation scenarios considering the complexity of natural language queries and other contextual factors.

Knowledge graphs (KGs) play a crucial role in the integration and organization of heterogeneous data and knowledge, enabling advanced data analytics and decision-making across various industries. This vision paper addresses critical challenges in managing KGs, emphasizing their relevance in integrating information from disparate sources. We propose the concept of knowledge graph ecosystems and life cycles to systematically manage tasks, e.g., data integration, standardization, continuous updates, efficient querying, and provenance tracking. By adopting our approach, organizations can enhance the accuracy, consistency, and reliability of KGs, thus improving knowledge management, enabling the extraction of valuable insights, and ensuring transparency and accountability.

Recent advancements in the field of Natural Language Processing, particularly the development of large-scale language models that are pretrained on vast amounts of knowledge, are creating novel opportunities within the realm of Knowledge Engineering. In this paper, we investigate the usage of large language models (LLMs) in both zero-shot and in-context learning settings to tackle the problem of extracting procedures from unstructured PDF text in an incremental question-answering fashion. In particular, we leverage the current state-of-the-art GPT-4 (Generative Pre-trained Transformer 4) model, accompanied by two variations of in-context learning that involve an ontology with definitions of procedures and steps and a limited number of samples of few-shot learning. The findings highlight both the promise of this approach and the value of the in-context learning customisations. These modifications have the potential to significantly address the challenge of obtaining sufficient training data, a hurdle often encountered in deep learning-based Natural Language Processing techniques for procedure extraction.

The ability to extract valuable information from documents and convert it into knowledge is crucial for driving technological innovation across industries. While adding metadata to manuals enhances their searchability, the real knowledge is still hidden in the procedural information they contain, which offers vital guidance for operators. Therefore, the approach of extracting and transforming unstructured human-readable information into machine-interpretable data is fundamental for establishing cutting-edge digital knowledge-based platforms. This paper presents a methodology tailored to the specific requirements of users who are seeking support in extracting and representing procedural knowledge from documents. We introduce a tool designed to support users in manually annotating procedures within PDF documents and generating a corresponding procedural knowledge graph. We assess the tool in real-world scenarios, aimed at evaluating its effectiveness in accomplishing various tasks. Finally, we generate a procedural knowledge graph that can facilitate knowledge discovery.

Abstract In recent years, Cyber Physical Production Systems and Digital Threads opened the vision on the importance of data modelling and management to lead the smart factory towards a full-fledged vertical and horizontal integration. Vertical integration refers to the full connection of smart factory levels from the work centers on the shop floor up to the business layer. Horizontal integration is realised when a single smart factory participates in multiple interleaved supply chains with different roles (e.g., main producer, supplier), sharing data and services and forming a Cyber Physical Production Network. In such an interconnected world, data and services become fundamental elements in the cyberspace to implement advanced data-driven applications such as production scheduling, energy consumption optimisation, anomaly detection, predictive maintenance, change management in Product Lifecycle Management, process monitoring and so forth. In this paper, we propose a methodology that guides the design of a portfolio of data-oriented services in a Cyber Physical Production Network. The methodology starts from the goals of the actors in the network, as well as their requirements on data and functions. Therefore, a data model is designed to represent the information shared across actors according to three interleaved perspectives, namely, product, process and industrial assets. Finally, multi-perspective data-oriented services for collecting, monitoring, dispatching and displaying data are built on top of the data model, according to the three perspectives. The methodology also includes a set of access policies for the actors in order to enable controlled access to data and services. The methodology is tested on a real case study for the production of valves in deep and ultra-deep water applications. Experimental validation in the real case study demonstrates the benefits of providing a methodological support for the design of multi-perspective data-oriented services in Cyber Physical Production Networks, both in terms of usability of the data navigation through the services and in terms of service performances in presence of Big Data.

This article investigates the evolution of data quality issues from traditional structured data managed in relational databases to Big Data. In particular, the paper examines the nature of the relationship between Data Quality and several research coordinates that are relevant in Big Data, such as the variety of data types, data sources and application domains, focusing on maps, semi-structured texts, linked open data, sensor & sensor networks and official statistics. Consequently a set of structural characteristics is identified and a systematization of the a posteriori correlation between them and quality dimensions is provided. Finally, Big Data quality issues are considered in a conceptual framework suitable to map the evolution of the quality paradigm according to three core coordinates that are significant in the context of the Big Data phenomenon: the data type considered, the source of data, and the application domain. Thus, the framework allows ascertaining the relevant changes in data quality emerging with the Big Data phenomenon, through an integrative and theoretical literature review.