Skip to main content
Top

2023 | Book

Knowledge Graphs and Semantic Web

5th Iberoamerican Conference and 4th Indo-American Conference, KGSWC 2023, Zaragoza, Spain, November 13–15, 2023, Proceedings

Editors: Fernando Ortiz-Rodriguez, Boris Villazón-Terrazas, Sanju Tiwari, Carlos Bobed

Publisher: Springer Nature Switzerland

Book Series : Lecture Notes in Computer Science

insite
SEARCH

About this book

This book constitutes the refereed proceedings of the 5th Iberoamerican Conference and 4th Indo-American Conference on Knowledge Graphs and Semantic Web, KGSWC 2023, held jointly in Zaragoza, Spain, during November 13–15, 2023.
The 18 full and 2 short papers presented were carefully reviewed and selected from 50 submissions. They focus on the following topics: knowledge representation; natural language processing/text mining; and machine/deep learning research.

Table of Contents

Frontmatter
An Ontology for Tuberculosis Surveillance System
Abstract
Existing epidemiological surveillance systems use relational databases to store data and the SQL language to get information and automatically build statistics tables and graphics. However, a lack of logical and machine-readable relations among relational databases prevent computer-assisted automated reasoning and useful information may be lost. To overcome this difficulty, we propose the use of an ontology based-approach. Given that existing ontologies for epidemiological surveillance of TB does not exist, in this article, we present how we developed with the help of an epidemiologist an ontology for TB Surveillance System (O4TBSS). Currently, this ontology contains 807 classes, 117 ObjectProperties, 19 DataProperties.
Azanzi Jiomekong, Hippolyte Tapamo, Gaoussou Camara
Should I Stay or Should I Go
A New Reasoner for Description Logic
Abstract
We present the Emi reasoner, based on a new interpretation of the tableau algorithm for reasoning with Description Logics with unique performance characteristics and specialized advantages. Emi turns the tableau inside out, solving the satisfiability problem by adding elements to expressions rather than adding expressions to element node labels. This strategy is inspired by decidable reasoning algorithms for Horn Logics and \(\mathcal{E}\mathcal{L}^{++}\) that run on a loop rather than recursive graph-based strategies used in a tableau reasoner. Because Emi solves the same problem there will be a simple correspondence with tableaux, yet it will feel very different during execution, since the problem is inverted. This inversion makes possible many unique and straightforward optimizations, such as paralellization of many parts of the reasoning task, concurrent ABox expansion, and localized blocking techniques. Each of these optimizations contains a design trade-off that allows Emi to perform extremely well in certain cases, such as instance retrieval, and not as well in others. Our initial evaluations show that even a naive and largely un-optimized implementation of Emi is performant with popular reasoners running on the JVM such as Hermit, Pellet, and jFact.
Aaron Eberhart, Joseph Zalewski, Pascal Hitzler
An Intelligent Article Knowledge Graph Formation Framework Using BM25 Probabilistic Retrieval Model
Abstract
Knowledge graphs are considered as the best practice to illustrate semantic relationships among the collection of documents. The research article presents an intelligent article knowledge graph formation framework that utilizes the BM25 probabilistic retrieval model for constructing a knowledge graph to express conditional dependency structure between articles. The framework generates conditional probability features by employing the Okapi BM25 Score and rank documents according to the search query. Also, the CRF model is trained to perform Named Entity Recognition to empower semantic relationships. This indicates that a BM25 score feature matrix is created of all named entity in the collection which comes out as high dimensional. This problem is rectified using singular value decomposition dimension reduction technique and pairwise cosine similarity is calculated on reduced dimensions. Based on a similarity value, the documents are connected and construct the knowledge graph structure. The results represent the outcome of the search engine for different types of queries on IMDB database and constructed knowledge graphs based on similarity between articles in BBC and IMDB Dataset.
Jasir Mohammad Zaeem, Vibhor Garg, Kirti Aggarwal, Anuja Arora
Defeasible Reasoning with Knowledge Graphs
Abstract
Human knowledge is subject to uncertainties, imprecision, incompleteness and inconsistencies. Moreover, the meaning of many everyday terms is dependent on the context. That poses a huge challenge for the Semantic Web. This paper introduces work on an intuitive notation and model for defeasible reasoning with imperfect knowledge, and relates it to previous work on argumentation theory. PKN is to N3 as defeasible reasoning is to deductive logic. Further work is needed on an intuitive syntax for describing reasoning strategies and tactics in declarative terms, drawing upon the AIF ontology for inspiration. The paper closes with observations on symbolic approaches in the era of large language models.
Dave Raggett
Recommendation of Learning Paths Based on Open Educational Resources
Abstract
Open Educational Resources include different types of material for learning and teaching. Over time the number of them has been increasing to a great extent. Although the availability of educational material is beneficial for teachers and learners; however, the search for relevant material becomes a complex task due to the limited availability of specialized tools to locate content that meets the learners’ level of knowledge. The current research presents a recommendation service that provides a learning path based on Open Educational Resources. The learning path is created according to the topic of interest of users, and the level of understanding that they have about a particular topic. The recommendation method is based on a knowledge graph, that is created based on the metadata of educational resources obtained from an academic repository. Then, the graph is enriched by three methods: 1) keyword reconciliation using Spanish DBPedia as a target, 2) semantic annotation to find semantic resources, and 3) identification of the level of knowledge of each OER associated with a particular topic. The enriched graph is stored in GraphDB, a repository that provides the creation of semantic similarity indexes to generate recommendations. Results are compared with the TF-IDF measure and validated with the precision metric.
Jonathan Yaguana, Janneth Chicaiza
An Enterprise Architecture for Interpersonal Activity Knowledge Management
Abstract
In today’s economy, knowledge is essential for organizations’ growth as it allows them to solve problems, be productive, make decisions, and be competitive. Moreover, personal know-how on organizations’ activities possessed by individuals, which can benefit other persons within the organization and contribute to the organization’s growth, needs to be better managed within information systems generally not designed for these purposes.
Nevertheless, the efficient use of explicit and implicit personal know-how of organizations’ activities requires an adequate enterprise architecture to perform tasks such as collecting, transforming, sharing, and using interpersonal activity knowledge. However, existing enterprise architectures that support explicit knowledge do not offer efficient means to capitalize on this humans’ interpersonal activity knowledge.
This study provides a holistic view of an enterprise architecture that allows organizations to acquire, transform, share, and use interpersonal activity knowledge for persons and the organization’s growth. It describes the proposed architecture through the prism of the Zachman enterprise architecture framework and an information system.
Serge Sonfack Sounchio, Laurent Geneste, Bernard Kamsu-Foguem, Cédrick Béler, Sina Namaki Araghi, Muhammad Raza Naqvi
Unlocking the Power of Semantic Interoperability in Industry 4.0: A Comprehensive Overview
Abstract
As Industry 4.0 continues to transform the current manufacturing scene, seamless integration and intelligent data use have emerged as important aspects for increasing efficiency, productivity, and creativity. Semantic interoperability, a critical notion in this disruptive era, enables machines, systems, and humans to comprehend and interpret data from disparate sources, resulting in improved cooperation and informed decision-making. This article presents a thorough overview of semantic interoperability in the context of Industry 4.0, emphasizing its core concepts, problems, and consequences for smart manufacturing. Businesses may unlock the full power of interoperability and promote a new level of data-driven insights and optimizations by investigating the potential of semantic technologies such as ontologies, linked data, and standard data models. The goal of this paper is to provide a full knowledge of the role of semantic interoperability in Industry 4.0, enabling enterprises to embrace the latest advances and propel themselves toward a more intelligent and connected industrial landscape.
Fatima Zahra Amara, Meriem Djezzar, Mounir Hemam, Sanju Tiwari, Mohamed Madani Hafidi
On the Representation of Dynamic BPMN Process Executions in Knowledge Graphs
Abstract
Knowledge Graphs (KGs) are a powerful tool for representing domain knowledge in a way that is interpretable for both humans and machines. They have emerged as enablers of semantic integration in various domains, including Business Process Modeling (BPM). However, existing KG-based approaches in BPM lack the ability to capture dynamic process executions. Rather, static components of BPM models, such as Business Process Model and Notation (BPMN) elements, are represented as KG instances and further enriched with static domain knowledge. This poses a challenge as most business processes exhibit inherent degrees of freedom, leading to variations in their executions. To address this limitation, we examine the semantic modeling of BPMN terminology, models, and executions within a shared KG to facilitate the inference of new insights through observations of process executions. We address the issue of representing BPMN models within the concept or instance layer of a KG, comparing potential implementations and outlining their advantages and disadvantages in the context of a human-AI collaboration use case from a European smart manufacturing project.
Franz Krause, Kabul Kurniawan, Elmar Kiesling, Heiko Paulheim, Axel Polleres
Towards a Framework for Seismic Data
Abstract
Over the past decade, Knowledge Graphs (KGs) gained significant attention as a powerful method for knowledge representation. Driven by increasing interest, a paradigm shift has occurred, where the technology of KGs has transitioned from the research domain to the industry and public sector, with companies and organizations increasingly representing their data as Linked Open Data, gaining in that way significant traction for this technology. This paper focuses on KGs in the context of environmental challenges. More specifically, this work concerns KGs that contain seismic event data, such as location, timestamp, magnitude, depth, target date, as well as images before and after the event occurrence. Moreover, a Natural Language Processing (NLP) module is integrated to enhance the KG. That module enables users to query for seismic events in a free-text manner, before addressing them with a relevant response through a dedicated Information Retrieval (IR) component. The KG was constructed with data retrieved from the Instituto Nazionale di Geofisica e Vulcanologia, a rich resource that comes with earthquake-related information, such as magnitude, depth, occurrence location, and timestamp. Additionally, public APIs from the Copernicus Open Access Data Hub and ONDA DIAS are leveraged to provide access to sentinel data, such as images of the event location before and after its occurrence.
Valadis Mastoras, Alexandros Vassiliades, Maria Rousi, Sotiris Diplaris, Thanassis Mavropoulos, Ilias Gialampoukidis, Stefanos Vrochidis, Ioannis Kompatsiaris
Using Pregel to Create Knowledge Graphs Subsets Described by Non-recursive Shape Expressions
Abstract
Knowledge Graphs have been successfully adopted in recent years, existing general-purpose ones, like Wikidata, as well as domain-specific ones, like UniProt. Their increasing size poses new challenges to their practical usage. As an example, Wikidata has been growing the size of its contents and their data since its inception making it difficult to download and process its data. Although the structure of Wikidata items is flexible, it tends to be heterogeneous: the shape of an entity representing a human is distinct from that of a mountain. Recently, Wikidata adopted Entity Schemas to facilitate the definition of different schemas using Shape Expressions, a language that can be used to describe and validate RDF data. In this paper, we present an approach to obtain subsets of knowledge graphs based on Shape Expressions that use an implementation of the Pregel algorithm implemented in Rust. We have applied our approach to obtain subsets of Wikidata and UniProt and present some of these experiments’ results.
Ángel Iglesias Préstamo, Jose Emilio Labra Gayo
ITAQ: Image Tag Recommendation Framework for Aquatic Species Integrating Semantic Intelligence via Knowledge Graphs
Abstract
In the era of Web 3.0, there is an increasing demand for social image tagging that incorporates knowledge-centric paradigms and adheres to semantic web standards. This paper introduces the ITAQ framework, a recommendation framework specifically designed for tagging images of aquatic species. The framework continuously integrates strategic knowledge curation and addition at various levels, encompassing topic modelling, metadata generation, metadata classification, ontology integration, and enrichment using knowledge graphs and sub graphs. The ITAQ framework calculates context trees from the enriched knowledge dataset using AdaBoost classifier which is a lightweight machine learning classifier. The CNN classifier handles the metadata, ensuring a well-balanced fusion of learning paradigms while maintaining computational feasibility. The intermediate derivation of context trees, computation of KL divergence, and Second Order Co-occurrence PMI contribute to semantic-oriented reasoning by leveraging semantic relatedness. The Ant Lion optimization is utilized to compute the most optimal solution by building upon the initial intermediate solution. Finally, the optimal solution is correlated with image tags and categories, leading to the finalization of labels and annotations. An overall precision of 94.07% with the lowest value of FDR of 0.06% and accuracy of 95.315 % has been achieved by the proposed work.
S. S. Nitin Hariharan, Gerard Deepak, Fernando Ortiz-Rodríguez, Ronak Panchal
Automatic Semantic Typing of Pet E-commerce Products Using Crowdsourced Reviews: An Experimental Study
Abstract
This paper considers the problem of semantically typing pet products using only independent and crowdsourced reviews provided for them on e-commerce websites by customers purchasing the product, rather than detailed product descriptions. Instead of proposing new methods, we consider the feasibility of established text classification algorithms in support of this goal. We conduct a detailed series of experiments, using three different methodologies and a two-level pet product taxonomy. Our results show that classic methods can serve as robust solutions to this problem, and that, while promising when more data is available, language models and word embeddings tend both to be more computationally intensive, as well as being susceptible to degraded performance in the long tail.
Xinyu Liu, Tiancheng Sun, Diantian Fu, Zijue Li, Sheng Qian, Ruyue Meng, Mayank Kejriwal
A Modular Ontology for MODS – Metadata Object Description Schema
Abstract
The Metadata Object Description Schema (MODS) was developed to describe bibliographic concepts and metadata and is maintained by the Library of Congress. Its authoritative version is given as an XML schema based on an XML mindset which means that it has significant limitations for use in a knowledge graphs context. We have therefore developed the Modular MODS Ontology (MMODS-O) which incorporates all elements and attributes of the MODS XML schema. In designing the ontology, we adopt the recent Modular Ontology Design Methodology (MOMo) with the intention to strike a balance between modularity and quality ontology design on the one hand, and conservative backward compatibility with MODS on the other.
Rushrukh Rayan, Cogan Shimizu, Heidi Sieverding, Pascal Hitzler
A Topic Model for the Data Web
Abstract
The usage of knowledge graphs in industry and at Web scale has increased steadily within recent years. However, the decentralized approach to data creation which underpins the popularity of knowledge graphs also comes with significant challenges. In particular, gaining an overview of the topics covered by existing datasets manually becomes a gargantuan if not impossible feat. Several dataset catalogs, portals and search engines offer different ways to interact with lists of available datasets. However, these interactions range from keyword searches to manually created tags and none of these solutions offers an easy access to human-interpretable categories. In addition, most of these approaches rely on metadata instead of the dataset itself. We propose to use topic modeling to fill this gap. Our implementation LODCat automatically creates human-interpretable topics and assigns them to RDF datasets. It does not need any metadata and solely relies on the provided RDF dataset. Our evaluation shows that LODCat can be used to identify the topics of hundreds of thousands of RDF datasets. Also, our experiment results suggest that humans agree with the topics that LODCat assigns to RDF datasets. Our code and data are available online.
Michael Röder, Denis Kuchelev, Axel-Cyrille Ngonga Ngomo
KnowWhereGraph-Lite: A Perspective of the KnowWhereGraph
Abstract
KnowWhereGraph (KWG) is a massive, geo-enabled knowledge graph with a rich and expressive schema. KWG comes with many benefits including helping to capture detailed context of the data. However, the full KWG can be commensurately difficult to navigate and visualize for certain use cases, and its size can impact query performance and complexity. In this paper, we introduce a simplified framework for discussing and constructing perspectives of knowledge graphs or ontologies to, in turn, construct simpler versions; describe our exemplar KnowWhereGraph-Lite (KWG-Lite), which is a perspective of the KnowWhereGraph; and introduce an interface for navigating and visualizing entities within KWG-Lite called KnowWherePanel.
Cogan Shimizu, Shirly Stephen, Antrea Christou, Kitty Currier, Mohammad Saeid Mahdavinejad, Sanaz Saki Norouzi, Abhilekha Dalal, Adrita Barua, Colby K. Fisher, Anthony D’Onofrio, Thomas Thelen, Krzysztof Janowicz, Dean Rehberger, Mark Schildhauer, Pascal Hitzler
Automating the Generation of Competency Questions for Ontologies with AgOCQs
Abstract
Competency Questions (CQs) are natural language questions drawn from a chosen subject domain and are intended for use in ontology engineering processes. Authoring good quality and answerable CQs has been shown to be difficult and time-consuming, due to, among others, manual authoring, relevance, answerability, and re-usability. As a result, few ontologies are accompanied by few CQs and their uptake among ontology developers remains low. We aim to address the challenges with manual CQ authoring through automating CQ generation. This novel process, called AgOCQs, leverages a combination of Natural Language Processing (NLP) techniques, corpus and transfer learning methods, and an existing controlled natural language for CQs. AgOCQs was applied to CQ generation from a corpus of Covid-19 research articles, and a selection of the generated questions was evaluated in a survey. 70% of the CQs were judged as being grammatically correct by at least 70% of the participants. For 12 of the 20 evaluated CQs, the ontology expert participants deemed the CQs to be answerable by an ontology at a range of 50\(\%\)-85\(\%\) across the CQs, with the remainder uncertain. This same group of ontology experts found the CQs to be relevant between 70\(\%\)-93\(\%\) across the 12 CQs. Finally, 73\(\%\) of the users group and 69\(\%\) of the ontology experts judged all the CQs to provide clear domain coverage. These findings are promising for the automation of CQs authoring, which should reduce development time for ontology developers.
Mary-Jane Antia, C. Maria Keet
Ontology-Based Models of Chatbots for Populating Knowledge Graphs
Abstract
Knowledge graphs and graph databases are nowadays extensively used in various domains. However, manually creating knowledge graphs using existing ontology concepts presents significant challenges. On the other hand, chatbots are one of the most prominent technologies in the recent past. In this paper, we explore the idea of utilizing chatbots to facilitate the manual population of knowledge graphs. To implement these chatbots, we generate them based on other special knowledge graphs that serve as models of chatbots. These chatbot models are created using our modelling ontology (specially designed for this purpose) and ontologies from a specific domain. The proposed approach enables the manual population of knowledge graphs in a more convenient manner through the use of automatically generated conversational agents based on our chatbot models.
Petko Rutesic, Dennis Pfisterer, Stefan Fischer, Heiko Paulheim
An Ontology for Reasoning About Fairness in Regression and Machine Learning
Abstract
As concerns have grown about bias in ML models, the field of ML fairness has expanded considerably beyond classification. Researchers now propose fairness metrics for regression, but unlike classification there is no literature review of regression fairness metrics and no comprehensive resource to define, categorize, and compare them. To address this, we have surveyed the field, categorized metrics according to which notion of fairness they measure, and integrated them into an OWL2 ontology for fair regression extending our previously-developed ontology for reasoning about concepts in fair classification. We demonstrate its usage through an interactive web application that dynamically builds SPARQL queries to display fairness metrics meeting users’ selected requirements. Through this research, we provide a resource intended to support fairness researchers and model developers alike, and demonstrate methods of making an ontology accessible to users who may be unfamiliar with background knowledge of its domain and/or ontologies themselves.
Jade S. Franklin, Hannah Powers, John S. Erickson, Jamie McCusker, Deborah L. McGuinness, Kristin P. Bennett
SCIVO: Skills to Career with Interests and Values Ontology
Abstract
While getting a doctorate degree, new skills are acquired, opening up multiple traditional and non-traditional avenues of future employment. To help doctoral students explore the available career paths based on their skills, interests and values, we built a findable, accessible, interoperable, and reusable Skills to Career with Interests and Values ontology (SCIVO). It is a compact ontology of seven classes to harmonize the heterogeneous resources available providing information related to career paths. We demonstrate the interoperability and usability of SCIVO through building a knowledge graph using the web scraped publicly available data from the Science Individual Development Plan tool for current doctoral students - myIDP and the National Science Foundation Survey of Earned Doctorates (NSF SED) 2019 data. The generated knowledge graph (named SCIVOKG) consists of one hundred and sixteen classes and one-thousand seven-hundred and forty instances. An evaluation is conducted using application-based competency questions generated by analyzing data collected through surveys and individual interviews with current doctoral students. SCIVO provides an ontological foundation for building a harmonized resource as an aid to doctoral students in exploring the career options based on their skills, interests and values.
Resource Website:
Neha Keshan, James A. Hendler
Topic Modeling for Skill Extraction from Job Postings
Abstract
With an increase in the number of online job posts, it is becoming increasingly challenging for both job searchers and employers to navigate this large quantity of information. Therefore, it is crucial to use natural language processing techniques to analyze and draw inferences from these job postings. This study focuses on the most recent job postings in Turkiye for Computer Engineering and Management Information Systems. The objective is to extract skills for the job postings for job seekers wanting to apply for a new position. LSA is a statistical method for figuring out the underlying characteristics and meanings of sentences and words in natural language. Frequency analysis was also performed in addition to the LSA analyses with the goal of conducting a thorough examination of job postings. This study was conducted to determine and evaluate the skills that the sector actually needs. Thus, job seekers will have the chance to develop themselves in a more planned way. The findings indicate that the job postings for the two departments reflect various characteristics in terms of social and technical abilities. A higher requirement for social skills is thought to exist in the field of Management Information Systems rather than Computer Engineering. It has been discovered that employment involving data have been highly popular in recent years, and both departments often list opportunities involving data analysis.
Ekin Akkol, Muge Olucoglu, Onur Dogan
Backmatter
Metadata
Title
Knowledge Graphs and Semantic Web
Editors
Fernando Ortiz-Rodriguez
Boris Villazón-Terrazas
Sanju Tiwari
Carlos Bobed
Copyright Year
2023
Electronic ISBN
978-3-031-47745-4
Print ISBN
978-3-031-47744-7
DOI
https://doi.org/10.1007/978-3-031-47745-4

Premium Partner