Article
Biochemical Research Methods
Juan J. Lastra-Diaz, Alicia Lara-Clares, Ana Garcia-Serrano
Summary: This paper introduces an updated version of the HESML Java software library for the biomedical domain, which implements efficient and scalable ontology representation methods and proposes a new shortest-path algorithm for taxonomies. The algorithm allows for real-time computation of path-based semantic similarity measures.
BMC BIOINFORMATICS
(2022)
Article
Biochemical Research Methods
Fernando Zhapa-Camacho, Maxat Kulmanov, Robert Hoehndorf
Summary: Motivation: This research developed a Python library called mOWL for machine learning with ontologies formalized in the Web Ontology Language (OWL). MOWL implements ontology embedding methods to map information in formal knowledge bases and ontologies into vector spaces, preserving properties and relations in ontologies. It also provides methods for similarity computation, deductive inference, and zero-shot learning. The library was demonstrated on knowledge-based predictions of protein-protein interactions using the gene ontology and gene-disease associations using phenotype ontologies.
Article
Computer Science, Artificial Intelligence
Yuanfei Deng, Wen Bai, Yuncheng Jiang, Yong Tang
Summary: Semantic similarity is a fundamental task in natural language processing that determines the similarity between two concepts in a taxonomy. This paper proposes a method that utilizes heterogeneous knowledge graphs and multi-view features to improve concept representation and calculate semantic similarity.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Abdullah Almuhaimeed, Mohammed A. Alhomidi, Mohammed N. Alenezi, Emad Alamoud, Saad Alqahtani
Summary: With the widespread of data resources on the internet, overlapping between these resources can provide researchers with more information. Extracting and calculating the semantic similarity between these resources is a challenging task due to their varying descriptions. To address this issue, the paper presents a new semantic similarity method that considers different factors to calculate the semantic similarity between different resources. By utilizing node descriptions and relations from multiple ontologies, this method strengthens the similarity relations between resources and discovers new semantic similarities.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Samira Babalou, Alsayed Algergawy, Birgitta Koenig-Ries
Summary: Computing the semantic similarity between pairs of terms is crucial for various shared data applications. One way to determine term similarity is to assess their word similarity using knowledge resources like ontologies or corpora. Information-theoretic approaches have shown promise by computing concept information content from ontologies. Choosing a suitable subsumer, called Consensus Common Subsumer (CCS), among common ancestors of two concepts can impact the quality of term similarity assessment.
DATA & KNOWLEDGE ENGINEERING
(2023)
Article
Mathematical & Computational Biology
Nicolas Matentzoglu, Damien Goutte-Gattat, Shawn Zheng Kai Tan, James P. Balhoff, Seth Carbon, Anita R. Caron, William D. Duncan, Joe E. Flack, Melissa Haendel, Nomi L. Harris, William R. Hogan, Charles Tapley Hoyt, Rebecca C. Jackson, HyeongSik Kim, Huseyin Kir, Martin Larralde, Julie A. McMurry, James A. Overton, Bjoern Peters, Clare Pilgrim, Ray Stefancsik, Sofia M. C. Robb, Sabrina Toro, Nicole A. Vasilevsky, Ramona Walls, Christopher J. Mungall, David Osumi-Sutherland
Summary: This article discusses the complex workflows involved in managing the ontology life cycle and the diverse set of tools required for this process. Standardizing release practices and quality standards are crucial in the biomedical domain to enable interoperability. The article also provides an overview of the Ontology Development Kit (ODK) and its practical applications.
DATABASE-THE JOURNAL OF BIOLOGICAL DATABASES AND CURATION
(2022)
Article
Computer Science, Information Systems
Muhammad Jawad Hussain, Heming Bai, Shahbaz Hassan Wasti, Guangjian Huang, Yuncheng Jiang
Summary: This paper proposes a comprehensive method for semantic similarity and relatedness based on WordNet and Wikipedia. By integrating the semantic knowledge of both resources at the feature level, the proposed method combines semantic similarity and relatedness into a single measure. Experimental results demonstrate its effectiveness over existing measures on various benchmarks.
INFORMATION SCIENCES
(2023)
Article
Psychology, Mathematical
Tyler M. Ensor, Molly B. MacMillan, Ian Neath, Aimee M. Surprenant
Summary: The study conducted three experiments to evaluate various measures of semantic relatedness and their ability to predict the recall of related and unrelated word lists in immediate memory tests. The results showed that lists of semantically related words are better recalled than lists of unrelated words. Different measures had slightly different predictions on the recall of related and unrelated word lists.
BEHAVIOR RESEARCH METHODS
(2021)
Article
Computer Science, Information Systems
Muhammad Jawad Hussain, Heming Bai, Yuncheng Jiang
Summary: This paper presents a model called WBLM that improves the link-based vector representations of concepts by exploring the Wikipedia link structure as a semantic graph and assigning weights to connected links. Experimental results show that the proposed WBLM model significantly improves the SS and SR computation accuracy of the WOLM and WTLM models.
INFORMATION PROCESSING & MANAGEMENT
(2023)
Article
Computer Science, Information Systems
Rita. T. T. Sousa, Sara Silva, Catia Pesquita
Summary: Semantic similarity plays a crucial role in bioinformatics applications such as protein-protein interaction prediction and disease-gene association discovery. However, existing semantic similarity measures are general-purpose and may not align well with specific biological perspectives. In this study, we introduce a supervised machine learning approach to tailor aspect-oriented semantic similarity measures for different biological views. The results demonstrate the superiority of our method in fitting semantic similarity models to diverse biological perspectives compared to commonly used manual combinations of semantic aspects.
Article
Computer Science, Artificial Intelligence
Mohannad AlMousa, Rachid Benlamri, Richard Khoury
Summary: This paper discusses the benefits of using all types of non-taxonomic relations to enhance semantic similarity measures, proposing a comprehensive poly-relational approach. Experimental results show significant improvements over existing methods on various gold standard datasets.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Information Systems
Reza Khajavi, Sholeh Arastoopoor
Summary: This study proposes a method to assess the similarity of scientific outputs based on keyword co-occurrence matrices, which can be applied to ranking, research monitoring, and scientific policy-making. The method involves transforming keyword co-occurrence networks into scientosemantic domains and calculating fuzzy distances between domains using frequency, development, and investment appeal indicators. The bibliometric data from appropriate queries on SCOPUS is used to derive scientosemantic domains. The results show that the distances based on investment appeal are greater than those based on frequency and development, with the largest distances observed for technology-related keywords.
INFORMATION PROCESSING & MANAGEMENT
(2023)
Article
Computer Science, Artificial Intelligence
Ouafa Ajarroud, Ahmed Zellou, Ali Idri
Summary: Most mediation systems use caching policies, with semantic caching being a widely adopted strategy. However, the current semantic caching approach compares syntax rather than semantics, leading to delays when multiple requests are stored in the cache. This work proposes a new ontology-based semantic approach and algorithm to filter regions in the cache that do not semantically cover a user query, optimizing cache usage for faster retrieval.
KNOWLEDGE AND INFORMATION SYSTEMS
(2023)
Review
Green & Sustainable Science & Technology
Dipti Pawar, Shraddha Phansalkar, Abhishek Sharma, Gouri Kumar Sahu, Chun Kit Ang, Wei Hong Lim
Summary: Biomedical text summarization (BTS) is an emerging area of work and research that supports sustainable healthcare applications. However, due to the rapid growth in biomedical literature and its diversities, effective text summarization is becoming more challenging. This study aims to conduct a comprehensive literature review of significant works in BTS and analyze the relevance and efficacy of deep learning and context-aware feature extraction techniques.
Article
Biochemical Research Methods
Benjamin Elsworth, Tom R. Gaunt
Summary: The field of literature-based discovery is rapidly developing, with triples extracted from biomedical literature proving to be useful in representing knowledge. By implementing efficient search methods and an application programming interface, it is possible to explore mechanistic knowledge from the literature quickly and conveniently.
Article
Chemistry, Multidisciplinary
Emmanuel Vallance, Nicolas Sutton-Charani, Abdelhak Imoussaten, Jacky Montmain, Stephane Perrey
APPLIED SCIENCES-BASEL
(2020)
Article
Engineering, Environmental
Lucie Jacquin, Abdelhak Imoussaten, Francois Trousset, Didier Perrin, Jacky Montmain
Summary: With the increase in waste streams, industrial sorting has become a major issue. Some industrialists have designed sorting machines that can reliably discriminate between several types of plastics, but are not suitable for dark plastics. Mid-wavelength infrared technology can be used instead of near infrared technology, but the new spectral range is poorer in terms of wavelength for some plastics of interest. Cautious sorting enables containers dedicated to specific materials to contain fewer impurities, leading to higher-quality secondary raw materials.
RESOURCES CONSERVATION AND RECYCLING
(2021)
Article
Multidisciplinary Sciences
Yu Du, Nicolas Sutton-Charani, Sylvie Ranwez, Vincent Ranwez
Summary: Recommender systems aim to predict users' preferences for unrated items, with collaborative filtering being a widely used mechanism; the way user neighborhoods are identified significantly impacts prediction accuracy, the proposed method in this paper improves similarity measures by considering the number of co-ratings.
Article
Computer Science, Information Systems
Yu Du, Sylvie Ranwez, Nicolas Sutton-Charani, Vincent Ranwez
Summary: The diversity of item lists suggested by recommender systems significantly impacts user satisfaction. Existing diversity optimization approaches may not be effective for different recommendation approaches due to the diversity level of candidate lists depending on the recommender system used. Individual users' diversity needs are often ignored in post-processing diversification. This study systematically compares diversity performances of recommendation models in different domains and proposes a diversification post-processing objective that considers specific users' diversity needs.
INFORMATION PROCESSING & MANAGEMENT
(2021)
Article
Computer Science, Artificial Intelligence
Yu Du, Sylvie Ranwez, Nicolas Sutton-Charani, Vincent Ranwez
Summary: This paper investigates the use of knowledge graphs for post-hoc recommendation explanations. Existing approaches rely on overlap properties to describe user liked items and recommended ones, but do not fully leverage the property hierarchy of knowledge graphs, which may lead to flawed explanations. The authors propose an approach that considers the whole property hierarchy and validate its effectiveness through a user study.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Sport Sciences
Frank Imbach, Nicolas Sutton-Charani, Jacky Montmain, Robin Candau, Stephane Perrey
Summary: The emergence of Fitness-Fatigue impulse response models (FFMs) has allowed for the investigation of training effects on athletic performance. However, their effectiveness in describing and predicting performance is limited due to simplified physiological processes and a univariate consideration of factors. Therefore, machine-learning methods can enhance performance prediction by incorporating physiological representations and multivariate algorithms.
SPORTS MEDICINE-OPEN
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Charles Condevaux, Sebastien Harispe, Stephane Mussard
Summary: The interpretability of predictive machine learning models is crucial for applications where end-users need to understand the decisions made. A new attribution method called FESP is introduced as an alternative to the popular Shapley value. Results show that FESP and ES produce better attribution maps compared to existing approaches in image and text classification settings.
MACHINE LEARNING AND KNOWLEDGE DISCOVERY IN DATABASES, ECML PKDD 2022, PT I
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Ali Yaddaden, Sebastien Harispe, Michel Vasquez
Summary: This work presents a hybrid approach of deep neural network and dynamic programming for solving the Capacity Constrained Vehicle Routing Problem. Experimental results demonstrate the ability to learn an implicit algorithm that generates competitive solutions.
OPTIMIZATION AND LEARNING, OLA 2022
(2022)
Article
Computer Science, Artificial Intelligence
Ali Yaddaden, Sebastien Harispe, Michel Vasquez
Summary: This paper studies the benefit of Transfer Learning for Neural Combinatorial Optimization (NCO) by evaluating the improvement in model training and efficiency achieved by leveraging knowledge learned from similar tasks. The focus is on the Traveling Salesman Problem (TSP) and the Vehicle Routing Problem (VRP), and the results show that Transfer Learning can speed up the training process and improve sample efficiency.
COMPUTING AND INFORMATICS
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Gerard Dray, Pierre-Antoine Jean, Yann Maheu, Jacky Montmain, Nicolas Sutton-Charani
Summary: This paper describes the implementation of machine learning methods in the AffectMove challenge for classifying body movements in physical rehabilitation, mathematical problem solving, and interactive dance contexts. The methods, results, and future work are presented.
2021 9TH INTERNATIONAL CONFERENCE ON AFFECTIVE COMPUTING AND INTELLIGENT INTERACTION WORKSHOPS AND DEMOS (ACIIW)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Karim Radouane, Andon Tchechmedjiev, Binbin Xu, Sebastien Harispe
Summary: This study evaluates the performance of different baseline systems for protective behavior detection, with PA-ResGCN-N51 demonstrating the best overall performance. The lower performance of the Transformer and LSTM baseline systems could be attributed to issues with handling class imbalances.
2021 9TH INTERNATIONAL CONFERENCE ON AFFECTIVE COMPUTING AND INTELLIGENT INTERACTION WORKSHOPS AND DEMOS (ACIIW)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Abdelhak Imoussaten, Pierre Couturier, Jacky Montmain
2020 IEEE INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS (FUZZ-IEEE)
(2020)
Proceedings Paper
Computer Science, Artificial Intelligence
Lucie Jacquin, Abdelhak Imoussaten, Sebastien Destercke, Francois Trousset, Jacky Montmain, Didier Perrin
2020 IEEE INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS (FUZZ-IEEE)
(2020)
Article
Mathematics, Interdisciplinary Applications
Gildas Tagny-Ngompe, Stephane Mussard, Guillaume Zambrano, Sebastien Harispe, Jacky Montmain
