Annotated Research Articles

A Sociotechnical Framework for Semantic Biomedical Content Authoring and Publishing

Due to the ubiquity of unstructured biomedical data, significant obstacles still remain in achieving accurate and fast access to online biomedical content. In lieu of the growing volume of biomedical content on the web, embedding semantic annotations has become key to enhancing search engine context-aware indexing, thereby improving search speeds and retrieval accuracy. We introduce Semantically: a socio-technical framework for semantic biomedical content authoring and publishing. Identifying the appropriate semantic vocabulary for biomedical content annotation is a time-consuming and technically challenging process. Semantically automates this search by recommending highly accurate annotations from a wide range of biomedical ontologies. Furthermore, the framework is integrated with a knowledge-sharing system which allows biomedical authors to collaborate on identifying precise annotations during the content authoring process. Similarly, preserving content-level semantics during and after publishing to foster semantic search remains a research challenge. Semantically addresses this barrier by extending Schema.org, a community-agreed and research engine endorsed guideline for publishing structured content on the web. gosemantically.com

A catalogue with semantic annotations makes multilabel datasets FAIR

Multilabel classification (MLC) is a machine learning task where the goal is to learn to label an example with multiple labels simultaneously. It receives increasing interest from the machine learning community, as evidenced by the increasing number of papers and methods that appear in the literature. Hence, ensuring proper, correct, robust, and trustworthy benchmarking is of utmost importance for the further development of the field. We believe that this can be achieved by adhering to the recently emerged data management standards, such as the FAIR (Findable, Accessible, Interoperable, and Reusable) and TRUST (Transparency, Responsibility, User focus, Sustainability, and Technology) principles. We introduce an ontology-based online catalogue of MLC datasets originating from various application domains following these principles. The catalogue extensively describes many MLC datasets with comprehensible meta-features, MLC-specific semantic descriptions, and different data provenance information. The MLC data catalogue is available at: http://semantichub.ijs.si/MLCdatasets.

A Statistical Model for Morphology Inspired by the Amis Language

We introduce a statistical model for analysing the morphology of natural languages based on their affixes. The model was inspired by the analysis of Amis, an Austronesian language with a rich morphology. As words contain a root and potential affixes, we associate three vectors with each word: one for the root, one for the prefixes, and one for the suffixes. The morphology captures semantic notions and we show how to approximately predict some of them, for example the type of simple sentences using prefixes and suffixes only. We then define a Sentence vector s associated with each sentence, built from the prefixes and suffixes of the sentence and show how to approximately predict a derivation tree in a grammar.

A Semantically Data-Driven Classification Framework for Energy Consumption in Buildings

Encouraged by the European Union, all European countries need to enforce solutions to reduce non-renewable energy consumption in buildings. The reduction of energy (heating, domestic hot water, and appliances consumption) aims for the vision of near-zero energy consumption as a requirement goal for constructing buildings. In this paper, we review the available standards, tools and frameworks on the energy performance of buildings. Additionally, this work investigates if energy performance ratings can be obtained with energy consumption data from IoT devices and if the floor size and energy consumption values are enough to determine a dwellings’ energy performance rating. The essential outcome of this work is a data-driven prediction tool for energy performance labels that can run automatically. The tool is based on the cutting edge kNN classification algorithm and trained on open datasets with actual building data such as those coming from the IoT paradigm. Additionally, it assesses the results of the prediction by analysing its accuracy values. Furthermore, an approach to semantic annotations for energy performance certification data with currently available ontologies is presented. Use cases for an extension of this work are also discussed in the end.

Application of Rough Concept Lattice Model in Construction of Ontology and Semantic Annotation in Semantic Web of Things

In order to solve the problem of interoperability in Internet of Things, the Semantic Web technology is introduced into the Internet of Things to form Semantic Web of Things. Ontology construction is the core of Semantic Web of Things. Firstly, this paper analyzes the shortcomings of ontology construction methods in the Semantic Web of Things. Then, this paper proposes construction of semantic ontology based on improved rough concept lattice, which provides theoretical basis for semantic annotation of the sensing data attributes. In addition, this paper describes the semantic annotation system for the Internet of Things based on semantic similarity of ontology. The system consists of three steps: ontology mapping integration module, information extraction module, and semantic annotation of sensing data. Finally, the experimental results show that this semantic annotation method effectively improves the flexibility of sensor information and data attributes and effectively enhances the expression ability of sensor information and the use value of data.

Geospatial Web Services Discovery through Semantic Annotation of WPS

This paper presents an approach to GWS (GeospatialWeb Service) discovery through the semantic annotation of WPS (Web Processing Service) service descriptions. The rationale behind this work is that search engines that use appropriate semantic-based similarity measures in the matching process are more accurate in terms of precision and recall than those based on syntactic matching alone. The lack of semantics in the description of services using a standard such as WPS prevents the use of such a matching process and is considered a limitation of GWS discovery. The GWS discovery approach presented is based on the consideration of semantics in the service description method and in the matching process. The description of services is based on a semantic lightweight meta-model instantiated in the WPS 2.0 standard, extending the description of the service through metadata tags. The matching process is performed in three steps (functionality matching step, I/O (Input/Output) matching step and non-functional matching step). Its core is a semantic similarity measure that combines logical and non-logical matching methods. Finally, the paper presents the results of an experiment applying the proposed discovery approach on a GWS corpus, showing promising results and the added value of the three-step matching process.

Indoor Positioning and Navigation Model Based on Semantic Grid

Traditional GPS positioning technology cannot be used in indoor space. With the development of the new positioning technology and the Internet of things, the indoor mobile object positioning and navigation model have been the focus of the relevant research institutions at home and abroad. Based on this, indoor positioning technology was studied starting from Wi-Fi, RFID, and iBeacon technology in this paper. However, the accuracy of indoor positioning and navigation needs to be further improved. This paper presents a semantic space model based on artificial intelligence technology, through semantic pattern matching, semantic concept extension, semantic reasoning and semantic mapping, and interior semantic localization is realized. The indoor semantic network and indoor grid navigation model are constructed, and the indoor semantic path is modeled from time, location, user, and congestion. At the same time, the improved Term Frequency-Inverse Document Frequency is combined with the Hidden Markov Model to improve the accuracy of matching the stay area with the most likely location to visit and improve the accuracy of semantic annotation. It was found that the research on the indoor positioning and navigation model based on the semantic grid can realize the uniform expression of the complex spatial semantics of the theme, geometry, connectivity, and distance, which can promote the development of indoor positioning and navigation.

Real-time wildfire detection with semantic explanations

Wildfire detection is an indispensable component of many resilient platforms, preventing environmental disasters from damaging life. Online detection, which refers to situations in which wildfire events need to be detected in real time, is an essential tool, as the consequences of wildfires might already be irreversible by the time they are detected. Recently, wildfire detection methods based on machine learning and deep learning have been proposed. However, most of these methods take raw data as the input and then aggregate it into predictive features, which requires all of the historical data or a large part of a stream and violates the timeliness requirement of online detection. Moreover, these methods scarcely discuss the explainability (e.g., how the model comes up with the final detection). The lack of explainability reduces human trust in the system and may hinder further applications of the system, especially in high-stakes situations where decisions can have significant consequences. Unlike existing works that treat the timeliness and explainability problems separately, we propose a real-time wildfire detection system that is built upon the streaming capability of complex event processing and the expressiveness of semantic annotation. The system continuously processes raw data streams, transforms them into semantic events, and learns CEP queries that are both predictive and interpretable for humans at the same time. Experiments on four real datasets and one synthetic dataset show that our approach outperforms the baselines in terms of efficiency, accuracy, explainability, and adaptivity.

Teacher-student approach for lung tumor segmentation from mixed-supervised datasets.

PurposeCancer is among the leading causes of death in the developed world, and lung cancer is the most lethal type. Early detection is crucial for better prognosis, but can be resource intensive to achieve. Automating tasks such as lung tumor localization and segmentation in radiological images can free valuable time for radiologists and other clinical personnel. Convolutional neural networks may be suited for such tasks, but require substantial amounts of labeled data to train. Obtaining labeled data is a challenge, especially in the medical domain.MethodsThis paper investigates the use of a teacher-student design to utilize datasets with different types of supervision to train an automatic model performing pulmonary tumor segmentation on computed tomography images. The framework consists of two models: the student that performs end-to-end automatic tumor segmentation and the teacher that supplies the student additional pseudo-annotated data during training.ResultsUsing only a small proportion of semantically labeled data and a large number of bounding box annotated data, we achieved competitive performance using a teacher-student design. Models trained on larger amounts of semantic annotations did not perform better than those trained on teacher-annotated data. Our model trained on a small number of semantically labeled data achieved a mean dice similarity coefficient of 71.0 on the MSD Lung dataset.ConclusionsOur results demonstrate the potential of utilizing teacher-student designs to reduce the annotation load, as less supervised annotation schemes may be performed, without any real degradation in segmentation accuracy.

Revisiting the Boundary between ASR and NLU in the Age of Conversational Dialog Systems

Abstract As more users across the world are interacting with dialog agents in their daily life, there is a need for better speech understanding that calls for renewed attention to the dynamics between research in automatic speech recognition (ASR) and natural language understanding (NLU). We briefly review these research areas and lay out the current relationship between them. In light of the observations we make in this article, we argue that (1) NLU should be cognizant of the presence of ASR models being used upstream in a dialog system’s pipeline, (2) ASR should be able to learn from errors found in NLU, (3) there is a need for end-to-end data sets that provide semantic annotations on spoken input, (4) there should be stronger collaboration between ASR and NLU research communities.

Transductive Zero-Shot Hashing for Multilabel Image Retrieval.

Hash coding has been widely used in the approximate nearest neighbor search for large-scale image retrieval. Given semantic annotations such as class labels and pairwise similarities of the training data, hashing methods can learn and generate effective and compact binary codes. While some newly introduced images may contain undefined semantic labels, which we call unseen images, zero-shot hashing (ZSH) techniques have been studied for retrieval. However, existing ZSH methods mainly focus on the retrieval of single-label images and cannot handle multilabel ones. In this article, for the first time, a novel transductive ZSH method is proposed for multilabel unseen image retrieval. In order to predict the labels of the unseen/target data, a visual-semantic bridge is built via instance-concept coherence ranking on the seen/source data. Then, pairwise similarity loss and focal quantization loss are constructed for training a hashing model using both the seen/source and unseen/target data. Extensive evaluations on three popular multilabel data sets demonstrate that the proposed hashing method achieves significantly better results than the comparison methods.

A SURVEY ON SEMANTIC ANNOTATION TOOLS FOR KNOWLEDGE MANAGEMENT

Support for information and knowledge exchange is a key issue in the information society. To reduce the time wasted in searching and to reduce associated user frustration much more selective user access is needed. This is possible by semantic information processing of online documents.Knowledge management in an organisation are used for managing knowledge resources in order to facilitate access and reuse of knowledge.Semantic annotation is about assigning to the entities in the text, links to their semantic descriptions. This sort of metadata provides both class and instance information about the entities. Semantic annotation is applicable for any type of textweb pages, regular documents etc. For semantic annotation, there are various manual, semiautomatic and full automatic tools are developed by various organizations like mindswap.org, ontotext.org etc. In this paper, we are presenting analysis and review of some of these tools according to their applicability for an application domain in knowledge management.Our review and analysis may help the research community in selecting an appropriate tool for extracting the relevent and desired information from huge knowledge base of an organisation.

Multi-modal temporal attention models for crop mapping from satellite time series

Optical and radar satellite time series are synergetic: optical images contain rich spectral information, while C-band radar captures useful geometrical information and is immune to cloud cover. Motivated by the recent success of temporal attention-based methods across multiple crop mapping tasks, we propose to investigate how these models can be adapted to operate on several modalities. We implement and evaluate multiple fusion schemes, including a novel approach and simple adjustments to the training procedure, significantly improving performance and efficiency with little added complexity. We show that most fusion schemes have advantages and drawbacks, making them relevant for specific settings. We then evaluate the benefit of multimodality across several tasks: parcel classification, pixel-based segmentation, and panoptic parcel segmentation. We show that by leveraging both optical and radar time series, multimodal temporal attention-based models can outmatch single-modality models in terms of performance and resilience to cloud cover. To conduct these experiments, we augment the PASTIS dataset (Garnot and Landrieu, 2021a) with spatially aligned radar image time series. The resulting dataset, PASTIS-R, constitutes the first large-scale, multimodal, and open-access satellite time series dataset with semantic and instance annotations. (Dataset available at: https://zenodo.org/record/5735646)

A typology of proportional quantifiers: Evidence from Polish partitives

AbstractIn this paper, I investigate the distribution and semantic behavior of various proportional quantifiers (PQs) in Polish. Based on novel evidence including corpus data, I conclude that Polish PQs do not constitute a uniform category, but rather can be divided into four distinct classes based on the following properties: i) (in)compatibility with numerals and measure words, ii) (in)compatibility with approximative modifiers, iii) (in)compatibility with cumulative predicates and iv) (non)occurrence of spatial integrity effects. I propose that such a typology results from an interplay between more primitive semantic notions. In particular, the data call for combining degree semantics with a mereotopological approach to the meaning of PQs.

Learning invariant semantic representation for long-term robust visual localization

Long-term visual localization is a challenging problem in practice, which depends on the observed images captured by airborne camera to make the mobile robot perform the task of pose estimation during continuously navigating in complex scenes. Semantic information in the image has great invariance in changing environment and can be used to generate robust scene descriptor, but the performance of Convolutional Neural Network (CNN) based semantic segmentation highly depends on semantic labels, the generalization ability of the trained model is weak and labeling process for large-scale scene images is labor-intensive. To solve these problems, this paper proposes a new long-term visual localization method which fuses depth and semantic information in the scene, and its novelty lies in: (1) using a module of fusing depth and semantic information in the scene aims to extract the invariant scene representation when the environment changes, which effectively improves the robustness of long-term visual localization task and (2) using a domain adaptation module with the adversarial loss has adaptation ability from virtual dataset to real dataset, which require labor-free semantic labels annotation and generalize to more realistic application scenarios. Finally, the results show that our method outperforms state-of-the-art baselines under various challenging environments on the Extended CMU Seasons and RobotCar Seasons datasets in specific precision metrics.

Multimodal Narratives for the Presentation of Silk Heritage in the Museum

In this paper, a representation based on digital assets and semantic annotations is established for Traditional Craft instances, in a way that captures their socio-historic context and preserves both their tangible and intangible Cultural Heritage dimensions. These meaningful and documented experiential presentations are delivered to the target audience through narratives that address a range of uses, including personalized storytelling, interactive Augmented Reality (AR), augmented physical artifacts, Mixed Reality (MR) exhibitions, and the Web. The provided engaging cultural experiences have the potential to have an impact on interest growth and tourism, which can support Traditional Craft communities and institutions. A secondary impact is the attraction of new apprentices through training and demonstrations that guarantee long-term preservation. The proposed approach is demonstrated in the context of textile manufacturing as practiced by the community of the Haus der Seidenkultur, a former silk factory that was turned into a museum where the traditional craft of Jacquard weaving is still practiced.

Releasing Differentially Private Trajectories with Optimized Data Utility

The ubiquity of GPS-enabled devices has resulted in an abundance of data about individual trajectories. Releasing trajectories enables a range of data analysis tasks, such as urban planning, but it also poses a risk in compromising individual location privacy. To tackle this issue, a number of location privacy protection algorithms are proposed. However, existing works are primarily concerned with maintaining the trajectory data geographic utility and neglect the semantic utility. Thus, many data analysis tasks relying on utility, e.g., semantic annotation, suffer from poor performance. Furthermore, the released trajectories are vulnerable to location inference attacks and de-anonymization attacks due to insufficient privacy guarantee. In this paper, to design a location privacy protection algorithm for releasing an offline trajectory dataset to potentially untrusted analyzers, we propose a utility-optimized and differentially private trajectory synthesizer (UDPT) with two novel features. First, UDPT simultaneously preserves both geographical utility and semantic utility by solving a data utility optimization problem with a genetic algorithm. Second, UDPT provides a formal and provable guarantee against privacy attacks by synthesizing obfuscated trajectories in a differentially private manner. Extensive experimental evaluations on real-world datasets demonstrate UDPT’s outperformance against state-of-the-art works in terms of data utility and privacy.

A SEMANTICALLY ENRICHED MULTIMODAL IMAGING APPROACH DEDICATED TO CONSERVATION AND RESTORATION STUDIES

Abstract. For the past decades, Cultural Heritage (CH) is commonly documented by digital-based imaging and analytical techniques. This documentation is used as a support by heritage scientists to study and help the preservation of CH objects. Multiple techniques or modalities are usually required and applied to complete the documentation and the possible diagnostic from it. In this paper we explored multimodal imaging strategies to survey, analyse and share semantically enriched digital replicas. Three challenging case-studies from the SUMUM research project aims to illustrate efficient multi-source approaches in multi-scalar, multi-temporal and multi-spectral contexts. From multimodal data acquisitions, a photogrammetric-based registration method (TACO) has been developed in order to exploit a 2D/3D semantic annotation process implemented into a CH oriented collaborative web platform (AIOLI). In the exemples showed, the structure and the content of the annotations work is based from condition reports provided by conservation and restoration experts. To this end, all the documentation gathered on CH objects are either directly merged by image based registration while complementary analysis can be spatially anchored to annotations as linked resources. The dissemination part is explored by built-in AIOLI’s collaborative features or external Potree-based viewer, to enhance the accessibility of the final 3D annotated scenes for further expertises or wide-public events and purposes.

Providing Adverse Outcome Pathways from the AOP-Wiki in a Semantic Web Format to Increase Usability and Accessibility of the Content.

Introduction:The AOP-Wiki is the main platform for the development and storage of adverse outcome pathways (AOPs). These AOPs describe mechanistic information about toxicodynamic processes and can be used to develop effective risk assessment strategies. However, it is challenging to automatically and systematically parse, filter, and use its contents. We explored solutions to better structure the AOP-Wiki content, and to link it with chemical and biological resources. Together, this allows more detailed exploration, which can be automated.Materials and Methods:We converted the complete AOP-Wiki content into resource description framework (RDF) triples. We used >20 ontologies for the semantic annotation of property–object relations, including the Chemical Information Ontology, Dublin Core, and the AOP Ontology.Results:The resulting RDF contains >122,000 triples describing 158 unique properties of >15,000 unique subjects. Furthermore, >3500 link-outs were added to 12 chemical databases, and >7500 link-outs to 4 gene and protein databases. The AOP-Wiki RDF has been made available at https://aopwiki.rdf.bigcat-bioinformatics.orgDiscussion:SPARQL queries can be used to answer biological and toxicological questions, such as listing measurement methods for all Key Events leading to an Adverse Outcome of interest. The full power that the use of this new resource provides becomes apparent when combining the content with external databases using federated queries.Conclusion:Overall, the AOP-Wiki RDF allows new ways to explore the rapidly growing AOP knowledge and makes the integration of this database in automated workflows possible, making the AOP-Wiki more FAIR.

NLIMED: Natural Language Interface for Model Entity Discovery in Biosimulation Model Repositories.

Semantic annotation is a crucial step to assure reusability and reproducibility of biosimulation models in biology and physiology. For this purpose, the COmputational Modeling in BIology NEtwork (COMBINE) community recommends the use of the Resource Description Framework (RDF). This grounding in RDF provides the flexibility to enable searching for entities within models (e.g., variables, equations, or entire models) by utilizing the RDF query language SPARQL. However, the rigidity and complexity of the SPARQL syntax and the nature of the tree-like structure of semantic annotations, are challenging for users. Therefore, we propose NLIMED, an interface that converts natural language queries into SPARQL. We use this interface to query and discover model entities from repositories of biosimulation models. NLIMED works with the Physiome Model Repository (PMR) and the BioModels database and potentially other repositories annotated using RDF. Natural language queries are first “chunked” into phrases and annotated against ontology classes and predicates utilizing different natural language processing tools. Then, the ontology classes and predicates are composed as SPARQL and finally ranked using our SPARQL Composer and our indexing system. We demonstrate that NLIMED's approach for chunking and annotating queries is more effective than the NCBO Annotator for identifying relevant ontology classes in natural language queries.Comparison of NLIMED's behavior against historical query records in the PMR shows that it can adapt appropriately to queries associated with well-annotated models.