Domain-specific Concepts Research Articles

Search Query Refinement Using Context, Knowledge and Long-term Memorization

In the era of information overload, search tools are crucial, yet users often struggle to articulate their precise dataneeds, resulting in sub-optimal search outcomes. This is often due to users associating products with influencers or celebrities rather than knowing the specific brand or product name. This research aims to enable users to find products based on real-life associ- ations, emphasizing the importance of upgrading search query refinement for accuracy and relevance. A significant challenge faced by existing web tools is refining queries involving unrelated entities. This research addresses this gap by proposing a compre- hensive approach that integrates context, knowledge represen- tation, and long-term memorization. The framework combines contextual information with advanced knowledge representation techniques, enhancing the system’s understanding of user intent and domain-specific concepts. Long-term memorization reduces the time complexity of query refinement by leveraging past search experiences. This study underscores the potential of incorporating context, knowledge representation, and long-term memorization in refining search queries, offering more accurate results. As the digital landscape evolves, our approach has the potential to upgrade the search engine experience, providing personalized and contextually relevant results globally.

Supporting the working life exposome: Annotating occupational exposure for enhanced literature search.

An individual's likelihood of developing non-communicable diseases is often influenced by the types, intensities and duration of exposures at work. Job exposure matrices provide exposure estimates associated with different occupations. However, due to their time-consuming expert curation process, job exposure matrices currently cover only a subset of possible workplace exposures and may not be regularly updated. Scientific literature articles describing exposure studies provide important supporting evidence for developing and updating job exposure matrices, since they report on exposures in a variety of occupational scenarios. However, the constant growth of scientific literature is increasing the challenges of efficiently identifying relevant articles and important content within them. Natural language processing methods emulate the human process of reading and understanding texts, but in a fraction of the time. Such methods can increase the efficiency of both finding relevant documents and pinpointing specific information within them, which could streamline the process of developing and updating job exposure matrices. Named entity recognition is a fundamental natural language processing method for language understanding, which automatically identifies mentions of domain-specific concepts (named entities) in documents, e.g., exposures, occupations and job tasks. State-of-the-art machine learning models typically use evidence from an annotated corpus, i.e., a set of documents in which named entities are manually marked up (annotated) by experts, to learn how to detect named entities automatically in new documents. We have developed a novel annotated corpus of scientific articles to support machine learning based named entity recognition relevant to occupational substance exposures. Through incremental refinements to the annotation process, we demonstrate that expert annotators can attain high levels of agreement, and that the corpus can be used to train high-performance named entity recognition models. The corpus thus constitutes an important foundation for the wider development of natural language processing tools to support the study of occupational exposures.

Weakly-Supervised Scientific Document Classification via Retrieval-Augmented Multi-Stage Training.

Scientific document classification is a critical task for a wide range of applications, but the cost of collecting human-labeled data can be prohibitive. We study scientific document classification using label names only. In scientific domains, label names often include domain-specific concepts that may not appear in the document corpus, making it difficult to match labels and documents precisely. To tackle this issue, we propose WanDeR, which leverages dense retrieval to perform matching in the embedding space to capture the semantics of label names. We further design the label name expansion module to enrich its representations. Lastly, a self-training step is used to refine the predictions. The experiments on three datasets show that WanDeR outperforms the best baseline by 11.9%. Our code will be published at https://github.com/ritaranx/wander.

Knowledge representation of the state of a cloud-native application

Cloud Computing has revolutionized the way applications are developed, deployed, and maintained. Over the past decade, we have observed dynamically growing interest in Cloud Computing. The benefits of the cloud approach caused the increasing popularity of Cloud-native applications. Cloud-native is an approach to developing and deploying applications according to the concepts of DevOps, Continuous Integration/Continuous Delivery (CI/CD), containers and microservices. The knowledge about Cloud Computing has become extensive and complex. Fortunately, before Cloud-native applications development, there was a great deal of effort to develop tools for effective knowledge representation. Ontologies are a convenient way to show the relations between domain-specific concepts. In this paper, we propose an ontology named CNOnt that describes the state-of-the-art of Cloud-native applications. CNOnt covers aspects from the clusterization perspective. First, this paper presents the engineering perspective of building the CNOnt ontology. Second, we demonstrate a use case of our ontology that proves the correctness of CNOnt development. This ontology is exhausted in CNOnt Broker. It is a system that applies the information in the OWL file into the Kubernetes cluster and in reverse. The knowledge representation makes Cloud-native applications understandable to third-party systems and increases interoperability between different microservices.

BPMN4MOOC: A BPMN extension for the design of connectivist MOOC

ABSTRACT Business Process Model and Notation (BPMN) is a standard formalism for business process modeling that is very popular in professional practices due to its expressiveness, the well-defined meta-model, and its easiness of use by non-technical users. For instance, BPMN2.0 is used for business processes in commercial areas such as banks, shops, production, and supply industry. Due to its flexible notation, BPMN is increasingly being used in non-traditional business process domains like education. The presented research work aims to develop a valid extension by applying the BPMN representation of pedagogical scenario in connectivist cMOOC context. BPMN provides a set of generic process modeling elements that make it necessary to extend the language and notation by domain-specific concepts from connectivist pedagogical approach. Therefore, BPMN2.0 extensibility mechanism was explored to include the representation of specific cMOOC pedagogical concepts. Moreover, this paper provides an analysis of the requirements of those concepts. Based on a cMOOC-oriented pedagogical scenario model, the need for an extension is identified and the valid BPMN extension meta-model is designed by the construction of a conceptual domain-specific model and the corresponding BPMN extension model.

Do Neural Transformers Learn Human-Defined Concepts? An Extensive Study in Source Code Processing Domain

State-of-the-art neural networks build an internal model of the training data, tailored to a given classification task. The study of such a model is of interest, and therefore, research on explainable artificial intelligence (XAI) aims at investigating if, in the internal states of a network, it is possible to identify rules that associate data to their corresponding classification. This work moves toward XAI research on neural networks trained in the classification of source code snippets, in the specific domain of cybersecurity. In this context, typically, textual instances have firstly to be encoded with non-invertible transformation into numerical vectors to feed the models, and this limits the applicability of known XAI methods based on the differentiation of neural signals with respect to real valued instances. In this work, we start from the known TCAV method, designed to study the human understandable concepts that emerge in the internal layers of a neural network, and we adapt it to transformers architectures trained in solving source code classification problems. We first determine domain-specific concepts (e.g., the presence of given patterns in the source code), and for each concept, we train support vector classifiers to separate points in the vector activation spaces that represent input instances with the concept from those without the concept. Then, we study if the presence (or the absence) of such concepts affects the decision process of the neural network. Finally, we discuss about how our approach contributes to general XAI goals and we suggest specific applications in the source code analysis field.

TUCAN: A molecular identifier and descriptor applicable to the whole periodic table from hydrogen to oganesson

TUCAN is a canonical serialization format that is independent of domain-specific concepts of structure and bonding. The atomic number is the only chemical feature that is used to derive the TUCAN format. Other than that, the format is solely based on the molecular topology. Validation is reported on a manually curated test set of molecules as well as a library of non-chemical graphs. The serialization procedure generates a canonical “tuple-style” output which is bidirectional, allowing the TUCAN string to serve as both identifier and descriptor. Use of the Python NetworkX graph library facilitated a compact and easily extensible implementation.Graphical

Implementation and evaluation of a multivariate abstraction-based, interval-based dynamic time-warping method as a similarity measure for longitudinal medical records

We extended dynamic time warping (DTW) into interval-based dynamic time warping (iDTW), including (A) interval-based representation (iRep): [1] abstracting raw, time-stamped data into interval-based abstractions, [2] comparison-period scoping, [3] partitioning abstract intervals into a given temporal granularity; (B) interval-based matching (iMatch): matching partitioned, abstract-concepts records, using a modified DTW. Using domain knowledge, we abstracted the raw data of medical records, for up to three concepts out of four or five relevant concepts, into two interval types: State abstractions (e.g. LOW, HIGH) and Gradient abstractions (e.g. INCREASING, DECREASING). We created all uni-dimensional (State or Gradient) or multi-dimensional (State and Gradient) abstraction combinations. Tasks: Classifying 161 oncology patients records as autologous or allogenic bone-marrow transplantation; classifying 125 hepatitis patients records as B or C hepatitis; predicting micro- or macro-albuminuria in the next year for 151 Type 2 diabetes patients. We used a k-Nearest-Neighbors majority, k = an odd number from 1 to SQRT(N), N = set size. 75,936 10-fold cross-validation experiments were performed: 33,600 (Oncology), 28,800 (Hepatitis), 13,536 (Diabetes). Measures: Area Under the Curve (AUC), optimal Youden's Index. Paired t-tests compared result vectors for equivalent configurations other than a tested variable, to determine a significant mean accuracy difference (P<0.05). Mean classification and prediction using abstractions was significantly better than using only raw time-stamped data. In each domain, at least one abstraction combination led to a significantly better mean performance than raw data. Increasing feature number and using Multi-dimensional abstractions enhanced performance. Unlike when using raw data, optimal mean performance was often reached with k=5, using abstractions.

PyFlies: A Domain-Specific Language for Designing Experiments in Psychology

The majority of studies in psychology are nowadays performed using computers. In the past, access to good quality software was limited, but in the last two decades things have changed and today we have an array of good and easily accessible open-source software to choose from. However, experiment builders are either GUI-centric or based on general-purpose programming languages which require programming skills. In this paper, we investigate an approach based on domain-specific languages which enables a text-based experiment development using domain-specific concepts, enabling practitioners with limited or no programming skills to develop psychology tests. To investigate our approach, we created PyFlies, a domain-specific language for designing experiments in psychology, which we present in this paper. The language is tailored for the domain of psychological studies. The aim is to capture the essence of the experiment design in a concise and highly readable textual form. The editor for the language is built as an extension for Visual Studio Code, one of the most popular programming editors today. From the experiment description, various targets can be automatically produced. In this version, we provide a code generator for the PsychoPy library while generators for other target platforms are planned. We discuss the language, its concepts, syntax, some current limitations, and development directions. We investigate the language using a case study of the implementation of the Eriksen flanker task.

What Do Programmers Discuss About Blockchain? A Case Study on the Use of Balanced LDA and the Reference Architecture of a Domain to Capture Online Discussions About Blockchain Platforms Across Stack Exchange Communities

Blockchain-related discussions have become increasingly prevalent in programming Q&A websites, such as Stack Overflow and other Stack Exchange communities. Analyzing and understanding those discussions could provide insights about the topics of interest to practitioners, and help the software development and research communities better understand the needs and challenges facing developers as they work in this new domain. Prior studies propose the use of LDA to study the Stack Exchange discussions. However, a simplistic use of LDA would capture the topics in discussions blindly without keeping in mind the variety of the dataset and domain-specific concepts. Specifically, LDA is biased towards larger sized corpora; and LDA-derived topics are not linked to higher level domain-specific concepts. We propose an approach that combines balanced LDA (which ensures that the topics are balanced across a domain) with the reference architecture of a domain to capture and compare the popularity and impact of discussion topics across the Stack Exchange communities. Popularity measures the distribution of interest in discussions, and impact gauges the trend of popularity over time. We made a number of interesting observations, including: (1) Bitcoin, Ethereum, Hyperledger Fabric and Corda are the four most commonly-discussed blockchain platforms on the Stack Exchange communities. (2) A broad range of topics are discussed across the various platforms of distinct layers in our derived reference architecture. (3) The Application layer topics exhibit the highest popularity (33.2 percent) and fastest growth in topic impact since November 2015. (4) The Application, API, Consensus and Network layer topics are discussed across the studied blockchain platforms, but exhibit different distributions in popularity. (5) The impact of architectural layer topics exhibits an upward trend, but is growing at different speeds across the studied blockchain platforms. The breakdown of the topic impact across the architectural layers is relatively stable over time except for the Hyperledger Fabric platform. Based on our findings, we highlighted future directions and provided recommendations for practitioners and researchers.

The Development and Evaluation of a Tool to Determine the Characteristics of STEM Activities

STEM activities integrate various disciplines and are organized around domain-specific concepts and skills. What is achieved in terms of learning is related to how well the important STEM education characteristics are reflected in the implemented and integrated activities. This study aims to develop and evaluate a measurement tool for determining the characteristics of STEM activities by using a case study method. The STEM Activity Evaluation Form was developed based upon a literature review and experts’ opinions. The observation-based form was applied to 21 different STEM activities, grouped in four categories: curiosity arousing activities, encouraging and upskilling activities, research and design projects, and one science fair. Students in science education partook in the activities in an eight-month period. The evaluation form has four categories: STEM learning environment, activitation of students, STEM content and practices, and connecting STEM. Application of the form revealed that the tool is useful to determine the appropriateness of activities with respect to the nature of STEM education.

Automatic construction of academic profile: A case of information science domain

To provide junior researchers with domain-specific concepts efficiently, an automatic approach for academic profiling is needed. First, to obtain personal records of a given scholar, typical supervised approaches often utilise structured data like infobox in Wikipedia as training dataset, but it may lead to a severe mis-labelling problem when they are utilised to train a model directly. To address this problem, a new relation embedding method is proposed for fine-grained entity typing, in which the initial vector of entities and a new penalty scheme are considered, based on the semantic distance of entities and relations. Also, to highlight critical concepts relevant to renowned scholars, scholars’ selective bibliographies which contain massive academic terms are analysed by a newly proposed extraction method based on logistic regression, AdaBoost algorithm and learning-to-rank techniques. It bridges the gap that conventional supervised methods only return binary classification results and fail to help researchers understand the relative importance of selected concepts. Categories of experiments on academic profiling and corresponding benchmark datasets demonstrate that proposed approaches outperform existing methods notably. The proposed techniques provide an automatic way for junior researchers to obtain organised knowledge in a specific domain, including scholars’ background information and domain-specific concepts.

Adapting TDL to Provide Testing Support for Executable DSLs.

Testing is one of the most prevalent and successful verification and validation (V&V) techniques used in the field of software engineering. While a large number of testing frameworks exist for general-purpose programming languages, providing testing facilities for any given executable Domain Specific Language (xDSL) remains a costly and challenging task. In this context, a standard such as the Test Description Language (TDL) appears as a suitable foundation for the definition of a generic testing approach for xDSLs. Unfortunately, TDL does not provide the domain-specific concepts required to write test cases for a given xDSL and does not include any model execution facilities. Our contribution addresses these limitations and thereby provides a fully generic testing approach for xDSLs based on TDL. Required TDL domain-specific concepts are automatically inferred from the xDSL definition through a model transformation into TDL. Model execution facilities are provided through the definition of a refined operational semantics for TDL. The application of our approach on 5 different xDSLs shows its generality and that it can successfully be used for testing executable models.

Role of Content Analysis in Improving the Curation of Experimental Data

&#x0D; As researchers are increasingly seeking tools and specialized support to perform research data management activities, the collaboration with data curators can be fruitful. Yet, establishing a timely collaboration between researchers and data curators, grounded in sound communication, is often demanding. In this paper we propose manual content analysis as an approach to streamline the data curator workflow. With content analysis curators can obtain domain-specific concepts used to describe experimental configurations in scientific publications, to make it easier for researchers to understand the notion of metadata and for the development of metadata tools. We present three case studies from experimental domains, one related to sustainable chemistry, one to photovoltaic generation and another to nanoparticle synthesis. The curator started by performing content analysis in research publications, proceeded to create a metadata template based on the extracted concepts, and then interacted with researchers. The approach was validated by the researchers with a high rate of accepted concepts, 84 per cent. Researchers also provide feedback on how to improve some proposed descriptors. Content analysis has the potential to be a practical, proactive task, which can be extended to multiple experimental domains and bridge the communication gap between curators and researchers.&#x0D; [This paper is a conference pre-print presented at IDCC 2020 after lightweight peer review.]&#x0D;

Scientific Reasoning in Biology - the Impact of Domain-General and Domain-Specific Concepts on Children's Observation Competency.

Research on the development of scientific reasoning has put the main focus on children’s experimentation skills, in particular on the control-of-variables strategy. However, there are more scientific methods than just experimentation. Observation is defined as an independent scientific method that includes not only the description of what is observed, but also all phases of the scientific inquiry, such as questioning, hypothesizing, testing, and interpreting. Previous research has shown that the quality of observations depends on specific knowledge in the domain. We argue that observation competency shares the domain-general ability to differentiate hypotheses from evidence with other scientific methods. The present study investigates the relations of both domain-general scientific thinking and domain-specific knowledge in biology with observation competency in grade K children. We tested relations between observation competency, domain-general scientific reasoning, domain-specific knowledge, and language abilities of 75 children (age 4;9 to 6;7). Both scientific reasoning and domain-specific knowledge proved to be significant predictors of observation competency, explaining 35% of the variance. In a mediation analysis, we found a significant indirect effect of language via these two predictors. Thus, the present results indicate that observation skills require not only domain-specific knowledge but also domain-general scientific reasoning abilities.

Seven Properties of Self-Organization in the Human Brain

The principle of self-organization has acquired a fundamental significance in the newly emerging field of computational philosophy. Self-organizing systems have been described in various domains in science and philosophy including physics, neuroscience, biology and medicine, ecology, and sociology. While system architecture and their general purpose may depend on domain-specific concepts and definitions, there are (at least) seven key properties of self-organization clearly identified in brain systems: (1) modular connectivity, (2) unsupervised learning, (3) adaptive ability, (4) functional resiliency, (5) functional plasticity, (6) from-local-to-global functional organization, and (7) dynamic system growth. These are defined here in the light of insight from neurobiology, cognitive neuroscience and Adaptive Resonance Theory (ART), and physics to show that self-organization achieves stability and functional plasticity while minimizing structural system complexity. A specific example informed by empirical research is discussed to illustrate how modularity, adaptive learning, and dynamic network growth enable stable yet plastic somatosensory representation for human grip force control. Implications for the design of “strong” artificial intelligence in robotics are brought forward.

Reusing Static Analysis across Different Domain-Specific Languages using Reference Attribute Grammars

Context: Domain-specific languages (DSLs) enable domain experts to specify tasks and problems themselves, while enabling static analysis to elucidate issues in the modelled domain early. Although language workbenches have simplified the design of DSLs and extensions to general purpose languages, static analyses must still be implemented manually. Inquiry: Moreover, static analyses, e.g., complexity metrics, dependency analysis, and declaration-use analysis, are usually domain-dependent and cannot be easily reused. Therefore, transferring existing static analyses to another DSL incurs a huge implementation overhead. However, this overhead is not always intrinsically necessary: in many cases, while the concepts of the DSL on which a static analysis is performed are domain-specific, the underlying algorithm employed in the analysis is actually domain-independent and thus can be reused in principle, depending on how it is specified. While current approaches either implement static analyses internally or with an external Visitor, the implementation is tied to the language's grammar and cannot be reused easily. Thus far, a commonly used approach that achieves reusable static analysis relies on the transformation into an intermediate representation upon which the analysis is performed. This, however, entails a considerable additional implementation effort. Approach: To remedy this, it has been proposed to map the necessary domain-specific concepts to the algorithm's domain-independent data structures, yet without a practical implementation and the demonstration of reuse. Thus, to make static analysis reusable again, we employ relational Reference Attribute Grammars (RAGs) by creating such a mapping to a domain-independent overlay structure using higher-order attributes. Knowledge: We describe how static analysis can be specified on analysis-specific data structures, how relational RAGs can help with the specification, and how a mapping from the domain-specific language can be performed. Furthermore, we demonstrate how a static analysis for a DSL can be externalized and reused in another general purpose language. Grounding: The approach was evaluated using the RAG system JastAdd. To illustrate reusability, we implemented two analyses with two addressed languages each: a cycle detection analysis used in a small state machine DSL and for detecting circular dependencies in Java types and packages, and an analysis of variable shadowing, applied to both Java and the Modelica modelling language. Thereby, we demonstrate the reuse of two analysis algorithms in three completely different domains. Additionally, we use the cycle detection analysis to evaluate the efficiency by comparing our external analysis to an internal reference implementation analysing all Java programs in the Qualitas Corpus and thereby are able to show that an externalized analysis incurs only minimal overhead. Importance: We make static analysis reusable, again, showing the practicality and efficiency of externalizing static analysis for both DSLs and general purpose languages using relational RAGs.

Learning domain ontologies from engineering documents for manufacturing knowledge reuse by a biologically inspired approach

The evolution of any given product family in manufacturing enterprises follows the epicycles in product lifecycle development, thus generating many useful but unmanageable engineering documents. These documents record nearly all the information about product lifecycle and thus provide domain-specific concepts and significant relations for ontology extraction. A qualified ontology, in turn, facilitates the management and reuse of manufacturing knowledge. Consequently, this paper proposes a biologically inspired adaptive growth (BIAG) approach to learn domain ontologies (DO) from these documents. BIAG is modeled on the growth and development mechanisms of a tree, where genes and three key stages of tree lifecycle including seed formation, root development, and tree growth are reference for DO learning. In BIAG, axioms and constrains that serve the same function as genes of a tree are first generated by a semi-supervised learning algorithm, to control the construction process of DO; then three adaptive growth strategies inspired by the above three stages of tree lifecycle are proposed to extract DO from engineering documents. The results of the DO construction and application examples demonstrate the feasibility and effectiveness of BIAG, which could, therefore, be a good choice for DO construction towards manufacturing knowledge reuse.

Discovering generalized design knowledge using a multi-objective evolutionary algorithm with generalization operators

The early-phase design of complex systems is a challenging task, as a decision maker has to take into account the intricate relationships among different design variables. A popular way to help decision makers easily identify important design features is to use data mining. However, many of the existing algorithms output design features that are too complex (e.g., conjunction of many literals with unrelated predicates), making it difficult for a user to understand, remember, and apply these features to find better designs. In this paper, we introduce a new data mining method that extracts compact design features through knowledge generalization. The proposed method performs a search over the space of features using a multi-objective evolutionary algorithm that contains a set of generalization operators in addition to conventional evolutionary operators. Both variables and feature types are generalized by using an ontology defining a set of domain-specific concepts and relationships. Generalization leads to more compact and insightful features, as generalized knowledge encompasses wider concepts. A comparative experiment is conducted on a real-world system architecting problem to demonstrate the gain in compactness of the extracted features without significant reductions in predictive power.

Concept Extraction and Prerequisite Relation Learning from Educational Data

Prerequisite relations among concepts are crucial for educational applications. However, it is difficult to automatically extract domain-specific concepts and learn the prerequisite relations among them without labeled data.In this paper, we first extract high-quality phrases from a set of educational data, and identify the domain-specific concepts by a graph based ranking method. Then, we propose an iterative prerequisite relation learning framework, called iPRL, which combines a learning based model and recovery based model to leverage both concept pair features and dependencies among learning materials. In experiments, we evaluated our approach on two real-world datasets Textbook Dataset and MOOC Dataset, and validated that our approach can achieve better performance than existing methods. Finally, we also illustrate some examples of our approach.