A Method for Materials Knowledge Extraction from HTML Tables Based on Sibling Comparison

Millions of HTML tables containing structured data can be found on the Web. With their wide coverage, these tables are potentially very useful for filling missing values and extending cross-domain knowledge bases such as DBpedia, YAGO, or the Google Knowledge Graph. As a prerequisite for being able to use table data for knowledge base extension, the HTML tables need to be matched with the knowledge base, meaning that correspondences between table rows/columns and entities/schema elements of the knowledge base need to be found. This paper presents the T2D gold standard for measuring and comparing the performance of HTML table to knowledge base matching systems. T2D consists of 8 700 schema-level and 26 100 entity-level correspondences between the WebDataCommons Web Tables Corpus and the DBpedia knowledge base. In contrast related work on HTML table to knowledge base matching, the Web Tables Corpus (147 million tables), the knowledge base, as well as the gold standard are publicly available. The gold standard is used afterward to evaluate the performance of T2K Match, an iterative matching method which combines schema and instance matching. T2K Match is designed for the use case of matching large quantities of mostly small and narrow HTML tables against large cross-domain knowledge bases. The evaluation using the T2D gold standard shows that T2K Match discovers table-to-class correspondences with a precision of 94%, row-to-entity correspondences with a precision of 90%, and column-to-property correspondences with a precision of 77%.

As the Internet is a global network, there is a demand on accessing closely related data without browsing through different Web documents. A significant amount of these data are presented in HTML documents. Since data contents of HTML documents are intervened by markups, it is not trivial to integrate and provide a unified view of closely related data in different HTML documents. In this paper we present an approach for integrating semantically related data in any HTML tables that belong to a particular domain of interest (ID), such as house/apartment rental, by using the semantic hierarchies generated from the tables and the predefined meta-data sets that indicate related column names in ID. In our approach, we capture each data source as semi-structured data, called semantic hierarchy, and the end result of integrating different HTML tables of ID is a unified view of data in the tables, which is presented in an XML document. Besides HTML tables, our approach can be adopted by any system that integrates semi-structured data across different platforms.

This paper investigates the impact of knowledge extraction methods on the effectiveness of RAG (Retrieval-Augmented Generation) systems that utilize knowledge graphs. It highlights that the quality of the knowledge graph, which is formed using various knowledge extraction methods, is crucial for overcoming limitations of large language models (LLMs), such as “hallucinations”. The paper analyzes the architectures of LightRAG and GraphRAG, emphasizing that the selection of an optimal knowledge extraction strategy depends on specific tasks and the subject area. LLMs have advanced significantly, but they have limitations, including generating factually incorrect information (“hallucinations”) and possessing “outdated knowledge”. RAG systems were proposed to address these issues by combining LLMs with external knowledge bases. This approach reduces hallucinations, ensures factual accuracy, solves the problem of outdated knowledge, and increases transparency. Knowledge graphs are powerful tools for structuring information, consisting of entities (nodes) and relations (edges). They enhance RAG systems by enabling more precise and contextually grounded retrieval compared to keyword-based searches. The quality of a knowledge graph depends on the knowledge extraction methods used, which include named entity recognition (NER), relation extraction (RE), entity linking (EL), and event extraction. Different methods, such as rule-based, classical machine learning, and deep learning approaches, have varying trade-offs in terms of accuracy, completeness, and scalability. Entity linking and knowledge graph completion are also crucial for accuracy and richness. LightRAG and GraphRAG are two main graph-based RAG systems. LightRAG uses the knowledge graph as a quick reference, requiring high precision in knowledge extraction to avoid noise degradation. GraphRAG uses the knowledge graph as a domain model, where completeness of extracted knowledge is more critical, though systematic errors are still harmful. Both systems rely on LLMs for knowledge extraction, which makes them dependent on the LLM’s quality and the size of document fragments processed. The theoretical analysis confirms that the effectiveness of RAG systems is critically dependent on knowledge extraction methods. The quality, completeness, and accuracy of the knowledge graph directly influence the RAG system’s ability to provide relevant, accurate, and truthful answers. Different RAG system architectures like LightRAG and GraphRAG have distinct requirements for knowledge graph characteristics, prioritizing either accuracy or completeness in knowledge extraction.

The power industry usually records equipment failures, defects and other information in the form of text, which contains lots of regular patterns. Knowledge extraction in fault text is of great significance to improve efficiency and reduce the labor cost in the power industry. However, the research for knowledge extraction of text information in this field is rare, it is even more difficult to use machine learning algorithms to mine the deep patterns. To solve this problem, a method of knowledge extraction is proposed in this field. We use power equipment fault texts and relevant guidance as raw materials. Firstly, the knowledge base of this field is designed and constructed based on the ontology concepts, including ontology concept base, description base and regular expression base. Then, the knowledge extraction algorithm is designed according to the knowledge base. After that we conduct the knowledge merge operation to make the extraction results more accurate. Experiments on the real fault texts shows the feasibility and the high accuracy of our method when compared with artificial extraction.

Methods for knowledge extraction from a clinical database on liver diseases

Knowledge graph technology belongs to the field of artificial intelligence. It is widely used in semantic search and intelligent question answering. Construction of Uyghur's knowledge graph has the great value of Uyghur information processing and Uyghur application software development. Firstly, this paper describes the definition and structure of the knowledge graph, then it reviews the related research and construction process of the Uyghur's knowledge graph, and analyses the characteristics of word formation of Uighur entities in depth, focuses on the related concepts and methods of knowledge extraction, knowledge fusion, and knowledge update, Introduced the Uyghur semantic disambiguation technology in detail, Finally, looking forward to the future works.

The paper explores the tasks of situational management and modeling, considers modern situational management methods and tools. The possibility and effectiveness of using the CBR approach for finding solutions to atypical, unique problems, especially in conditions of limited available resources and uncertainty of the environment, are investigated. A set of modern models of knowledge representation and extraction in situational management systems are considered. Methods of knowledge representation and extraction based on precedents have been analyzed.

Urban Data management represents a major challenge in the field of Smart Cities. Its understanding is essential for the development of better smart services, which are a persistent demand in urban policies. From all the sources of data available, those that involve a collective processing of urban information (by the citizens or other collectives) deliver in fact, useful insights into social perception. Such is the case, for example, of data collected from mobile networks. Prior to the design of sociotechnical artifacts in cities, it seems important to extract the qualitative, quantitative opinions, sentiment, feedbacks present in these data. In this paper we present three solutions for mining these contents through Knowledge Extraction methods, as a previous step to the prospection of new smart services.

A human-assisted knowledge extraction method for machining operations

Knowledge extraction is the core work of constructing a knowledge graph, but most knowledge extraction methods assume perfect data support. Therefore, this paper analyzes the characteristics of big data in the cultural industry. In addition to the consensus characteristics of big data, these data also highlight the features of sectors such as low resources and intense data boundary fuzziness. Therefore, this paper proposes a knowledge extraction method for cultural industry data (CT-DA). Firstly, design a labeling strategy for big data in the cultural industry. Secondly, according to the low resource characteristics of data, create the counter transfer learning layer to realize resource transfer. Considering the intense fuzziness of data, design the dynamic attention mechanism layer for learning the critical attention of entities in the cultural field. Finally, build an experimental platform. The experiments show that this method has performance advantages in accuracy, recall, and F1.

Facing the network security problem of power monitoring system, this paper studies the knowledge graph-based network security knowledge extraction and management method. In the entity recognition module, an entity recognition algorithm based on the idea of multi task learning is adopted, which decomposes the entity recognition process into entity boundary perception and entity classification to reduce the impact of error accumulation. Meanwhile, the introduction of comprehensive description information of entity categories into entity classification tasks can not only be widely used in general fields, but also be targeted to specific scenes in the field of power monitoring system. In the relation extraction module, based on the idea of multi-instance learning, the method of piecewise convolution neural network is employed, which reduces the impact of some instance annotation errors. The relation extraction method can extract text feature expression and inter entity structure information, which has a good effect on specialized field knowledge. Finally, based on the established knowledge graph, this paper studies the application problems of network security knowledge management and intelligent question answering of power monitoring system, which shows the effectiveness of knowledge graph technology in the field of network security management.

The fragmentation of infrastructure information systems has long been an obstacle to integrated transportation asset management (TAM). This paper presents a novel method for automatic knowledge extraction and ontology modelling from heterogeneous TAM databases using large language models (llMs). The method adopts a Chain-of-Thought framework to decompose the complex ontology modelling process into atomic tasks, harnessing the semantic understanding and reasoning capabilities of LLMs. As a result, class entities, class hierarchies, and relations are generated to construct an ontology model that supports semantic interoperability across diverse TAM systems. The method's performance was evaluated using four sets of TAM database schemas from UK road agencies. The results show that the overall recall rate for entity generation reaches 89.5% compared to the standard ontology. Furthermore, the accuracy rates for entity classification and relation classification are 82.1% and 75.6%, respectively, demonstrating the effectiveness of the proposed LLM-based approach in addressing data fragmentation issues in transportation information systems.

Software programming requires both API reference (know-what) knowledge and programming task (know-how) knowledge. Lots of programming know-what and know-how knowledge is documented in text, for example, API reference documentation and programming tutorials. To improve knowledge accessibility and usage, several recent studies use Natural Language Processing (NLP) methods to construct API know-what knowledge graph (API-KG) and programming task know-how knowledge graph (Task-KG) from software documentation. Although being promising, current API-KG and Task-KG are independent of each other, and thus are void of inherent connections between the two types of knowledge. Our empirical study on Stack Overflow questions confirms that only 36% of the API usage problems can be answered by the know-how or the know-what knowledge alone, while the rest questions requires a fusion of both. Inspired by this observation, we make the first attempt to fuse API-KG and Task-KG by API entity linking. This fusion creates nine categories of API semantic relations and two types of task semantic relations which are not present in the stand-alone API-KG or Task-KG. According to the definitions of these new API and task semantic relations, our approach dives deeper than surface-level API linking of API-KG and Task-KG, and infer nine categories of API semantic relations from task descriptions and two types of task semantic relations with the assistance of API-KG, which enrich the declaration or syntactic relations in the current API-KG and Task-KG. Our fused and semantically-enriched API-Task KG supports coherent API/Task-centric knowledge search by text or code queries. We have implemented our approach on Java programming documentation and built a web tool to search and explore API and programming task knowledge. Our evaluation confirms the high-accuracy of our knowledge extraction, fusion and enrichment methods, and the effectiveness and usefulness of our API-Task KG for answering Stack Overflow questions.

Most current Machine Learning based recommender systems act like black boxes, not offering the user any insight into the system logic or justification for the recommendations. Thus, risking losing trust with users and failing to achieve acceptance. The goal of this work is to improve the explainability of recommender systems by using a knowledge extraction method.

By the careful rearrangement and analysis, various meaningful information could be extracted from the published biological literature, which is of great significance for the related follow-up research. Since the rapid accumulation of literature publications, manual-based curation method is too inefficient to cope with the massive biological literature data. Knowledge extraction methods in computer science seem to be able to process the biology literature more efficiently, however, most of them are based on supervised learning and greatly limited by the annotation quality of the species corpora. Here we present a new named entity recognition algorithm named biolitNER, which features unsupervised domain adaptation and accordingly has the capability to recognize named entities on biomedical literature across domains. Considering the lack of well-annotated corpora of many species, biolitNER is of great utility for named entity recognition and bears critical significance for biomedical literature curation in various species. Experimental verification shows that, compared with traditional programs, biolitNER produces higher quality results and has a satisfactory runtime performance.KeywordsDomain adaptation methodName entity recognitionLiterature miningUnsupervised learningBiology big data