Rule-enhanced iterative complementation for knowledge graph reasoning

Lin, Qika; Liu, Jun; Pan, Yudai; Zhang, Lingling; Hu, Xin; Ma, Jie

doi:10.1016/j.ins.2021.06.040

Cited by 25 publications

(10 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• RNNLogic [118] Learning Rule Weights • ExpressGNN [176] • pLogicNet [117] • pGAT [64] • BioGRER [177] Iterative Rule Mining • SN-Hybrid [145] • Rule-IC [88] • IterE [174] New Rules: Bridging the Discrete and Continuous Spaces (NEURO;SYMBOLIC)…”

Section: A Logically-informed Embedding Approachesmentioning

confidence: 99%

“…b) Iterative Rule Mining: While pLogicNet, pGAT, and BioGRER learn weights for rules, none of them update the rule set with new rules. In contrast, Suresh and Neville's hybrid method (SN-Hybrid) [145], Lin et al 's rule-enhanced iterative complementation (Rule-IC) [88], and Zhang et al's IterE [174] mine new rules over iterations, rather than using the same rules and adjusting confidences. In other words, the neural modules of these methods guide the rule-mining processes.…”

Section: Learning Rules For Graph Reasoningmentioning

confidence: 99%

“…Then, triples of the relevant relations are randomly sampled, and axioms with more than one instance are considered eligible [174]. Rule-IC also operates in a similar fashion, but it uses a computed confidence threshold rather than a frequency threshold [88]. In SN-Hybrid, however, a pool of candidate rules is generated using a more sophisticated algorithm in which partial rules are queued and then extended until they become proper rules.…”

Section: Learning Rules For Graph Reasoningmentioning

confidence: 99%

See 2 more Smart Citations

Neurosymbolic AI for Reasoning on Graph Structures: A Survey

DeLong¹,

Fernández²,

Whyte³

et al. 2023

Preprint

View full text Add to dashboard Cite

Neurosymbolic AI is an increasingly active area of research which aims to combine symbolic reasoning methods with deep learning to generate models with both high predictive performance and some degree of human-level comprehensibility. As knowledge graphs are becoming a popular way to represent heterogeneous and multi-relational data, methods for reasoning on graph structures have attempted to follow this neurosymbolic paradigm. Traditionally, such approaches have utilized either rule-based inference or generated representative numerical embeddings from which patterns could be extracted. However, several recent studies have attempted to bridge this dichotomy in ways that facilitate interpretability, maintain performance, and integrate expert knowledge. Within this article, we survey a breadth of methods that perform neurosymbolic reasoning tasks on graph structures. To better compare the various methods, we propose a novel taxonomy by which we can classify them. Specifically, we propose three major categories: (1) logicallyinformed embedding approaches, (2) embedding approaches with logical constraints, and (3) rule-learning approaches. Alongside the taxonomy, we provide a tabular overview of the approaches and links to their source code, if available, for more direct comparison. Finally, we discuss the applications on which these methods were primarily used and propose several prospective directions toward which this new field of research could evolve.

show abstract

Section: A Logically-informed Embedding Approachesmentioning

confidence: 99%

Section: Learning Rules For Graph Reasoningmentioning

confidence: 99%

Section: Learning Rules For Graph Reasoningmentioning

confidence: 99%

See 1 more Smart Citation

Neurosymbolic AI for Reasoning on Graph Structures: A Survey

DeLong¹,

Fernández²,

Whyte³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The completion process is developed through adding new triples to a knowledge graph, including the three link, entity and relation prediction subtasks (Shi & Weninger, 2018). It can be realized by different approaches, such as embedding-based models, relation-path inference, path finding reasoning, metarelational learning, and rule-based reasoning (Lin et al, 2015;Lin et al, 2021;Trouillon et al, 2017). Each method has its specific advantages and disadvantages, and can be used according to the demands of practical applications.…”

Section: Geokg Completionmentioning

confidence: 99%

Geoscience Knowledge Graph (GeoKG): Development, construction and challenges

Zhang

Huang

Zhang

et al. 2022

Transactions in GIS

View full text Add to dashboard Cite

Big earth data is a cross‐domain of geoscience and information science, which provides a novel perspective for solving geoscience problems. Most contemporary research is driven by data but neglect the potential value of knowledge. As a new scientific language in Geoscience, GeoKG is essential for understanding, representing, and mining geoscience knowledge, and can contribute to the integration of big earth data, geoscience knowledge, and geoscience models. However, research on GeoKG lack spatiotemporal perspectives in knowledge cognition, representation, acquisition and management. To this end, this article first outlines a cognitive mechanism from the human–machine double perspective and categorizes the characteristics and content of geoscience knowledge. To express evolution and complex natural rules, a knowledge representation framework is proposed through ‘state‐process’ and ‘condition‐result’ models. Aiming at multimodal data, a workflow is put forward to acquire knowledge from a small sample, a knowledge graph, a map, and a schematic diagram. Furthermore, we discuss the organization of GeoKG by improving existing data models, developing spatiotemporal correlation indexing and advancing knowledge graph completion. The concrete construction process of GeoKG is analyzed thoroughly in this study, which can support the synthetic analysis of big earth data, promote the development of knowledge engineering and provide a foundation for improving intelligent geoscience.

show abstract

“…It is adopted as a cost-effective way to provide accurate evaluation and system optimization, which stores a set of rules for management support, expert knowledge, and so on [ 25 ]. By determining the strong relationship between the data collected during the process execution and the knowledge rules [ 26 ], an optimized quality evaluation standard and prediction model is constructed [ 27 ] to effectively ensure the quality and safety of products in the actual process and provide decision-makers and users with a more accurate evaluation and prediction results.…”

Section: Introductionmentioning

confidence: 99%

Intelligent Dynamic Quality Prediction of Chilled Chicken with Integrated IoT Flexible Sensing and Knowledge Rules Extraction

Stankovski

et al. 2022

Foods

View full text Add to dashboard Cite

With the enhancement of consumers’ food safety awareness, consumers have become more stringent on meat quality. This study constructs an intelligent dynamic prediction model based on knowledge rules and integrates flexible humidity sensors into the non-destructive monitoring of the Internet of Things to provide real-time feedback and dynamic adjustments for the chilled chicken cold chain. The optimized sensing equipment can be attached to the inside of the packaging to deal with various abnormal situations during the cold chain, effectively improving the packaging effect. Through correlation analysis of collected data and knowledge rule extraction of critical factors in the cold chain, the established quality evaluation and prediction model achieved detailed chilled chicken quality level classification and intelligent quality prediction. The obtained results show that the accuracy of the prediction model is higher than 90.5%, and all the regression coefficients are close to 1.00. The relevant personnel (workers and cold chain managers) were invited to participate in the performance analysis and optimization suggestion to improve the applicability of the established prediction model. The optimized model can provide a more efficient theoretical reference for timely decision-making and further e-commerce management.

show abstract

Rule-enhanced iterative complementation for knowledge graph reasoning

Cited by 25 publications

References 16 publications

Neurosymbolic AI for Reasoning on Graph Structures: A Survey

Neurosymbolic AI for Reasoning on Graph Structures: A Survey

Geoscience Knowledge Graph (GeoKG): Development, construction and challenges

Intelligent Dynamic Quality Prediction of Chilled Chicken with Integrated IoT Flexible Sensing and Knowledge Rules Extraction

Contact Info

Product

Resources

About