Citation

Shah, Nathar and Messom, Christopher (2026) A Parse Table for εL+ Language Using Top-Down Parsing Approach. IEEE Access, 14. pp. 100036-100053. ISSN 2169-3536

Text 6.pdf - Published Version Restricted to Repository staff only Download (3MB)

Abstract

The traditional MapReduce paradigm, while widely adopted for its scalability, fault tolerance, and ease of distributed computing, lacks native support for expressive reasoning capabilities. This limitation becomes increasingly significant in domains such as healthcare, and finance, where ontologies and description logics are essential for enabling intelligent data interpretation and decision-making. Motivated by this gap, we propose an Expressive Hadoop MapReduce Framework (EHMR) based on the EL+ description logic, aiming to bridge the divide between scalable data processing and semantic reasoning. The investigation developed a frontend for an EHMR Framework consisting of an EL+ description logic based parse table. We achieved this by formulating a new normal form taking advantage of an inherent tree-like structure present in EL+ . A set of normalization rules were developed to convert complex EL+ expressions to its normalized forms. We showcased how modified FIRST and FOLLOW rules can be used to build a parse table for EL+ . We evaluated the time performance of the parse table to several workloads generated by custom benchmarking suites. The parse table is having better time performance results in the range between O(n) and O(log n) compared to theoretical polynomial complexity evidenced by EL+ based CEL algorithm. These findings are significant in providing more responsive EL+ query answering. The proposed EHMR framework has broad applicability across various domains. In healthcare, it can be used to reason over complex medical ontologies such as SNOMED CT, enabling real-time clinical decision support. In finance, it can model relationships between economic indicators, asset classes, and market trends, allowing analysts to query the soundness of investment strategies. The framework’s ability to handle large-scale knowledge bases with precomputed reasoning results makes it suitable for big data environments where responsiveness is critical.

Item Type:	Article
Uncontrolled Keywords:	Big data, distributed computing, expressiveness, knowledge-based systems, map reduce, parse table
Subjects:	P Language and Literature > P Philology. Linguistics
Divisions:	Faculty of Computing and Informatics (FCI)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	31 Jul 2026 05:55
Last Modified:	31 Jul 2026 05:55
URII:	http://shdl.mmu.edu.my/id/eprint/16405

Downloads

Edit (login required)