Community-Oriented Sentence Simplification: Towards Accessible Language Processing

Sentence simplification aims to convert complex sentences into simpler forms while preserving their original meaning, improving accessibility for diverse audiences, including second-language learners and individuals with cognitive disabilities. This paper introduces a communityoriented sentence simplification (COSS) framework that tailors simplifications to the linguistic needs of specific user groups by incorporating community-specific constraints, such as predefined vocabulary lists and grammar rules. The hybrid framework integrates rule-based and data-driven approaches, leveraging tools like WordNet, BART (Bidirectional and Auto-Regressive Transformers), and SBERT (Sentence-Bidirectional Encoder Representations from Transformers) to balance lexical and syntactic simplifications while maintaining semantic consistency. Four progressively enhanced approaches were devised and implemented, and then evaluated using standard metrics, including FKGL (Flesch-Kincaid Grade Level) and SARI (System output Against References and against the Input sentence). The results demonstrate that the Hybrid Simplification approach achieves the best balance between readability, semantic retention, and adherence to communityspecific constraints. Additionally, a Gradio-based user interface ensures accessibility and interactivity for end users. This work highlights the importance of tailoring simplifications to diverse linguistic needs and lays the foundation for future advancements in adaptive simplification systems.