EUVIMEDEUVIMED entdecken
Crossref

Relevant Infinite Content Genesis with Vector Store and AI agents: A PERT-Driven Hybrid Framework for NLP - Enhanced Semantic Retrieval and Evaluation

Ayushman Pranav, Rajesh Kumar Modi, Ankit Dubey, Umesh Gupta, Ankit Dubey, Pankaj Mishra

International Conference on Computers in Education · 01.12.2025

DOI: 10.58459/icce.2025.5923 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Conference on Computers in Education, online erschienen am 01.12.2025. Verfasst von Ayushman Pranav und 5 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automated content generation systems face significant challenges in ensuring quality, coherence, and relevance while maintaining scalability. This paper introduces a PERT (Program Evaluation and Review Technique)-driven workflow that integrates advanced natural language processing (NLP), semantic retrieval, and collaborative multi-agent frameworks to address these challenges. Leveraging transformer-based embeddings (MiniLM-L12-v2) and a Qdrant vector database, the system performs context-aware retrieval of topically relevant articles, enabling dynamic content planning via a simulated language model. A trio of specialized agents'ontent Planner, Writer, and Editor'ollaborate to produce polished articles adhering to SEO and stylistic guidelines. Rigorous evaluation is conducted through a hybrid framework: quantitative metrics (lexical diversity, readability scores) are computed via automated scripts, while qualitative dimensions (coherence, persuasiveness, tone) are assessed using a fine-tuned LLaMA-3-70B model. Results from five representative topics (e.g., Quantum Computing, AI in Healthcare) demonstrate the system' efficacy, with articles achieving an average Flesch-Kincaid Grade Level of 8.2 (SD = 0.23) and high qualitative scores (4.6/5 for coherence, 4.5/5 for persuasiveness). Consensus evaluation approved all outputs, requiring only minor revisions (mean = 1.4 per article). The framework' modular architecture reduces workflow latency by 32% compared to linear approaches, while semantic retrieval improves source relevance by 41% over traditional TF-IDF methods. This study establishes a benchmark for scalable, high-quality content generation, offering insights into the integration of structured workflows, NLP advancements, and multidimensional evaluation for applications in technical journalism, education, and SEO-driven media.

Zitieren

APA

Pranav, A., Modi, R. K., Dubey, A., Gupta, U., Dubey, A., & Mishra, P. (2025). Relevant Infinite Content Genesis with Vector Store and AI agents: A PERT-Driven Hybrid Framework for NLP - Enhanced Semantic Retrieval and Evaluation. International Conference on Computers in Education. https://doi.org/10.58459/icce.2025.5923

Vancouver

Pranav A, Modi RK, Dubey A, Gupta U, Dubey A, Mishra P. Relevant Infinite Content Genesis with Vector Store and AI agents: A PERT-Driven Hybrid Framework for NLP - Enhanced Semantic Retrieval and Evaluation. International Conference on Computers in Education. 2025. doi:10.58459/icce.2025.5923

BibTeX
@article{Pranav2025,
  author = {Pranav, Ayushman and Modi, Rajesh Kumar and Dubey, Ankit and Gupta, Umesh and Dubey, Ankit and Mishra, Pankaj},
  title = {Relevant Infinite Content Genesis with Vector Store and AI agents: A PERT-Driven Hybrid Framework for NLP - Enhanced Semantic Retrieval and Evaluation},
  journal = {International Conference on Computers in Education},
  year = {2025},
  doi = {10.58459/icce.2025.5923}
}
Kostenlos starten →