Journal of Advanced Research in Computational Linguistics: Journal of Computer Science Language

Retrieval-Augmented Generation for Large Language Models: Architectures, Applications, and Challenges

2026-09-16T08:57:23+00:00

Large Language Models (LLMs) exhibit impressive generative capabilities but remain limited by static parametric knowledge, hallucination, and the inability to access domain-specific or up-to-date information. Retrieval-Augmented Generation (RAG) addresses these limitations by coupling LLMs with external retrieval mechanisms that ground generation in retrieved evidence. This paper presents a systematic study of RAG architectures, tracing their evolution from naive retrieve-then-read pipelines to advanced modular and agentic designs incorporating iterative retrieval, reranking, and self-reflection. We categorize architectural variants along retrieval, indexing, and generation dimensions; survey applications across question answering, healthcare, legal, and enterprise domains; and report an experimental comparison of sparse (BM25), dense (bi-encoder), and hybrid retrieval configurations on a domain-specific question answering benchmark. Our results show that hybrid retrieval with cross-encoder reranking improves answer accuracy by up to 11.4% over naive dense retrieval while reducing hallucination rates. We conclude by analyzing open challenges—retrieval noise, context window limitations, evaluation methodology, latency, and privacy—and outline directions for future research.

Self-Evolving Large Language Model Agent with Adaptive Long-Term Memory and Autonomous Knowledge Updating

2026-09-16T09:03:27+00:00

Large Language Model (LLM) agents are increasingly being used for multi-step reasoning, tool use, information retrieval, and autonomous task execution. However, conventional LLM agents are largely session-oriented: they have limited persistent memory, may repeatedly relearn previously observed information, and can generate responses from stale knowledge. This paper proposes a self-evolving LLM agent architecture that combines adaptive long-term memory, autonomous knowledge updating, confidence-aware retrieval, and controlled self-reflection. The proposed framework separates short-term conversational context from long-term episodic, semantic, and procedural memory. A memory controller determines which experiences should be stored, strengthened, summarized, replaced, or forgotten. An autonomous knowledge updater periodically discovers candidate information from trusted external sources, evaluates evidence, detects conflicts with stored knowledge, and updates the knowledge base only when predefined confidence and consistency conditions are satisfied. The framework also introduces temporal validity, provenance tracking, contradiction detection, and rollback mechanisms to reduce knowledge drift and memory corruption. A comprehensive evaluation methodology is proposed using long-horizon task completion, memory precision and recall, knowledge freshness, update accuracy, contradiction rate, context efficiency, and recovery from incorrect updates. The research aims to provide a practical architecture for LLM agents that can continuously improve while maintaining traceability and reliability

An Efficient Small Language Model with Retrieval-Augmented Generation for Educational Information Systems

2026-09-17T07:35:31+00:00

Educational institutions maintain a large amount of information in notices, academic regulations, syllabi, faculty profiles, examination instructions, admission documents, and other institutional records. Students and staff often need this information in a simple question-and-answer form, but the information is usually distributed across many documents. Large language models can provide natural conversations, but their size, computing requirements, and tendency to produce unsupported answers make direct use difficult for a college-level information system. This paper presents a practical framework that combines a Small Language Model (SLM) with Retrieval-Augmented Generation (RAG) for answering educational and institution-specific questions. The proposed system keeps the college information in an external knowledge base and retrieves relevant content before generating an answer. This separates the changing college information from the language model and makes the system easier to update. The paper describes the data preparation process, retrieval pipeline, prompt construction, SLM integration, evaluation plan, and possible deployment on limited hardware. The proposed framework is intended for academic institutions that want a conversational information assistant without depending completely on a large hosted language model. The paper also discusses accuracy, relevance, response time, hallucination, privacy, and maintenance as important evaluation areas.

An Overview of Discoverable and Reusable Intelligence for Collaborative Edge Devices Without Full Model Transfer

2026-09-16T09:08:12+00:00

The rapid growth of Artificial Intelligence (AI), Internet of Things (IoT), and edge computing has transformed the development of intelligent applications across healthcare, transportation, manufacturing, agriculture, and smart cities. Although cloud-based AI provides powerful computational capabilities, it often suffers from high latency, bandwidth limitations, privacy concerns, and dependency on continuous internet connectivity. Edge AI addresses these limitations by performing computation closer to where data is generated. However, conventional collaborative learning techniques such as Federated Learning still require repeated exchange of complete model parameters, resulting in high communication overhead and increased resource consumption.

Recent research has introduced the concept of discoverable and reusable intelligence, where edge devices collaborate by exchanging reusable knowledge instead of transferring complete AI models. This paradigm enables devices to discover suitable knowledge sources and reuse learned intelligence through techniques such as knowledge distillation, prototype sharing, and lightweight transfer learning. As a result, communication cost is reduced while maintaining privacy, scalability, and learning performance.

This review paper presents an overview of discoverable and reusable intelligence for collaborative edge devices without full model transfer. It discusses the evolution of collaborative Edge AI, its architecture, major application areas, existing challenges, research gaps, and future research directions. The paper also highlights how this emerging paradigm can improve collaboration among heterogeneous edge devices and support the development of efficient and intelligent distributed AI systems.

Unlocking Sentiment Insights on Twitter through Quantum Computing

2026-06-08T06:58:41+00:00

This study looks at the role played by quantum influence and entangled by machine learning methods to improve sentiment analysis in social media type content. With the digital arena being on the verge of huge transformations, the interpretation of implicit feelings conveyed on the Internet has in recent years become an essential skill necessary for many areas of applications that include marketing efforts, public opinion analysis, and trend detection. Through the fusion of quantum computing principles alongside with state-of-the-art machine learning methods, our research work ultimately aims at that sensing sentiments permeate the entire social media ecosystem. We intend to span the idea gap between quantum mechanics and the feelings in the data analysis process, exploring the new path for understanding the complex emotions into the online interaction data. By implementing empirical analysis and experimental research, we strive to provide an illustration of the interdisciplinary nature of our methods and to eventually demonstrate the efficacy of this approach in reaching higher sentiment analysis accuracy and uncovering unknown ontological patterns within social media content. This will ultimately bring us closer to a better understanding of the dynamics of digital discourse.

DOI: https://doi.org/10.24321/3051.424X.202604

How to cite this article:
Lateshwari, Bansal S K. Unlocking Sentiment Insights on Twitter Through Quantum Computing. J Adv Res Comput Linguist : J Comput Sci Lang. 2026; 2(2): 1-9.