دانلود کتاب تحلیل زبان طبیعی با مدل‌های مولد زبان‌های بزرگ – یک رویکرد عملی با Ollama و LLMهای متن‌باز

The inception of this book arose from the authors’ collective ambition to explore and utilize generative Large Language Models (LLMs) as “language calculators” to infer and extract soft data, providing semantic analyses and insights expressed in the form of Key Soft Indicators (KSIs). While LLMs are often employed for tasks such as translation and summarization, this book seeks to extend their use to a more sophisticated form called Natural Language Analytics (NLA). This book includes semiotic extraction, sentiment analysis, emotion inference, mindset analysis, and cultural understanding—tasks that enable the processing of complex and multistable textual data to uncover patterns and offer insights into human decision-making. The goal of this book is to show how generative LLMs can be employed in a downstream process to extract useful data from natural language artifacts in a manner that integrates seamlessly into the data science pipeline, ultimately delivering actionable intelligence to stakeholders. By leveraging these models, analyses can be conducted across a variety of fields, including strategic communication, marketing, human resources, journalism, branding, customer satisfaction, social media, civil defense, risk reduction, and cognitive defense, among others. This work represents a transdisciplinary endeavor, bringing together expertise from computer science, artificial intelligence, cognitive semiotics, linguistics, management, and information systems, spanning both industry and academia. The convergence of these diverse fields is essential to effectively address the complexities of using LLMs for NLA. Consequently, this book is intended not only for AI specialists but also for professionals across a wide range of fields who are familiar with information systems, decision-making, data science, and natural language processing and who are interested in employing LLMs for natural language analysis, particularly using open-source models on local systems. Understanding the utility of LLMs as tools for NLA involves recognizing both their strengths and limitations. The book adopts the metaphor of LLMs as “language calculators” to explain how they function by mapping input sentences to output soft data artifacts, driven by complex semantic embeddings. While this metaphor has its limits, it provides a useful framework for understanding the strengths and weaknesses of LLMs in language processing tasks. The book discusses and provides solutions for the inherent variability in LLM outputs and addresses the implications for tasks that require high precision versus those where utility and nuanced interpretation are more critical. Additionally, it tackles challenges related to the ambiguity and variability of natural language and illustrates how LLMs can be effectively set up and applied in this context. A case study is presented on the development of a KSI derived from text data, showcasing real-world applications of these techniques. Finally, this book does not aim to be an exhaustive guide, but rather to provide practical solutions and guidance based on the authors’ experiences. As such, it serves as a starting point for those interested in the evolving field of NLA using LLMs. We hope this book will inspire and equip readers with the tools and knowledge to explore the potential of LLMs in their own work, fostering innovation and advancing understanding in this dynamic area.