Answering system based on langchain : intelligent system
Du, Yifan (2026)
Kandidaatintyö
2026
School of Engineering Science, Tietotekniikka
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe2026052151383
https://urn.fi/URN:NBN:fi-fe2026052151383
Tiivistelmä
This paper introduces the design and implementation of a question answering system based on langchain, which is a framework to facilitate application development based on language models. This study explores the integration of langchain with external knowledge sources to leverage knowledge graphs to improve the accuracy and reliability of generated responses. The main goal is to build a QA system that can effectively retrieve and process information to provide accurate and contextual answers.
This study begins with a comprehensive analysis of existing QA systems and their limitations, highlighting challenges in information retrieval, response generation, and system scalability. Langchain's modular architecture enables seamless interaction with large language models, external databases, and apis, and is then investigated as a solution to address these challenges. This paper details the design of the system, including data preprocessing, prompt engineering, and the implementation of retrieval augmented generation to improve response accuracy.
We conclude with a discussion of practical applications, limitations, and possible areas for future improvement of the system. The results show that langchain provides a flexible and extensible framework for building robust QA systems, providing valuable insights for further developments in natural language processing and AI-driven information retrieval.
This study begins with a comprehensive analysis of existing QA systems and their limitations, highlighting challenges in information retrieval, response generation, and system scalability. Langchain's modular architecture enables seamless interaction with large language models, external databases, and apis, and is then investigated as a solution to address these challenges. This paper details the design of the system, including data preprocessing, prompt engineering, and the implementation of retrieval augmented generation to improve response accuracy.
We conclude with a discussion of practical applications, limitations, and possible areas for future improvement of the system. The results show that langchain provides a flexible and extensible framework for building robust QA systems, providing valuable insights for further developments in natural language processing and AI-driven information retrieval.