On July 25, 2025, the thirdlessonof “AI SummerCamp” course series,with the topic of "AI Empowering Literature Reading", was conducted by Daxia College. This lesson was delivered by Xia Yang, an intermediate librarian of Inter-disciplinary Support Center and hosted by Zhan Song, the Teaching Tutor of Daxia, with ECNU library as the co-organizer.
In the lecture, Xia emphasized the importance of literature reading in scientific research and the low efficiency of traditional keyword search. He introduced the evolution of AI technology: from early keyword matching to modern semantic search. Relying on Retrieval-Augmented Generation (RAG) technology, which combines semantic retrieval and large-model summarization, search accuracy and efficiency are improved. He also explained the principles, advantages and disadvantages of Embedding models and Large Language Models (LLMs).
He recommended practical AI tools, including Scopus AI (for English databases), Mita AI Search (for Chinese databases, offering intelligent retrieval and analysis of academic literature) and CNKI Enhanced Search (for Chinese databases, with semantic search and knowledge graphs). He demonstrated Mita AI on-site, showing its ability to quickly understand literature and generate structured summaries.
In addition, Xia reminded that AI tools have issues like omissions in long-text processing and possible identification of generated content by paper detection systems. He suggested reasonable segmentation, keyword classification and manual revision for optimization. He also mentioned the limitations of large models, such as higher error rates of Transformer models with more context, but noted that a specific workflow can improve the summarization effect of large models by 25 times. In the interactive session, many students said they would try these tools to enhance research efficiency.
