Seminars

Home Centers & Programs AI and Natural Sciences Seminars

TITLE: Natural Language Processing-based Literature and Text Data Analysis for Efficient Scientific Experiments

ABSTRACT: Scientific literature is composed of a vast array of documents written in natural language, representing a carefully curated collection of knowledge that is intricately intertwined. This extensive source of knowledge has great potential to yield valuable information when text-mined and analyzed. Research in scientific literature mining has been conducted across various disciplines to automate the process of knowledge acquisition and to address challenges associated with handling large volumes of textual data. One practical application is the automated retrieval of relevant information, which can streamline scientific research by eliminating unnecessary variables and proposing new hypotheses, thereby making costly and time-consuming experiments more efficient. Given the rapid growth of digitized research papers, establishing data collection from literature facilitates scientific discovery while reducing manual labor and human error. To automatically process literature data, natural language processing (NLP) techniques have been utilized across different fields. With the advent of large language models (LLMs), these processes have become even more effective. This presentation will introduce studies applying NLP techniques and LLMs that can be utilized in various research experiments, including those in biology and materials science.