Independent Office Meeting Minutes Text Extraction Dataset

In modern enterprises, meeting minutes are a crucial component of daily operations. However, extracting key content poses challenges due to the richness of information and the need for structured data. Existing solutions largely rely on manual review, which is inefficient and prone to errors. This dataset is constructed to support automated text information extraction, enabling efficient information classification and management. Data collection involved collaboration with multiple enterprises, using transcription devices and software to capture real meeting minutes in natural office environments. During data annotation, multiple rounds of annotation and consistency checks were conducted, and a team of experienced linguists reviewed the data to ensure high accuracy and reliability. The annotation process involved more than 5 experts with a linguistic background to maintain the dataset's professional level. Data preprocessing involved text cleaning, syntactic analysis, and semantic annotation, utilizing cutting-edge natural language processing techniques to enhance model training effectiveness. Data is stored in a uniform TXT format, with entries classified using embedded labels, facilitating easy access to information and model training.