An open dataset of Chinese duration expressions

Area of Science:

• Linguistics
• Natural Language Processing
• Psychology

Background:

• Duration information is crucial for understanding text, presented numerically (e.g., 1 hour) or verbally (e.g., shortly).
• Existing lexicons lack comprehensive mappings between verbal duration expressions and numerical values.
• Databases of temporal expressions often omit word frequency data, hindering processing analysis.

Purpose of the Study:

• To create a comprehensive dataset of Chinese duration expressions with corresponding numerical values and word frequencies.
• To support research in natural language processing, psychology, and linguistics by providing a valuable resource for temporal information analysis.
• To address the gap in lexicons for converting verbal duration expressions to numerical durations.

Main Methods:

• Compiled an open dataset of 2,101 Chinese duration expressions.
• Annotated each expression with its corresponding numerical duration.
• Obtained word frequencies from a 10 billion character corpus (BLCU Corpus Center) and computed adjusted frequencies.

Main Results:

• A dataset of 2,101 Chinese duration expressions, each linked to a numerical duration.
• Inclusion of adjusted word frequencies for each expression, derived from a large-scale corpus.
• The dataset provides a foundational resource for analyzing temporal expressions in Chinese.

Conclusions:

• The newly created dataset offers a significant resource for researchers studying temporal information in Chinese.
• This dataset facilitates advancements in natural language processing, psychology, and linguistics by enabling quantitative analysis of duration expressions.
• The inclusion of word frequency data enhances the dataset's utility for understanding information processing related to time.