抖音爆款商品评价中英双语平行语料集

本数据集是由百图科技工作室虚拟生成的中英双语平行语料集，模拟抖音电商平台商品评价的内容风格，共计100,000条有效数据。每条语料同时提供地道的中文和英文版本，中文长度控制在2-120字，英文控制在1-80词。数据涵盖正面（40%）、中性（30%）、负面（30%）三种情感倾向，涉及100款抖音爆款商品类别。所有数据均附有细粒度自动标注，包含情感倾向、情感强度（1-5级）、主题关键词、语料意图、语气风格、传播指数、语言美感评分、翻译方式及双语对齐质量等13个标注字段，可为自然语言处理相关研究提供高质量训练与评测资源。This dataset is a Chinese-English parallel corpus virtually generated by Baitu Technology Studio, which simulates the content style of product reviews on the Douyin E-commerce platform and contains a total of 100,000 valid entries. Each entry is provided with both authentic Chinese and English versions, where the Chinese text is limited to 2-120 characters and the English text is limited to 1-80 words. The dataset covers three sentiment orientations: positive (40%), neutral (30%), and negative (30%), involving 100 categories of Douyin best-selling products. All entries are equipped with fine-grained automatic annotations, including 13 labeled fields such as sentiment orientation, sentiment intensity (1-5 scale), topic keywords, corpus intent, tone style, propagation index, linguistic beauty score, translation method, and bilingual alignment quality. It can serve as high-quality training and evaluation resources for natural language processing-related research.

搜集汇总

背景与挑战

背景概述

本数据集是一个由百图科技工作室虚拟生成的、包含10万条有效数据的中英双语平行语料集，模拟抖音电商平台的商品评价内容风格。数据涵盖正面、中性和负面三种情感倾向，涉及100款爆款商品类别，每条语料均附带情感倾向、强度、主题关键词等13个细粒度自动标注字段，适用于自然语言处理相关研究与开发。