2026-10-12T18:14:25Z https://www.scidb.cn/oai

10.57760/sciencedb.44643 2026-09-20T09:57:30Z

2026-09-20 Dataset of Vendor Participation Behavior and Transaction Information in Skill-Based Crowdsourcing Platforms doi:10.57760/sciencedb.44643 en This dataset is collected from EPWK (https://www.epwk.com/), one of the largest skill-based crowdsourcing platforms in China. Python was used to crawl completed bidding task data over approximately six years. After data cleaning, deduplication, and outlier removal, the final dataset contains 8,933 task records, involving 7,835 employers and 69,643 vendors.The dataset includes task information, employer characteristics, vendor participation behavior, and textual semantic features. Task-related information contains task descriptions, task budgets, and deadlines. Employer-related information includes authentication status, credit ratings, and employer scale. Vendor participation behavior includes the number of participants and vendor capability levels. Furthermore, natural language processing techniques were applied to extract textual features, including BERT-based classification models for identifying employer preferences, large language model-based semantic analysis for evaluating task complexity, and LDA topic modeling for extracting latent topics from task descriptions.This dataset can support research on user behavior analysis, transaction mechanisms, network effects, and text mining in skill-based crowdsourcing platforms. Skill-based crowdsourcing platform; Vendor participation behavior; Crowdsourcing transaction data; User behavior analysis; Text mining Wang Xiaolun lin xin lan Zhao Yuxiang Yao Qian RESTRICTED dataset Science Data Bank