A Dataset Quality Evaluation Algorithm for Data Trading on Blockchain

As a vital solution for data trading, blockchain often carries a substantial amount of critical data. Consequently, data quality, as a crucial evaluation metric, plays a key role in facilitating data trading on the blockchain. In the process of comprehensively evaluating the quality of multi-field datasets, it is crucial to determine the importance coefficient of each data item. The importance evaluation methods of data items include subjective methods that depend on expert scoring, as well as objective methods based on statistical methods. Objective methods evaluate field importance based on the laws of data itself. However, such methods start from the quality of data to be evaluated, and cannot reflect the mutual dependency between data items and the resulting field importance. To solve this problem, this article starts from the business dependency between data items, constructs a directed graph, namely dependency network. On this basis, according to the characteristics and quantity of data items that depend on fields, we propose the FieldRank algorithm. The algorithm objectively quantifies and evaluates the importance of each field, so that the field importance weight obtained based on the algorithm is directly related to the actual business and is more close to the essence of data business. Therefore, this algorithm can provide objective data quality evaluation results for data use. Practice has proved that in the case of large amounts of big data and closely related businesses, the data quality evaluation results obtained by this method are more close to actual business practice.