Task-oriented method for improving data availability

Abstract

The invention discloses a task-oriented method for improving data availability. Based on a two-part graph theory and the correlation between data attributes and task attributes, a task-oriented potential available attribute mining model is constructed; based on the two-part graph theory and the task-oriented data attribute correlation, a task-oriented multi-source data mining model with complementary attributes is constructed; then potential available attributes and complementary multi-source data of an existing data set are mined through the constructed task-oriented potential available attribute mining model; and then other multi-source data sets with the complementary attributes of the existing data set are mined through the constructed task-oriented multi-source data mining model withthe complementary attributes. According to the task-oriented method for improving the data availability, the provided potential attribute mining model and the complementary multi-source data mining model can screen out available attributes and multi-source data exceeding users' expectations for specific tasks, and then the implementation efficiency of the specific tasks can be improved.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a task-oriented method for improving data availability. Background technique [0002] With the development of information technology, the ability to obtain data has been greatly improved. We can obtain massive, multi-source, and heterogeneous data in a timely manner. That is, many existing available data attributes are not related to specific mission objectives; on the other hand, due to information islands, data privacy security and other reasons, many data attributes related to specific prediction or decision-making tasks cannot be obtained in time. [0003] Thus, there is an inherent contradiction in the process of data availability analysis for specific tasks: specific tasks require specific data properties, but we cannot obtain these properties from the available data; available data has many properties, but these properties are not related to specific Tasks are not directly ...

AI Technical Summary

Problems solved by technology

[0005] (1) The research on data quality mainly focuses on the accuracy and correlation of data, while the research on data usability for specific tasks is less;

[0006] (2) The research on data set correlation is mainly concentrated in the field of information retrieval, and its application is mainly reflected in the precision marketing and personalized recommendation of e-commerce, while the current data correlation research mostly focuses on the correlation between existing data attributes and task requirements. There is still a lack of correlation mining for potentially valuable attributes of data;

[0007] (3) The research on data set completeness is mainly concentrated in the field of data fusion, and its application is mainly reflected in the field of data transactions. However, the current research on data completeness mostly focuses on the integrity of existing data attributes, and there is still a lack of specific task requirements. , Complementarity Mining of Multi-source Heterogeneous Data Attributes

Images