MDV: A Multi-Factors Data Valuation Method

Data valuation plays a key role in information lifecycle management (ILM). There are two kinds of methods to assess data's value: policy-based method and non-policy-based method. A policy-based method depends on the administrators' principles and the relations between applications in upper layer which makes data value assessment comprehensive. UT (usage over time) which is a typical non-policy-based method, uses visit time to assess data's value. Such ways make data valuation simple and objective. However, as the appearance of big data, the existing algorithms cannot be appropriate to assess data value quickly and accuracy. In this paper, a new one called MDV based on UT method is proposed. MDV takes more factors into consideration, including data's size, age and usages. Such synthetic idea makes the method more suitable for the features of big data. We build an experiment platform based on Hadoop and test the algorithm. The preliminary results show that MDV could achieve the valuation goal.