MEASURING SIMILARITY OF NUMERIC CONCEPT VALUES WITHIN A CORPUS

A method, computer system, and computer program product for measuring similarity of numeric concept values within a corpus are provided. The embodiment may include retrieving numerical values associated with a concept in a corpus. The embodiment may also include converting the numerical values to a standard unit. The embodiment may further include computing a distribution value of the converted numerical values. The embodiment may also include determining a tolerance value based on the distribution value, wherein the tolerance value is the maximum allowable distance between two numerical values. The embodiment may further include determining a distance function based on the determined tolerance value, wherein the distance function is defined by dividing a difference between two numerical values by the determined tolerance value. The embodiment may also include computing a similarity distance between the numerical values.