Relationship discovering method and device

Abstract

The invention discloses a relationship discovering method and device, and relates to the field of data mining. The method comprises the steps that a union of entities of a first entity type and entities of a second entity type is obtained, and information entropy of the entities contained in the union is calculated; according to the first attributes of the first entity type and the second attributes of the second entity type, clustering is conducted on the entities in the union, so that the entities containing same attribute values are classified into same clusters, and information entropy ofthe entities contained in the union after clustering is calculated; according to the information entropy before and after clustering, the changing situation of the information entropy before and afterclustering is determined, and then whether or not the first attributes of the first entity type and the second attributes of the second entity type are suitable for connection relationship building is determined. Therefore, by means of the method and device, connection relationships can be automatically discovered; moreover, based on the changing situation of the information entropy before and after entity clustering, attribute pairs with appropriate connection numbers which are suitable for connection relationship building can be discovered.

Description

technical field [0001] The invention relates to the field of data mining, in particular to a method and device for discovering relationships. Background technique [0002] At present, when using structured data to build a connection relationship, the data describing two types of entities is usually extracted first. The entity data contains attribute information of several entities, and then it is judged whether there is a certain relationship between the two entities according to whether the attribute values ​​are the same. . [0003] For example, if the value of the attribute "children's name" of entity A is "Zhang San", and the value of the attribute "name" of entity B is "Zhang San", then the same attribute value appears between entity A and entity B In this case, a "child relationship" connection relationship can be constructed. [0004] However, if the user does not know the meaning of the attribute or the distribution of the attribute, the connection relationship exp...

AI Technical Summary

Problems solved by technology

[0004] However, if the user does not know the meaning of the attribute or the distribution of the attribute, the connection relationship expected by the user cannot be constructed, especially when performing relationship mining in massive data, relying only on experience or manual observation and judgment , it is often difficult to construct the connection relationship expected by the user

Images