Identity Linking Method Based on Multilevel Attribute Embedding and Constrained Canonical Correlation Analysis
A typical correlation analysis and attribute technology, applied in neural learning methods, biological neural network models, instruments, etc., can solve problems such as the difficulty of capturing the implicit connection of different user attributes, and the difficulty of uniformly dealing with various types of attribute texts. The effect of data acquisition cost and method training cost, reducing the amount of prior information, and strong robustness
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0035] see figure 1 , an identity linking method based on multi-level attribute embedding and constrained canonical correlation analysis, including the following steps:
[0036] (a) Preprocess social network user data; represent social network users as nodes, and relationships between users (such as friends, followers / fans, etc.) as edges, and construct an undirected and unweighted graph G=(V, E , A), where V represents the set of users in the network, E represents the set of relationships between users (such as friend relationship, follower / fan relationship, etc.), and A represents the set of user attributes, such as user name, occupation and educational experience, etc.
[0037] (b) Embedding multi-level text attributes; first divide the text attributes of each network into three parts A = (A c ,A w ,A t ), where A c Represents a character-level attribute, A w Represents word-level attributes, A t represents topic-level attributes; then three corresponding user feature...
Embodiment 2
[0075] The present invention will be further described below in conjunction with specific examples. This example is two real social networks collected from the Internet, Sina Weibo and Douban. The specific information is shown in Table 1.
[0076] Table 1 Weibo-Douban Network Data Statistics
[0077]
[0078] Step (a): preprocessing social network user data. ;
[0079] Consider the users in the two social networks Weibo and Douban to be matched as network G X / G Y = node V in (V, E, A), and use different numbers to distinguish different users. For example, users in Weibo network correspond to numbers 0 to 9713, and users of Douban network correspond to numbers 9714 to 19239.
[0080] The following / fan relationship between users is regarded as an edge E in the network, that is, if two users have a following or fan relationship, an edge (u) is constructed between them. i ,u j ) ∈ E.
[0081] Use the respective screen names (ie nicknames) of users in the two networks as ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More