Visual analysis framework for understanding missing links in bipartite networks
地域：
Tokyo
摘要
Example implementations described herein involve an interface for calculating and displaying missing links for data represented as a bipartite network, along with novel methods for improving link prediction algorithms in the related art. Through example implementations described herein, the accuracy of link prediction algorithms can be improved upon, thereby providing the user with a more accurate understanding of the data in the bipartite network.
說明書
The present disclosure relates generally to data analytics, and more specifically, for determining and visualizing missing links within bipartite networks.
Many real-world complex systems can be modeled as bipartite networks (two-mode networks), where there are two types of nodes in a network and links only exist between different node types. Analyses of bipartite relationships have been used for data analytics in various application domains, such as studying political leanings with voter-vote networks based on roll call vote records, and investigating gene-expression networks in bioinformatics.
One analysis problem for such networks is link prediction (e.g., detecting missing links), which infers the existence of new relationships between nodes based on currently observed links. Such link production is valuable because real-world data may be noisy or incomplete. But normally the output of link prediction algorithms just contains a list of scores or probabilities for all predicted missing links, which is difficult to interpret, and these results can be inaccurate.
權(quán)利要求
What is claimed is:1. A method, comprising:for data represented as a bipartite network and for a set of missing links in the bipartite network:calculating a weight for each of the missing links in the set based on bicliques of the bipartite network; executing a link prediction algorithm configured to incorporate the weight for each of the missing links; and providing ones from the set of missing links selected by the link prediction algorithm as predicted missing links of the bipartite network, wherein the providing by the link prediction algorithm comprises providing a score for each predicted missing link that indicates a probability that a respective predicted missing links exists between respective nodes, and presenting the bipartite network as a bi-adjacency matrix comprising rows that represent a first type of node in the bipartite network, and columns that represent rows of a second type of node in the bipartite network, each of the entries in the matrix representing one link between a node of the first type and a node of the second type, wherein the providing ones from the set of missing links selected by the link prediction algorithm as the predicted missing links of the bipartite network comprises providing, by the link prediction algorithm, a score for each of the predicted missing links that indicates a probability that a respective predicted missing link exists between a respective node of the first type and a respective node of the second type, and wherein the presenting the bipartite network comprises providing an interface configured to represent each of the entries in the matrix as a color hue according to the score, and configured to order the rows and columns of the bi-adjacency matrix according to a selected criteria. 2. The method of claim 1 , wherein the calculating the weight for the each of the missing links in the set based on bicliques of the bipartite network comprises:for each pair of bicliques having a score based on a number of overlapping nodes meeting a threshold and a size of the each pair of bicliques, calculating the weight for ones of the set of missing links between the pair of bicliques. 3. The method of claim 2 , wherein the calculating the weight for ones of the set of missing links between the each pair of bicliques is based on the number of overlapping nodes and the size of the each pair of bicliques. 4. The method of claim 1 , wherein providing the ones from the set of missing links selected by the link prediction algorithm as the predicted missing links of the bipartite network comprises presenting the predicted missing links linearly by probability. 5. A non-transitory computer readable medium, storing instructions for executing a process, the instructions comprising:for data represented as a bipartite network and for a set of missing links in the bipartite network:calculating a weight for each of the missing links in the set based on bicliques of the bipartite network; executing a link prediction algorithm configured to incorporate the weight for each of the missing links; and providing ones from the set of missing links selected by the link prediction algorithm as predicted missing links of the bipartite network, wherein the providing by the link prediction algorithm comprises providing a score for each predicted missing link that indicates a probability that a respective predicted missing links exists between respective nodes, and presenting the bipartite network as a bi-adjacency matrix comprising rows that represent a first type of node in the bipartite network, and columns that represent rows of a second type of node in the bipartite network, each of the entries in the matrix representing one link between a node of the first type and a node of the second type, wherein the providing ones from the set of missing links selected by the link prediction algorithm as the predicted missing links of the bipartite network comprises providing, by the link prediction algorithm, a score for each of the predicted missing links that indicates a probability that a respective predicted missing link exists between a respective node of the first type and a respective node of the second type, and wherein the presenting the bipartite network comprises providing an interface configured to represent each of the entries in the matrix as a color hue according to the score, and configured to order the rows and columns of the bi-adjacency matrix according to a selected criteria. 6. The non-transitory computer readable medium of claim 5 , wherein the calculating the weight for the each of the missing links in the set based on bicliques of the bipartite network comprises:for each pair of bicliques having a score based on a number of overlapping nodes meeting a threshold and a size of the each pair of bicliques, calculating the weight for ones of the set of missing links between the pair of bicliques. 7. The non-transitory computer readable medium of claim 6 , wherein the calculating the weight for ones of the set of missing links between the each pair of bicliques is based on the number of overlapping nodes and the size of the each pair of bicliques. 8. The non-transitory computer readable medium of claim 5 , wherein providing the ones from the set of missing links selected by the link prediction algorithm as the predicted missing links of the bipartite network comprises presenting the predicted missing links linearly by probability. 9. An apparatus, comprising:a processor configured to:for data represented as a bipartite network and for a set of missing links in the bipartite network:calculate a weight for each of the missing links in the set based on bicliques of the bipartite network; execute a link prediction algorithm configured to incorporate the weight for each of the missing links; and provide ones from the set of missing links selected by the link prediction algorithm as predicted missing links of the bipartite network, wherein the providing by the link prediction algorithm comprises providing a score for each predicted missing link that indicates a probability that a respective predicted missing links exists between respective nodes, and present the bipartite network as a bi-adjacency matrix comprising rows that represent a first type of node in the bipartite network, and columns that represent rows of a second type of node in the bipartite network, each of the entries in the matrix representing one link between a node of the first type and a node of the second type, wherein the providing ones from the set of missing links selected by the link prediction algorithm as the predicted missing links of the bipartite network comprises providing, by the link prediction algorithm, a score for each of the predicted missing links that indicates a probability that a respective predicted missing link exists between a respective node of the first type and a respective node of the second type, and wherein the presenting the bipartite network comprises providing an interface configured to represent each of the entries in the matrix as a color hue according to the score, and configured to order the rows and columns of the bi-adjacency matrix according to a selected criteria. 10. The apparatus of claim 9 , wherein the processor is configured to calculate the weight for the each of the missing links in the set based on bicliques of the bipartite network by:for each pair of bicliques having a score based on a number of overlapping nodes meeting a threshold and a size of the each pair of bicliques, calculating the weight for ones of the set of missing links between the pair of bicliques. 11. The apparatus of claim 10 , wherein the processor is configured to calculate the weight for ones of the set of missing links between the each pair of bicliques is based on the number of overlapping nodes and the size of the each pair of bicliques. 12. The apparatus of claim 9 , wherein the processor is configured to provide the ones from the set of missing links selected by the link prediction algorithm as the predicted missing links of the bipartite network by presenting the predicted missing links linearly by probability. 13. The apparatus of claim 9 , wherein the processor is configured to:in response to a selection on an interface of one of the predicted missing links:conducting at least one of motif analysis or metric analysis on the selected one of the predicted missing links through adding the one of the predicted missing links in the bipartite network; and providing a result of the at least one of the motif analysis or the metric analysis for the selected one of the predicted missing links. 14. The method of claim 1 , wherein each of the entries in the matrix for the missing links are represented according to a first color scale, wherein the interface is configured to represent existing links between a node of the first type and a node of the second type according a second color scale different than the first color scale. 15. The method of claim 1 , wherein the interface is configured to reorder the rows and columns of the bi-adjacency matrix according to selected criteria, each selected criteria corresponding to a different order, wherein the selected criteria comprises one of a node label, an average prediction score, and a total number of missing links. 16. The method of claim 1 , wherein the interface is configured to add a missing link as an entity in response to an input identifying an entity of the matrix.
意見反饋
使用該功能遇到問題