Mitigating topic diffusion
View Patent ↗Determining the relevance of a web node is disclosed. A seed score value of a first type is assigned to a seed set of nodes. A score value of a second type is derived for the web node based on a mapping of a reachability relationship between one or more seed nodes and the web node. A score value of the first type is derived for the web node based on a mapping of a reachability relationship between the web node and one or more evaluation nodes having derived weight values of the second type. Topic diffusion is mitigated.
1. A method of determining the relevance of a web node comprising:
assigning one or more source score values to one or more seed nodes included in a seed set;
deriving a destination score value for the web node based on a mapping of a reachability relationship between one or more seed nodes and the web node, wherein the reachability relationship is associated with a random surfer model;
deriving a source score value for the web node based on a mapping of a reachability relationship between the web node and one or more evaluation nodes and based on a derived destination score associated, respectively, with each of the one or more evaluation nodes; and
mitigating a topic diffusion comprising at least one of a topic drift and a topic generalization, wherein a topic drift occurs when the seed set is contaminated by nodes that are not relevant to a topic associated with the seed set, and wherein a topic generalization occurs when a set of evaluation nodes is recursively enlarged;
wherein the web node, the seed nodes, and the one or more evaluation nodes are documents included in a document collection.
2. The method of claim 1 wherein mitigating topic diffusion further includes removing at least one universal source from the seed set.
3. The method of claim 1 wherein mitigating topic diffusion further includes removing at least one universal destination from the set of evaluation nodes.
4. The method of claim 1 wherein mitigating topic diffusion includes mitigating topic drift.
5. The method of claim 1 wherein mitigating topic diffusion includes mitigating topic generalization.
6. The method of claim 1 wherein mitigating topic diffusion includes preventing at least one web node that is only marginally relevant to a topic from entering a seed set of nodes.
7. The method of claim 1 wherein mitigating topic diffusion includes using a source score threshold.
8. The method of claim 1 wherein mitigating topic diffusion includes using a destination score threshold.
9. The method of claim 1 wherein mitigating topic diffusion includes pruning based at least in part on selectivity.
10. The method of claim 9 wherein pruning is based at least in part on a parameter γ.
11. The method of claim 9 wherein pruning is based at least in part on a parameter δ.
12. The method of claim 1 wherein mitigating topic diffusion includes pruning based at least in part on relative selectivity.
13. The method of claim 12 wherein pruning is based at least in part on a parameter γ′.
14. The method of claim 12 wherein pruning is based at least in part on a parameters δ′.
15. A system for determining the relevance of a web node comprising:
a processor configured to:
assign one or more source score values to one or more seed nodes included in a seed set of nodes;
derive a destination score value for the web node based on a mapping of a reachability relationship between one or more seed nodes and the web node, wherein the reachability relationship is associated with a random surfer model;
derive a source score value for the web node based on a mapping of a reachability relationship between the web node and one or more evaluation nodes and based on a derived destination score associated, respectively, with each of the one or more evaluation nodes; and
mitigate a topic diffusion comprising at least one of a topic drift and a topic generalization, wherein a topic drift occurs when the seed set is contaminated by nodes that are not relevant to a topic associated with the seed set, and wherein a topic generalization occurs when a set of evaluation nodes is recursively enlarged; and
a memory coupled with the processor, wherein the memory provides the processor with instructions;
wherein the web node, the seed nodes, and the one or more evaluation nodes are documents included in a document collection.
16. The system of claim 15 wherein the processor is configured to mitigate topic diffusion including by removing at least one universal source from the seed set.
17. The system of claim 15 wherein the processor is configured to mitigate topic diffusion including by removing at least one universal destination from the set of evaluation nodes.
18. The system of claim 15 wherein the processor is configured to mitigate topic diffusion including by mitigating topic drift.
19. The system of claim 15 wherein the processor is configured to mitigate topic diffusion including by mitigating topic generalization.
20. A computer program product for determining the relevance of a web node, the computer program product being embodied in a computer readable storage medium and comprising computer instructions for:
assigning one or more source score values to one or more seed nodes included in a seed set;
deriving a destination score value for the web node based on a mapping of a reachability relationship between one or more seed nodes and the web node, wherein the reachability relationship is associated with a random surfer model;
deriving a source score value for the web node based on a mapping of a reachability relationship between the web node and one or more evaluation nodes and based on a derived destination score associated, respectively, with each of the one or more evaluation nodes; and
mitigating a topic diffusion comprising at least one of a topic drift and a topic generalization, wherein a topic drift occurs when the seed set is contaminated by nodes that are not relevant to a topic associated with the seed set, and wherein a topic generalization occurs when a set of evaluation nodes is recursively enlarged;
wherein the web node, the seed nodes, and the one or more evaluation nodes are documents included in a document collection.
21. A method of determining the relevance of a web node comprising:
assigning one or more source score values to one or more seed nodes included in a seed set;
deriving a destination score value for the web node based on a mapping of a reachability relationship between one or more seed nodes and the web node, wherein the reachability relationship is associated with a random surfer model;
deriving a source score value for the web node based on a mapping of a reachability relationship between the web node and one or more evaluation nodes and based on a derived destination score associated, respectively, with each of the one or more evaluation nodes; and
mitigating topic diffusion, at least in part by excluding either one universal source or one universal destination from the seed set;
wherein the web node, the seed nodes, and the one or more evaluation nodes are documents included in a document collection.