Back to results
US 8,046,360 B2Granted
Reduction of annotations to extract structured web data
AssigneeYahoo! Inc.
InventorMahesh Tiyyagura
Life of this patent
Estimating the term…
Loading patent…
Loading patent…
AssigneeYahoo! Inc.
InventorMahesh Tiyyagura
Estimating the term…
Document, such as web pages of a domain, are annotated to facilitate extracting structured information from the documents. The documents are clustered. Each cluster is such that the documents within that cluster are similar to each other at least with respect to a first threshold, such as according to a shingling metric, where the first threshold is an 8/8 shingling match. There is at least one overlap cluster, each overlap cluster including at least one of the plurality of clusters such that documents of the at least one cluster included in that overlap cluster are similar to each other at least with respect to a second threshold that is lower than the first threshold. A particular overlap cluster is designated, as is a particular cluster of the particular overlap cluster. For the particular designated cluster, an obtained annotation is transferred to other clusters included in the designated particular overlap cluster.
A method of annotating documents of a corpus to facilitate extracting structured information from the corpus, comprising:
A computer program product for annotating documents of a corpus to facilitate extracting structured information from the corpus, the computer program product comprising at least one computer-readable medium having computer program instructions stored therein which are operable to cause at least one computing device to:
A computing system including at least one computing device, configured to annotate documents of a corpus to facilitate extracting structured information from the corpus, the at least one computing device configured to:
Dependencies are inferred from the claim text.