Advertisement placement method and system using semantic analysis
First Claim
1. A machine-readable non-transitory medium having instructions stored thereon, where the instructions, when read by the machine, cause the machine to perform the steps of:
- accessing a semantic representation associated with a first dataset and a semantic representation associated with a second dataset, wherein at least one of the semantic representation associated with the first dataset and the semantic representation associated with the second dataset is a trainable semantic vector generated based on at least one data point included in a respective dataset and known relationships between predetermined data points and predetermined categories to which the predetermined data points may relate;
determining a similarity between the semantic representation associated with the first dataset and the semantic representation associated with the second dataset; and
selectively relating the first dataset to the second dataset based on a result of the determining step;
wherein each attribute in a trainable semantic vector indicates how likely a dataset represented by the trainable semantic vector belongs to one of the predetermined categories, andthe trainable semantic vector has a dimension equal to the number of the predetermined categories.
2 Assignments
0 Petitions
Accused Products
Abstract
An advertisement placement method and system for relating an advertisement to a dataset based on a trainable semantic vector (TSV) associated with the dataset and respective semantic representations of the advertisements. The trainable semantic vector associated with the dataset is generated based on at least one data point included in the dataset and known relationships between predetermined data points and predetermined categories. A comparison process is performed to determine a similarity between the trainable semantic vector associated with the dataset and the semantic representation associated with each of the plurality of advertisements. The system selectively relates one or more of the advertisements with the dataset based on a result of the comparison process. The selected advertisement or advertisements may be displayed with the dataset.
-
Citations
22 Claims
-
1. A machine-readable non-transitory medium having instructions stored thereon, where the instructions, when read by the machine, cause the machine to perform the steps of:
-
accessing a semantic representation associated with a first dataset and a semantic representation associated with a second dataset, wherein at least one of the semantic representation associated with the first dataset and the semantic representation associated with the second dataset is a trainable semantic vector generated based on at least one data point included in a respective dataset and known relationships between predetermined data points and predetermined categories to which the predetermined data points may relate; determining a similarity between the semantic representation associated with the first dataset and the semantic representation associated with the second dataset; and selectively relating the first dataset to the second dataset based on a result of the determining step; wherein each attribute in a trainable semantic vector indicates how likely a dataset represented by the trainable semantic vector belongs to one of the predetermined categories, and the trainable semantic vector has a dimension equal to the number of the predetermined categories. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
20. An advertisement placement system for relating one of a plurality of advertisements to a dataset, the system comprising:
-
a data processor configured to process data; and a data storage system configured to store instructions which, upon execution by the data processor, control the data processor to perform the steps of; accessing a semantic representation associated with a first dataset and a semantic representation associated with a second dataset, wherein at least one of the semantic representation associated with the first dataset and the semantic representation associated with the second dataset is a trainable semantic vector generated based on at least one data point included in a respective dataset and known relationships between predetermined data points and predetermined categories to which the predetermined data points may relate; determining a similarity between the semantic representation associated with the first dataset and the semantic representation associated with the second dataset; and selectively relating the first dataset to the second dataset based on a result of the determining step; wherein each attribute in a trainable semantic vector indicates how likely a dataset represented by the trainable semantic vector belongs to one of the predetermined categories, and the trainable semantic vector has a dimension equal to the number of the predetermined categories. - View Dependent Claims (21)
-
-
22. A method implemented on a computer having a storage, a processor, and a communication platform connected to a network for ascertaining the relatedness between a first dataset to a second dataset, the method comprising the steps of:
-
accessing, from the storage of the computer, a first semantic representation associated with the first dataset and a second semantic representation associated with the second dataset; obtaining, by the processor, a similarity between the first dataset and the second dataset based on the first and second semantic representations, where the similarity is assessed in terms of how each of the first and second datasets relates to a plurality of predetermined categories; and determining, by the processor, how the first dataset is related to the second dataset based on the similarity obtained, wherein at least one of the semantic representations associated with the first and second datasets is a trainable semantic vector, each attribute in the trainable semantic vector indicates how likely a dataset represented by the trainable semantic vector belongs to one of a plurality of predetermined categories, and the trainable semantic vector has a dimension equal to the number of the predetermined categories.
-
Specification