Organization system for ad campaigns

• US 8,180,760 B1
• Filed: 12/20/2007
• Issued: 05/15/2012
• Est. Priority Date: 12/20/2007
• Status: Expired due to Fees

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Carver, Anton, Ashton, Will, Jakobi, Nick
• Primary Examiner(s): Trujillo, James
• Assistant Examiner(s): Singh, Amresh
• Application Number:
  US11/962,021
• Time in Patent Office: 1,608 Days
• Field of Search: 707/3, 707/708, 707/6
• US Class Current: 707/708
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06Q 30/0241 Advertisements

Abstract

An automatic account organization tool is provided to organize a large adgroup into smaller adgroups with semantically meaningful names. For example, a set of input keywords is received, semantically related pairs of keywords are identified from the set of input keywords, and hierarchical clustering is applied to the pairs of keywords to identify a set of clusters of keywords, each cluster having semantically related keywords. A name can be determined for each of the clusters.

Claims

1. A computer-implemented method comprising:

receiving targeting keywords for an advertisement campaign, each of the targeting keywords being a keyword with which advertisements in the advertisement campaign is selectively distributed;

assigning, by a computer, each targeting keyword to one or more topic clusters, each targeting keyword being assigned to the one or more topic clusters based on a measure of semantic relatedness between the targeting keyword and a topic to which the topic cluster has been determined relevant, wherein at least some of the targeting keywords are assigned to multiple topic clusters;

for each topic cluster, determining, by a computer and for all possible pairs of targeting keywords for the topic cluster, a semantic distance between targeting keywords in the pair, the semantic distance being determined based at least in part on a measure of similarity between cluster vectors for the targeting keywords in the pair, the cluster vector for each of the targeting keywords in the pair specifying topic clusters to which the targeting keyword was assigned;

identifying, by a computer and based on the semantic distances, semantically related keyword pairs, each semantically related keyword pair being a keyword pair for which the semantic distance is less than a specified semantic distance threshold;

clustering, by a computer, the identified semantically related keyword pairs into a plurality of sets of pair clusters, the clustering being exclusive of the pairs of targeting keywords for which the semantic distance is greater than the specified semantic distance threshold, the clustering being performed according to a hierarchical clustering technique; and

creating, by a computer and based on the clustering, new ad groups for the advertisement campaign, each ad group specifying at least one advertisement that is selectively distributed using a set of the pair clusters.

2. The method of claim 1, further comprising removing a second instance of a particular keyword pair from at least one of the topic clusters, the second instance of the particular keyword pair being removed from a different topic cluster than the topic cluster in which a first instance of the particular keyword was identified.

3. The method of claim 1, wherein assigning each targeting keyword to one or more topic clusters comprises assigning each targeting keyword to one or more topic clusters for predetermined classifications that are specified in a semantic database.

4. The method of claim 3, wherein assigning each targeting keyword to one or more topic clusters for predetermined classifications specified in a semantic database comprises, for each of the targeting keywords, identifying one or more classifications in the semantic database that are associated with the keyword.

5. The method of claim 1, wherein clustering the semantically related keyword pairs comprises applying single link hierarchical clustering to the semantically related keyword pairs.

6. The method of claim 5, wherein applying the single link hierarchical clustering comprises joining a targeting keyword to a cluster if the targeting keyword and another targeting keyword in the cluster belong to a same keyword pair.

7. The method of claim 6, wherein applying the single link hierarchical clustering comprises joining the targeting keyword to the cluster if the semantic distance between the targeting keyword and the other targeting keyword is less than a threshold.

8. The method of claim 5, wherein clustering the semantically related keyword pairs comprises, for each topic cluster, applying complete link hierarchical clustering to the targeting keywords in the topic cluster to generate one or more sets of clustered semantically related keywords.

9. The method of claim 8, wherein applying the complete link hierarchical clustering to the targeting keywords in the topic cluster comprises determining a semantic distance for every keyword pair in the topic cluster.

10. The method of claim 9, wherein applying the complete link hierarchical clustering comprises removing a targeting keyword from a-topic cluster if the semantic distance between the targeting keyword and any other targeting keyword in the topic cluster is above a threshold.

11. The method of claim 8, wherein applying the complete link hierarchical clustering to the topic clusters is performed in parallel.

12. The method of claim 1, further comprising:

sorting the keyword pairs according to the semantic distances prior to the clustering.

13. The method of claim 1, further comprising selecting a name for each of the pair clusters, the name for each pair cluster being selected based on one or more topics to which the pair cluster has been identified as relevant.

14. An apparatus comprising:

a database storing a plurality of keywords for an advertisement campaign, each keyword specifying a phrase with which an advertisement in the advertisement campaign is selectively distributed;

one or more computers configured to interact with the database, and further being configured to perform operations comprising;

assigning each targeting keyword to one or more topic clusters based on a measure of semantic relatedness between the targeting keyword and the one or more topic clusters, wherein at least some of the targeting keywords are assigned to multiple topic clusters;

for each topic cluster, determining, for all possible pairs of targeting keywords for the topic cluster, a semantic distance between targeting keywords in the pair, the semantic distance being determined based at least in part on a measure of similarity between cluster vectors for the targeting keywords in the pair, the cluster vector for each of the targeting keywords in the pair specifying topic clusters to which the targeting keyword was assigned;

identifying, based on the semantic distances, semantically related pairs keyword pairs, each semantically related keyword pair being keyword pair for which the semantic distance is less than a specified semantic distance threshold;

clustering the identified semantically related keyword pairs into a plurality of sets of pair clusters, the clustering being exclusive of the pairs of targeting keywords for which the semantic distance is greater than the specified semantic distance threshold, the clustering being performed according to a hierarchical clustering technique; and

creating, based on the clustering, new ad groups for the advertisement campaign, each ad group specifying at least one advertisement that is selectively distributed using a set of pair clusters.

15. The apparatus of claim 14, wherein the one or more computers are further configured to query the database for information about which common classifications are associated with a given keyword, and for scores indicating strengths of associations between the keyword and the common classifications.

16. The apparatus of claim 14, wherein the one or more computers are further configured to remove keyword pairs for which the semantic distance is greater than a threshold.

17. The apparatus of claim 14, wherein the one or more computers are further configured to sort the keyword pairs according to semantic distances of the keyword pairs.

18. The apparatus of claim 14, wherein the one or more computers are further configured to cluster the keywords based on single link hierarchical agglomerative clustering.

19. The apparatus of claim 14, wherein the one or more computers are further configured to cluster the keywords based on complete link hierarchical agglomerative clustering.

20. The apparatus of claim 14, wherein the one or more computers are further configured to generate a name for each cluster of keywords.

21. An apparatus comprising:

one or more computers including;

means for storing a plurality of keywords for an advertisement campaign, each keyword specifying a phrase with which an advertisement in the advertisement campaign is selectively distributed;

means for assigning each targeting keyword to one or more topic clusters, each targeting keyword being assigned to the one or more topic clusters based on a measure of semantic relatedness between the targeting keyword and a topic to which the topic cluster has been determined relevant, wherein at least some of the targeting keywords are assigned to multiple topic clusters;

means for determining, for all possible pairs of targeting keywords for the topic cluster, a semantic distance between targeting keywords in the pair, the semantic distance being determined based at least in part on a measure of similarity between cluster vectors for the targeting keywords in the pair, the cluster vector for each of the targeting keywords in the pair specifying topic clusters to which the targeting keyword was assigned;

means for identifying, based on the semantic distances, semantically related keyword pairs, each semantically related keyword pair being a keyword pair for which the semantic distance is less than a specified semantic distance threshold;

means for clustering the identified semantically related keyword pairs into a plurality of sets of pair clusters, the clustering being exclusive of the pairs of targeting keywords for which the semantic distance is greater than the specified semantic distance threshold, the clustering being performed according to a hierarchical clustering technique; and

means for creating, based on the clustering, new ad groups for the advertisement campaign, each ad group specifying at least one advertisement that is selectively distributed using a set of the pair clusters.