Click through rate prediction system and method
A computer implemented method comprises analyzing a plurality of attributes of a sample of online documents using a boosted decision tree and generating a model from it. The model is used to predict a click through rate (CTR) of an additional online document based on the analyzing. The predicted CTR is output to a display device, storage medium or network.
1. A computer implemented method comprising:
analyzing a plurality of attributes of a sample of online documents using a boosted decision tree and generating a machine learning model therefrom;
using the machine learning model to predict a click through rate (CTR) of an additional online document based on the analyzing;
outputting the predicted CTR to a display device, storage medium or network,
wherein the plurality of attributes include a CTR of other documents having titles that are the same as or similar to a title of the additional online document;
determining similarity of any two documents by cardinality of a difference set that includes non-overlapping terms included in the title of one of the two documents but not included in the title of the other of the two documents;
using the machine learning model to predict respective CTR of a plurality of additional online documents based on the analyzing; and
ranking the plurality of additional online documents by using the respective predicted CTR of each of the plurality of additional online documents as an input factor.
2. The method of claim 1 , further comprising:
rendering the additional online documents or portions thereof for display in order of their respective rankings.
3. The method of claim 1 , wherein the additional online document is one of the group consisting of an advertisement, a response to a search query, a photograph, a movie, an image, an audio clip, a video clip and a commercial document product.
4. The method of claim 1 , wherein the plurality of attributes include at least one of the group consisting of CTR of other documents by the same author as the additional online document, CTR of other documents published by the same publisher of the additional online document, and CTR of other documents having the same type as the additional online document.
5. The method of claim 1 , wherein the plurality of attributes includes a measure indicating an amount of spam feedback received relating to the document.
6. A system comprising:
a machine readable storage medium storing a sample of online documents; and
a processor configured to analyze a plurality of attributes of the sample of online documents using a boosted decision tree and generating a model therefrom;
the processor configured to use the model to predict a click through rate (CTR) of an additional online document based on the analyzing;
the processor configured to output the predicted CTR to a display device, storage medium or network,
wherein the plurality of attributes include a CTR of other documents having titles that are the same as or similar to a title of the additional online document;
the processor configured to determine similarity of any two documents by cardinality of a difference set that includes non-overlapping terms included in the title of one of the two documents but not included in the title of the other of the two documents;
the processor configured to use the model to predict respective CTR of a plurality of additional online documents based on the analyzing; and
the processor configured to rank the plurality of additional online documents by using the respective predicted CTR of each of the plurality of additional online documents as an input factor.
7. The system of claim 6 , wherein:
the processor is configured to render the additional online documents or portions thereof for display in order of their respective rankings.
8. The system of claim 6 , wherein the plurality of attributes include at least one of the group consisting of CTR of other documents by the same author as the additional online document, CTR of other documents published by the same publisher of the additional online document, and CTR of other documents having the same type as the additional online document.
9. A machine readable storage medium encoded with computer program code, wherein when the computer program code is executed by a processor, the processor performs a machine implemented method comprising the steps of:
analyzing a plurality of attributes of a sample of online documents using a boosted decision tree and generating a model therefrom;
using the model to predict a click through rate (CTR) of an additional online document based on the analyzing;
outputting the predicted CTR to a display device, storage medium or network,
wherein the plurality of attributes include a CTR of other documents having titles that are the same as or similar to a title of the additional online document;
determining similarity of any two documents by cardinality of a difference set that includes non-overlapping terms included in the title of one of the two documents but not included in the title of the other of the two documents;
using the model to predict respective CTR of a plurality of additional online documents based on the analyzing; and
ranking the plurality of additional online documents by using the respective predicted CTR of each of the plurality of additional online documents as an input factor.
10. The machine readable storage medium of claim 9 , further comprising:
rendering the additional online documents or portions thereof for display in order of their respective rankings.
11. The machine readable storage medium of claim 9 , wherein the plurality of attributes include at least one of the group consisting of CTR of other documents by the same author as the additional online document, CTR of other documents published by the same publisher of the additional online document, and CTR of other documents having the same type as the additional online document.