System and method for ranking search engine results

Patent metadata

Field	Value
Publication number	US-10642905-B2
Application number	US-201615368925-A
Country	US
Kind code	B2
Filing date	Dec 5, 2016
Priority date	Dec 28, 2015
Publication date	May 5, 2020
Grant date	May 5, 2020

Abstract

There are discloses methods and systems for generating a search engine results page (SERP). The method is executable at a server executing a search engine, the server being accessible via a communication network by at least one electronic device. The method comprises, as part of generating a search result list, the search result list containing a first search result and a second search result, predicting a first interest parameter for the first search result; predicting a second interest parameter for the second search result; predicting a usefulness parameter for the first search result, the predicting being at least partially based on the first interest parameter and the second interest parameter; adjusting a position of the first search result within the ranked search result list based on the predicted usefulness parameter, the adjusting resulting in the first search result being at an adjusted position within the ranked search result list.

First claim

The invention claimed is:

1. A method of generating a search engine results page (SERP), the method executable at a server executing a search engine, the server being accessible via a communication network by at least one electronic device, the method comprising:
receiving, from the at least one electronic device, a search query;
generating a search result list, the search result list containing a first search result and a second search result, both of which are responsive to the search query;
ranking the search result list such that as a result of the ranking, the first search result and the second search result are at a first SERP position and a second SERP position, respectively, the first SERP position and the second SERP position being adjacent therebetween, the ranking resulting in a ranked search result list;
predicting a first interest parameter for the first search result, the predicting based on a first machine learning algorithm;
the first interest parameter being indicative of a likelihood of a user clicking on the first search result;
the first machine learning algorithm having been trained to predict the likelihood of the user clicking on a given search result positioned on a given position of a given SERP,
the first machine learning algorithm having been trained based on a first training set of data;

predicting a second interest parameter for the first search result, the predicting based on a second machine learning algorithm, the second machine learning algorithm being different from the first machine learning algorithm;
the second interest parameter being indicative of a likelihood of the user clicking not on the first search result and clicking on the second search result instead;
the second machine learning algorithm having been trained to predict a likelihood of the user not clicking on the given search result and clicking on an other search result instead, the other search result being positioned lower than the given search result, and
the second machine learning algorithm having been trained based on a second training set of data different from the first training set of data;

the first machine learning algorithm and the second machine learning algorithm having been trained independently and separately from each other;

predicting a usefulness parameter for the first search result, the predicting being based on a third machine learning algorithm, the third machine learning algorithm being different from the first machine learning algorithm and the second machine learning algorithm, the predicting being at least partially based on the first interest parameter and the second interest parameter;
adjusting the position of the first search result within the ranked search result list based on the predicted usefulness parameter, the adjusting resulting in the first search result being at an adjusted position within the ranked search result list;
generating the SERP including the first search result and the second search result, the first search result being positioned at the adjusted position of the SERP.

2. The method of claim 1 , wherein the first search result is a vertical search result.

3. The method of claim 1 , wherein the second search result is a web search result.

4. The method of claim 1 , wherein the first machine learning algorithm and the second machine learning algorithm have been trained independently therebetween.

5. The method of claim 4 , wherein the third machine learning algorithm has been trained using parameters predicted by the first machine learning algorithm and the second machine learning algorithm.

6. The method of claim 5 , wherein parameters predicted by the first machine learning algorithm and the second machine learning algorithm include a first factor indicative of a click-through rate of a given search result and a second factor indicative of a click-through rate of another search result following, in a ranked order, the given search result.

7. The method of claim 6 , wherein the third machine algorithm is based at least in part on a modified “win-loss” algorithm.

8. The method of claim 1 , wherein the adjusting the position of the first search result further comprises an additional adjusting the position of the second search result in line with the adjusted position of the first search result.

9. The method of claim 1 , wherein the search result list further comprises a third search result ranked at a third SERP position, the third SERP position being adjacent to the second SERP position, the method further comprises:
predicting a third interest parameter for the first search result for the second SERP position, the predicting based on the first machine learning algorithm;
the third interest parameter being indicative of a likelihood of a user clicking on the first search result;

predicting a fourth interest parameter for the first search result, the predicting based on the second machine learning algorithm, second machine learning algorithms being different from the first machine learning algorithm;
the fourth interest parameter being indicative of a likelihood of the user clicking not on the first search result and clicking on the third search result instead; and wherein

the predicting the usefulness parameter for the first search result, further comprises predicting a second usefulness parameter for the first search result, the predicting being based on the third machine learning algorithm, the predicting being at least partially based on the third interest parameter and the fourth interest parameter; and wherein
the adjusting the position of the first search result further comprises selecting one of the first SERP position and the second SERP position for placing the first search result based on comparison of the usefulness parameter and the second usefulness parameter.

10. The method of claim 9 , wherein the predicting the first usefulness parameter and the predicting the second usefulness parameter are executed substantially simultaneously.

11. The method of claim 10 , wherein each of the first interest parameter and the second interest parameter is a respective predicted click-through rate.

12. The method of claim 1 , wherein the first machine learning algorithm and the second machine learning algorithm have been trained independently based on different training sets of data.

13. A method of generating a search engine results page (SERP), the method executable at a server executing a search engine, the server being accessible via a communication network by at least one electronic device, the method comprising:
receiving, from the at least one electronic device, a search query;
generating a search result list, the search result list containing a first search result, a second search result and a third search result, the first search result being a vertical search result and the second search result and the third search result being web search results, all of the first, second and third search results being responsive to the search query;
ranking the second search result and the third search result in a ranked order such that the second search result is at a first ranked position and the third search result is in a second ranked position; the first ranked position and the second ranked position being adjacent therebetween;
for each of the first ranked position and the second ranked position, determining a respective usefulness parameter for the first search result, the determining being executed by:
predicting a first interest parameter for the first search result at a respective one of each of the first ranked position and the second ranked position, the predicting based on a fir

Assignees

• Yandex Europe Ag

Inventors

• Gusev Gleb Gennadievich
• Nikulin Vadim Andreevich
• Ustinovskiy Yury Mikhailovich

Classifications

• G06F16/24578 Primary

  using ranking · CPC title

• G06F16/951 Primary

  Indexing; Web crawling techniques · CPC title

• G06F16/00

  Information retrieval; Database structures therefor; File system structures therefor · CPC title

• G06N20/00 Primary

  Machine learning · CPC title