METHOD AND SYSTEM FOR GENERATION, ADJUSTMENT AND UTILIZATION OF WEB PAGES SELECTION RULES

• US 20110196821A1
• Filed: 12/30/2010
• Published: 08/11/2011
• Est. Priority Date: 02/10/2010
• Status: Active Grant

• Current Assignee: DSNR Media Group Limited
• Original Assignee: DSNR Media Group Limited
• Inventors: Peles, Tsafrir, Levin, Boris, Scalosub, Daniel Reuven, HELLERMAN, Gilad
• Granted Patent: US 8,601,093 B2
• Time in Patent Office: Days
• US Class Current: 707/600
• CPC Class Codes: G06F 16/958 Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking

Abstract

According to some embodiments of the present invention there is provided a system of rule generation for web pages selection mechanism comprising a pattern recognition unit which analyzes a plurality of network requests and identifies a plurality of repeated patterns of characterizing features therein, a calculation unit which computes, for each said repeated pattern, a plurality of statistic features, and a rule construction unit which generates a plurality of web pages selection rules for at least some of said repeated patterns, wherein each of the web pages selection rules comprises a rule condition defining at least one criterion for matching it with characterizing features associated with at least one additional network request and a target web pages selection distribution set according to respective statistic features.

Claims

1. A rule generator for web pages selection mechanism comprising:

a pattern recognition unit which analyzes a plurality of network requests and identifies a plurality of repeated patterns of characterizing features therein;

a calculation unit which computes, for each said repeated pattern, a plurality of statistic features; and

a rule constructing unit which defines a plurality of web pages selection rules for at least some of said plurality of repeated patterns, each said at least some web pages selection rules comprises a rule condition defining at least one criterion for matching with characterizing features associated with at least one additional network request and a target web pages selection distribution set according to respective said statistic features.

2. The rule generator of claim 1, wherein each network request is associated with a plurality of indications comprising, a selected web page indication specifying a web page that been selected responsive to said each network request, and a desired user action indication specifying occurrence of a desired user action responsive to provision of said web page.

3. The rule generator of claim 1, wherein the at least one criterion is based on one of the plurality of repeated patterns.

4. The rule generator of claim 2, further comprising a database hosting the web pages selection rules and a plurality of data records, each said data record comprises a plurality of characterizing features identified for the each network request and the plurality of indications associated with it.

5. The rule generator of claim 1, wherein the plurality of statistic features comprise a plurality of probabilities, each said probability is a probability of a rate of occurrence of desired user actions responsive to selection of a specific web page.

6. The rule generator of claim 5, wherein the target web pages selection distribution is defined based on the plurality of probabilities.

7. The rule generator of claim 1, wherein the statistic features comprise a plurality of predicted rates of occurrence of desired user actions computed for the plurality of web pages selection rules.

8. The rule generator of claim 7, wherein the rule constructing unit sorts the plurality of web pages selection rules according to their predicted rate of occurrence of desired user actions.

9. The rule generator of claim 1, wherein the rule constructing unit validates the web pages selection rules based on a predetermined rule approval policy and removes non-valid web pages selection rules.

10. The rule generator of claim 7, wherein the rule constructing unit removes a new web pages selection rule which rule condition includes or covers a rule condition of another web pages selection rule, if the predicted rate of occurrence of desired user actions of said new web pages selection rule is smaller than a rule maintenance threshold.

11. A web pages selection system, comprising:

the rule generator of claim 1; and

a redirection unit which receives the at least one additional network request, identifies a plurality of characterizing features associated with it, matches said plurality of characterizing features with at least one of the plurality of web pages selection rules, selects a web page from a plurality of web pages based on target web pages selection distributions of said at least one of the plurality of web pages selection rules, and redirects a user terminal from which said at least one additional network request been received to said selected web page.

12. The system of claim 11, further comprising a network interface which receives the plurality of network requests from a plurality of user terminals over a computer network.

13. The system of claim 11, wherein the plurality of web pages comprise campaign landing web pages.

14. The system of claim 11, further comprising a database hosting the plurality of web pages selection rules and a plurality of data records, each said data record comprises the plurality of characterizing features of one of the plurality of network requests and a plurality of indications associated with it,

wherein the redirection unit updates said database with the plurality of characterizing features of the additional network request and with its plurality of indications.

15. The system of claim 14, wherein the calculation unit routinely or periodically computes for the web pages selection rules a plurality of effectiveness evaluations reflecting the increase in desired user actions occurred since said web pages selection rules being used.

16. The system of claim 15, wherein the rule generator issues alerts whenever one of the plurality of effectiveness evaluations is of negative value.

17. The system of claim 15, wherein the plurality of effectiveness evaluations comprises a plurality of predicted effectiveness evaluations reflecting the expected increase in desired user actions.

18. The system of claim 15, wherein the rule constructing unit approves web pages selection rules which effectiveness evaluation is greater than the average of their previously computed effectiveness evaluation and predicted effectiveness evaluation.

19. A web page selection unit for directing a user terminal to a web page selected from a plurality of web pages, comprising:

a feature identifier unit which receives a network request sent from said user terminal and identifies a plurality of characterizing features associated with it;

a rule matcher unit which matches said network request with at least one web pages selection rule from a plurality of web pages selection rules, each said web pages selection rule comprises a rule condition defining at least one criterion for matching it with plurality of characterizing features associated with network requests, and a desired web pages selection distribution; and

a distributor unit which selects a web page from said plurality of web pages based on target web pages selection distributions of said at least one web pages selection rule, and redirects said user terminal to said web page.

20. A method of rule generation and adjustment for a web page selection mechanism, comprising:

monitoring a plurality of network requests received from a plurality of user terminals;

identifying a plurality of characterizing features in each said network request;

analyzing said plurality of network requests to identify a plurality of repeated patterns of characterizing features;

computing a plurality of statistic features for each said repeated pattern; and

generating a plurality of web pages selection rules for at least some of said plurality of repeated patterns, each said web pages selection rule comprises a rule condition defining at least one criterion for matching with plurality of characterizing features of at least one additional network request and a target web pages selection distribution set according to respective said statistic features.

21. The method of claim 20, further comprising recording in a database the plurality of characterizing features with plurality of indications associated with it, said plurality of indications comprise a selected web page indication specifying a web page that been selected responsive to said plurality of characterizing features, and a desired user action indication specifying occurrence of a desired user action responsive to provision of said web page,

wherein a web pages selection rule is generated for each of the at least some of the plurality of repeated patterns whenever a total number of complying desired user action indications is greater than a predetermined rule creation threshold.

22. The method of claim 21, wherein the predetermined rule creation threshold is in the range obtained by multiplying the number of landing pages by a predetermined parameter, said predetermined parameter an integer number between 5 and 9, inclusive, selected based on the campaign rate.

23. The method of claim 21, wherein the statistic features comprise a plurality of probabilities, each said probability is a probability of a rate of occurrence of desired user actions responsive to selection of a specific web page, and wherein said probabilities are computed by raising rates of desired user actions related to the repeated pattern to a power of a rule reliability factor.

24. The method of claim 23, wherein the rule reliability factor is computed for each web pages selection rule based on an average number of desired user action indications.

25. The method of claim 24, wherein the rule reliability factor is set by a piecewise linear function of the average number of desired user action indications.

26. The method of claim 23, wherein the statistic features comprise a plurality of instances of the plurality of probabilities, each said instance of plurality of probabilities is based on plurality of indications associated with network requests received within a time window selected from a plurality of time windows,

wherein the target web pages selection distribution is based on a weighted average of said instances of plurality of probabilities.

27. The method of claim 26, wherein the statistic features comprise a plurality of predicted rates of occurrence of desired user actions, each said predicted rate is based on plurality of indications associated with network requests received within one of the time windows,

wherein said plurality of predicted rates are used as the weights in the weighted average.

28. The method of claim 27, wherein the plurality of time windows comprises two predetermined time windows, a first time window defined to include all network requests received, and a second time window defined to include network requests received within a predetermined recent time window.

29. The method of claim 28, wherein the predetermined recent time window is determined as a function of the number of web pages.

30. The method of claim 20, wherein the plurality of statistic features comprise a plurality of predicted rates of occurrence of desired user actions computed for each web pages selection rule, further comprising sorting said web pages selection rules in a descending order according to said plurality of predicted rates.

31. The method of claim 30, further comprising computing for each of the plurality of web pages selection rules a new target web pages selection distribution, and new predicted rate of occurrence of desired user actions, whenever a new web page is added.

32. The method of claim 21, wherein the web pages selection rules are generated whenever the total number of the desired user action indications is greater than a predefined number.

33. The method of claim 32, wherein the predefined number is in the range of 200 to 300.

34. The method of claim 21, wherein a new set of web pages selection rules is generated whenever a new web page is added and number of the desired user action indications accumulated since said new web page was added is greater than a preset value.

35. The method of claim 34 wherein the preset value is in the range of 200 to 300.

36. The method of claim 34 wherein the preset value is set according to a rule creation threshold.

37. The method of claim 20 further comprising:

receiving at least one additional network request;

identifying a plurality of characterizing features associated with said additional network request;

determining at least one of the plurality of web pages selection rules which rule condition matches said plurality of characterizing features;

selecting a web page based on target web pages selection distributions of said at least one of the plurality of web pages selection rules; and

directing a user terminal from which said additional network request originated to access said web page.

38. The method of claim 37, further comprising determining at least one partially matching web pages selection rule and selecting a web page based on target web pages selection distributions of said at least one partially matching web pages selection rule, if no matching web pages selection rules are determined.

39. The method of claim 38, further comprising selecting a default web pages selection rule and selecting a web page based on target web pages selection distribution of said default web pages selection rule, if no partially matching web pages selection rules are determined.

40. The method of claim 37, comprising, whenever more than one web pages selection rules are determined, computing an interim target web pages selection distribution based on target web pages selection distributions of said more than one web pages selection rules.

41. The method of claim 30, wherein the interim target web pages selection distribution is computed as a weighted average of the target web pages selection distributions of the more than one web pages selection rules wherein the plurality of predicted rates of occurrence of desired user actions are used as weights.

42. The method of claim 20, wherein the target web pages selection distribution of each of the plurality of web pages selection rules defines a percentage of the plurality of network requests for which said each of the plurality of web pages be selected.

43. The method of claim 40, wherein the interim target web pages selection distribution defines for each of the plurality of web pages a percentage of the network requests for which said each of the plurality of web page be selected.

44. The method of claim 21, wherein the plurality of statistic features comprises at least one feature selected from a group consisting of:

a number of selected web page indications associated with a specific web page, a number of desired user action indications associated with a specific web page, a total number of selected web page indications, a total number of desired user action indications, an average number of desired users actions per number of web pages, and a rate of desired user actions occurred responsive to selection of each web page.