WEB PAGE ANALYSIS SYSTEM FOR COMPUTERIZED DERIVATION OF WEBPAGE AUDIENCE CHARACTERISTICS

• US 20120209795A1
• Filed: 05/23/2011
• Published: 08/16/2012
• Est. Priority Date: 02/12/2011
• Status: Active Grant

• Current Assignee: RED Contexto, Ltd.
• Original Assignee: RED Contexto, Ltd.
• Inventors: GLICKMAN, Oren
• Granted Patent: US 8,700,543 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06F 16/958 Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/0241 Advertisements
• CPC Class Codes: G06Q 30/0242 Determining effectiveness of advertisements

Abstract

A system for generating a demographic profile for a set of at least one webpage, the system comprising a webpage audience information gatherer operative for providing, for at least one webpage, training data including demographic information characterizing an audience of the webpage; a predictor developing system operative to compute at least one content characteristic of said webpage and to develop a prediction process which if applied to said content characteristic would have predicted said training data; and a webpage audience predictor operative, for at least one new webpage, whose audience is unknown, to compute at least one content characteristic of the new webpage and to generate predicted demographic information predicted to characterize said unknown audience of said new webpage by applying said prediction process to said new webpage'"'"'s content characteristic.

Claims

1. A system for generating a demographic profile for a set of at least one webpage, the system comprising:

a webpage audience information gatherer operative for providing, for at least one webpage, training data including demographic information characterizing an audience of the webpage; and

a predictor developing module using a processor to compute at least one content characteristic of said webpage and to develop a prediction process which if applied to said content characteristic would have predicted said training data,

wherein the prediction process is constructed and operative to be employed by an audience predicting processor operative, for at least one new webpage whose audience is unknown, to compute at least one content characteristic of the new webpage and to generate predicted demographic information predicted to characterize said unknown audience of said new webpage by applying said prediction process to said new webpage'"'"'s content characteristic.

2. A system according to claim 1 and also comprising a webpage audience predicting processor operative, for at least one new webpage whose audience is unknown, to compute at least one content characteristic of the new webpage and to generate predicted demographic information predicted to characterize said unknown audience of said new webpage by applying said prediction process to said new webpage'"'"'s content characteristic.

3. A system according to claim 1 wherein said combining comprises averaging.

4. A system according to claim 1 wherein said demographic information characterizing an audience of the web page data is associated in computer memory with web page data comprising actual html content.

5. A system according to claim 1 wherein said demographic information characterizing an audience of the web page data is associated in computer memory with a uniform resource identifier identifying at least one webpage.

6. A system according to claim 1 wherein webpage audience information gatherer is run as a onetime process.

7. A system according to claim 1 wherein webpage audience information gatherer is run as a periodical batch job.

8. A system according to claim 1 wherein webpage audience information gatherer is run continuously, generating a continuous stream of input training data.

9. A system according to claim 1 wherein webpage audience information gatherer fetches demographic information characterizing an audience of the webpage from a supplier of demographic information characterizing webpage audiences.

10. A system according to claim 1 wherein webpage audience information gatherer fetches demographic information characterizing an audience of the webpage from click-stream data of a user-panel with known demographics.

11. A system according to claim 1 wherein webpage audience information gatherer provides demographic information characterizing an audience of the webpage by monitoring social media streams.

12. A system according to claim 11 wherein said webpage audience information gatherer utilizes information provided by social media platform users'"'"' public homepages on which at least one demographic characteristic is explicitly stated.

13. A system according to claim 1 and also comprising a feature extraction unit serving at least one of the predictor developing module and the webpage audience predicting processor.

14. A system according to claim 13 wherein said feature extraction unit is operative to extract at least one word from webpage textual content.

15. A system according to claim 14 wherein said feature extraction unit is operative to extract at least one stylistic writing feature.

16. A system according to claim 15 wherein said stylistic writing feature includes use of parts of speech.

17. A system according to claim 15 wherein said stylistic writing feature includes sentence length.

18. A system according to claim 14 wherein said feature extraction unit is operative to extract at least one html specific feature.

19. A system according to claim 14 wherein said feature extraction unit comprises a web content fetching tool.

20. A system according to claim 14 wherein said feature extraction unit caches content thereby to enhance efficiency and/or reduce web traffic volume.

21. A system according to claim 19 wherein said content fetching unit performs page cleaning to remove sections of pages which do not contain main content essence of the pages.

22. A system according to claim 14 wherein said feature extraction unit is operative to extract at least one design feature representing visual design aspects of content and derived from markup language.

23. A system according to claim 14 wherein said feature extraction unit is operative to extract at least one feature relating to a content identifier of a web page.

24. A system according to claim 14 wherein said feature extraction unit is operative to extract at least one feature relating to a content identifier of a web-site.

25. A system according to claim 14 wherein said feature extraction unit is operative to extract at least one feature from meta-data supplied with a web-page.

26. A system according to claim 14 wherein said feature extraction unit extracts at least one feature from information about a web-site obtained from a 3rd party provider.

27. A system according to claim 14 wherein said feature extraction unit extracts at least one feature from site information mined from the Internet.

28. A system according to claim 14 wherein said feature extraction unit extracts a typification of a website from an existing website categorization information repository.

29. A system according to claim 14 wherein said feature extraction unit extracts overall site demographic statistics obtained from a web-site demographic statistics information repository.

30. A system according to claim 1 wherein said webpage audience predicting processor includes a plurality of classifiers respectively trained for a plurality of types of features.

31. A system according to claim 30 wherein said prediction process includes a plurality of individual models for each of said plurality of types of features and wherein said classifiers apply said individual models respectively thereby to generate a plurality of outputs.

32. A system according to claim 31 wherein said plurality of outputs is combined by majority voting.

34. A system according to claim 1 wherein said webpage audience predicting processor is operative to perform a new webpage learning process including comparing output of said prediction process as applied to said new webpage, to new webpage audience information, if known, and updating the prediction process accordingly.

35. A system according to claim 1 wherein said webpage audience predicting processor is also operative to perform website-level processing including combining said demographic information over all pages in the site.

37. A system according to claim 1 wherein said audience information gatherer is operative to crawl user profile pages of social media.

38. A system according to claim 37 wherein extraction is customized by specifying where the user profile information is found on the page.

39. A system according to claim 1 wherein said audience information gatherer employs social media public APIs to provide user profile information in a structured format.

40. A system according to claim 39 wherein the audience information gatherer runs iteratively over users, by following suitable links, so as to obtain a larger set of different users.

41. A system according to claim 1 wherein said audience information gatherer is operative for:

identifying user profiles with explicit demographic characteristics stated in a profile page and adding these to a user panel; and

monitoring data stream in social media platforms of users in panel, identifying posts referencing at least one web page; and

adding web page URI with demographic characteristics of user to a training data repository.

42. A system according to claim 41 wherein said predictor developing module is operative for:

obtaining training data from the training data repository;

fetching any missing content;

passing individual content examples to a feature extraction unit which is operative to provide representative features of each training content item; and

using content items represented as features along with corresponding target demographic characteristics as input to a supervised learning process thereby to obtain a prediction model predicting demographic characteristics of a page audience, as a function of page features.

43. A system according to claim 42 wherein said obtaining training data includes obtaining available input from said training data repository and fetching any missing content.

44. A system according to claim 42 and wherein said fetching comprises calling a content fetcher and cleaning at least one page fetched thereby, leaving only a content section of the page.

45. A system according to claim 1 wherein said webpage audience predicting processor is operative for:

obtaining input content whose audience is unknown from an input data repository;

passing at least one individual content example to a feature extraction unit which is operative to provide representative features of each training content item; and

passing content items represented as features to a classifier which runs a prediction model and obtains demographic characteristic predictions.

46. A system according to claim 45 wherein, if multiple items from a single site were passed as input, the webpage audience predicting processor is operative for computing a central tendency of the individual content predictions to obtain site-level statistics.

33. A method for generating a demographic profile for a set of at least one webpage, the system comprising:

providing, for at least one webpage, training data including demographic information characterizing an audience of the webpage;

computing at least one content characteristic of said webpage and developing a prediction process which if applied to said content characteristic would have predicted said training data; and

for at least one new webpage whose audience is unknown, computing at least one content characteristic of the new webpage and generating predicted demographic information predicted to characterize said unknown audience of said new webpage by applying said prediction process to said new webpage'"'"'s content characteristic.

36. A system for generating a demographic profile for a set of at least one webpage, the system comprising:

a audience predicting processor operative, for at least one new webpage whose audience is unknown, to compute at least one content characteristic of the new webpage and to generate predicted demographic information predicted to characterize said unknown audience of said new webpage by applying a prediction process to said new webpage'"'"'s content characteristic,

the prediction process being characterized in that if applied to at least one content characteristic of a webpage whose audience'"'"'s demographic information is known, said prediction process predicts training data including known demographic information characterizing an audience of the webpage.

47. (canceled)