U.S. patent application number 17/592634 was filed with the patent office on 2022-05-19 for providing additional digital content or advertising based on analysis of specific interest in the digital content being viewed.
The applicant listed for this patent is Gil Emanuel Fuchs. Invention is credited to Gil Emanuel Fuchs.
Application Number | 20220156793 17/592634 |
Document ID | / |
Family ID | 1000006121845 |
Filed Date | 2022-05-19 |
United States Patent
Application |
20220156793 |
Kind Code |
A1 |
Fuchs; Gil Emanuel |
May 19, 2022 |
PROVIDING ADDITIONAL DIGITAL CONTENT OR ADVERTISING BASED ON
ANALYSIS OF SPECIFIC INTEREST IN THE DIGITAL CONTENT BEING
VIEWED
Abstract
Systems and methods are described to provide additional relevant
content to a viewer of digital media such as a webpage. A webpage
being viewed is divided into regions and in each region, statistics
are compiled on pertinent words and phrases. Statistically
significant words and phrases are compared with semantically
similar words and phrases in the additional content. Where there is
a significant match between the viewed content and available
additional content, the additional content is provided to the
user.
Inventors: |
Fuchs; Gil Emanuel; (Nes
Tziona, IL) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Fuchs; Gil Emanuel |
Nes Tziona |
|
IL |
|
|
Family ID: |
1000006121845 |
Appl. No.: |
17/592634 |
Filed: |
February 4, 2022 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
14843407 |
Sep 2, 2015 |
11257115 |
|
|
17592634 |
|
|
|
|
62044904 |
Sep 2, 2014 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06Q 30/0277 20130101;
G06Q 30/0255 20130101 |
International
Class: |
G06Q 30/02 20060101
G06Q030/02 |
Claims
1. A method to supply relevant internet-based content to at least
one specific viewer of one or more internet pages, comprising:
dividing the one or more internet pages into regions using one or
more processing units, wherein the regions are defined by a form of
the content of the one or more internet pages; monitoring and
tracking the frequency and duration of viewing by a plurality of
viewers of each region using input devices functionally connected
to the one or more processing units; compiling textural information
for each region using the one or more processing units and for each
region: sorting out words or phrases that are pre-determined to be
of no value for determining user interest; compiling a list of the
most frequently used words and phrases from words that were not
sorted out; and for regions viewed by the plurality of viewers
above a specified threshold viewing frequency and duration,
semantically comparing the most frequently used words and phrases
to the most frequently used words and phrases of internet-based
content that can be supplied to the specific viewer; and selecting
and providing relevant internet-based content with semantic
similarity above a pre-specified threshold value to the specific
viewer on a display screen.
2. The method of claim 1, wherein said providing relevant
internet-based content to the specific viewer on a display screen
comprises at least one of a display of one of the relevant content
and a link to the relevant content in one or more of a specified
region or little used region of the one or more webpages, a new
page, and a popup.
3. The method of claim 1, wherein the form defining the regions is
based on html tags and elements.
4. The method of claim 1, wherein the form defining the regions is
based on rectangular sections of the content being displayed.
4. The method of claim 1, wherein the compiled textural information
for each region comprises at least one of: displayed text;
non-visible text within an underlying webpage definition, image
captions and internet links.
6. The method of claim 1, wherein the sorted out words that are
pre-determined to be of no value for determining interest comprise
words other than nouns and verbs.
7. The method of claim 1, wherein the semantically comparing of the
most frequently used words and phrases comprises: statistical
tokenizing and parsing of the frequently used words and phrases;
placing, the tokenized and parsed words and phrases in a database
operatively connected to the one or more processing; and comparing
the tokenized and parsed words in the database to tokenized and
parsed words from the additional content.
8. The method of claim 7, wherein said placing in the database
comprises placing the tokenized and parsed words and phrases in an
associative database using conceptual graph formalism, and said
comparing comprises comparing the tokenized and parsed words in
said associative database to tokenized and parsed words from the
additional content using a transformational algebra-based semantic
rule-set.
9. The method of claim 1, wherein the tracking of the frequency and
duration of viewing by the plurality of viewers comprises at least
one of (i) monitoring the location and duration time of a pointing
device within the confines of each region, and (ii) measuring the
location of the gaze of one or more of said plurality of viewers by
monitoring at least one of head and eye movement of the viewer with
respect to the location of a display and regions being viewed by
the viewer, and recording a duration of the gaze on each
region.
10. The method of claim 9, wherein for the tracking method of
monitoring of the pointing device intervals during which the
pointing device does not move for periods longer than a pre-defined
threshold time period are not recorded.
11. The method of claim 1, wherein one said region comprises an
entire webpage.
12. The method of claim 1, wherein the relevant internet-based
content is one or more of advertising, images, and internet page
links.
13. The method of claim 1, wherein the analysis for the one or more
internet pages is performed collectively for all of the plurality
of viewers combined.
14. The method of claim 1, wherein the compiling of the list of
most frequently used words and phrases of the one or more internet
pages is done prior to the viewing of the one or more internet
pages by the specific viewer.
15. The method of claim 1, further comprising determining whether
the additional content supplied to the specific viewer is also
viewed by the specific viewer and for how long and to further
utilize this information to adjust the pre-specified threshold
value used for selecting and providing relevant internet-based
content.
16. The method of claim 1, wherein the monitoring and tracking the
frequency and duration of viewing by the plurality of viewers is
supplemented by observing, using a camera facing one or more of
said plurality of viewers, the facial expressions of the monitored
viewers during the viewing of the one or more internet pages
wherein a captured facial expression is correlated with interest in
the viewing area.
17. A computer-based system implemented to supply relevant
internet-based content to at least one specific viewer of one or
more internet pages using collective analysis of viewing behavior
of a plurality of viewers of the one or more internet pages,
comprising: at least one memory or storage device configured to
store: internet content; semantic relationships between two or more
words and phrases within the internet content; counts of groupings
of semantically related word and phrases within each page of the
internet content; at least one display device configured to display
internet content; at least one input device configured to monitor
regions of the displayed internet content a viewer is looking at
and for how long; at least one processing unit functionally
connected to the at least one memory or storage device, the input
device and the display device, said at least one processing unit
configured to: divide the one or more internet pages displayed on
the display device and viewed by the plurality of viewers, into
regions, wherein the regions are defined by a form of the content
of the one or more internet pages; count the number of words and
phrases in each semantic group in each region that are the same
semantic groups that are counted from within the stored internet
content; for regions viewed by the plurality of viewers above a
specified threshold viewing frequency and duration, comparing the
highest count semantic groups within the stored internet content
with the most frequently viewed semantic groups; and select and
provide relevant internet-based content with semantic similarity to
the internet content being viewed on the display screen accessed by
the specific viewer.
18. The computer-based system of claim 17, wherein the input device
is configured to at least one of: monitor the location and duration
time of a pointing device within the confines of each region; and
measure the location of the gaze of a viewer by monitoring at least
one of head and eye movement of the viewer with respect to the
location of a display and regions being viewed by the viewer, and
recording a duration of the gaze on each region.
19. The computer-based system of claim 17, wherein the input device
comprises a camera facing the viewer, configured to monitor the
facial expressions of the viewer during the viewing of the one or
more internet pages wherein a captured facial expression is
correlated with interest in the viewing area.
20. The computer-based system of claim 17, wherein said at least
one processing unit is further configured to not record periods of
non-movement of the pointing device longer than a pre-defined
threshold time period.
21. A method to supply relevant internet-based content to at least
one specific viewer of one or more internet pages based on
collective analysis of views of the one or more internet pages by a
plurality of viewers, said method comprising: dividing the one or
more internet pages into regions using at least one processing
unit, wherein the regions are defined by a form of the content of
the one or more internet pages, the form defining the regions based
on html tags and elements; monitoring and tracking the frequency
and duration of viewing by the plurality of viewers of each region
using input devices functionally connected to the at least one
processing unit, wherein the tracking of the frequency and duration
of viewing by the viewer comprises at least one of (i) monitoring
the location and duration time of a pointing device within the
confines of each region and (ii) measuring the location of the gaze
of a viewer by monitoring at least one of head and eye movement of
the monitored viewer with respect to the location of a display and
regions being viewed by the monitored viewer, and recording a
duration of the gaze on each region; compiling the textural
information for each region using the at least one processing unit,
and for each region: sorting out words or phrases that are
pre-determined to be of no value for determining user interest;
compiling a list of the most frequently used words and phrases from
words that were not sorted out; for regions viewed by the plurality
of viewers above a specified threshold viewing frequency and
duration, semantically comparing the most frequently used words and
phrases to the most frequently used words and phrases of
internet-based content that can be supplied to the specific viewer,
wherein the semantically comparing of the most frequently used
words and phrases comprises (i) statistical tokenizing and parsing
of the frequently used words and phrases, (ii) placing, in one of a
memory or a storage device operably connected to the processing
unit, the tokenized and parsed words and phrases in an associative
database contained using conceptual graph formalism, and (iii)
comparing the tokenized and parsed words in the associative
database to tokenized and parsed words from the additional content
using a transformational algebra-based semantic rule-set; and
selecting and providing relevant internet-based content with
semantic similarity above a pre-specified threshold value to the at
least one specific viewer on a display screen as at least one of a
display of one of the relevant content, a link to the relevant
content in one or more of a specified region or little used region
of the one or more webpages, a new page, or a popup.
Description
RELATED APPLICATION DATA
[0001] This application is a continuation of U.S. application Ser.
No. 14/843,407, filed Sep. 2, 2015 and entitled "PROVIDING
ADDITIONAL DIGITAL CONTENT OR ADVERTISING BASED ON ANALYSIS OF
SPECIFIC INTEREST IN THE DIGITAL CONTENT BEING VIEWED", which
application claims priority to U.S. Provisional Application
62/044,904 filed Sep. 2, 2014. Each of these applications is
incorporated herein by reference.
[0002] The invention is related to U.S. Pat. No. 8,880,537, granted
Nov. 4, 2014, entitled "SYSTEM AND METHOD FOR USE OF SEMANTIC
UNDERSTANDING IN STORAGE, SEARCHING AND PROVIDING DATA OR OTHER
CONTENT INFORMATION" and is incorporated herein by reference.
"PRACTICAL NATURAL LANGUAGE PROCESSING QUESTION ANSWERING USING
GRAPHS", PhD dissertation by Gil Emanuel Fuchs, University of
California Santa Cruz, December 2004, is also related and is herein
incorporated by reference.
FIELD OF INVENTION
[0003] Embodiments of the invention are generally related to
identification of the main topics associated with digital media
with at least part represented as text and more specifically what
information within a displayed text and/or associated graphics is
of importance to the viewer and then utilizing this identified
information for tasks such as targeted advertising or to direct a
user to similar content.
BACKGROUND
[0004] Often when viewing a website or other form of digital media,
the underlying information being provided can be used to give an
indication of what the user is interested in. This can in-turn be
used to provide further content in the form of advertising of
products or services that are targeted based on the content.
[0005] Web pages and similar media can be complex and provide a
variety of information, therefore to more specifically target
follow-on information, there is a need to know what content that a
user is looking at and is of primary interest. One way to do this
is to monitor the precise location that a user is looking and for
how long they are viewing to more precisely target areas of
interest. This can be achieved by tracking the physical interaction
of the user with the digital media, by for example tracking mouse
movements and/or or eye movement relative the page. The location
and resulting content can be analyzed using a variety of techniques
such as semantic analysis of the text being viewed to predict what
specific content the view is interested in.
[0006] Another method is to simply count keywords or phrases in the
text and text associated with included graphics and supply
additional content or advertising with similar word counts.
Glossary
[0007] Text Based Digital Media: Any form of media that can be
presented on a screen and is made up of characters. The digital
media may also include imbedded images, videos, graphs or charts.
Example of text based digital media are: web pages, blogs, e-reader
content, pdfs, spreadsheets, text editor documents, and emails.
Also of importance is imbedded text associated with the viewable
media and images, that is hidden--for example the underlying HTML
of a webpage or XML or CSS.
[0008] Interest Indicators: Any measurable activity that may be
related to how interested a user of digital media is in the media
they are viewing. These activities could include the duration of
viewing and the location within a text that a user is viewing or
activities such as mouse clicks that could indirectly point to
this; clicking on certain buttons, imbedded information in images
about content; semantic parsing of key words or phrases that occur
in text being viewed; search terms utilized when scanning text,
simple or weight word counts in the digital media and many
others.
[0009] Normalized: With respect to a word count, and comparing more
than one word count, the basis for the count must be the same. If a
document had 2000 words (10 pages) and twenty instances of the word
"dog", then word count would be normalized by putting it in terms
of an average word count per page or the average number of
instances of a word per 100 words for example. So in this case, the
word count could be expressed as 2 per page or 1 in every 200
words.
BRIEF SUMMARY OF THE INVENTION
[0010] It is an object of this invention to determine both the
location and the content within digitally presented text/image
based media that a user is looking at.
[0011] Location within a digital media can be determined by
tracking eye movement, mouse or other pointing device position,
scroll bar position/s and duration and the like.
[0012] It is a further object of this invention to take this
information and determine appropriate follow-on targeted
advertising and/or addition similar content or links to similar
content.
[0013] Determination of how useful the targeted advertising and
content and links are is tracked by how often users proceed to the
additional content or links.
[0014] Based on the above, the targeting for additional content or
advertising is refined by weighting indications of what are
important to the user once metrics are determined on how often a
user follows suggested links or looks at additional content or
advertising. For example, it may be determined that a mouse being
stationary over a certain point on a website may be less important
of an indication of interest than if the scroll bar has not been
manipulated for the same amount of time.
[0015] It an object of this invention to filter indications of
interest in a particular portion of digital media based on when a
user is engaged vs when they are not. For example, if a user's
mouse is hovering in a certain location for an amount of time, but
yet the user is not at the computer, then the user is not engaged
and therefore the mouse location information is of no use.
[0016] A function of the present invention is to rate factors
associated with interest with respect to digital media.
[0017] It is an embodiment of this invention, to monitor word
counts of media presented to a user and the average duration of
those word counts over time to be used as a predictor of interest
in the content. The relative word count in the media being viewed
can then be compared to the relative word count of additional media
and where the relative word counts are similar in the additional
media, present that media to the user.
[0018] It is an aspect of this invention to survey what content is
of most importance to a collection of users and use this
information to better target the content in future revisions.
[0019] It is an aspect of this invention to utilize the above
techniques to determine the value of advertising content.
[0020] Other aspects of the invention will become apparent after
reading this document.
[0021] In one implementation, the present disclosure is directed to
a method to supply relevant internet-based content to at least one
specific viewer of one or more internet pages. The method includes
dividing the one or more internet pages into regions using one or
more processing units, wherein the regions are defined by a form of
the content of the one or more internet pages; monitoring and
tracking the frequency and duration of viewing by a plurality of
viewers of each region using input devices functionally connected
to the one or more processing units; compiling textural information
for each region using the one or more processing units and for each
region: sorting out words or phrases that are pre-determined to be
of no value for determining user interest; compiling a list of the
most frequently used words and phrases from words that were not
sorted out; and for regions viewed by the plurality of viewers
above a specified threshold viewing frequency and duration,
semantically comparing the most frequently used words and phrases
to the most frequently used words and phrases of internet-based
content that can be supplied to the specific viewer; and selecting
and providing relevant internet-based content with semantic
similarity above a pre-specified threshold value to the specific
viewer on a display screen.
[0022] In another implementation, the present disclosure is
directed to a computer-based system implemented to supply relevant
internet-based content to at least one specific viewer of one or
more internet pages using collective analysis of viewing behavior
of a plurality of viewers of the one or more internet pages. The
system includes at least one memory or storage device configured to
store: internet content; semantic relationships between two or more
words and phrases within the internet content; counts of groupings
of semantically related word and phrases within each page of the
internet content; at least one display device configured to display
internet content; at least one input device configured to monitor
regions of the displayed internet content a viewer is looking at
and for how long; at least one processing unit functionally
connected to the at least one memory or storage device, the input
device and the display device, the at least one processing unit
configured to: divide the one or more internet pages displayed on
the display device and viewed by the plurality of viewers, into
regions, wherein the regions are defined by a form of the content
of the one or more internet pages; count the number of words and
phrases in each semantic group in each region that are the same
semantic groups that are counted from within the stored internet
content; for regions viewed by the plurality of viewers above a
specified threshold viewing frequency and duration, comparing the
highest count semantic groups within the stored internet content
with the most frequently viewed semantic groups; and select and
provide relevant internet-based content with semantic similarity to
the internet content being viewed on the display screen accessed by
the specific viewer.
[0023] In yet another implementation, the present disclosure is
directed to a method to supply relevant internet-based content to
at least one specific viewer of one or more internet pages based on
collective analysis of views of the one or more internet pages by a
plurality of viewers. The method includes dividing the one or more
internet pages into regions using at least one processing unit,
wherein the regions are defined by a form of the content of the one
or more internet pages, the form defining the regions based on html
tags and elements; monitoring and tracking the frequency and
duration of viewing by the plurality of viewers of each region
using input devices functionally connected to the at least one
processing unit, wherein the tracking of the frequency and duration
of viewing by the viewer comprises at least one of (i) monitoring
the location and duration time of a pointing device within the
confines of each region and (ii) measuring the location of the gaze
of a viewer by monitoring at least one of head and eye movement of
the monitored viewer with respect to the location of a display and
regions being viewed by the monitored viewer, and recording a
duration of the gaze on each region; compiling the textural
information for each region using the at least one processing unit,
and for each region: sorting out words or phrases that are
pre-determined to be of no value for determining user interest;
compiling a list of the most frequently used words and phrases from
words that were not sorted out; for regions viewed by the plurality
of viewers above a specified threshold viewing frequency and
duration, semantically comparing the most frequently used words and
phrases to the most frequently used words and phrases of
internet-based content that can be supplied to the specific viewer,
wherein the semantically comparing of the most frequently used
words and phrases comprises (i) statistical tokenizing and parsing
of the frequently used words and phrases, (ii) placing, in one of a
memory or a storage device operably connected to the processing
unit, the tokenized and parsed words and phrases in an associative
database contained using conceptual graph formalism, and (iii)
comparing the tokenized and parsed words in the associative
database to tokenized and parsed words from the additional content
using a transformational algebra-based semantic rule-set; and
selecting and providing relevant internet-based content with
semantic similarity above a pre-specified threshold value to the at
least one specific viewer on a display screen as at least one of a
display of one of the relevant content, a link to the relevant
content in one or more of a specified region or little used region
of the one or more webpages, a new page, or a popup.
BRIEF DESCRIPTION OF THE DRAWINGS
[0024] The drawings constitute a part of this specification and
include exemplary embodiments to the invention, which may be
embodied in various forms. It is to be understood that in some
instances various aspects of the invention may be shown exaggerated
or enlarged to facilitate an understanding of the invention.
[0025] FIG. 1 is an example portion of a webpage pointing out word
counts.
[0026] FIG. 2 is a flow chart of an embodiment of determining
additional internet content of interest.
[0027] FIG. 3 is an example of processing hardware.
DETAILED DESCRIPTION OF THE FIGURES
Overview
[0028] One overall goal of this invention is to provide targeted
advertising or additional content to a viewer of text based digital
media that will be of interest to the viewer based on what they are
viewing.
[0029] To achieve the above goal one aspect of this invention is to
determine what portion or aspect of a text based digital media is
most important to a user or group of users. This information is
then used to find additional content to display to the user which
can include targeted advertising based on the above, additional
information similar in nature to the user's interest, or links to
additional information.
[0030] In an embodiment where it is desired to determine specifics
concerning when and where a viewer is viewing text based digital
media, the following must be known: [0031] A mapping of the areas
associated with each subset of the information displayed on a
display screen and the time and duration the information was
displayed. [0032] Where a user is looking on a display device at
any given time. This could be broken into quadrants that, for
example could be a specific part of a webpage or simply a grid in a
grid pattern or certain lines of text. [0033] The actual content of
each area.
[0034] Armed with the above information, then it can be determined
what information the user was looking at and for how long.
[0035] Depending on where the user is looking and for how long,
indicators can be developed to infer what is most important to the
user.
[0036] An embodiment is shown in FIG. 2. If a viewer of an internet
page or web page or other digital media is looking at a particular
web page or textural digital content, the page/s is divided into
regions 202. The regions could be based on a particular element of
html, for example a title or a frame; the regions could be based on
a rectangular region of a display presented to the user.
[0037] When a user is viewing the page or pages above, how it is
tracked how long a user is looking at a particular region 204.
[0038] Textural information is compiled 206 which could, for
example, be the visible text in the region, the text associated
with the region both visible or imbedded in HTML or other
non-visible information. Words or phrases found not to be of
semantic importance are thrown out.
[0039] Previous to this analysis, pertinent words and phrases have
been counted and put in a semantic database.
[0040] The frequency of words and phrases are compared with the
frequency of semantically similar words in phrases in the content
to be provided using means described later in this document 208.
Based on the strongest correlations between the page being viewed
and the additional content, additional content is selected and
displayed to the user in a variety of ways 210.
Determining Location on the Display
[0041] In an embodiment, whatever text content is displayed or for
that matter, text associated with the view but is hidden, must be
subdivided into sections. If the document or page or active window
being viewed is configured according to some standard, then the
subdivision could be based on that standard.
[0042] If the digital media is a webpage, for example, the page
might consist of a header, a footer, and navigation bar (either on
the top, bottom , left side or right side). In addition if the page
does not fit on the display screen, it may have scroll bars on the
right and/or bottom of the screen generally. These subsections will
be described in the code for the webpage, for example in html or
cascading style sheets (CSS). In this code, it would describe what
pixels are associated with what portion of the page.
[0043] Using the webpage example, assume there is a standard
configuration and, for example, the page is a news feed. Further
assume the user frequently clicks on a button that is a link for
sports news in the index section of the page and sports is selected
more often than a button for world news. It could be inferred that
the user is more interested in sports new than world news. The
number of clicks on the sports news buttons could be determined by
tracking when the mouse is over the location of the button (as
described in the HTML) and when the mouse button is clicked.
[0044] If the display is simply a text document or a section of the
display that the user is looking at is a text area, there may be no
subsections of the page or subsections of the text that are
meaningful. In this case, then it may be necessary to semantically
map the text for given lines or a quadrant of the display area and
to determine key words that may indicate what the user is
interested in. This will be described more in a later section.
[0045] Ideally to determine where on a display screen a user is
looking, it would be desirable to track eye movements and barring
that, head movements. Devices that interface with a computer that
track eye or head movement exist in the art. If one is available
and the user is willing to wear a device or turn on a free standing
device, then this is one of the most desirable methods for
determining what portion of digital media a user is viewing. One
simply needs to know the relative position of the screen to the
eyes/head, the orientation of the gaze with respect to the screen,
and what content is were on the screen and when the content is
there.
[0046] Other methods to detect the viewing location for a user
include: [0047] Tracking the mouse (or other pointing device)
location on the viewing screen [0048] Tracking the relative
location of scroll bars (for pages that spill off the viewing
screen) [0049] Monitoring search words used when viewing text
[0050] If images are imbedded with the text, determining the
subject of the image
[0051] With respect to scrolling, it can be assumed that in general
the point of interest will be in the center of the window or screen
that is active. By knowing the relative location of the scroll
bars, this can be determined. Likewise the duration that the scroll
bars are in one position can also be monitored with techniques
known in the art.
Determining Preferred Content Based on Duration of Use and Location
Within a Display
[0052] Through methods described above, it can be determined what
text is on a display and for how long. It will also be known how
much text is on the display based on the font size and the size of
the display. How long individual lines of text are displayed can be
determined based on what portion of a document is displayed at any
given time. By performing semantic analysis of the information
displayed on a screen, key phrases or words (and words or phrases
that are synonymous) that appear can be tallied. Key words or
phrases that appear the most can be inferred to be of the most
interest.
[0053] In analysis in the art, if for example, a user was looking
at a website concerning movies, it may be assumed that the user may
be interested in anything to do with movies.
[0054] In an embodiment of the present invention, if the text that
was being viewed was analyzed, it may be found that the phrase
"Johnny Depp" comes up frequently and for long duration relative to
other phrase during the viewing of the article. This information
could then be used to search for additional content or advertising
on movies associated with "Johnny Depp" instead of movies in
general.
[0055] As mentioned previously, there may be mixed media on the
display, including not only text, but images and/or video and/or
other forms of graphics. If a topic can be inferred from the mixed
media (in addition to the analysis of the text), this may also be
of value in determining primary interest to the viewer. For
example, certain types of images may have imbedded tags within the
image file that describe what the image depicts. Or in some
situations, the title of the image itself may be descriptive. So if
any embedded tags within the image or the title of the image are
analyzed, this will provide additional input that could imply
primary interest on the user's part.
Determining Preferred Content by Counting Words
[0056] In an embodiment of the present invention, the assumption is
made that a viewer is viewing an entire document. Words counts in a
basic form would consist of counting each word, or phrase
(generally ignoring common words that would not be associated with
the topic of the text--for example "as", "the", "and", etc.) in a
text based media, then comparing a normalized count of words in the
viewed media, with the normalized count of words in proposed
additional media or advertising. Also compare the relative ratio of
each word count between the viewed media and the proposed
additional media or advertising. When the proposed media matches
closely to the viewed media, then present the proposed media or
link to the viewer.
[0057] It can readily be identified by the underlying structure of
digital media, what content is on a display screen or in an open
and active window on a display screen. It is also readily known in
the art how to measure how long a particular display is visible.
For example, if a user is viewing a text document on a display, the
display is generally static, until such time as the users inputs
via a pointing device or other means that they would like to scroll
or go to a different part of the text that is not visible. This can
happen gradually such as scrolling down a text or could happen
abruptly by, for example, entering a search string and moving to
sequential locations that contain that search string.
[0058] In an embodiment the prediction of user interest in subject
matter is a function of the words in the view of the user and the
duration those words are there. For example, FIG. 1 is a typical
portion of a webpage 102. Notice that there are several instances
of the word "wave" visible 106, 108, 112, and if you include words
with a root word of "wave" also 116 and 120. This makes 8 instances
of the word "wave" on a stationary page. Assume that while
monitoring user activity, it was determined that this screen shot
was up for 30 seconds. If the user scrolled down the page (ignore
the scrolling time in this example), then, for example, 4 of the
previously viewed instances of "wave" may still be visible and
perhaps 2 new instances appear further down the text. If this
modified view was up for 20 seconds then the new tally of 6
instances of "wave" over 20 seconds. Combing the two measurements,
on average the user was exposed to ((8*30)+(6*20))/(30+20)=7.2
instances of "wave".
[0059] Similar word counts are made for all other words on the
visible page. In an embodiment, only nouns and verbs are counted as
they are more likely to be indicative of user interest. In an
embodiment, all words that have a similar root are collectively
counted--for example, surf, surfing, surfer.
[0060] In one form, provided you have a database of additional
content, that has word counts associated with it, you could compare
the word count normalized based on the number of words on the page
that was viewed by the user to the normalized word count of
proposed additional content--whether that content is a web page, an
article or advertising material. If the additional material had
word counts in a predetermined range close to the word counts for
key words of the viewed material, and the numerical ordering of the
word counts were similar, then either the material or a link to the
material could be presented to the user for possible viewing.
[0061] In FIG. 1, notice the high instances of the words (or root
words) "wave" 106, 108, 112, 116 and 120 and "surf" 110, 114, 118
and 122. If there is a high percentage of the terms "wave" and
"surf" it is more likely that the user is interested in water
sports than if there was a high percentage of the terms "wave" and
"destruction" where the user may be more interested in natural
disasters.
[0062] Taking the method still further, different instances of the
same word on a page could have different weights associated with
them. The weights could be a function of, for example, font size,
font type (bold, italic), use in a header, use in a caption for a
figure or a graph, use in a link, embedded in an image and the
like. The more prominently displayed the word is on a page, the
more weight it has in prediction of interest of the user--provided
of course that the user takes time to view the prominently
displayed material. By weighting, it is meant that word counts are
modified based on the prominence of the word. Therefore, for
example, the same word with an instance in a title may be
classified as being equivalent to 10 instances in a text body. Note
in FIG. 1. the image of a wave 104. The caption of the image 104
does not have the word "wave" in it, however if you look at the
underlying html that specifies the source of the image <img
src="../../images/ALL%20HAWAII/allhawaii_resized/allhawaii/WatchBigWavePH-
07_9108.jpg "class="photo" width="516" height="234" alt="all hawaii
images" style="top: 19px; left: 10px;">) notice that the word
"wave" appears. Once again, the weight of the image name may be
assigned to be higher than normal text. In an embodiment the word
count associated with the image name and/or caption of the image
may be weighted according to the relative space the image takes up
on the page (in this case 516 by 234 pixels).
[0063] It is understood that how to weight word counts for certain
types of display material is a function of a predetermined
weighting criteria and this criteria may vary for types of media
and/or change once feedback is established on how relevant
additional content that is provided to a user is utilized and/or
viewed.
[0064] In an embodiment word counts are only made on a subsection
of the displayed material where the user is determined to be
looking.
Determining User Interest by Observation of User Reaction
[0065] There exists in the art methods of monitoring individual
characteristics and reactions. Many of these reactions when
interpreted in relationship to displayed content may be indicative
of interest in the content. For example, it is well known that
facial expressions can express joy, frustration, anger,
concentration and other state of emotion. Using techniques to
monitor facial expressions, such as image analysis, for example, it
can be postulated that strong emotion was related to interest in
the content.
[0066] Similarly brain wave activity monitoring could show
indications of interest in content. Subtle changes in skin
temperature could indicated increased interest.
[0067] In the case of facial expressions, these could be detected
by a camera located on top of a computer display or smartphone
facing the user of the device.
[0068] Various sensors exist in the art for monitoring brain waves
and skin temperature variations.
Finding Relevant Additional Content, Links or Advertising Media to
Present to the User
[0069] Embodiments of the invention require that proposed
additional content and/or advertising media be indexed or have
indicators such as word count, in the same manner as the text being
viewed. Further text being compared must have indicators normalized
in the same manner. For example, if the word counts are per page is
used for the text being viewed, all word counts must be normalized
per page.
Semantic Decomposition of Text and Finding Relevant Additional
Content or Advertising Media
[0070] Techniques exist in the art to allow capturing and
interpreting semantics or meaning from text. In one example, a
conceptual graph can be created from text being viewed. A plurality
of conceptual graphs, or portions thereof, can be compared to
determine matches with conceptual graphs of other text. An
associative database can be generated and/or searched using a
hierarchy of conceptual graphs, so that the data storage and
searching of such database is optimized. The associative database
can be used to find additional content or advertising from multiple
different sources which can be presented to a viewer of digital
media.
[0071] Additional information describing various aspects and
examples of conceptual graphs, link grammars, and associative
databases are described in "PRACTICAL NATURAL LANGUAGE PROCESSING
QUESTION ANSWERING USING GRAPHS", PhD dissertation by Gil Emanuel
Fuchs, University of California Santa Cruz, December 2004, which is
herein incorporated by reference.
[0072] In an embodiment, a system and method for use of semantic
understanding in searching and providing of additional content to a
viewer of digital media is presented. In accordance with an
embodiment, the system comprises a Syntactic Parser (SP) or
statistical word tokenizer for data retrieval and parsing; a Syntax
To Semantics (STS) transformational algebra-based semantic rule
set, and an Associative Database (ADB), utilizing a conceptual
graph formalism. Data can be represented within the ADB, enabling
both fast data retrieval in the form of semantic objects and a
broad ranging taxonomy of content, e.g. advertising categories.
Each semantic object contains all the related terms and phrases
articulating a specific subject, enabling automatic categorization
of any given page. This semantic approach can be used in a variety
of ways, for example to improve the ability to serve ads based on
the meaning of a website's page content. By semantically analyzing
the web pages, the system can properly understand and classify the
meaning and sentiment of any given digital text, and accordingly
ensure that the web page receives the most appropriate advertising.
The system can also ensure that campaigns are placed on pages which
are contextually relevant to them, whatever the format and medium.
For example, the semantic approach can be used to analyze an
advertiser's ad and the website it links to, in order to identify
the most relevant matches.
Filtering Out Idle Periods from Location/Content Statistics
[0073] Unless there is a system for monitoring eye or head motion
in place, it may not be possible to determine if a user is actually
viewing digital media. Therefore inferences must be made based on
user activity or lack thereof. For example, if a user does not
utilize a pointing device for a set period of time, a rule could be
made that does not include the duration of any media being on a
viewing screen for a prescribed time prior relative to the
inactivity. If a screen saver is enabled, a similar rule could be
made based on when the screen saver comes on. Similar rules could
be made for lack of keystrokes on a keyboard.
Rating Indicators of Relevant Content
[0074] In a section above describing the determination of preferred
content, key phrases and/or words were mentioned in an example.
Also mentioned were descriptive titles of images and/or video's and
embedded descriptive tags in this media. After much analysis of
these indicators is performed, and subsequent analysis of whether
additional media that is presented based on these indicators is
viewed, it can be determined which indicators are more
statistically valid as pointers to preferred additional content.
For example, continuing the example posed earlier, if an image
associated with "Johnny Depp" is in view for a long duration, this
may be more indicative of the user being interested in "Johnny
Depp" than if the phrase "Johnny Depp" appeared in visible text 10
times, and for the same duration as the image. If this was the
case, the method for rating interest may include a relative
weighting of the various indicators being used.
[0075] The argument can be extended to include other indicators
including mouse clicks and scroll bar position and duration and the
like.
[0076] A combination of indicators could be used simultaneously,
for example in a polynomial equation where the probability of a
user being highly interested in content was a function of the
summation of individual indicators multiplied by their respective
weighting factors.
Rating Content from Crowd Sourced Indicators
[0077] In an embodiment, rating of interest of particular media
could be collectively performed by observing indicators of interest
from several users that view the same media. By have a collective
rating of the media, this information could be used to modify the
media to be more directed to more users, or the media could be
subdivided to provided content targeted to certain interest groups.
The information could also be used to sell advertising to show that
certain type of interest groups read the content that is
provided.
[0078] The present invention may be conveniently implemented using
one or more conventional general purpose or specialized digital
computers or microprocessors programmed according to the teachings
of the present disclosure. Appropriate software coding can readily
be prepared by skilled programmers based on the teachings of the
present disclosure, as will be apparent to those skilled in the
software art.
[0079] In some embodiments, the present invention includes a
computer program product which is a non-transitory storage medium
(media) having instructions stored thereon/in which can be used to
program a computer to perform any of the processes of the present
invention. The storage medium can include, but is not limited to,
any type of disk including floppy disks, optical discs, DVD,
CD-ROMs, microdrive, and magneto-optical disks, ROMs, RAMs, EPROMs,
EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic or optical
cards, nanosystems (including molecular memory ICs), or any type of
media or device suitable for storing instructions and/or data.
[0080] The foregoing description of the present invention has been
provided for the purposes of illustration and description. It is
not intended to be exhaustive or to limit the invention to the
precise forms disclosed. The embodiments were chosen and described
in order to best explain the principles of the invention and its
practical application, thereby enabling others skilled in the art
to understand the invention for various embodiments and with
various modifications that are suited to the particular use
contemplated. For example, although the illustrations provided
herein primarily describe embodiments using vehicles, it will be
evident that the techniques described herein can be similarly used
with, e.g., trains, ships, airplanes, containers, or other moving
equipment. It is intended that the scope of the invention be
defined by the following claims and their equivalence.
[0081] FIG. 3 is schematic of a typical processor an embodiment of
the method can run on. The system comprises some form of memory to
store both programming and information being accessed 302. One of
more processors are used to compute sematic relationships 312.
Additional storage for data and programming that does not need
immediate access can be either removable or remote storage (cloud
based) 304, or non-removable storage 306. Several input devices can
be connected to the system which including various pointing devices
such as a mouse, a camera facing the user of the system used to
monitor facial expressions for example and some form of head or eye
tracker which knows the relative position of eyes or the head to
the display screen 310.
* * * * *