ENABLING PREDICTIVE WEB BROWSING
BACKGROUND OF THE INVENTION
Field of the Invention
This invention relates generally to the field of computer networks, and, in particular,
to web browsing.
Description of the Related Art
The World-Wide Web (the "web") is a collection of web pages and other viewable
files that are interconnected to one another over the network known as the Internet. A user
typically browses web pages using a web browser program running on a client computing
device, such as a laptop or desktop computer, mobile phone, or other wired or wireless device
connected to the Internet. The address of the web page currently being viewed, known as the
uniform resource locator (URL), is generally displayed within the user's web browser.
A given web page (known as a "referring" web page) can permit a user to navigate to
one or more other web pages (known as "linked" web pages) by means of one or more links
(also called "hyperlinks"). A link is an item on the referring web page that, when selected,
transfers a user directly to view another location, either on the referring web page, or on a
different linked web page. Accordingly, a link specifies at least the location (e.g., URL) of
the linked web page. A link can also specify the name or title of the linked web page, and
some links also specify a specific location within the linked web page. It should be
understood that the word "link" is used in the art to refer both to the location (e.g., URL) of
the linked web page and to the selectable item (e.g., an underlined group of words, an icon, or
a picture) on the referring web page on which the user clicks to navigate to the linked web
page.
Links to linked web pages are typically graphically denoted within the referring web
page via underlining or bold text, and sometimes by a graphic item (e.g., an icon or a picture)
that the user selects. By selecting a link (typically by mouse click), the user directs the
browser to load, acquire, and display the corresponding linked web page to which the link
points.
Using a web browser typically entails a combination of two different methods, both
web browsing and web searching.
The process of web browsing involves the user moving from web page to web page
by selecting links on currently-displayed web pages, completely at the whim of the user. This
process is often performed with the user having little or no information as to the relevance of
a given linked web page to the user's particular topic of interest. Moreover, the user often
cannot discern whether a linked web page is relevant until the user actually clicks on the link
and subsequently views the linked web page. One reason this happens is that many links are short or contain little description about t e linked web pages to which they refer. For
example, some links might simply state "Click Here" or "More." Thus, it is not always
possible during the process of web browsing for the user to obtain a sense of the relevance of
a linked web page to which a given link points, and it can therefore be difficult for a user to
select manually, by way of browsing, pages that are actually relevant.
In contrast to t e relatively undirected process of web browsing, the process of web
searching generally involves the user employing a search engine, e.g., Google
(http://www.google.com ) or Yahoo (http://www.yahoo.com ), to find information available
about a given topic of interest. When web searching, the user inputs, into a search engine,
relevant keywords about the topic, and the search engine then returns a list of links to linked
web sites that may relate to the topic.
Although search engines provide the most prevalent method for users to obtain
relevant information on the web today, as the volume of web pages on the web continuously
increases, the relevance of search results tends to decrease. Significant research is being
conducted in an attempt to address this difficulty, including improving web page-rank
evaluations, methods of refining searches, and methods of categorizing information based on
input from other users. All of these methods require an active query from the browsing user
to determine relevant sites. Further, these search engine-based solutions provide only the
selection of a single entry point into a web site. In other words, the user enters search
keywords and receives a list of links to potentially-relevant web pages returned as search
results. However, once the user clicks on one of the links in the list and is taken to the
corresponding linked web page, the user is left without any guidance as to which link on the
linked web page and subsequent linked web pages the user should select, and the user is
forced to return to the undirected process of web browsing to navigate further within the
linked web pages returned as search results.
SUMMARY OF THE INVENTION
Problems in the prior art are addressed in accordance with the principles of the
present invention by providing a system and method for determining and displaying, in a
predictive manner, relevant "next clicks" based upon historical web-usage patterns of the
browsing user and/or other users. In certain embodiments of the invention, identification of
one or more links selected by previous visitors on a plurality of referring web pages is stored
in a database. When the browsing user initiates a request to view a referring web page for
which one or more links exist in the database, the appearance of those links on the referring
web page is altered to suggest those links to the browsing user. Thus, the browsing user is
given suggestions as to the most probable path (based upon their own history and/or thehistory of others) through each referring web page for which one or more links are stored in
the database, thereby streamlining the browsing process.
In one embodiment, the invention provides a computer-implemented method for
displaying web pages to a current user of a web browser running on a computing device. The
method includes: (a) storing identification of (i) one or more referring web pages visited by
one or more previous visitors and, (ii) for each of the one or more referring web pages, one or
more links on the referring web page selected by the one or more previous visitors who
visited the referring web page; and (b) in response to the initiation by the current user of a
request to view one of the one or more referring web pages, and based on the stored
identification of the one or more links on the requested referring web page selected by the one
or more previous visitors who visited the referring web page, altering the appearance of the
requested referring web page displayed to the current user.
In another embodiment, the invention provides an apparatus for displaying web pages
to a current user of a web browser running on a computing device. The apparatus includes a
database and a processor. The database is adapted to store identification of (i) one or more
referring web pages visited by one or more previous visitors and, (ii) for each of the one or
more referring web pages, one or more links on the referring web page selected by the one or
more previous visitors who visited the referring web page. The processor is adapted, in
response to the initiation by the current user of a request to view one of the one or more
referring web pages, and based on the stored identification of the one or more links on the
requested referring web page selected by the one or more previous visitors who visited the
referring web page, to alter the appearance of the requested referring web page displayed to
the current user.
In a further embodiment, the invention provides a non-transitory machine-readable
storage medium, having encoded thereon program code, wherein, when the program code is
executed by a machine, the machine implements a method for displaying web pages to a
current user of a web browser running on a computing device. The method includes: (a)
storing identification of (i) one or more referring web pages visited by one or more previous
visitors and, (ii) for each of the one or more referring web pages, one or more links on the
referring web page selected by the one or more previous visitors who visited the referring web
page; and (b) in response to the initiation by the current user of a request to view one of the
one or more referring web pages, and based on the stored identification of the one or more
links on the requested referring web page selected by the one or more previous visitors who
visited the referring web page, altering the appearance of the requested referring web page
displayed to the current user.BRIEF DESCRIPTION OF THE DRAWINGS
Other aspects, features, and advantages of the present invention will become more
fully apparent from t e following detailed description, the appended claims, and the
accompanying drawings in which like reference numerals identify similar or identical
elements.
FIG. 1 is a block diagram depicting an exemplary predictive web-browsing system in
accordance with a first embodiment of the invention;
FIG. 2 is a block diagram depicting an exemplary predictive web-browsing system in
accordance with a second embodiment of the invention; and
FIG. 3 is a flow diagram depicting an exemplary method of predictive web browsing
in accordance with certain embodiments of the invention.
DETAILED DESCRIPTION
FIG. 1 depicts an exemplary predictive web-browsing system 100, consistent with a
first embodiment of the invention. System 100 includes one or more clients 102(1), 102(2),
. . . , 102(N) in communication with a predictive-click processor 106 via the Internet (not
shown). Each client 102(i) is a computing device on which a respective web browser 104(i)
runs. Although, in the discussion below, the operations of only client 102(1) and browser
104(1) are described, it should be understood that, in embodiments with multiple clients, the
operations of clients 102(2) through 102(N) and browsers 104(2) through 104(N) are
substantially the same as those described below for client 102(1) and browser 104(1),
respectively.
Predictive-click processor 106 includes a historical page and click tracker 108, a page
and click statistics database 110, and a next-click generator 112.
A user operates browser 104(1) on client 102(1) to browse the web. During the
browsing process, browser 104(1), pursuant to the HyperText Transfer Protocol (HTTP),
requests pages from web servers (not shown) by sending packets of text called HTTP
requests, and those web servers reply by sending the requested pages, preceded by packets of
text called HTTP responses. It is noted that, for clarity, the HTTP requests sent to web
servers by browser 104(1) and the HTTP responses and web pages received from web servers
by browser 104(1), all of which are exchanged during normal operation of browser 104(1),
are omitted from FIG. 1.
In this embodiment, browser 104(1) provides, to historical page and click tracker 108
of predictive-click processor 106, information about which link(s) on each referring page the
user has clicked (i.e., selected) during browsing. Specifically, for each link clicked by the
user, browser 104(1) provides to historical page and click tracker 108 both (i) page data, i.e.,the URL identifying the referring web page containing the link that was clicked and (ii) click
data, i.e., identification of which link on t e referring web page was clicked.
In various embodiments of the invention, the transmission of page data and click data
from browser 104( 1) to historical page and click tracker 108 can be accomplished by several
different methods, including, e.g., the use of a customized browser having this capability or a
plug-in or add-on software module for an existing browser to provide this capability. (In
alternative embodiments, as will be discussed in further detail with respect to FIG. 2, an
ordinary, unmodified browser can be used for browsing, in which case, packet inspection is
used to extract page data and click data from the browser's outgoing traffic stream.)
Although, in the embodiment of FIG. 1, browser 104(1) provides page data and click
data in real time while the user browses the web, it is contemplated that, to reduce network
congestion and processing overhead, in other embodiments, page data and click data could be
provided to historical page and click tracker 108 either at periodic intervals (e.g., once per
hour or per day), or alternatively, during periods when browser 104( 1) and/or client 102(1)
are determined to be idle or when sufficient unused network bandwidth is detected.
Page and click statistics database 110 contains one or more data structures for storing
(i) a plurality of URLs identifying a plurality of referring web pages containing links that
were clicked by one or more users and, (ii) for each URL in the database, data identifying the
frequency with which one or more links on the corresponding referring web page were
previously clicked by the one or more users. Historical page and click tracker 108 receives
and aggregates the page data and click data from browser 104(1) and uses that data to update
page and click statistics database 110. In this embodiment, database 110 is updated using an
algorithm implemented in historical page and click tracker 108 that tracks and stores in
database 110 data identifying the number of times a plurality of links on each referring web
page were previously clicked by the one or more users. From this information, the most
frequently clicked link can be determined, which will typically be the link that a user is most
likely to select when viewing that referring web page. Therefore, it is desirable to suggest this
link to one or more subsequent users who land on that referring web page, as will now be
described.
During a normal operation of browser 104( 1) on client 102(1) to browse the web, the
user will view a series of referring web pages provided by various web servers. In addition to
browser 104(1) transmitting to historical page and click tracker 108 information about the
referring web pages being viewed, browser 104( 1) also transmits to next-click generator 112,
in real time, the URL of every referring web page being viewed. Next-click generator 112
uses this URL to perform a query on page and click statistics database 110. If there is an
entry for the URL in database 110, then next-click generator 112 receives, as a result of the
query, the most frequently clicked link for the referring web page identified by that URL. (Ifmore than a single link on a referring web page has previously been selected with t e same
highest frequency, then t e "tie" is broken by randomly selecting any one of those highest-
frequency links.) If there is no entry for the URL in page and click statistics database 110,
then next-click generator 112 receives, as a result of the query, an error flag.
Next-click generator 112 then provides to browser 104(1) either a suggested "next
click," which is the most frequently clicked link on the referring web page identified by the
URL, or else an error flag indicating that a most frequently clicked link does not exist for the
URL. If browser 104(1) receives an error flag, then no action is taken by browser 104( 1).
However, if browser 104(1) receives a suggested next click, then browser 104( 1) indicates
that suggested next click to the user by altering the manner in which the corresponding link on
the referring web page is displayed within the browser, such as by changing that link's color
or size or otherwise modifying the appearance of that link. Such alteration can be made either
by modifying the underlying code (e.g., HTML code) of the referring web page or by
manipulating the appearance of the link on the screen only, without modifying the underlying
web page code.
The result of the foregoing operations is that the browsing user is given suggestions as
to the most probable path (based upon their own history and/or the history of others) through
each referring web page that has a corresponding most frequently clicked link stored in the
database, thereby streamlining the browsing process by reducing the number of decisions that
the user makes while browsing, or, at a minimum, assisting the user in making decisions
while browsing. Each click by the browsing user will update the database based upon that
user's browsing history to improve the suggested links subsequently provided to the user.
Moreover, in embodiments in which data from multiple users is aggregated, a user can
essentially access the collective knowledge of others, e.g., by being able to determine rapidly
which links are the most useful when encountering a new web page or web site.
FIG. 2 depicts an exemplary predictive web-browsing system 100, consistent with a
second embodiment of the invention. System 200 is similar to system 100 of FIG. 1, with the
exception that, in this embodiment, the predictive-click processor functionality is integrated
into a proxy server that also handles incoming and outgoing HTTP web traffic for one or
more web browsers, as will be described in further detail below.
System 200 includes one or more clients 202(1), 202(2), . . . , 202(N) coupled to the
Internet via a predictive -click proxy server 206, although, for clarity, only a single client
202(1) and browser 204(1) are shown in FIG. 2 . Each client 202(i) is a computing device on
which a respective web browser 204(i) runs. Although, in the discussion below, only the
operations of client 202( 1) and browser 204(1) are described, it should be understood that, in
embodiments with multiple clients, the operations of clients 202(2) through 202(N) andbrowsers 204(2) through 204(N) are substantially the same as those described below for client
202(1) and browser 204(1).
Predictive-click proxy server 206 includes a historical page and click tracker 208, a
page and click statistics database 210, a next-click generator 212, and a packet inspection and
modification module 220. The operation of historical page and click tracker 208, page and
click statistics database 210, and next-click generator 212 are substantially the same as that of
historical page and click tracker 108, page and click statistics database 110, and next-click
generator 112 of FIG. 1, respectively, except as otherwise described below.
In the embodiment of FIG. 2, browser 204(1) is an ordinary, unmodified browser,
which t e user configures to use predictive-click proxy server 206 as a proxy server for
browsing by entering the IP address of predictive-click proxy server 206 into either a setting
within browser 204(1) or an operating-system setting on client 202(1). Once this setting has
been configured, all HTTP requests and responses between browser 204(1) and the Internet
will pass through predictive-click proxy server 206, which is configured to inspect packets
containing HTTP requests and modify certain packets containing HTTP responses.
A user operates browser 204(1) on client 202(1) to browse the web. During the
browsing process, browser 204(1) requests pages from web servers (not shown) by sending
HTTP requests. Browser 204(1) provides the HTTP requests to predictive-click proxy server
206, where packet inspection and modification module 220 receives and forwards those
HTTP requests to historical page and click tracker 208 and next-click generator 212. Packet
inspection and modification module 220 also forwards those HTTP requests to the
appropriate web servers on the Internet. Those web servers reply by sending the requested
pages, preceded by HTTP responses, to predictive -click proxy server 206, where the web
pages and HTTP responses are received by packet inspection and modification module 220.
As will be described in further detail below, packet inspection and modification module 220
forwards to browser 204(1), without modification, referring web pages for which no most
frequently clicked link exists. However, if a most frequently clicked link exists for a referring
web page, then packet inspection and modification module 220 modifies that referring web
page prior to forwarding it to browser 204(1).
Historical page and click tracker 208 receives the HTTP requests from packet
inspection and modification module 220 and extracts information about which link on each
referring web page the user has clicked (i.e., selected) during browsing. Specifically,
historical page and click tracker 208 extracts, in real time, both (i) page data, i.e., the URL
identifying the referring web page containing the link that was clicked and (ii) click data, i.e.,
identification of which link on the referring web page was clicked. Historical page and click
tracker 208 uses the page data and click data from browser 204(1) data to update page and
click statistics database 210.Next-click generator 212 receives the HTTP requests and extracts page data from
those HTTP requests to determine the URL identifying the referring web page containing the
link that was clicked. Once t e URL has been determined, next-click generator 212 uses this
URL to perform a query on page and click statistics database 210. If there is an entry for the
URL in database 210, then next-click generator 212 receives, as a result of the query, the most
frequently clicked link for the referring web page identified by that URL. If there is no entry
for the URL in page and click statistics database 210, then next-click generator 212 receives,
as a result of the query, an error flag. Next-click generator 212 then provides to browser
packet inspection and modification module 220 either a suggested next click, i.e., the most
frequently clicked link on the referring web page identified by the URL, or else an error flag
indicating that a most frequently clicked link does not exist for the URL.
If packet inspection and modification module 220 receives an error flag for a given
referring web page, then no action is taken by packet inspection and modification module
220, and that referring web page is forwarded to browser 204(1) intact, i.e., without
modification of the referring web page. However, if packet inspection and modification
module 220 receives a suggested next click, then packet inspection and modification module
220 modifies the underlying code (e.g., HTML code) of the referring web page identified by
the URL prior to forwarding that referring web page to browser 204(1), so that the suggested
next click is indicated to the browsing user. Specifically, this modification of the referring
web page's code is performed so that the referring web page visually suggests the most
frequently clicked link to the user by altering the manner in which that link is displayed
within the browser, such as by changing that link's color or size or otherwise modifying the
appearance of that link (e.g., by adding HTML "bold" tags and around the link).
While the foregoing embodiment of FIG. 2 permits the use of an ordinary, unmodified
browser, it is contemplated that, in certain embodiments, instead of altering the HTML code
of referring web pages so that the alteration of the code itself changes the manner in which the
most frequently clicked link is displayed, the HTML code could be altered to add a hidden
field or flag (e.g., by adding an HTML "hidden" tag in the form of: )
that a custom browser (or browser plug-in or add-on) detects to initiate manipulation by the
browser of the appearance of the link on the screen only.
FIG. 3 is a flow diagram illustrating an exemplary method 300 for displaying web
pages to a current user of a web browser running on a computing device. First, at step 30 1,
identification of (i) one or more referring web pages visited by one or more visitors and, (ii)
for each of the one or more referring web pages, data identifying the frequency with which
one or more links on the referring web page were previously selected by the one or more
visitors who visited the referring web page, is stored in a database. Next, at step 302, a
request initiated by the current user to view one of the referring web pages is received. Next,at step 303, a determination is made whether the database contains one or more links on t e
referring web page that were identified as having been selected by the one or more previous
visitors who visited the referring web page. If, at step 303, it is determined that the database
does not contain any such links for the referring web page, then, at step 304, an unaltered
version of the referring web page is displayed in the user's web browser. If, at step 303, it is
determined that the database contains one or more such links, then, at step 305, an altered
version of the referring web page, in which the appearance of those links contained in the
database is altered, is displayed in the user's web browser. Following either step 304 or 305,
the method returns to step 302 to receive another request from the current user to view
another referring web page.
In alternative embodiments of the invention, in addition to, or instead of, the browser
altering the appearance of the most frequently clicked link, the browser could be configured
to move the cursor or mouse pointer automatically to "hover" over the most frequently
clicked link after loading a referring web page, so that the user merely clicks the mouse or
presses a predetermined keystroke to navigate to that link.
In other alternative embodiments of the invention, the most frequently clicked link
could be indicated in other ways, such as by displaying the link itself in a specified location
(such as at the top or bottom of the screen, or in a separate window), or by displaying the
actual linked web page corresponding to that link in a specified location (such as at the top or
bottom of the screen, or in a separate window).
In further alternative embodiments, even if the most frequently clicked link is not
actually displayed on-screen, the browser could be configured so that a predefined keystroke,
mouse gesture, click of an on-screen button in a predetermined location, or similar action by
the user automatically causes the browser to navigate to the most frequently clicked link on
the currently-displayed referring web page, regardless of the then-current location of the
mouse cursor or pointer.
Moreover, although the foregoing described embodiments involve the page and click
statistics database returning only a single most frequently clicked link corresponding to each
URL, it should be recognized that additional information could be retrieved from the database
in other embodiments of the invention. For example, identification of two or more selectable
links corresponding to each URL, coupled with statistical information comprising, e.g., (i) the
number of "clicks" by the one or more users for each of those links and/or (ii) the frequency
with which each of those links was selected by the one or more users, could be maintained in
and retrieved from the database and presented to the user. In this scenario, the database could
provide such statistical information reflecting the popularity of one or more links appearing
on a given referring web page to the next-click generator as part of the query results. The
browser would receive this statistical information (either in the form of a "click" count or apercentage) from t e next-click generator and would be configured (either instead of or in
addition to displaying or otherwise indicating the single most frequently clicked link) to
indicate to the browsing user the percentage of time each of the selectable links was selected
by the one or more users who visited a given referring web page. The popularity for a given
link could be displayed as a number (indicating either "click" count or percentage), such as a
number appearing in a "tool tip" when the user hovers over a link, or indicated using other
means, such as by color-coding the most frequently clicked links with a predetermined color
scheme (e.g., green for the most frequently-clicked links, yellow for less frequently-clicked
links, and red for infrequently-clicked links), so that the user can easily discern, at a glance,
the ranking of links most frequently selected by the one or more users. If the number
displayed for a link represents frequency (e.g., percentage of time the link was previously
selected), then that number inherently reflects not only the popularity of that link based on
historical data, but also the probability that the current browsing user will select that link.
This arrangement provides users with the ability, by simply "mousing" over a link on a
referring web page, to determine the percentage (or number) of people who have selected that
link in the past.
It should also be recognized that certain embodiments of the invention involve
suggesting next clicks to a user based only on that single user's browsing history (or, e.g., the
browsing history of all users who share a particular IP address), while other embodiments
involve aggregating browsing histories for a plurality of users at different client computers to
determine which next click should be suggested to a user. The plurality of users from which
browsing histories are aggregated could be all browsing users on the Internet, or select groups
of users based on, e.g., a social network consisting of the user's personally-chosen friends, a
geographic location (e.g., users in the same neighborhood, state, or country), one or more
common interests shared by a group of users (e.g., chess enthusiasts or users sharing a
particular religion), a particular age group, or the like. In some embodiments, the links
selected by the browsing user himself/herself are weighted more heavily so as to have greater
influence over the aggregated link-selection information than links selected by other users.
When aggregating browsing history data from users, some embodiments of the
invention might count multiple instances of selection of the same link by a single user as
multiple selections, while other embodiments might count multiple instances of selection of
the same link by a single user as only a single selection.
Additionally, different embodiments of the invention involve suggesting next clicks
based on the aggregation of browsing histories from different time frames. For example,
browsing histories from only the past week or the past month might be used to suggest next
clicks. The time windows from which data is used to suggest links can also be adjusted to
emphasize past browsing history (e.g., using historical data aggregated over a long period oftime), to emphasize current browsing history (e.g., for the current session, for t e current
week, for t e last N clicks by that user), or various combinations of past and current browsing
history. In some embodiments, different time periods are weighted differently. For
example, more recently selected links could be weighted more heavily so as to have greater
influence over the aggregated link-selection information than links selected during less recent
time periods.
In different embodiments of the invention, different types of information might be
stored to indicate historical link selections by users. For example, the page and click statistics
database might contain entire specific URLS to indicate links chosen (e.g.,
http://host.domain.com/page.htm), or might contain merely a domain name (e.g.,
host.domain.com) or portion of a domain name (e.g., domain.com) corresponding to a
selected URL, to indicate a most relevant link or set of links. In the latter scenario, the
appearance of all links to that domain name (or portion thereof) on a referring web page
viewed by the browsing user could be manipulated to suggest a plurality of links to the user.
Although embodiments of the invention are described above as being implemented in
a predictive-click processor or a predictive-click proxy server, it is contemplated that the
invention could also be embodied in a router owned by an Internet service provider (ISP) or in
a single personal computer client containing a browser.
Additionally, the information in page and click statistics database could alternatively
be stored in other locations and could be obtained by other methods, such as bulk transfer
from external entities, such as data warehouses, that aggregate historical information
regarding links selected by one or more users during browsing.
It is further contemplated that, in certain embodiments of the present invention that
employ custom software (e.g., a plug-in or add-on software module) to modify a browser,
such custom software could provide additional functionality. For example, the user could
have selectable options to shape the set of historical data used to suggest links, or to add
contextual information to improve the suggestion of links, such as permitting the user to
provide feedback regarding the quality of links that are suggested. If a user indicates (e.g.,
using a predetermined keystroke or mouse gesture) that a suggested link on a referring web
page is not relevant to that user, then such feedback could be used, e.g., to prevent that link
(or other links with the same domain name) from being suggested to the user in the future
when arriving at that referring web page, or even when arriving at any other referring web
pages on which that link appears (or on which other links with the same domain name
appear).
Further, although the data representing information about links selected by users does
not necessarily need to be stored in a standardized format, standardization could be used toenhance the ability to share browsing history data across multiple users and multiple types of
software and platforms.
Methods consistent with embodiments of the invention could alternatively be
implemented in a web site, without the use of a proxy server or any browser customization
(such as a plug-in or add-on software module). In this scenario, the web site would track the
links selected by one or more visitors to various referring pages within the web site and would
provide suggested next clicks to one or more current users of the web site by visually altering
the way certain links on referring pages within the web site appear to those current users (e.g.,
as described above with respect to FIG. 1). A similar implementation could be used for a set
of web sites, a domain, or a set of domains. Links on referring pages could be suggested to a
user either (i) based on links selected by the same user during the user's previous visits to
those referring pages, (ii) based on links selected by other visitors to those referring pages, or
(iii) a combination of links selected by the same user and by other visitors.
In certain embodiments of the invention, next clicks are suggested based on statistics
characterizing the frequency with which links are selected, as described above. However, in
other embodiments of the invention, a user's past or recent browsing history could be used to
determine suggested next clicks based on Bayesian inference, i.e., an approach that makes
statistical inferences in cases where some of the probabilities are interpreted as representing
beliefs, or knowledge, rather than a purely frequency-based interpretation. In other words,
based on the links a user clicks, that user's behavior can be learned and better predicted, so
that future suggested next clicks are tailored to that particular user. Such behavior can be
stored in the form of a user profile, and the stored profile information can be used either alone
or in conjunction with page data and click data to improve the suggestions of next clicks. The
profiles could be user-specific or could characterize certain categories of users.
In the case of profiles that characterize certain categories of users, such profiles are
used to bias suggestions toward content more relevant for the browsing user and away from
content less relevant to the browsing user. For example, certain content-based user-selectable
profiles would be provided so that, e.g., a user who is a new parent of a child but has no pets
could select a profile for new parents and deselect a profile for pet owners, resulting in
improved next-click suggestions. In the foregoing example, suggestions are biased toward
new parent-related links and away from pet-related links, such that the user is presented with
suggestions for links that a typical new parent would select, but is not presented with
suggestions for links that a pet owner would select, even if a pet-related link is otherwise that
"best" next click. By observing the browsing patterns of the user and/or the browsing
patterns of other users who share certain characteristics, improved next-click results can be
provided.In alternative embodiments, instead of next clicks being individually chosen on t e
basis of the frequency with which a single link is selected, joint probabilities are used to
suggest next clicks. In such embodiments, for each web page, the page and click statistics
database stores one or more paths consisting of sequences of links that are frequently selected.
For example, once a user clicks link A on a first web page, that user is likely to select link B
on the next web page, followed by selecting link on the next web page, and so forth.
Another example involves a web site having four pages (A, B, C, and D), with each
page containing four links that have equal probabilities of being selected (i.e., page A has
links A A2, A3, A4, each of which is selected 25% of the time by all users landing on page A;
page B has links Bi, B2, B3, B4, each of which is selected 25% of the time by all users landing
on page B; and so forth). Without the use of additional information, it would be difficult to
suggest any one of links Bi, B2, B3, B4 to a user who lands on page B by clicking link A i on
page A, because of the equal probabilities of selection for links Bi, B2, B3, and B4. This
additional information can be obtained using stored sequences of next clicks in the page and
click statistics database. An example of a stored sequence of next clicks for this exemplary
four-page web site might be A i→B3→Ci, which represents the fact that users who selected
link A i on page A also selected link B3 on page B most (e.g., 100%) of the time, and then
further selected link most (e.g., 80%) of the time. Based on this stored information, it can
be inferred that a user who has just selected link A i on page A will most likely select link B3
on page B (even though only 25% of all users landing on page B from all referring pages
selected link B3) and will then most likely select link d on page C (even though only 25% of
all users landing on page C from all referring pages selected link d). In this manner, the
selection of link A i on page A , link B3 on page B, and link d on page C are correlated based
on the stored sequence of next clicks. Thus, in the above-described scenario of a user landing
on page B, on which four links (Bi, B , B3, B4) would normally have equal probabilities of
being selected, next-click generator would suggest link B3 as the best choice (based on stored
sequences of next clicks in the page and click statistics database), given the fact that the user
landed on page B by clicking on link A . Although storing a large number of sequences of
next clicks in the page and click statistics database could be impractical, storage space can be
reduced by storing only a portion of the sequence of links selected by one or more users, e.g.,
by storing only the most recent 4 or 5 links in a sequence of links selected by a given user.
By storing sequences of links in the foregoing manner, pathways through various web pages
can be suggested to a user based on pathways previously selected by that user or other users,
instead of each link suggestion being made independently based on the current page being
viewed by the user.
Embodiments of the present invention present a different way for a user to find
relevant information on the Internet, different from the experience provided by searchengines. This is done by providing browsing methods that indicate to the browsing user, on a
link-by-link basis, which links through a series of referring web pages are most likely to lead
to t e most relevant path or most popular path. While search engines provide a list of only
one navigable specific link at a time, such as a relevant link buried deep within a complex
web site, search engines provide little or no indication of the path through referring pages
used to arrive at that navigable specific link. The process of link-by-link browsing and
suggestion consistent with embodiments of the invention, which provides the user with
information about paths used to arrive at various links, would be difficult, if not impossible,
for a search engine to provide.
The present invention can be embodied in the form of methods and apparatuses for
practicing those methods. The present invention can also be embodied in the form of program
code embodied in tangible media, such as magnetic recording media, optical recording media,
solid state memory, floppy diskettes, CD-ROMs, hard drives, or any other non-transitory
machine-readable storage medium, wherein, when the program code is loaded into and
executed by a machine, such as a computer, the machine becomes an apparatus for practicing
the invention. The present invention can also be embodied in the form of program code, for
example, stored in a non-transitory machine-readable storage medium including being loaded
into and/or executed by a machine, wherein, when the program code is loaded into and
executed by a machine, such as a computer, the machine becomes an apparatus for practicing
the invention. When implemented on a general-purpose processor, the program code
segments combine with the processor to provide a unique device that operates analogously to
specific logic circuits.
The present invention can also be embodied in the form of a bitstream or other
sequence of signal values stored in a non-transitory recording medium generated using a
method and/or an apparatus of the present invention.
It will be further understood that various changes in the details, materials, and
arrangements of the parts which have been described and illustrated in order to explain the
nature of this invention may be made by those skilled in the art without departing from the
scope of the invention as expressed in the following claims.
It should be understood that the steps of the exemplary methods set forth herein are
not necessarily required to be performed in the order described, and the order of the steps of
such methods should be understood to be merely exemplary. Likewise, additional steps may
be included in such methods, and certain steps may be omitted or combined, in methods
consistent with various embodiments of the present invention.
Although the elements in the following method claims are recited in a particular
sequence with corresponding labeling, unless the claim recitations otherwise imply aparticular sequence for implementing some or all of those elements, those elements are not
necessarily intended to be limited to being implemented in that particular sequence.
Also, for purposes of this description, the terms "couple," "coupling," "coupled,"
"connect," "connecting," or "connected" refer to any manner known in the art or later
developed in which energy is allowed to be transferred between two or more elements, and
the interposition of one or more additional elements is contemplated, although not required.
Conversely, the terms "directly coupled," "directly connected," etc., imply t e absence of
such additional elements.
Reference herein to "one embodiment" or "an embodiment" means that a particular
feature, structure, or characteristic described in connection with the embodiment can be
included in at least one embodiment of the invention. The appearances of the phrase "in one
embodiment" in various places in the specification are not necessarily all referring to the same
embodiment, nor are separate or alternative embodiments necessarily mutually exclusive of
other embodiments. The same applies to the term "implementation."
The embodiments covered by the claims in this application are limited to
embodiments that (1) are enabled by this specification and (2) correspond to statutory subject
matter. Non-enabled embodiments and embodiments that correspond to non-statutory subject
matter are explicitly disclaimed even if they fall within the scope of the claims.CLAIMS
We claim:
1. A computer-implemented method for displaying web pages to a current user of a
web browser running on a computing device, t e method comprising:
(a) storing identification of (i) one or more referring web pages visited by one or
more previous visitors and, (ii) for each of t e one or more referring web pages, one or more
links on the referring web page selected by the one or more previous visitors who visited the
referring web page;
(b) in response to the initiation by the current user of a request to view one of the
one or more referring web pages, and based on the stored identification of the one or more
links on the requested referring web page selected by the one or more previous visitors who
visited the referring web page, altering the appearance of the requested referring web page
displayed to the current user.
2 . The invention of claim 1, wherein step (b) comprises:
(bl) altering the appearance of at least one of the one or more links on the referring
web page selected by the one or more previous visitors who visited the referring web page.
3. The invention of claim 1, wherein step (b) comprises:
(b 1) moving a cursor or pointer to at least one of the one or more links on the
referring web page selected by the one or more previous visitors who visited the referring web
page.
4 . The invention of claim 1, further comprising:
(c) displaying, in a predetermined location in the web browser display, at least one
of:
(i) at least one of the one or more links on the referring web page selected by the
one or more previous visitors who visited the referring web page, and
(ii) at least one linked web page corresponding to at least one of the one or more
links on the referring web page selected by the one or more previous visitors who visited the
referring web page.
5. The invention of claim 1, wherein:step (a) comprises storing at least one sequence of links comprising two or more links
sequentially selected by t e one or more previous visitors who visited the referring web page,
and
the method comprises performing step (b) two or more times for two or more
respective links in the at least one sequence of links.
6. The invention of claim 1, further comprising:
(c) storing, for each of the one or more links on the referring web page selected by
the one or more previous visitors who visited t e referring web page, statistical information
reflecting the number of times the one or more users selected each of the one or more links;
and
(d) in response to the initiation by the current user of the request to view the one of
the one or more referring web pages, and based on the statistical information, altering the
appearance of the requested referring web page displayed to the current user.
7. The invention of claim 6, wherein step (d) comprises displaying the statistical
information for the one or more links corresponding to the one of the one or more referring
web pages.
8. The invention of claim 6, wherein step (d) comprises displaying one or more
colors for the one or more links corresponding to the one of the one or more referring web
pages, wherein the one or more colors are selected based on the statistical information.
9. The invention of claim 1, further comprising receiving at a proxy server, prior to
step (a), the identification of (i) the one or more referring web pages visited by the one or
more previous visitors and, (ii) for each of the one or more referring web pages, the one or
more links on the referring web page selected by the one or more previous visitors who
visited the referring web page.
10. Apparatus adapted to perform a method as recited in any of claims 1-9