U.S. patent application number 14/867077 was filed with the patent office on 2016-09-01 for identification of user segments based on video viewership activity and analysis.
The applicant listed for this patent is Visible Measures Corp.. Invention is credited to Brian J. Shin.
Application Number | 20160255415 14/867077 |
Document ID | / |
Family ID | 54149767 |
Filed Date | 2016-09-01 |
United States Patent
Application |
20160255415 |
Kind Code |
A1 |
Shin; Brian J. |
September 1, 2016 |
Identification of user segments based on video viewership activity
and analysis
Abstract
The techniques disclosed herein facilitate online audience
targeting based on brand and product interest/enthusiasm as
determined by a measurement metric that is a function of one of
more (and preferably all) of the following: recency data, frequency
data, intensity data, consumption data, site data, meta data
related to given videos, data, and other demographic data.
Inventors: |
Shin; Brian J.; (Boston,
MA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Visible Measures Corp. |
Boston |
MA |
US |
|
|
Family ID: |
54149767 |
Appl. No.: |
14/867077 |
Filed: |
September 28, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
12987620 |
Jan 10, 2011 |
9148706 |
|
|
14867077 |
|
|
|
|
61293327 |
Jan 8, 2010 |
|
|
|
Current U.S.
Class: |
725/34 |
Current CPC
Class: |
H04N 21/8456 20130101;
H04N 21/25891 20130101; G06Q 30/00 20130101; H04N 21/44222
20130101; H04N 21/4782 20130101; H04N 21/812 20130101; G06Q 30/0269
20130101; H04N 21/4667 20130101 |
International
Class: |
H04N 21/81 20060101
H04N021/81; H04N 21/4782 20060101 H04N021/4782; H04N 21/442
20060101 H04N021/442; H04N 21/466 20060101 H04N021/466; G06Q 30/02
20060101 G06Q030/02; H04N 21/258 20060101 H04N021/258 |
Claims
1. Apparatus, comprising: a processor; computer memory holding
computer program instructions that when executed by the processor
perform an advertising method, comprising: receiving data generated
from tracking video viewing behavior and consumption of related
content; based on the received data, generating at least one metric
that correlates content being viewed with a given product/brand
that is desired to be promoted; and using the metric to facilitate
a behavioral targeting decision.
2. The apparatus as described in claim 1 wherein the metric is a
function of one of: recency data, frequency data, intensity data,
consumption data, site data, metadata related to given videos, and
combinations thereof.
3. The apparatus as described in claim 1 wherein the method further
includes generating a profile associated with the metric.
4. The apparatus as described in claim 3 wherein the profile is
forwarded to an ad server or an ad network.
5. Apparatus, comprising: a processor; computer memory holding
computer program instructions that when executed by the processor
perform a method, comprising: receiving video viewing data; using
the video viewing data to determine an enthusiasm metric for one
of: a particular brand, a product, an activity, and an interest;
and using that enthusiasm metric for ad targeting.
6. The apparatus as described in claim 5, wherein the video viewing
data includes user-generated video content.
Description
[0001] This application includes subject matter that is protected
by copyright.
[0002] This application is based on and claims priority to Ser. No.
61/293,327, filed Jan. 8, 2010.
BACKGROUND OF THE INVENTION
[0003] 1. Technical Field
[0004] The present invention relates generally to Internet video
audience behavior measurement, consumption, tracking, and
reporting.
[0005] 2. Description of the Related Art
[0006] Many destination web sites and aggregators offer web-based
services that host videos for content publishers and that allow
audiences to directly consume those video clips either on their
websites, or via other sites, blogs, or social networks to which
their content is linked and or embedded. To be successful, it is
desirable for site owners to understand their audience and their
consumption habits so that a site's appeal and stickiness to end
users and others (such as potential advertisers) can be tracked,
managed, and optimized. Additionally, video and rich media formats
are being used more and more for interactive advertising campaigns.
Creatives that are provided in such formats need to be measured for
audience behavior as well to ascertain the effectiveness of such
campaigns.
[0007] Behavioral targeting has long held the promise of improving
the delivery of suitable advertising to a specific consumer.
Traditional methods of behavioral targeting online generally
involve tracking which websites and which web pages a web user
visits, and then utilizing this information to infer things about
the user, such as the user's level of interest in buying a car, the
user's interest in given information, and the like.
BRIEF SUMMARY
[0008] This disclosure relates to behavioral targeting based on a
deep analytics into user behavior around video (viewing patterns)
combined with an approach to categorizing and organizing videos
into relevant segments related to brands, products, activities,
lifestyles or otherwise. The technique enables targeting of users
based on their video viewing patterns and the content and subject
matter of the videos they watch.
[0009] Preferably, the technique leverages "learning"
user-targeting and scoring algorithms that combine various inputs,
such as the number of videos consumed, how recently these videos
were consumed, the subject matter of the videos, the intensity and
engagement patterns of the interaction that the user had with the
videos, as well as relevant metadata around the videos including,
without limitation, the channels on which the videos were shown,
the environment of the user platform (web, mobile, or otherwise) on
which the videos were watched, and the users themselves (if no
permission, then no personally identifiable information used; with
permission, user profile information then is leveraged to improve
targeting).
[0010] In this manner, a user can be identified as a target user of
interest and be served an advertisement in conjunction with
watching videos on one or more sites or as he or she visits other
sites distinct from those where videos were watched.
[0011] The techniques disclosed herein facilitate online audience
targeting based on the "enthusiasm" they exhibit for a particular
brand, product, activity, interest or other criteria. Preferably,
enthusiasm is determined by one or more algorithms that are a
function of one of more (and preferably all) of the following:
recency data, frequency data, consumption data, intensity data,
site data, content data, meta data related to the videos and
related content, and other demographic data.
[0012] The techniques disclosed herein provide a novel technique to
segment and target an audience brand/interest/product enthusiasm
based on video viewing behavior across video, sites, and the like,
in a privacy-compliant manner.
[0013] The foregoing has outlined some of the more pertinent
features of the invention. These features should be construed to be
merely illustrative. Many other beneficial results can be attained
by applying the disclosed invention in a different manner or by
modifying the invention as will be described.
BRIEF DESCRIPTION OF THE DRAWINGS
[0014] FIG. 1 is a block diagram of an audience tracking system for
use to provide behavioral targeting according to the subject matter
herein;
[0015] FIG. 2 illustrates how a measurement metric is generated
according to the techniques herein;
[0016] FIG. 3 illustrates how an audience overlap analysis tool
identifies unduplicated unique users;
[0017] FIG. 4 illustrates how an audience overlap analysis tool
generates a cross-site matrix;
[0018] FIG. 5 is a representative content affinities matrix
generated by an audience analysis tool;
[0019] FIG. 6 illustrates a first view of segmentation of usage
intensity, which is data representing relevant videos viewed per
user;
[0020] FIG. 7 illustrates a second view of segmentation of usage
intensity, based on sites on which relevant videos are viewed per
viewer;
[0021] FIG. 8 illustrates a third view of segmentation of usage
intensity, based on sessions per unique viewer;
[0022] FIG. 9 illustrates a fourth view of segmentation of usage
intensity, based on videos viewed session;
[0023] FIG. 10 is an illustrative service provider infrastructure
that provides analytics as a managed service; and
[0024] FIG. 11 is an illustrative system for collecting end user
video viewing usage data.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
[0025] Marketers are always searching for better and more effective
ways of reaching their target audience. It would be desirable to
provide a new type of targeting that can enable the identification
of users who may (but need not) be "enthusiasts" of a particular
brand, in-market buyers for a particular product, or practitioners
of a particular activity (or other use case where audiences
exhibiting a particular level of enthusiasm or interest is being
targeted). Preferably, the identification of such enthusiasts is
carried out by tracking and analyzing video viewing habits, as will
be described.
[0026] "Enthusiast marketing," for example, can provide significant
benefits. As an example, if an advertising (ad) targeting system
could identify users who were very interested in a certain product
or brand, e.g., a user who is very interested in a specific Lexus
automobile, that information might be leveraged in many useful
ways. Thus, e.g., the manufacturer itself (Lexus) could demonstrate
"gratitude" and offer some sort of loyalty incentive. A competitor
(e.g., Audi) might like to be "considered" as an alternative option
for later purchase and could thus provide some comparative
advertisement. Lastly, if an entity (e.g., a specific dealership)
is interested in converting that user to become a customer, it
could employ "conquest marketing" (or other such) techniques in an
effort to convert the interest to a real purchase. All of these
options, and many more, could be available to marketers and the
agencies that serve them if they had the ability to more granularly
and accurate identify user interest, segments, enthusiasm, and
behavior. The approach for leveraging video viewing data that is
now described enables this type of improved targeting.
[0027] As seen in FIG. 1, an audience tracking system as described
herein provides the high-level functions including audience
behavior tracking 101 and category classification 103, with the
result being an audience profile database 105. The database stores
data in the form of a unique algorithm calculation (preferably
based on many inputs) referred to herein as a VScore. This
designation is provided solely for convenience of explanation and
not by way of limitation. In general, a VScore can be thought of as
a measure of how strongly a user is interested in a particular
thing, whether it is a brand, a product, an activity or otherwise.
One possible use case of this VScore data would be as follows; Ad
server(s) 107 access the audience tracking system through an
audience filter 109 to obtain an audience profile based on a
VScore, which profile is then used to inform the ad server as to
which ad or ads should be delivered to the ad network or publisher
that will display the ad or cause the ad to be displayed to the
user.
[0028] As seen in FIG. 2, a VScore 201 preferably is influenced by
a number of factors, including, without limitation, recency data
203, frequency data 205, engagement data 207, content metadata,
site data, and other demographic data (e.g., geography or other
metadata 209). It is not required that data from each factor is
included to generate the VScore, but in a preferred embodiment this
is the case. Generally, recency data 203 is a measure of how
"active" the user has been based on his or her recent video viewing
behavior over given time periods (e.g., 1, 2, 30, 60, 90, 120 and
180 days). The frequency data 205 measures a number of video clips
viewed within a given category (e.g., manufacturer's brand, model,
make, interest, lifestyle attribute, and the like). The engagement
data 207 measures intensity, such as how long the user spent
viewing clips within a given category, brand loyalty (e.g.,
established through referral behavior), what actions the user took
within a session, and the like. The geography and other metadata
209 measures the location of the target audience, the user's
platform or other local environment, the channels on which the
videos have been consumed, and other demographic data and metadata
(e.g., title, content type) if available.
[0029] As a skilled artisan will appreciate, analysis and
categorization of the content/advertising viewed is critical to
understanding the interests and intent of the users. Ideally, a
targeting system such as this should take into account videos that
are: user generated, professionally-generated or otherwise;
advertisements, editorial content, news, or otherwise; original
versions of content, copies of the originals, or derivatives,
remixes, mash-ups, user reviews, fan-submitted videos, and the
like. Essentially, if all such videos are tagged and categorized
appropriately, and all the viewing behavior against these videos is
similarly organizing and analyzed, the resultant data can provide
insights into the interests/intent and characteristics of the
viewing audience. This is the goal of the subject matter disclosed
herein.
[0030] One of the important implications of this type of targeting
system is that essentially all video on the web can have value.
Currently, a significant percent of web video clips are
"user-generated" (UGC) and as such are not generally easily
monetizable. These videos, however, often contain by their nature
tremendous information about the passions and interests of their
creators, viewers and distributors. These videos, as well as
others--together with related content/tweets/blog posts, etc.--all
can provide the inventive system relevant data to be used for
algorithmic learning about audience enthusiasm.
[0031] FIG. 1 illustrates how the inventive targeting technique may
be employed. Of course, it is assumed that users watch videos (in
this case, videos preferably are watched all across the web, on any
internet-enabled device). This user behavior is monitored by a
measurement and analytics system (as will be described below) and
tracked against which content was watched, where it was watched,
when, in what type of environment, how it was watched, and so
forth. This is the audience behavior tracking 101. Next, content
and advertising are organized into segments, preferably based on
the content, target segment, or other metadata associated with the
content. This is the category classification 103. Some examples
would include organizing videos based on which products they
reference, or which types of activities are featured. Thereafter,
one or more audience profiles are developed, e.g., and these
profiles are based upon many inputs and involve algorithms that are
improved and trained over time based upon feedback mechanisms. The
profiles are generated and stored in the database 105. The data
points that could be leveraged in calculations may include one or
more of: recency, frequency, intensity, consumption, content, user
metadata, and the like, which are used by the VScore calculation,
as has been described. The audience profile data and scoring are
then called by an ad server 107 or ad network 113 to determine if
the user is of a particular type or should be served a particular
type of ad (in this case, show a Ford ad to an enthusiast of a
competing brand). The resultant user activity is then tracked, by
conversion tracking system 111, to provide feedback into the system
for enhancement.
[0032] The advertising server (ad server), on-line advertising
network (ad network), and conversion tracking systems referenced in
FIG. 1 are well-known in the prior art. Familiarity with such
technologies, systems, and business models are assumed in this
disclosure.
[0033] To truly gather behavioral data at scale, preferably the
system monitors the viewing behavior of large numbers (e.g.,
millions) of users across hundreds or thousands of video properties
for millions of videos, and billions of events and data points.
With all this data for this many users, it is desirable to develop
a set of algorithms that can help categorize the video content
viewed and the users doing the viewing. This is the purpose of the
VScore calculation algorithm, which factors in many data points to
determine how much a user is interested in a particular thing,
whether it is a brand, a product, activity or otherwise.
[0034] This widespread measurement of users and content provides
highly relevant data. As mentioned previously, preferably the
VScore combines such items as an understanding of the content, the
user, the environment, viewing patterns over time, as well as the
channels across which the content is viewed; the algorithm then
looks at the intersections of such various data.
[0035] Many different types of audience analysis tools may be used
in accordance with this invention to analyze behaviors at a "user"
dimension. An audience overlap analysis identifies what audiences
from other sites come to a particular site, e.g., "People who watch
on Site.sub.x also watch on Site.sub.y." As seen in FIG. 3, a first
aspect of an audience overlap analysis is to identify unduplicated
"uniques." In particular, given a site (or a domain, or sub-domain,
etc.), or a collection of sites, the system obtains a list of users
who have visited the site or sites. A baseline set may also be
used. As indicated in the table of FIG. 3, the system then compares
user lists to determine the total uniques and the uniques per list.
Another aspect of audience overlap may involve an investigation of
cross-site overlap, based on the premise that viewers may be more
likely to convert if they view videos on specific "pairs" of sites.
This type of audience overlap analysis evaluates which site pairs
drive the highest conversion rates. To that end, an analysis
involves creating an "n x n" matrix of all sites mapped against one
another. Each "cell" in the matrix then represents the overlapping
audience of a first Site and a second Site. Then, analyze common
metrics for pairs of sites (overlapping audience behavior) with
respect to one or more of: viewers, views, engagements, viewing
time, percentage completion, conversions, and the like. The goal of
the technique is to determine what "packages of sites address
similar audiences. An example matrix is shown in FIG. 4.
[0036] Another technique for audience analysis utilizes so-called
"Content Affinities," which determine what sites have audiences
that have strong affinity to a given site. One way of describing
this concept is that "Affinity is the increased likelihood of
viewing on Site.sub.x given went to Site.sub.y." Content
Affinities, as the name implies, are causal relationships between
content types, such as the following illustrative example: "People
who like content of type X also like content of type Y." The VScore
algorithm can make use of such data in the following, non-limiting
manner. Assume by way of example that the following affinity
assertions/data is available to the system:
[0037] "People who like X also like Y"
[0038] "Probability of viewing content on Site.sub.x is p(Sx)=Total
views on Site.sub.x/Total views on all sites"
[0039] "Probability of viewing content on Site is P(Sy)=Total views
on Site.sub.y/Total views on all sites"
[0040] "Probability of viewing on Site.sub.x & Site.sub.y is
p(Sx & Sy)=Total views on Site.sub.x & Site.sub.y/Total
views on all sites"
[0041] "Probability of viewing on Site.sub.x given viewed on
Site.sub.y is p(S.sub.x|S.sub.y)=p(S.sub.x &
S.sub.y)/p(S.sub.y), which is also equal to (p(S.sub.x) x
p(S.sub.y|S.sub.x))/p(S.sub.y)"
[0042] Therefore, Affinity=p(S.sub.x|S.sub.y)/p(S.sub.x)
[0043] Based on these assertions, an affinity matrix is then
developed, as shown in FIG. 5. For a given site, the affinity
matrix is used to develop a pareto graph of its "row."
[0044] Another form of audience analysis is audience
"segmentation," which groups visitors into (monetizable) buckets
based on their activities, e.g., to attempt to identify high versus
low intensity users. This determination may be made based on what
patterns each exhibit. Thus, for example, a set of "usage
intensity" segmentations of a viewer across many videos and
channels may be generated. This information can then be viewed in
many different ways, such as, without limitation, relevant videos
viewed per user (FIG. 6), sites on which relevant videos are viewed
(FIG. 7), sessions per user (FIG. 8), and videos viewed per session
(FIG. 9). Using this approach, a given user may be characterized as
"low" (passer-by), "medium" (regular) or "high" (enthusiast).
Audience segmentation may be investigated on any meta data
attribute (per account, per content type, etc.) by identifying
behaviors for viewers in each identified "segment" (or "bucket").
The "usage intensity" segmentation is not meant to be limiting, as
other techniques, such as content affinity segmentation, may be
used as well.
[0045] This approach provides deep analytics into user behavior
around video (viewing patterns). These analytics are then combined
with an approach to categorize and organize videos into relevant
segments related to brands, products, activities, lifestyles or
otherwise. The resulting metric is a VScore, which is then adapted
into a "profile" that is used to facilitate targeting of users
based on their video viewing patterns and the content and subject
matter of the videos they watch.
[0046] The VScore algorithm combines various inputs, such as the
number of videos consumed, how recently these videos were consumed,
the subject matter of the videos, the intensity and engagement
patterns of the interaction that the user had with the videos, as
well as relevant metadata around the videos including, without
limitation, the channels on which the videos were shown, the
environment of the user platform (web, mobile, or otherwise) on
which the videos were watched, and the users themselves (with or
without PII depending on permission). In this manner, a user can be
identified as a target user of interest and be served an
advertisement in conjunction with watching videos on one or more
sites or as he or she visits other sites distinct from those where
videos were watched.
[0047] A service provider or system architecture may be used to
implement the audience tracking system. FIG. 10 illustrates a
representative service provider or system architecture, which in
one embodiment is implemented in or across one or more data
centers. A data center typically has connectivity to the Internet.
The system provides a preferably web-based hosted solution through
which service customers (e.g., advertisers, content publishers,
publishing networks, site owners, video owners, and the like) view
Internet video and rich media end user experience analytics in an
online manner. Participants may interact with the platform as a
hosted service. In an alternative embodiment, the system may be
implemented over a private network, or as a product (as opposed to
a hosted or managed service).
[0048] A user of the service (e.g., an advertiser, content
publisher, publishing network, site owner, or the like) has an
Internet accessible machine such as a workstation or notebook
computer. Typically, the user accesses the service provider
architecture by opening a web browser on the machine to a URL
associated with a service provider domain or sub-domain. The user
then authenticates to the managed service in the usual manner,
e.g., by entry of a username and password. The connection between
the machine and the service provider infrastructure may be
encrypted or otherwise secure, e.g., via SSL, or the like. Although
connectivity via the publicly-routed Internet is typical, the user
may connect to the service provider infrastructure over any local
area, wide area, wireless, wired, private or other dedicated
network. As seen in FIG. 10, the service provider architecture 100
comprises an IP switch 102, a set of one or more web server
machines 104, a set of one more application server machines 106, a
database management system 108, and a set of one or more
administration server machines 110. A representative web server
machine 104 comprises commodity hardware (e.g., Intel-based), an
operating system such as Linux, and a web server such as Apache
2.x. A representative application server machine 106 comprises
commodity hardware, Linux, and an application server. The database
management system 108 may be implemented as an Oracle database
management package. In a high volume use environment, there may be
several web server machines, several application server machines,
and a number of administrative server machines. Although not shown
in detail, the infrastructure may include a name service, other
load balancing appliances, other switches, network attached
storage, and the like. The system typically will also include
connectivity to external data sources, such as third party
databases. Each machine in the system typically comprises
sufficient disk and memory, as well as input and output devices.
Generally, the web servers 104 handle incoming business entity
provisioning requests, and they export a display interface that is
described and illustrated in more detail below. The application
servers 106 manage the data and facilitate the functions of the
platform. The administrator servers 110 handle all back-end
accounting and reporting functions. The particular hardware and
software implementation details described herein are merely for
illustrative purposes are not meant to limit the scope of the
present invention.
[0049] As noted above, preferably the system includes a measurement
and analytics system. FIG. 11 illustrates a basic operation of an
Internet video experience measurement service that may be used for
this purpose, although this implementation is not meant to be
limiting. In this example, end user video players 200a and 200b
have been instrumented with a usage monitoring API. The usage
monitoring API is sometimes referred to as a remote node. Thus,
video player 200a has associated therewith the remote node 202a and
video player 200b has associated therewith the remote node 202b. Of
course, the use of two instances is merely illustrative, as the
system is designed to provide a highly scalable distributed logging
service wherein a large number of instances of the video player or
rich media application are instrumented and tracked. In operation,
the remote node 202a generates usage data set 204a, and remote node
202b generates usage data set 204b. This usage data is transported
to a central server (or to a set of servers), where the data sets
are aggregated (reference numeral 206) and processed within an
analytics and reporting engine 208.
[0050] Although not required, a usage monitoring API may be
instrumented into (or otherwise associated with) a video player to
collect end user usage data. A representative video player is a
Flash video player, although the principles of this disclosure are
not so limited, as the techniques described herein may be
implemented in any type of player. For Flash video players, a
monitoring component, preferably packaged as an extension, is
installed in the video player (e.g., using the Macromedia Extension
Manager). This component is implemented as an ActionScript library
that provides application programming interfaces (APIs) for
developers to integrate their players with the tracking mechanisms
of this disclosure. In one embodiment, developers directly insert
relevant stubs into the code to track whichever events they are
interested in. Using the monitoring system, events are dispatched,
aggregated automatically, processed, and reported. With this
approach (the remote node implemented as API calls), developers are
afforded control over the events and the level of granularity they
wish to track. An instrumented player typically collects data about
"system" events and "user" events. System events occur, for
example, when a player is loaded, when a video is loaded into the
player, when the video begins playing, when the video is paused,
when the video finishes scrubbing (i.e.., after it is forwarded or
rewound), or when the video ends. User events occur, for example,
when the user hits the pause button, scrubs the playhead (i.e.,
move the clip forward or backward), ends the scrub (i.e., stops
moving the clip forward or backward), emails the video or clicks a
button to get a video's embed code, or the like. In this approach,
a set of ActionScript methods are called whenever a system or user
event occurs. When an event occurs, API methods typically are
combined with one another. A system event and a user event often
occur simultaneously and might require two API methods be called at
once. A Statistics Collector Timer (SCT) keeps track of a clip's
playhead to ensure that accurate metrics are calculated. When a
clip is paused, the SCT is paused. When a clip is scrubbed, the SCT
is informed of the playhead's new position. The one or more system
or user events are heard using ActionScript event listeners. An
event listener is a piece of code that listens for a defined event
and invokes the API method associated therewith.
[0051] Preferably, the system has the capability for tracking video
viewing behavior of an individual viewer for instrumented video
players across given video sites, given content owners, all video
players, all versions of video players, and all browser types and
all browser versions. Preferably, a single tracking component is
used for all sites, content, players, and browsers, and preferably
a globally unique identifier for each user is used to track each
user's activity.
[0052] The hardware and software systems in which the invention is
illustrated are merely representative. The invention may be
practiced, typically in software, on one or more machines.
Generalizing, a machine typically comprises commodity hardware and
software, storage (e.g., disks, disk arrays, and the like) and
memory (RAM, ROM, and the like). The particular machines used in
the network are not a limitation of the present invention. A given
machine includes network interfaces and software to connect the
machine to a network in the usual manner. As illustrated in FIG. 8,
log server may be part of a managed service (e.g., in an ASP model)
using the illustrated set of machines, which are connected or
connectable to one or more networks. More generally, the service is
provided by an operator using a set of one or more
computing-related entities (systems, machines, processes, programs,
libraries, functions, or the like) that together facilitate or
provide the inventive functionality described above. In a typical
implementation, the service comprises a set of one or more
computers. A representative machine is a network-based server
running commodity (e.g. Pentium-class) hardware, an operating
system (e.g., Linux, Windows, OS-X, or the like), an application
runtime environment (e.g., Java, .ASP), and a set of applications
or processes (e.g., Java applets or servlets, linkable libraries,
native code, or the like, depending on platform), that provide the
functionality of a given system or subsystem. As described, the
service may be implemented in a standalone server, or across a
distributed set of machines. Typically, a server connects to the
publicly-routable Internet, a corporate intranet, a private
network, or any combination thereof, depending on the desired
implementation environment.
[0053] The hosted service may be implemented in a multi-server
cluster environment that is designed to scale efficiently. Each
server is designated with a primary and secondary series of tasks.
Preferably, one server is dynamically set to be a master server,
which server determines the secondary tasks to be performed by all
servers. All servers update their existence within a database, and
the servers cooperate to determine which server will be the master.
The servers in the cluster are assigned tasks (such as log import
and event processing) by the master server.
[0054] Having described the invention, what is now claimed is set
forth below.
* * * * *