(19) United States
`(12) Patent Application Publication (10) Pub. No.: US 2003/0083938 Al
`(43) Pub. Date: May 1, 2003
`
`Smith et al.
`
`US 20030083938A1
`
`(54) SYSTEM AND METHOD FOR PROFILING
`DIFFERENT USERS HAVING A COMMON
`COMPUTER IDENTIFIER
`
`(75)
`
`Inventors: Tucker L. Smith, Irmo, SC (US);
`James L. Browning, Columbia, SC
`(US)
`
`Correspondence Address:
`JAMES M. STOVER, ESQ.
`NCR CORPORATION LAW DEPARTMENT,
`WHQ-4
`1700 S. PATTERSON BOULEVARD
`DAYTON,, OH 45479 (US)
`
`(73) Assignee: NCR Corporation
`
`(21) Appl. No:
`
`10/008,259
`
`(22)
`
`Filed:
`
`Oct. 29, 2001
`
`Publication Classification
`
`Int. Cl.7 ..................................................... G06F 17/60
`(51)
`(52) US. Cl.
`................................................................ 705/14
`
`(57)
`
`ABSTRACT
`
`A system and method profiles different users having a
`common terminal identifier in communication system. The
`
`system comprises a user activity data analyzer for extracting
`profile data from user terminal activity data, a user identifier
`for determining Whether the profile data corresponds to a
`profile data history associated With the user terminal, and a
`user profile generator for generating a profile data history
`from the profile data for another user to be associated with
`the terminal in response to the profile data not corresponding
`to the profile data history associated with the user terminal.
`The user activity data analyzer extracts site addresses,
`URLs, click event data, metadata and other user activity
`from a session log to compile information useful for assess-
`ing a user’s interests. This extracted profile data may then be
`compared to a profile history previously generated and
`associated with the terminal identifier. A low level of cor-
`
`respondence between the extracted profile data and the
`profile history associated with the terminal identifier indi-
`cates that a different user is generating the user activity data.
`The user profile generator then builds a profile history from
`the extracted profile data and associates it With the terminal
`identifier. The profile histories are provided different user
`identifiers. Upon subsequent detection of the terminal iden-
`tifier,
`the profile data extracted from the user activity is
`compared to both profile histories to determine which user
`is navigating the site. Once sufficient profile data has been
`extracted to determine which profile history corresponds to
`the extracted data, advertising content that corresponds to
`the identified user may be selected and included in the
`content requested by the current user.
`
`RECEIVE
`USER
`TERMINAL
`DATA
`
`100
`
`104
`
`108
`
`
`
`EXTRACT
`PROFILE
`DATA
`
`LOCATE
`
`
`KEY DATA
`
`USER
`PROFILE
`?
`
`N
`
`
`
` GENERATE USER
`
`PROFILE KEYS
`
`134
`
`MERGE
`
`PROFILE
`DATA
`
`110
`
`GENERATE
`USER
`
`
`PROFILE
`
`STORE
`USER
`
`PROFILE
`133
`
`END
`
`COMPARE
`OF TERMINAL
`
`PROFILES
`DATA?
`
`
`MERGE
`
`PROFILES
`
`148
`
`STORE
`PROFILES)
`
`1 4
`
`COHRELATED
`?
`
`SELECT CONTENT
`BASED ON
`PREFERENCES
`
`
`
`
`
`
`
`END SESSION
`
`
`
`SONY - Ex.-1017
`
`1
`
`Sony Corporation - Petitioner
`
`SONY - Ex.-1017
`Sony Corporation - Petitioner
`
`1
`
`

`

`Patent Application Publication May 1, 2003
`
`Sheet 1 0f 3
`
`US 2003/0083938 A1
`
`E
`o
`
`9..
`
`P.
`o
`
`g
`
`LU
`:_4
`LL03
`00
`II
`CL
`
`(0
`CC)
`
`1—
`021
`0
`
`55
`
`_
`
`(5
`O
`
`N
`
`.3
`
`
`
`fi-
`l\
`
`O‘—
`
`0L
`
`o
`
`CC
`
`$9
`D
`
`O0
`
`0
`
`O l
`
`\
`
`(D
`B
`
`C)
`(O
`
`F03
`l—g
`
`o
`
`00.
`
`5E
`H:
`
`.C}
`(D
`N
`
`C)
`
`vLO
`
`3
`
`CO
`00
`
`33
`0.203>
`—0:
`Lu
`(/3
`
`fi'
`CO
`
`Lu
`*—
`8
`a:
`
`38
`
`288
`
`PI!
`
`:5
`
`h:
`
`III
`III"
`
`B
`
`Er:
`
`0 N
`
`m
`w-
`
`9
`E
`
`\
`
`@‘\
`
`CU
`
`§
`
`2
`
`

`

`Patent Application Publication May 1, 2003 Sheet 2 0f 3
`
`US 2003/0083938 A1
`
`USER
`ACRVRY
`DATA
`ANALYZE R
`
`CONTENT
`
`
` USER
`PROFILE
`
`
`
`SELECTOR RDBMS
`GENERATOR
`
`
`3
`
`

`

`Patent Application Publication May 1, 2003 Sheet 3 0f 3
`
`US 2003/0083938 A1
`
`RECBVE
`USER
`TERMINAL
`
`DATA
`
`100
`
`EXTRACT
`PRCRLE
`DATA
`
`104
`
`LOCATE
`KEYDATA
`
`108
`
`ITIIE}.23
`
`128
`
`134
`
`USER
`PRCELE
`?
`
`Y
`
`CORRELATED
`?
`
`MERGE
`N
`PROFKE
`
`DATA
`
`
`
`
`GENEEQTE
`GENERATE USER
`
`PROFILE KEYS
`PROFILE
`
`
`
`
`STORE
`USER
`
`120
`PROFILE
`
`
` END
`OF TERMINAL
`
`DATA?
`
`
`138
`
`140
`
`Y
`
`COMPARE
`
`pROFlES
`
`CORREUMED
`
`‘ MERGE
`
`SELECT CONTENT
`BASED ON
`PREFERENCES
`
`PROFLES
`
`148
`
`144
`
`STORE
`PROFILE(S)
`
`END SESSION
`
`4
`
`

`

`US 2003/0083938 A1
`
`May 1, 2003
`
`SYSTEM AND METHOD FOR PROFILING
`DIFFERENT USERS HAVING A COMMON
`COMPUTER IDENTIFIER
`
`FIELD OF THE INVENTION
`
`[0001] This invention relates generally to methods and
`systems for tracking user activity at a terminal on a com-
`munication network and, more particularly, to methods and
`systems for generating user profiles based on user activity a
`communication terminal.
`
`BACKGROUND OF THE INVENTION
`
`[0002] The Internet is a well-known computer network
`that supports the interaction of literally millions of comput-
`ers. Most of these computers communicate through a client/
`server scheme although the peer-to-peer communication
`model is used as well. Although the Internet was originally
`envisioned for the purpose of facilitating the dissemination
`of information between geographically divergent locations,
`many have tried to exploit the electronic commerce capa-
`bility of this network. Few companies, however, have suc-
`ceeded in establishing commercial Viability through revenue
`generated from sales occurring in Internet electronic trans-
`actions alone. Consequently, others have attempted to find a
`commercial opportunity in the prolific communication traf-
`fic that occurs over the Internet for the purposes of infor-
`mation acquisition and dissemination.
`
`[0003] Many Western nations are mass consumer markets
`and vendors Vie for the attention of consumers in order to
`
`interest them in their products and/or services. Advertising
`has moved into every media form as it has arisen, particu-
`larly in Western nations, to present products and services to
`potential consumers in an effort to influence their buying
`decisions. Advertising agencies and vendors spend signifi-
`cant amounts of resources to learn the attributes that identify
`those consumers most likely to purchase particular goods
`and services so the advertising can be placed in the media
`viewed most often these consumers.
`
`[0004] This targeting of consumers having a high degree
`of likelihood in purchasing a company’s products or services
`is very important. As the agencies and companies expend
`significant resources to develop advertising for various
`media, they want to present them where they are most likely
`to be effective. For example, car manufacturers do not buy
`television broadcast time for children’s programs and car-
`toons because the viewers of these shows are not consumers
`
`of automobiles. However, cereal manufacturers do purchase
`this broadcast time because these viewers do influence the
`
`buying decisions for breakfast cereals that are made by their
`parents. Thus, those persons responsible for buying access to
`media venues are very interested in the persons who view
`the various media venues.
`
`[0005] The problem with the Internet and indeed any
`communication network where people are able to view a
`media anonymously is that little or no knowledge about the
`demographics of the viewers is known. One particularly
`frustrating aspect of computer network communication is
`the inability to identify the computers accessing a server or
`peer over the network. For example, most users access the
`Internet through an Internet Service Provider (ISP). In this
`communication scheme, a user couples to an ISP’s commu-
`nication hub by communicating messages and files through
`
`a modem over an analog or digital telephone line. From the
`ISP’s hub, the messages and files may be routed internally
`through the ISP’s intranet
`to servers or peers that are
`subscribers to the ISPs services or the messages and files
`may be routed to the external Internet routers for delivery to
`servers or peers that are not ISP subscribers. In the message
`formats for
`the files and messages, a user’s computer
`includes an Internet Protocol (IP) address to identify the
`user’s computer. However, the ISP typically assigns a dif-
`ferent IP address to a computer each time a user accesses the
`ISP hub and the ISP may change the IP address before
`sending a message or file out to the external Internet in an
`effort to cloak the identity of the user’s computer. Likewise,
`computers known as firewalls and secure routers may also
`alter an IP address for external communications to frustrate
`
`efforts to compromise the integrity of a user’s computer.
`
`In an effort to pierce this level of anonymity, some
`[0006]
`sites utilize the cookie field of Transport Control Protocol/
`Internet Protocol (TCP/IP) messages. “Cookies,” as they
`have come to be commonly known, are identifiers assigned
`by a server or peer and placed in reply messages to another
`computer. Browsers, the application programs for commu-
`nicating over computer communication networks, particu-
`larly those using Hyper Text Transport Protocol (HTTP) for
`Hyper Text Markup Language (HTML) documents, use the
`cookie for subsequent messages to the site that assigned the
`cookie. At the conclusion of a communication session with
`
`a site, the communication program, such as a browser, stores
`the site address and cookie in a file on the user’s computer.
`If the user accesses the site during another subsequent
`browsing period, the browser determines that the site has
`been previously accessed by the user and inserts the stored
`cookie into the messages with the site. If the computer at the
`site has stored the cookies that it has previously assigned, it
`can detect the return of a computer that has previously
`visited. If any information was stored regarding the last time
`the computer identified by the cookie communicated with
`the site, then it can be retrieved and used. For example, if the
`user registered with the site by providing a user name and
`other user data during the previous visit, then, using the
`cookie as a key, the site computer can retrieve this data and
`welcome the user by name without requiring the user to
`login.
`
`If users were willing to register on computer sites
`[0007]
`then advertisers would be able to learn more about the
`
`people who visit computer sites. Unfortunately, most users
`are resistant
`to providing personal
`information. Conse-
`quently, a site computer may be able to detect the occurrence
`of another communication session with a computer to which
`a cookie was previously assigned but little or no information
`about the user can be ascertained. Not until the user provides
`personal data in response to a request
`to complete an
`information form or performs some transaction that accu-
`rately identifies the user will the site be able to obtain
`demographic data on the user. Once a user’s name and/or
`address can be obtained then personal information databases
`may be mined to obtain demographic data to further identify
`the user’s interests and attributes.
`
`[0008] Operators of Web sites or other computer sites on
`computer networks have attempted to exploit the potential of
`advertising to the users who view content delivered from a
`site. Site operators try to attract advertisers and vendors by
`touting the number of “clicks” or “hits” registered at their
`
`5
`
`

`

`US 2003/0083938 A1
`
`May 1, 2003
`
`site. They promote the advertising potential of the site by
`offering to run banner ads in their site content. Advertisers
`evaluate the advertising efiectiveness of a site by examining
`the amount of communication traffic at a site and the content
`
`of the site. Some content is more desirable to a particular age
`group, gender, or income level and, if that age group, gender,
`or income level is not an important consumer of the adver-
`tiser’s goods and services, advertising at the site is not a
`good investment.
`
`[0009] At least two problems make this evaluation diffi-
`cult. For one, some sites act as portals to a variety of content.
`Consequently, advertisers have a difficult time in assessing
`the content that is being most frequently viewed by users.
`Although a portal may have content that would attract new
`car buyers and after market car accessory buyers, if most of
`the hits demonstrate new car buyer visits then an after
`market accessory manufacturer would not do well in adver-
`tising at the site. To address this problem, systems have been
`developed that record and store session logs of a user’s
`activity with a site. These systems typically record the
`Universal Resource Locators (URL) that a user requests for
`viewing at a user’s computer. Using this information, a
`cookie may be associated with a session log and the content
`contained in the URLs of the log. An analysis of the most
`frequently requested logs may be performed to identify the
`most frequently obtained content from the site to assist in the
`identification of the users attracted by the site.
`
`[0010] The second problem for advertisers is information
`regarding the users. While a site operator may have suppo-
`sitions regarding the identity of a user accessing the site, the
`fact remains that all the site really knows is that a cookie
`identifier has been associated with a list of URLs accessed
`
`by the cookie-identified computer. However, there is rarely
`a one-to-one correspondence between a computer and a user.
`Many families use a common computer for all family
`members to access the Internet. Thus, when different family
`members use the computer during different browse periods
`to access the same site, a program at the site analyzing the
`user activity associates the activity with a single user. This
`assumption blurs the demographic distinctions that may be
`ascertained from the content previously viewed through the
`computer. If banner ads are selected based on this prior user
`activity then ads having little or no interest to the current
`user will be displayed. Because hits on banner ads are an
`important statistic for attracting advertisers to a site, a site
`operator wants to place ads before a viewer that are likely to
`result in a hit. Analyzing user activity from a single com-
`puter as if a single user generated it reduces the likelihood
`of accurately identifying the user during a session.
`
`[0011] What is needed is a method of analyzing user
`activity associated with a cookie that differentiates between
`users of a computer.
`
`[0012] What is needed is a system that recognizes different
`users of a single computer and selects appropriate advertis-
`ing for the recognized user.
`
`SUMMARY OF THE INVENTION
`
`[0013] The above-noted limitations of previously known
`systems and methods for analyzing user activity associated
`with a computer have been overcome by a system and
`method that operate in accordance with the principles of the
`present invention. The system for profiling different users
`
`having a common terminal identifier comprises a user activ-
`ity data analyzer for extracting profile data from user ter-
`minal activity data, a user identifier for determining whether
`the profile data corresponds to a profile data history asso-
`ciated with the user terminal, and a user profile generator for
`generating a profile data history from the profile data for
`another user to be associated with the terminal in response
`to the profile data not corresponding to the profile data
`history associated with the user terminal.
`
`site
`[0014] The user activity data analyzer extracts
`addresses, URLs, click event data, metadata and other user
`activity from a session log to compile information useful for
`assessing a user’s interests. This extracted profile data may
`then be compared to a profile history previously generated
`and associated with the terminal identifier. A low level of
`
`correspondence between the extracted profile data and the
`profile history associated with the terminal identifier indi-
`cates that a different user is generating the user activity data.
`The user profile generator then builds a profile history from
`the extracted profile data and associates it with the terminal
`identifier. The profile histories are provided different user
`identifiers. Upon subsequent detection of the terminal iden-
`tifier,
`the profile data extracted from the user activity is
`compared to both profile histories to determine which user
`is navigating the site. Once sufficient profile data has been
`extracted to determine which profile history corresponds to
`the extracted data, advertising content that corresponds to
`the identified user may be selected and included in the
`content requested by the current user.
`
`[0015] The system of the present invention may be imple-
`mented at a content communication site, such as a Web site,
`or at a content communication hub, such as an ISP hub. At
`a content communication site, the system analyzes session
`logs associated with a terminal identifier, such as a cookie.
`At a communication hub, the system may analyze a browse
`period log and the terminal identifier may be a cookie or an
`IP address. Abrowse period log is a database of user activity
`beginning with the initial access of the ISP hub and the
`user’s logout from the ISP hub. Within this browse period
`are the session logs containing the user activity data that has
`occurred between a user and the various content communi-
`
`cation sites visited by the user during the browse period. At
`the communication hub, therefore, the system of the present
`invention may obtain a wider range of profile data than is
`available at a content communication site alone. This addi-
`
`tional information is useful in identifying the interests of
`each user of a computer. Also, the IP addresses and infor-
`mation available to the ISP hub through its support software
`located on a terminal may be useful in further identifying the
`computer and user driving a browse period. For example, a
`laptop identified by the support software may indicate which
`family member is driving a browse period and that person’s
`identity may be used to supplement the data in the profile
`history. The stationary model used in the family house may
`indicate other users and help differentiate those users and
`their profiles from the laptop user even though they all
`access the computer through the same ISP account.
`
`[0016] The system of the present invention may be par-
`ticularly useful in cable television networks (CATV). To
`support view on demand and other user selection services,
`CATV networks are providing channels for receiving com-
`munication from subscriber sites. These channels may be
`used by the processors in set
`top boxes to send data
`
`6
`
`

`

`US 2003/0083938 A1
`
`May 1, 2003
`
`regarding the broadcast channel being viewed on a televi-
`sion set at a subscriber site. Messages indicating the power
`on and power off status of a television set may be used to
`identify browse periods and data regarding duration of View
`for each channel displayed may be evaluated as session data.
`These data may be analyzed to develop a parent profile, a
`husband profile, a wife profile, and group and individual
`child profiles for the viewers at a subscriber site. Comparing
`the extracted profile data to the profile histories associated
`with a subscriber identifier, the system at the head end of the
`CATV network may be able to identify the viewer or Viewer
`group at different television sets at a subscriber site. The
`system may then select different advertising content to be
`delivered to the different sets operating at the subscriber site.
`The ads or interest content may be displayed at a set as a
`scroll window along a border of the screen or as a window
`within screen display.
`[0017] The method of the present invention comprises
`extracting profile data from user terminal activity data,
`determining whether the profile data corresponds to a profile
`data history associated with the user terminal, and generat-
`ing a profile data history from the profile data for another
`user to be associated with the terminal in response to the
`profile data not corresponding to the profile data history
`associated with the user terminal. The method may be
`performed at a content communication site or a communi-
`cation hub in a communication network. The terminals may
`be computers identified by cookies or IP addresses or they
`may be television sets located at subscriber sites and further
`identified by data from set
`top processors. The method
`permits the different users and viewers of the terminals to be
`detected and associated with profile histories for the termi-
`nals. This data may also be used to better target advertising
`and further enhance the advertising potential of a commu-
`nication content site, an ISP service, or a CATV system.
`
`It is an object of the present invention to differen-
`[0018]
`tiate users and Viewers at terminals in a communication
`
`system without requiring registration information.
`
`It is an object of the present invention to identify a
`[0019]
`user during a session for purposes of targeting advertising to
`the user.
`
`It is an object of the present invention to build a
`[0020]
`user profile during a session in response to extracted profile
`data having a low degree of correspondence with existing
`profile histories associated with a terminal.
`
`It is an object of the present invention to mine
`[0021]
`session and browse period data to detect different users of a
`terminal.
`
`invention is the
`[0022] One advantage of the present
`ability to profile more users from hits at a content commu-
`nication site to improve the advertising potential of the site.
`
`[0023] Another advantage of the present invention is the
`ability to more accurately profile users from hits at a content
`communication site.
`
`[0024] Another advantage of the present invention is the
`ability to differentiate users or groups of users who access a
`content communication site through a secure access device
`such as a firewall.
`
`[0025] These and other objects, advantages and features of
`the present invention may be discerned from reviewing the
`accompanying drawings and the detailed description of the
`invention.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`[0026] The present invention may take form in various
`system and method components and arrangement of system
`and method components. The drawings are only for pur-
`poses of illustrating an exemplary embodiment and are not
`to be construed as limiting the invention.
`
`[0027] FIG. 1 depicts a diagram of communication net-
`work in which the present invention may be used;
`
`[0028] FIG. 2 is a block diagram of an exemplary embodi-
`ment of a system for differentiating users at a terminal that
`is shown in FIG. 1;
`
`[0029] FIG. 3 is a flowchart of an exemplary method for
`differentiating users at a terminal of the system shown in
`FIG. 2; and
`
`[0030] FIG. 4 is a diagram of a CATV network in which
`the present invention may be used.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`[0031] A system in which the system and method of the
`present invention may be used is shown in FIG. 1. System
`10 includes client computers 12, 16, and 20 that are coupled
`through telephone networks 24a and 24b and Point-of-
`Presence (POP) servers 28a and 28b to the local area
`network (LAN) or wide area network (WAN) 30 of an
`Internet Service Provider (ISP) 34.
`ISP 34 is coupled
`through a router 38 to a wide area network commonly
`known as the Internet and designated with the numeral 40 in
`the figure. By way of example, system 10 also includes
`terminals 44a-44m of a LAN/VVAN 48 that are coupled to
`network 40 through a firewall or other known secure server
`50. Terminals 44a-44m may also act as clients in system 10.
`Numeral 54 designates another ISP network that couples
`client computers C1-Cm to Internet 40. Web server 60 is
`coupled to network 40 to provide access to content stored in
`database servers 64 and 66 through Computer Gateway
`Interface (CGI) 70. CGI 70 is also coupled to a content
`server 74. Also coupled to CGI 70 is a user differentiating
`system 80 that includes a user profile database 84.
`
`In more detail, client computers 12, 16, 20, C1-Cm,
`[0032]
`and terminals T1-Tm may be any conventional known
`personal computers, work stations, desktop, portable, hand-
`held, set top, personal digital assistants, or wireless com-
`munication devices that may couple through an ISP, other
`service provider or communication hub to any wide area or
`local area network such as the Internet or an intranet. These
`
`devices typically include a processor, storage, programmed
`instruction for operating the device and a display. The
`examples shown in FIG. 1 may use a modem or network
`interface device to couple to a communication network such
`as a telephone or satellite network. Messages to and from the
`client devices may then be provided to an ISP or other
`service provider network. The ISP usually provides sub-
`scriber support services through one or more service servers
`36 such as email services, search engines, and other known
`services. Subscriber service providers include ISP compa-
`nies, commerce service providers (CSP), cellular and other
`wireless service providers who couple wireless communi-
`cation devices to communication networks such as the
`Internet or intranet communication networks. These various
`
`service providers couple users through computers and other
`
`7
`
`

`

`US 2003/0083938 A1
`
`May 1, 2003
`
`communication devices to a communication network for the
`
`purpose of exchanging information requests and the docu-
`ments and files responsive to those requests.
`
`[0033] Content for these information requests may come
`from other client users or from content communication sites
`
`such as Web server 60. Web servers typically establish and
`manage a session with a user. The requests for data and reply
`messages from a client device are typically provided to
`another computer for processing such as CGI 70. CGI 70
`typically executes script programs to locate resources for
`transmission to the requesting clients from databases located
`in proximity to CGI 70 such as databases 64 and 66 or from
`remote databases. These resources are typically identified by
`Universal Resource Locators (URL) that are included in the
`request and reply messages from client devices.
`
`In establishing the communication session with a
`[0034]
`client device, Web server 60 typically searches a data field
`in the request or reply for cookie data. Cookie data is
`identifying data placed in a message returned to a client
`device by a Web server. The communication application
`program in a client
`is commonly known as a browser.
`Browsers support communication over a communication
`network through a service provider in a variety of protocols.
`Although the HTTP protocol and documents in the HTML
`language/format are the ones most commonly used in many
`communication networks, other protocols and language/
`formats are typically supported. Most known browsers store
`in the client device the cookie data that is assigned by a
`server at a content communication site to the client device
`
`and insert the cookie data into data requests for subsequent
`communication sessions with the content communication
`site. In this manner, most servers at content communication
`sites may identify client devices communicating with them
`that have previously visited the site.
`
`In previously known systems, systems for gener-
`[0035]
`ating user profiles used the cookie data to recognize a client
`device that has subsequent communication sessions with the
`content communication site. These systems collect the user
`activity data such as length and time of a session, resources
`accessed in a session, a log of the mouse events occurring at
`the user client device, and transaction data entered by the
`user at the site. This data may be obtained by monitoring the
`messages returned to the site from the client device or the
`site may embed or otherwise download an application
`program with a document sent
`to the client device that
`executes on the client device and collects user activity. This
`user activity data may then be returned to the system at the
`site for data analysis and user profile generation. For
`example, a data object or a Java applet may be downloaded
`for user activity data collection purposes. If the user accepts
`a plug and play activity monitor, the application program for
`collecting user activity data may not only collect data but
`store the data on the user’s storage drives and monitor user
`activity data at other content communication sites. This
`stored data may then be returned to the content communi-
`cation site that downloaded the plug and play component.
`User activity components may also be part of the browser
`application program to obtain the same advantages as the
`plug and play modules. Regardless of the collection method,
`these previously known systems have evaluated the col-
`lected data as if they were generated by only one user. These
`systems perform Bayesian model analysis such as general
`Bayesian networks, hidden Markov models, and decision
`
`trees derived from Bayesian networks. Other statistical and
`probabilistic models are known and used to analyze the user
`activity data to construct a user profile for a single user. The
`limitation of these previously known systems is that they do
`not evaluate the user activity data to determine whether the
`user activity comports with a user profile associated with the
`client device. Thus, the analysis done on the user activity is
`used to alter the user profile history and that affects its
`accuracy. Only if the user has previously provided registra-
`tion data, such as a user name, account number or other
`specific user identifying data, are these previously known
`systems able to determine that a user activity corresponds to
`a user other than one for which a profile history has been
`previously generated. The system and method of the present
`invention overcomes this limitation of previously known
`systems.
`
`[0036] User identifying system 80 of FIG. 1 is shown as
`a separate system coupled to CGI 70 and Web server 60
`although such depiction is illustrative only. System 80 may
`be incorporated within a web server or CGI or distributed
`over one or more computers at a content communication
`site. Additionally, system 80 may be coupled to the ISP
`service server 36 or incorporated in such a server at the ISP
`site to collect and analyze user activity from subscriber sites
`to differentiate between users at the same site. When located
`
`at the ISP site or a content communication site, system 80
`may be used to identify which user is operating a client
`device during a session or browse period and the user’s
`profile history may be provided to a content server for
`purposes of providing targeted marketing to the user. Thus,
`a content server 74 may be provided at either or both ISP
`sites for communication with a system 80 at the site to
`provide targeted content such as advertising to the subscrib-
`ers of the ISP or
`to external users accessing publicly
`available documents from the ISP. System 80 may also be
`provided at any of the clients in FIG. 1. The user activity
`data extractor and analyzer may be program components or
`data objects that are downloaded from a content communi-
`cation site, a program component of the browser application,
`or a plug and play or other application program that pref-
`erably executes and communicates with the browser during
`network communication sessions. Of course, such an imple-
`mentation only extracts data and develops profiles for the
`users who access the computer on which system 80 executes
`but this data may be downloaded to a content server coupled
`to the communication network during browse periods. The
`content server may then select targeted content such as
`marketing documents for transmission to the client device.
`[0037] As shown in FIG. 1, system 80 is coupled to a
`profile history database 84 for storage of user profile histo-
`ries. Database 84 may be a relational database that
`is
`managed directly by system 80 or by a relational database
`management system (RDBMS). Database 84 may also be an
`object repository or other known managed data storage
`scheme. If memory or storage space on the components of
`system 80 permits, database 84 may be incorporated in
`system 80 or one of the other computers at the site where
`system 80 operates. By way of example, system 80 may be
`comprised of a Pentium II processor having sufficient RAM
`and hard disk storage, such as 64M of RAM and 20 GB hard
`drive. A read/write CD ROM drive may also be utilized for
`storage requirements. The operating system for system 80
`may be any known operating system such as Macintosh O/S,
`OS/2, AIX, BE OS, one of the Windows operating systems,
`
`8
`
`

`

`US 2003/0083938 A1
`
`May 1, 2003
`
`LINUX, UNIX, or the like. Communication between the
`computers at the site where system 80 is located may be
`performed through any known network method such as
`TCP/IP or other known Internet, intranet, LAN, or WAN
`implementation of a network protocol. The hardware speci-
`fication for the communication may be via Ethernet (IEEE
`802.3), Token Ring (IEEE 802.5), Fiber Distributed
`Dat