||||||l|||||||||||||||||||||||||||||||||||||||l||||||||||||||||||||||||||||
`
`U50064385?9Bl
`
`(12) United States Patent
`US 6,438,579 B1
`(10) Patent N0.:
`Hosken
`
`(45) Date of Patent: Aug. 20, 2002
`
`(S4) AU'H)MA’I'EI) CONTENT AND
`COLLABORATION-BASED SYSTEM AND
`METHODS FOR DETERMINING AND
`PROVIDING CONTENT
`RECOMMENDATIONS
`
`('15)
`
`Inventor: Benjamin H. l-Insken, Hawthorn (AU)
`
`(73) Assignee: Agent Arts, Inc., San Francisco, CA
`(US)
`
`(* ) Notice:
`
`J
`Y
`Sub'ect to an disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(6) by 0 days.
`
`(21) Appl. No.: 091616374
`
`(22
`
`Filed:
`
`Jul. 14, 2000
`
`(60)
`
`Related U.S. Application Data
`1299.
`Provisional application No. 60(‘144,3??. filed on Jul. 16,
`
`
`(test? [5116
`Int. (:1.7
`(51)
`7091203; 709;"202; 709.!217;
`(52) U.S.Cl.
`7t]9.t2t8; 709,219; 709.224; 7091229; 70732;
`7078; 707.5; 707110
`709t200—203,
`(58) Field of Search
`709(1217—229; ?05r’?—10, 14, 26—27; 70?.‘9—10,
`l-—6, 102403; 235.875, 380, 462
`
`(56)
`
`References Cited
`U.S. PA‘I‘EN'I‘ DOCUMENTS
`
`8,0993 Kaplan
`5,237,15? A "
`liltlciq? Baker ct al.
`5,678,041 A *
`TU'HZ
`3?]998 Rose cl 3|.
`5,?2456? A "‘
`
`?05.t'26
`6?]999 Bernard el ai.
`5,918,213 A *
`?05t26
`5,963,916 A * 10.0999 Kaplan
`
`709.3218
`5,991,799 A "‘ 11r'1999 Yen et a].
`3,-‘2000 Chislenko ct al.
`6,041,311 A *
`705,2?
`
`?05,.-'10
`
`4:2000 Sheena et al.
`6,049,??? A *
`7,-2000 Chislcnko et al.
`6,092,049 A "
`8,.“2000 Bergh el al.
`6,112,186 A *
`6,330,592 Bl
`* 12.900] Makuch ct al.
`6,334,127 Bl
`12mm Bieganskictal.
`FOREIGN PATENT DOCUMENTS
`
`..
`
`
`
`?El5{'10
`?05.-'10
`705.110
`.. ?t)‘)?21?
`rims
`
`EP
`W0
`W0
`
`A10969469
`W0 Al 9963458
`W()A2t’110130?
`
`701998
`5?1998
`6tl999
`
`006F?1?i'60
`(3061711780
`GOoFI'IWfifl
`
`* cited by examiner
`
`Primary Exerrtiner—Bharat Barot
`(74) Attorney, Agent, or Finn—Gerald B. Rosenberg; New
`Tech Law
`
`(57)
`
`ABSTRACT
`
`A content and collaborative filtering system for recommend-
`ing entertainment oriented content items, such as music and
`video, and other media content items to a user based on
`similarity in profile between the user and other users and
`between the content indexed in the user's profile and other
`content
`in the database. The system stores implicit and
`explicit ratings data for such content items provided by the
`users. Upon request of the user, the system accesses the
`user’s profile and corresponding content interests database.
`The system uses the relationships between the content items
`to determine a subset of the content items to he referred to
`the user. The system also correlates a similarity between the
`user’s ratings of the content items and other users' ratings.
`Based on the correlations. a subset of users is selected that
`is then used to provide recommendations to the user. The
`recommended items have a high probability of being sub—
`jectively appreciated by the user. The recommendations
`produced by the system will be represented to the user using
`a visual representation of the relationships between the
`content items allowing the user to explore the items related
`to the recommended items.
`
`22 Claims, 8 Drawing Sheets
`
`
`
`OTHER BEHAVIOURS
`
`
`GROUP
`
`
`BEHAvIouRs
`
`
`
`EXPERT
`FINAl
`
`
`
`WEIGHI‘ING
`WEIGHTING
`
`
`
`FILTER
`FILIER
`
`
`
`
`
`Usrrt
`BB
`
`
`BEHMOUR
`
`
`
`
`EXPLICIT
`
`BEHAVIOURS
`
`LYFT 1024
`
`LYFT 1024
`
`1
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 1 0f 8
`
`US 6,438,579 B1
`
`
`
`
`! ! ! ! !I i j ! 1| i j
`
`2
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 2 0f 8
`
`US 6,438,579 B1
`
` SYSTEM
`
`PROCESSES
`
`ITEMS
`
`INFO
`
`42
`
`
`RELATIONSHIP
`
`
`INFORMATION \
`
`36
`
`CONTENT
`
`26
`
`F'G- ‘8
`
`OTHER BEHAVIOURS
`
`60
`
`
`/’—\ 50
`
`COlLECTED
`GROUP
`BEHAVIOURS
`
`
`EXPERT
`
`
`WEIGHTING
`WEIGHTING
`PURCHASE
`
`
`
`FILTER
`REQUESTS
`FILTER
`
`
`52
`
`
`IMPLICIT
`USER BROWSE
`
`
`
`
`BEHAVIOURS
`ACTION
`
`7O
`
`
`68
`
`
`USER
`BEHAVIOUR
`
`USER
`PROFILE
`
`REFERRAL
`SYSTEM
`
`EXPLICIT
`BEHAVIOURS
`
`USER INPUT
`ACTION
`
`62
`REQUESTS
`
`RECOMENO
`SET
`
`72
`
`3
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 3 0f 8
`
`US 6,438,579 B1
`
`FIG. 3
`
`BROWSING
`Acnoms
`
`88
`
`BEHAVTOURS
`
`IMPLIED
`
`EXPLICIT
`BEHAVIOURS
`
`
`INTERVIEW!
`STATEMENTS
`84
`
`
`WELGHTE D
`
`BINARY
`
`
`WELGHTED
`
`WEIGHTED
`RELATION
`
`
`
`COLLECTION
`
`
`RELATION
`
`
`
`WEIGHTED
`
`WEIGHTED
`RELATTON
`
`BINARY
`RELATION
`
`WEIGHTED
`
`WEIGHTED
`
`..
`
`
`
`
`
`FIG. 6
`
`4
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 4 0f 8
`
`US 6,438,579 B1
`
`IMPLICIT RA'HNG INFORMATION SOURCES
`
`LL)
`
`".Al
`
`I-
`
`COLLECTION '
`
`FIG. 5
`
`U')
`L“
`UE
`I)
`OU)
`
`Z 9I
`
`—<
`
`(EE 0L
`
`I.
`
`2 (
`
`DZI
`
`— E'
`
`: QEX
`
`
`
`
`
`
`
`SHDEJHOSNOILVINUOdNISNILVE]HEIHIO
`
`5
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 5 0f 8
`
`US 6,438,579 B1
`
`
`
`
`CLEAR RESULT
`TABLES
`
`-
`
`GET LIST OF
`FAVOURITE ITEMS
`INTO FAVOURITE
`ITEMS TABLE
`
`FIG. 7A
`
`GET STYLE TO BASE
`RECOMMENDATION
`
`
`
`
`GET NEXT ITEM
`
`
` GET NEXT VECTOR
`No
`
`
`TRANSLATE FAVOURITE
`
`
`
`TABLE INTO VECTOR Fu
`
`START WETH FIRST ITEM IN
`
`FAVOURITES TABLE
`
`DETERMINE POSITION OF ITEM
`USING WEIGHT
`
`is IN TOP FAVOURITES?
`
`DD To INPUT
`TABLE
`
`START mm FIRST
`VECTOR IN CLUSTER
`TABLE AS C;
`
`
`
`RESTRICT VECTOR
`DEMENSIONS IF NEEDED
`
`DETERMTNE
`CORRELATION OF
`
`C| WITH FU
`
`IS CLUSTER TABLE
`
`A5 C"
`
`«YES
`
`ADD TO TARGET
`CLUSTER TABLE
`
`ES IT
`HIGHEST
`CORRELATION?
`
`NO
`
`.YE
`
`NO
`
`6
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 6 0f 8
`
`US 6,438,579 B1
`
`ADD TARGET CLUSTER
`LINKS TO USER PROFILE
`
`CLUSTER TABLE
`
`START WITH FIRST
`CLUSTER IN TARGET
`
`FIG. 7B
`
`
`SEARCH USER VECTOR
`TABLE FOR USER IN
`TARGET CLUSTER
`
`
`
`USE R
`
`
`VECTOR FOUND?
`
`
`
`GET NEXT CLUSTER IN
`
`TARGET CLUSTER TABLE
`
`START WITH FIRST USER
`
`VECTOR UVi
`
`DETERMINE
`GET NEXT USER VECTOR
`
`CORRELATION BETWEEN
`UV.
`
`
`
`UVIAND FLJ
`
`I5
`
`
`CORRELATION ABOVE
`
`THRESHOLD?
`
`ADD DIFFERENCES To
`CTOR RESULTS TABLE
`CORRELATTON *UV
`As WEIGHT
`K
`
`
`
`7
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 7 0f 8
`
`US 6,438,579 B1
`
`START WITH FIRST
`ITEM IN INPUTS
`
`SEfichrESCESENT
`RELATED ITEMS
`
`GET NEXT ITEM
`
`TABLE USING INPUT ITEM
`
`
`
`YES
`
`Is
`
`
`RELATIONSHIP WEIGHT ABOVE
`THRESHOLD?
`
`ADD CONTENT AND
`RELATIONSHIP WEIGHT TO
`CONTENT RESULTS TABLE
`
`YES
`
`NO
`
`ADD CONTENT RESULTS AND COLLABORATE.
`
`RESULTS USE WEIGHT AS
`RESULT WEIGHT
`
`F IG. 7 C
`
`EMOVE DUPLICATES WITH
`EXISTING RATED ARTISTS
`FROM RESULTS
`
`RESULTS
`
`SORT AND DISPLAY
`
`YES
`
`DISPLAY
`"NO RESULTS
`FOUN D”
`
`8
`
`

`

`US. Patent
`
`Aug. 20, 2002
`
`Sheet 8 of 8
`
`US 6,438,579 B1
`
`19803 DANCE
`
`/ BILINGUAL
`
`NEW ORDER
`
`PET SHOP BOYS
`
`FIG. 7D
`
`BRITISH POP
`
`WEAK RELATIONSHIP
`
`STRONG RELATIONSHIP
`
`VERY STRONG RELATIONSHIP
`
`9
`
`

`

`US 6,438,579 B]
`
`l
`AUTOMATED CONTENT AND
`COLLABORATION-BASED SYSTEM AND
`METHODS FOR DETERMINING AND
`PROVIDING CONTENT
`RECOMMENDATIONS
`
`This application claims the benefit of U.S. Provisional
`Application No. 6(Il14—4,377, filed Jul. 16, 1999.
`BACKGROUND OF THE INVENTION
`l. Field of the Invention
`
`invention is generally related to the
`The present
`collection, processing, and presentation of alternative in for-
`mation source content
`to a user and,
`in particular,
`the
`selective and automated generation of source content alter—
`natives based on content relationships and user behavioral
`patterns to support the recommendation of alternative eon—
`lBl'll SUUFCfiS.
`
`10
`
`15
`
`2
`Another known system recommends particular content
`items based on the given content or style of the item. Such
`systems are generally established by hand, requiring a broad,
`yet detailed, understanding of each media item. Establishing
`even basic knowledge—based systems requires a substantial
`investment in time and other costs. 'l‘t'terefore, these systems
`typically employ simplistic relationships between items,
`such as broad categories, such as Drama and Comedy, for
`relating content. Since these categories contain large num-
`bers of content items, any user selection against the catego-
`ries is likely to return an also large set of recommendations
`and, therefore, is unlikely to be significantly useful to a user.
`Finally, both of these existing systems produce recom—
`mendations that are eflectively final end-points in the rec-
`ommendation search. No clear ability is provided for users
`to explore further items related to the recommendations.
`Thus, the user is often left with recommendations, which are
`almost correct, but which don’t raise the user's propensity to
`consume to the level
`required to purchaselconsume the
`content.
`
`SUMMARY OF THE INVENTION
`
`Therefore, a general purpose of the present invention to
`provide a system that combines content-based filtering and
`progressively refined collaborativevbased filtering to deliver
`a set of media item recommendations that are consistent
`with a user’s personal media content interests.
`This purpose is achieved in the present
`invention by
`providing a system and method of providing media content
`recommendations through a computer sewer system eonv
`neeted to a network communications system. The computer
`server system preferably has access to a first database of
`media content
`items including media content and related
`information and a media content filter identifying and pro—
`viding qualifying attribute relationship data for media con-
`tent
`items within the first database. The media content
`recommendations are particularly tailored to the persona]-
`ized interests of a user through sequence of steps including
`presenting media content
`items through a network-
`connected interface to the user for review and consideration
`of potential personal interest, monitoring the consideration
`ofthe media content items implied through the user directed
`navigation among the presented media content
`items and
`user requests for related information; collecting the moni-
`tored data to develop a user weighted data set reflective of
`the user’s relative consideration of the media content items;
`and evaluating the user weighted data set in combination
`with the media content filter to identify a set of media
`content
`items accessible from the first database for
`re-presentation to the user.
`Thus,
`the operation of the present system reflects the
`consideration that media content
`items, such as music,
`video, and other forms of content, can be interrelated based
`on multiple characterizing attributes. The strength of these
`characterizing attributes, or similarities, is used to further
`define these content-based relationships, even as between
`quite diflerent forms or types of media content. An addi-
`tional aspect of the operation ot'the present invention allows
`for the progressive or continuing collaborative,
`including
`self—collaborative, development of such content-based rela—
`tionships.
`An advantage of the present invention, therefore, is that
`the provided combination of content and collaborative rec-
`ommendation systems enables the delivery of recommen-
`dations that are particularly tailored to the personalized
`interests of a user.
`
`2. Description of the Related Art
`There are an increasing number of typically entertainment
`oriented media items, such as music, books, videos, and
`other content sources, available for purchase by users. A
`currently existing system, available to at least some users, is
`capable of presenting the details of over 300,000 individual
`music compact discs alone for purchase by a user. The
`collection of source content is growing with the continual
`addition of new content titles as Well as the development and
`adoption of new content technologies, such as MP3, digital
`music software. Thus, a potential purchaser faces a signifi—
`cant investment of time and expense to comfortably select
`an appropriate item for purchase.
`Existing source content selection systems are quite inef-
`fective in supporting content searches much beyond using
`artist, collection, and title. Users therefore typically coniine
`their searches to just those media items that are indepen- ,
`dently known to them or are aware of through other sources
`of media information. These other sources are typically
`suflicient
`to provide indications of whether and which
`segments of the general population might appreciate parw
`ticular content items. No indication is given and none can be
`reliably inferred as to whether a particular user will enjoy or
`appreciate a given item.
`There is, at least for entertainment media content, some
`acceptance of the belief that a user’s appreciation of par-
`ticular content items can suggest the user’s likely apprecia~
`tion of other content
`titles. Systems built to exploit
`this
`belief have met with limiter] results. One known system,
`apparently a neural-net based expert system, determines and
`provides recommendation of other content
`titles based
`purely on the similarities between users without considering
`the relationships between the music items from a content or
`contextual point of view. These systems have the disadvan-
`tage that they require an initial “teaching” period where the
`recommendations given to users are likely to be inaccurate.
`Another disadvantage is that the user does not understand
`the reasoning behind the recommendations and therefore
`does not trust the recommendations. The absence of confi-
`
`25
`
`3f]
`
`40
`
`50
`
`55
`
`recommendations are given directly
`dence in whatever
`reduces the utility of the system. Additionally, such systems
`tend to generate recommendations that reflect
`the lowest
`common denominator between broad users tastes. As a
`result,
`these systems typically provide recommendations
`reflecting potential appreciation within a single generic
`style, such as only 1980’s pop music. These systems do not
`appear to be effectively capable of providing recommenda-
`tions across a diverse range of music, such as Death Metal
`and Classical.
`
`60
`
`65
`
`10
`
`10
`
`

`

`US 6,438,579 B1
`
`3
`the
`invention is that
`Another advantage of the present
`system flexibly determines a scope of applicable similarities
`between a particular and other users and recommends items
`within the applicable scope.
`A further advantage of the present invention is that the
`self—collaborative relationships developed for
`individual
`users of the system permit the development of individualn
`ized recommendations even where the group collaborative
`relationships reflect the choices of users with highly diverse
`media content interests.
`
`Stiil another advantage of the present invention is that the
`system enables multi—level media content relationship infor—
`mation to be captured and used as data evaluateable in
`providing particularized media content item recommenda-
`tions.
`
`invention is that
`Yet another advantage of the present
`implicit and explicit collaborative data is captured from and
`in consideration of particular users, supporting both the
`continuing development of both group and personal interest
`profiles. The implicit collaborative data is advantageously
`obtained from a user’s self-directed actions of reviewing and
`considering different media content items. Thus, the selec-
`tion of items to review and the length and nature of the
`consideration of such items inferentially reflects the user’s
`relative interest
`in particular media content
`items. Confi-
`dence levels in the inferences drawn can also be developed
`and refined through the continued monitoring of user actions
`in reviewing and considering the same and closely similar
`media content items. The explicit information provided by
`users regarding the level and nature of their interest
`in
`different media content
`items provides highvoonftdence
`information that can be incorporated into the group and
`individualized collaborative data.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`These and other advantages and features of the present
`invention will become better understood upon consideration
`of the following detailed description of the invention when
`considered in connection with the accompanying drawings,
`in which like reference numerals designate like parts
`throughout the Iigurcs thereof, and wherein:
`FIG. 1A provides an overview ofthe logical hardware and
`system implementation,
`including navigational user
`interface, of a preferred embodiment of the present inven—
`tion;
`IB provides a process overview of a preferred
`FIG.
`embodiment of the present invention;
`FIG. 2 is a detailed block diagram detailing the system
`operation ofthe personalized referral system implemented in
`accordance with a preferred embodiment of the present
`invention;
`FIG. 3 provides a block diagram detailing the collection
`and compilation of behavioral data in accordance with a
`preferred embodiment of the present invention;
`FIG. 4 illustrates the collection and correlation of infor-
`mation gathered from multiple information sources as may
`be utilized to establish profiles of individualized and group
`behaviors as a basis for determining and providing recom-
`mendation sets in accordance with a preferred embodiment
`of the present invention;
`FIG. 5 provides a representation of corresponding por-
`tions of individualized user profile data sets reflecting the
`strength and confidence of relationships between particular
`media content items and users;
`FIG. 6 is a graph representation of the media content
`characterization attribute network utilized in accordance
`
`10
`
`15
`
`25
`
`3f]
`
`40
`
`50
`
`55
`
`60
`
`65
`
`4
`with a preferred embodiment of the present invention to
`develop individualized media content
`item recommenda-
`tions; and
`FIGS. 7n, 7b, 7c, and 7D provide flowcharts of a preferred
`system operation, detailing the collection and processing of
`user input and the presentation of resulting recommenda—
`tions back to the user.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`The present invention operates to provide users with a
`source of recommendations for difierent media content
`items that may then he purchased or otherwise acquired by
`a user. These media content items are broadly any poten—
`tially consumable unit of content that can be characterized
`by content attributes. The content may be presented,
`sampled, used, and consumed in any of an open set of
`presentation formats, including audio and visual works,
`streaming and static pictorial images and clips, documents
`and reference materials alone or associated with other con—
`tent. In the exemplary case of audio content, media content
`items may be music samples, song tracks, and albums and
`CDs, which may also be referred to as collections. Music
`videos, cover art, and liner notes may be treated as inde-
`pendent media content items separately consumable or as
`components of song tracks and collections as may be
`appropriate.
`As illustrated in FIG. 1A, a preferred embodiment 10 of
`the present invention provides for the development of media
`content item recommendations within the scope of a trans
`action performed over a communications network, such as
`the Internet. The system and methods of the present inven-
`tion preferably provide for a user, operating a user computer
`system 12 with a network access supported interface 14,
`such as a conventional Web browser application, to access
`and navigate, via a communications network 16, through
`information presented by a server computer system 18.
`Preferably,
`the Web browser 14 operated by the user
`includes or is augmented with plug-ins and applications
`supporting the presentation of streaming audio and video
`data as may be returned from the server computer system 18
`to the user computer system 12.
`Recommendation and navigational requests are presented
`effectively by the user to a referral system 20 within the
`server system 18. Explicit profiling data provided by the user
`and implicitly derived from referral system 20 processes are
`preferably processed 22 and stored 24 by the server system
`18. This explicit and particularly the implicit profiling data
`gathered is then used to provide individualizing recommen-
`dations for particular users. The profiling data collected from
`individuals is also preferably combined to form a collabo—
`ratively developed basis for modifying and expanding on the
`individualized recommendations that might be otherwise
`produced by the referral system 20.
`In the currently preferred embodiment of the present
`invention, an expert compiled database 26 of content item
`relationship information is used as another basis for gener-
`ating media content item recommendations. This database
`26 preferably specifies logical connections between different
`media content items based on the sharing or similarity of
`characterizing attributes. In the case of music~type audio
`media content, these characterizing attributes maybe recog—
`nized as the empirically defined genre distinctions that occur
`between different music content
`items. These distinctions
`may be identified as belonging within some generic catego-
`ries or styles, such as orchestral, blues, and pop, and perhaps
`
`11
`
`11
`
`

`

`US 6,438,579 B1
`
`5
`
`ll]
`
`15
`
`5
`within somewhat more descriptive categories, such as 1980s
`Dance, Rock Anthems, and Techno-Ambient Synth Mixes.
`The level of distinction utilized in connection with the
`present invention is empirically determinable, based largely
`on the availability of detailed relationship characterization
`data and the processing power and throughput restrictions of
`the server computer system 18.
`The content
`item relationship database 26 preferably
`stores relative weighting factors that serve to establish the
`strength of the relationships identified in the database 26
`between different media content items. These weightings,
`along with the establishment of the difierent distinguishable
`characterizing attributes are also preferably compiled by
`experts or expert systems. A preferred database 26 suitable
`for use with the present
`invention may be obtained comfi
`mercially from All Media Guide, 301 East Liberty, Suite
`400, Ann Arbor, Mich. 48104, a subsidiary 01. Alliance
`Entertainment Corporation, 4250 Coral Ridge Drive, Coral
`Springs, Fla. 33065.
`The referral system 20 thus operates from a user provided
`request, typically identifying some media content item or
`artist,
`individual and collaborative profiles 24, and the
`content relations 26 to provide a set of recommended media
`content
`items that are believed likely to be of particular
`interest to the user. As preferably presented in the browser
`14,
`the user may variously navigate the set of
`recommendations, including requesting samples of particu-
`lar content items. A database of content samples 28 may be
`provided as part of the server computer system 18 directly
`or, in the contemplated preferred embodiment of the present
`invention, as a logical component of the server system 18
`supported or hosted externally by a source provider, content
`management, or other party. In either event,
`the content
`samples are returned to the user browser 14 for presentation
`to the user. Based on the review and consideration of the ‘
`recommendation set, including as applicable any presented
`content samples, the user may request a further search and
`production of a new recommendation set, typically identi-
`fying a prior recommendation media content item as part of
`the request, or request the purchase and delivery of a media
`content item.
`
`25
`
`3f]
`
`40
`
`In accordance with the present invention, the user navi-
`gation of a presented recommendation set and the user
`actions in reviewing and considering individual and groups
`of media content
`items are utilized in the progressive
`modification and refinement of the proliles data 24. The
`navigation events received by the server system 18 and the
`requests for additional information and content samples 28
`are readily monitored. Other information can be derived
`from periods of user non-action, particularly after some
`media content
`item information or content samples is
`requested. That
`is,
`the amount of time spent by a user
`apparently reviewing some biographical information about a
`particular media item, or the time spent listening to a music
`clip provides implicit
`information regarding the interest
`level of the user in a particular media content item. By
`extension, this implicit level of interest can also be used to
`imply a likely level of interest in other media content items
`with similar characteriring attributes. The implicit informa-
`tion gathered from user actions is preferably processed 22
`and stored as an addition and refinement of the profile data
`24 previously stored.
`Where the review and consideration of some recom—
`mended media content
`item prompts a user purchase
`decision, a user may execute an electronic purchase trans-
`action (not shown} leading to the delivery 30 of the chosen
`media content
`item to the user computer system 12. The
`
`50
`
`55
`
`60
`
`65
`
`12
`
`6
`delivered media content item is preferably obtained from a
`third-party Content l'ullillment server or other similar ser-
`vice. The delivery component 30 may be implemented by a
`Separate distribution service provider or by the content
`fulfillment service provider.
`An overview of the process implemented in a preferred
`embodiment of the present invention is shown in FIG. 2. The
`process 36 operates to encourage users 38 to provide source
`information 40 as at least the initial basis for directing the
`production of a recommendation set. This information 40
`may provide express indications of the interest
`level
`in
`different types and instances of media content and media
`content items, such as media tracks, artists, and collections.
`These indications or ratings are stored for both general use
`in connection with the production of recommendation sets
`for all users and specifically in regard to productions for the
`respective users. The ratings are preferably stored as user
`profiles 24.
`Input requests from a user 38, such as requests to lind
`media content in some way similar to an identified media
`content item, are submitted for processing through system
`processes 42 to produce a responsive recommendation set
`back to the use 38. Pre—defined content relationships 26 are
`retrieved and evaluated in connection with the system pro-
`cesses 42. The actions of the user 38 in browsing recomv
`mendalion sets are also considered by the system processes
`42 as retlecling, at least to some degree, the interests of users
`regarding particular media content
`items presented in the
`recommendations sets. These reflected and thus implied
`levels of interest are preferably quantified and qualified by
`the system processes 42 with the resulting information being
`incorporated into the user and group profiles 24.
`In a preferred embodiment of the present invention, the
`system processes 42 utilize a number of work tables 44
`through the process of preparing recommendation sets.
`These work tables 44 provide temporary and modifiable
`storage of intermediary relations between potentially rec—
`ommendable media content
`items. Thus,
`in a preferred
`embodiment of the present invention, the system processes
`42 may take multiple approaches to generating a recom-
`mendation set and subsequently combine the results of these
`approaches to produce the recommendation set presented to
`the user 38. Once such intermediary results approach may
`consider the content relationships between items the user 38
`has rated as highly interesting, or enjoyable, and other items
`identifiable through the content relations database 26 as
`having similar characterizing attributes. Another intermedi-
`ary results. approach may concentrate first on correlating
`user profiles as a basis of media content items rated highly
`or broadly that are not identifiably known to the user 38.
`In the first case, a user 38 selects a media content item
`known and of interest to the user from a master list of media
`content
`items. The selection is submitted to the system
`processes 42 for autonomous consideration against
`those
`items identifiable through the content relations database 26
`that are linked by some association, such as particular or
`cumulatively considered characterizing attributes,
`to the
`media content
`item selected by the user 38. The content
`relations database 26 provides qualifying information,
`reflecting the strength or weight of each attribute
`relationship, as well as identifying the linking relationships.
`As the product of this autonomous consideration, the system
`processes 42 produce a set of media content items that, as
`considered, have the strongest relationship connections to
`the user selected media content item or items based on this
`specific evaluation approach. Preferably, this intermediary
`set is temporarily maintained in the work tables 44.
`
`12
`
`

`

`US 6,438,579 B1
`
`8
`
`TABLE l-continued
`Data Tables
`
`
`
`Table Description
`
`10
`
`7
`In the second case, the user 38 may choose a known and
`well-regarded media content
`item from a master list of
`media content items. The system processes 42 operate on the
`selection by autonomously searching through the available
`users profiles 24 with the purpose of identifying those
`profiles reflecting similar ratings of media content items
`rated by the user 38. The identified profiles are then corre—
`lated against the profile of the present user on the basis of the
`commonly rated media content
`items. Preferably.
`these
`correlations may be represented as vector relationships,
`which are stored in the work tables 44-. The vector relation-
`ships are preferably then used as a basis for predicting the
`likely interest level of the user 38 to other media content
`items not
`rated by the user 38. The degree of profile
`correlation and the relative strength of the relationships
`between the media content
`items known and apparently
`unknowu to the user 38 aid in defining the likely level of
`interest represented in the vectors. This further processing
`preferably results in the generation of intermediary data sets
`temporarily maintained in the work tables 44.
`The intermediary result sets produced by these and any
`other intermediary approaches to generating the final rec-
`ommendation set are then considered as a group by the
`system processes 42. The intermediary sets are preferably
`joined and autonomously ordered based on any number of
`factors, potentially including frequency of inclusion in dif-
`ferent intermediary sets, the relative order of media content
`items in the different
`intermediary sets, and the attribute
`related strengths of the inter-relationships between the
`media content items.
`
`40
`
`Once the intermediary sets are joined and correspondingly
`ordered,
`the system processes 42 may trim the resulting
`recommendation set
`list
`to a manageable number and
`present the recommendations to the user 38. The presented
`recommendation list is preferably stored temporarily in the ”
`work tables 44. A history of the recommendation sets
`presented to a user may also be recorded in or stored in
`connection with the user prolile. Additionally, the level of
`interest
`in particular recommended media content items.
`particularly as can be inferred through the browsing of such
`recommendations in accordance with the present invention,
`is stored as part of the user profile. Thus, the media content
`items considered or reviewed in connection with the user
`browsing actions are at least implicitly rated by the user and
`stored to the user profile, which substantially extends and
`enhances both the individual and group—oriented basis for
`correlating user profiles. As user profiles are so extended, the
`effectiveness of the system processes 42 in generating
`recommendation sets improves.
`invention,
`For a preferred embodiment of the present
`multiple data tables within the work tables 44 are employed
`to store information used in formulating content-oriented
`and collaboration-oriented media content item recommenw
`dations. While other logical data representations can be
`readily used, the following table organized representations
`are preferred.
`
`TABLE l
`Data Tables
`
`60
`
`
`
`Table Description
`
`1. Favorites
`
`Internally stores identifying information about media
`content items selected andi'or rated by a user including
`rating weight and rating confidence information. For
`example, an external representation of this table or a set
`
`65
`
`2. Target
`Clusters
`
`Profile
`
`of linen tables may be used to store particular media
`content items, user lists of artists and collections
`of interest. and other tables generally organized by
`the user to reflect characterizing attributes of media
`content items and set of such items that are of some
`particular interest to the user. Preferably, an in-
`memory temporary table with persistent database storage.
`internally contains categorization details of user groups
`preferably on the basis of the strength of inlcrcst
`relative to some distinguishing characterizing attributes.
`This information can be used as an index to improve the
`performance of collaborative-oliented intermediary
`production of media content item recommendations.
`Preferably, an in-meniory temporary table with persistent
`database storage.
`internally contains identifying information reflecting the
`information contained in user profiles, including
`characterizing attribute and media content ratings, for the
`mcdia content and media content items linkcd to a user.
`The information in this table is preferably derived from
`explicit rating information provided by the user and
`through implicit observations performed by the system
`against user browsing acti