US20120311034A1

Movatterモバイル変換

Info

Publication number: US20120311034A1
Application number: US13/153,376
Authority: US
Inventors: Adam GOLDBAND; Andrew SHIREY; Adam HIATT
Original assignee: CBS Interactive Inc
Current assignee: CBS Interactive Inc
Priority date: 2011-06-03
Filing date: 2011-06-03
Publication date: 2012-12-06
Also published as: US20190005517A1; US10332133B2

Abstract

Systems, computer-implemented methods, and media for providing a graph of assets by one or more computing devices include building an asset graph from a data set of assets, the asset graph including plural assets, at least one connection connecting each asset to one or more other asset in the graph, and metadata associated with each connection storing details relating to the connection; identifying an asset in the asset graph that corresponds to an asset in a third party social media platform's asset graph; requesting information relating to connections to the identified asset; receiving the requested information relating to connections to the identified asset; and building out the asset graph according to the received information.

Description

BACKGROUND

Social networks are generally made up of individuals (i.e., “nodes”) connected by one or more specific types of connections (i.e., “ties”). Commercially available social network platforms, such as the FACEBOOK™ platform, allow users to connect to each other over the internet by “friending” each other (i.e., creating a “friend” connection between users). Some social network platforms have expanded to represent objects in addition to individuals and the connections between individuals and those objects. For example the FACEBOOK™ platform has expanded to include objects (e.g., users, photos, and webpages), and the connections between them (e.g., friendships, photo tags, and likes). In addition to allowing users to easily stay in touch with old friends and meet new friends through connections from existing friends, the connections involved in social networks may be useful for inferring interests of a user.

A social graph is a concept that describes the relationship between online users defined explicitly by the connections between the users. For example, the FACEBOOK™ graph Application Program Interface (“API”) allows third parties to access objects in the graph and the connections associated with objects. It has long been known that graph analysis can lead to insights. Social graphs may be mined, for example by collaborative filtering, to infer user preferences.

However, social graphs keep minimal data about the connections themselves. For example, the FACEBOOK™ platform provides that a connection between users may be a “friend” connection and that a connection between a user and a product may be a “like” connection; however connections other than those specifically defined by the platform must be inferred. This may lead to misguided inferences because the lack of data about the connection (i.e., connection metadata).

For example, mere knowledge of the fact that users “like” plural products may lead to misguided inferences without structured information relating the connections or to the plural products. As a simple example, a first user may like a PLAYSTATION 3™ and a second user, who is a friend of the first user, may like an LG™ Blu-ray player. Conventional methods of inferring user preferences may infer that the first user would like the LG™ Blu-ray player because his friend does. However, because the PLAYSTATION 3™ has an integrated Blu-ray player, the first user may not like any stand-alone Blue-ray player and instead find standalone Bly-ray plays to be unnecessary peripherals.

While increasingly large data sets and increasingly sophisticated data mining techniques may help to improve the quality of inferences mined from graphs, the improvement comes at a great cost of data gathering and processing. Further, individual connections between objects in a graph may seem anomalous simply because of a lack of understanding of the connection.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1ashows a conceptual diagram of an ontology logically connected to a graph from a third party social media platform.

FIG. 1bshows a conceptual diagram of a third party social media platform's graph coupled to an asset graph.

FIG. 1cshows a conceptual diagram an asset graph including assets from the asset ontology ofFIG. 1aorganized in an asset graph including assets and connections from a third party social media platform's graph.

FIG. 2 is a flow chart showing an exemplary method for building out an asset graph to include information from one or more third party social media platforms.

FIG. 3 shows a computing device useful for performing processes disclosed herein.

While systems and methods are described herein by way of example and embodiments, those skilled in the art recognize that systems and methods for using social media data to form connections between assets are not limited to the embodiments or drawings described. It should be understood that the drawings and description are not intended to be limiting to the particular form disclosed. Rather, the intention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the appended claims. Any headings used herein are for organizational purposes only and are not meant to limit the scope of the description or the claims. As used herein, the word “may” is used in a permissive sense (i.e., meaning having the potential to), rather than the mandatory sense (i.e., meaning must). Similarly, the words “include”, “including”, and “includes” mean including, but not limited to.

DETAILED DESCRIPTION

Disclosed embodiments provide computer-implemented methods and systems for receiving connections between assets from external sources and using that information for connecting assets in an internal graph. Methods and systems may use third party sources to identify relations between socially connected assets and assets arranged in an ontology.

Relational databases are known for storing vast amounts of structured data according to common relations. For example, a data set may store plural assets organized according to a hierarchical taxonomy having a tree structure. Assets may be technology products organized by a type of product, such as computers, mobile phones, televisions, gaming consoles, computer peripherals, etc. Within each type of product, products may be organized by manufacturer, for example computer type products may be divided into DELL™, APPLE™, HP™, etc. computers. Within each manufacturer, products may further be divided into form factor, such as laptop, desktop, tablet, etc. Products may continue to be divided into increasingly narrow categories by features such as processor speed, memory, etc.

Each of these categories may be thought of as an attribute-value pair. For example, a specific product may have an attribute “product type” and corresponding value “computer”, attribute “form factor” and value “laptop”, attribute “manufacturer” and corresponding value “LENOVO™”, attribute “product line” and value “IDEAPAD™”, attribute “product series” and value “Y series”, attribute “product model” and value “560d”, attribute “memory” and value “4 GB”, attribute “weight” and value “under 6 lbs”, and so on.

Of course, some products may have different attributes than others. For example, computer products having a “laptop” form factor may have a weight attribute while other products having a “desktop” form factor may not. Hierarchical taxonomies additionally may allow for inheritance of attributes and values, thereby saving resources and simplifying catalog build-outs.

Additionally, while a taxonomy having a tree structure provides efficient organization of assets and allows for quick searching of assets stored in a data set, alternative ontologies may arrange assets in one or more many-to-many relationships. For example, while a product type of computers may be divided by form factors at the highest level in a taxonomy and at every lower level be subdivided by an attribute and its corresponding values, many-to-many ontologies allow for a more flexibly structured organization of related assets. By arranging assets in such a fashion, assets may be navigated according to any of their attribute values, for example via an asset ontology index. A user navigating a catalog of assets may filter the assets according to any asset attribute value or range of values. For example, a user may narrow a catalog of technology products to include all assets having a product type “gaming console” and, in response, may receive a listing of assets including an “XBOX360™” and a “PLAYSTATION 3™”. Alternatively, the user may narrow the same catalog of technology products to include all assets having a drive type asset and value “Blu-ray” and, in response, receive a listing including “PLAYSTATION 3™” and “LG™ BD590”.

Arranging assets in a structured ontology thus may provide detailed information about the connections between assets. This connection information may be userful for arranging the assets in an asset graph having each connection between assets stored in a data set with metadata providing information about each connection. For example, a connection between an XBOX360™ asset and a PLAYSTATION 3™ asset may include metadata explaining that the assets are connected because they are both gaming consoles. At the same time, a connection between a PLAYSTATION 3™ asset and a LG™ BD590 asset may include metadata explaining that the assets are connected because they both include Blu-ray players. The metadata may include additional information about a connection, for example the strength of a connection.

While the above examples provide data sets arranging technology product assets such as computers, gaming consoles, and television peripherals, asset ontologies and asset graphs may arrange data sets of any type of asset. For example, streaming or downloadable content may be organized by types of content (e.g., TV shows, movies, video product reviews, user generated content (e.g., YOUTUBE™ videos), etc.), delivery of content (e.g., downloadable, streaming, etc.), genre of content (e.g., documentary, comedy, etc.) and any other appropriate attributes and values with each attribute-value pair determining an asset's classification within an asset ontology or providing a connection between assets in an asset graph. Likewise, textural content assets may be organized by, for example, types of content (e.g., product reviews, news articles, blog posts, novels, etc.), access rights of the content (e.g., subscription, freely available), and delivery of the content (e.g., website, e-book, print publication). Of course, these are only exemplary assets; assets may be any data that may be logically organized, such as people, computer games, services, etc.

Further, connections may span across data sets of assets or assets of different types. For example, an asset may be an online editor (i.e., a person) who writes reviews of products. Thus, a graph may connect the editor with the articles he or she wrote and each article may be connected to the products reviewed in the article. The connections themselves may include metadata explaining the respective connection. For example, the connection between an editor and a product review may indicate that the editor wrote the product review, thus differentiating the connection between the editor and the product review from a connection between a user who read and “liked” the product review and the product review (i.e., the connection between a user who read the product review and used a tool through the FACEBOOK™ platform to indicate their like of the review, thus creating a connection in the FACEBOOK™ graph between the user and the product review).

Additionally, connections may include metadata showing an opinion. For example, a product review article may have connections to each of the products reviewed therein. However, these connections may provide little value without knowing the opinion creating the connection. In other words, useful inferences cannot be drawn from merely knowing that a reviewer reviewed a product, but inferences may drawn based on whether the product received a positive review or a negative review and why. Thus, the metadata associated with a connection between a reviewer and a product may include an opinion, a degree or strength of the opinion, and like information. Similar connections may be made between the editor and each product reviewed.

In addition to assets having common attribute values being connected, allied assets may be connected. Allied assets may have different attribute values, but may be complementary assets. For example, a ream of paper may be connected to a printer and metadata associated with the connection may indicate that the assets are allied because the ream of paper is a supply used by the printer. Other allied assets may be accessories or other complementary assets, for example, a stroller and a cup holder configured to be mounted to a stroller may be allied assets. Still other allied assets may be products and related services, such as a mobile phone and a service (e.g., voice or data) plan for the mobile phone or a car and a bumper to bumper warranty. Of course, these allied assets are exemplary only. Any assets having substantially different attribute values but being allied may be connected and the connection metadata may include information about the connection such as the fact that the assets are allied (as opposed to similar), the strength of the alignment, how the assets are aligned, and the like.

As the foregoing describes, arranging assets into highly structured ontologies and into asset graphs may provide many uses. For example, by improving navigation of assets through online catalogs and using online filters, improving free language searching of assets by knowing the relation between assets and thus returning assets related to a search term rather than merely assets that explicitly contain a search term, suggesting assets related to or allied with an asset a user is browsing or searching, and providing ads to a user related to an asset.

While connections between assets arranged in an ontology are useful for organizing assets and recommending assets to users, such organization and recommendations based purely on the ontology may be limited to relations of closely related assets, closely allied assets, and the like. A website hosting content may acquire a great deal of user information based on their viewing habits, for example the content the user views and interacts with on the website, metadata received from a referring website (e.g., if the user navigated to the site through a GOOGLE™ search), cookies present on the user's computing device, express information provided by the user (e.g., user preferences and profile information), and the like. This information may be useful for recommending assets to a user, selecting directed advertisements to display to a user, and the like. Still, the quality of such recommendations may be further personalized by analyzing social media data relating to the user. Due to the rapid growth of social media and its integration into many websites, vast amounts of data about a user's connections may be retrieved from one or more third party social media platforms and this data may be used to personalize recommendations to the user based on an asset ontology. In other words, the quality (i.e., highly structured) connections of an asset ontology may be supplemented with the high quantity (i.e., vast number of connections, each providing little structure) connections of social media platforms to provide recommendations of assets of greater interest to the user.

This information may also be useful to infer connections between assets. This allows for a graph of assets to be built out which may include connections between assets inferred based on data extracted from a social media platform and may include at least one asset in the graph for a specific user.

FIG. 1ashows a conceptual diagram of an ontology logically connected to a graph from a third party social media platform. The right side of the diagram shows an exemplary portion of a simplified asset ontology. The asset ontology includes a category reviewers160 (i.e., editors who write reviews of products) and

plural reviewers

160₀and160_Xorganized under thecategory reviewers160. The asset ontology also includes a category ofMP3 players120 and withinMP3 players120 are categoriesmanufacturer APPLE™130 andSANDISK™140. WithinAPPLE™130 are the assetsIPOD™ touch132,IPOD™ nano134 andIPOD™ shuffle136. Additionally, withinSANDISK™140 areassets FUZE™142 andCLIP™144. The assets under theMP3 player120 category are shown in a hierarchical tree structure to ease illustration of sets of connections amongst plural assets, however, an ontology may organize such assets in alternative fashions.

Alternatively, in a graph structure assets within the APPLE™ category may be shown as each connected to each other based on their manufacturer. Likewise, in a graph structure the assets within the SANDISK™ category may be each connected to each other based on their manufacturer. In other words, a connection may be defined betweenIPOD™ shuffle136 andIPOD™ nano134 and the connection may have metadata describing an asset value pair for the connection (e.g., attribute=manufacturer; value=APPLE™). Additionally, all assetsIPOD™ touch132,IPOD™ nano134,IPOD™ shuffle136,FUZE™142, andCLIP™144 may have connections to each other and each connection may contain metadata indicating that the assets are connected because they are all MP3 players (e.g., attribute=producttype; value=mp3player).

Of course, an asset graph is not limited to the connections shown in the tree structure; additional connections may be made according to further attribute value pairs, for example theIPOD™ shuffle136 andCLIP™144 may be connected by virtue of having a “micro” form factor (e.g., attribute=formfactor; value=micro).

The left side ofFIG. 1ashows agraph101 of assets connected to a user retrieved from a third party social media platform. The exemplary graph shows auser102 “friend” connected to plural friends104₀,104₁, . . . ,104_Nand “like” connected to a product116_X. As shown, friends can be directly “friend” connected, forexample user102 is directly “friend” connected to friend104₀or they can be connected through a chain of “friend” connections, forexample user102 is “friend” connected to friend104₂via friend104₀.Graph101 also shows friend104₁“like” connected to product116₀. Of course, the asset ontology and third party social media platform shown each have only a few assets, but in practice asset ontologies and third party social medium platforms may include connections between thousands, millions, or more assets.

FIG. 1aadditionally showsconnections118 between known structured information within an asset ontology and information that may be gathered from a third party social media platform. For example, friend104₂ingraph101 may be areviewer160_Xin an asset ontology, product116₀ingraph101 may beIPOD™ shuffle136 in an asset ontology, and product116_Xingraph101 may beCLIP™144 in an asset ontology. The asset ontology may provide highly structured data regarding connections between the assets and a third party social media platform may provide more loosely structured data regarding connections between the assets. For example, the asset ontology may provide detailed information regarding one or more product the reviewer has reviewed, such asIPOD™ touch132. The detailed information may include not only whetherreviewer160_Xliked or disliked theIPOD™ touch132, but may include further metadata includingdetails indicating reviewer160_Xopinions regarding specific attributes of theIPOD™ touch132. For example, the metadata may indicate thatreviewer160_Xgave the IPOD™ touch132 a negative review and indicated that its size is inconveniently large in comparison to current “micro” form factor MP3 players according to the reviewer.

Additionally, information may be received from the third party social media platform, for example through the FACBEOOK™ graph API, indicating assets connected to friend104₂and the type of connection between friend104₂and each connected asset. For example, in the simple graph ofFIG. 1a, a request may be made to the third party social media platform's API for all users “like” connected to friend104₂and in response to the request the third party social media platform may identify friend104₀. Additionally, the type of connection may be identified by the third party social media platform's API. In like fashion, once friend104₀is identified, information about assets and connections connected to friend104₀may be retrieved from the third party social media platform as well. In this fashion, a system may continue to request information from the third party social media platform to discover greater portions of or the entire third party social media platform'sgraph101.

Of course, requests to a third party social media platform may be limited by security features put in place by the platform. For example, a user may expressly allow the third party social media platform to disclose their connections by logging into a system configured to interact with the third party social media platform. Alternatively, a user may adjust their settings on the third party social media platform to either allow systems to generally access their connections, such as through an API, or to specifically allow a system associated with an asset ontology to access their connections. Of course, embodiments may be designed to work with evolving security and privacy systems to ensure both privacy of user information and access to user connections.

When interacting with a third party social media platform that only allows access to information regarding users, objects, and the like directly connected to a user, embodiments may be configured to build out an asset graph by only requesting connections from assets specifically giving permission. For example, because friend104₂is alsoreviewer160_X,reviewer160_Xmay expressly allow the third party social media platform to share information relating to their connections with a system. Thus, a system may request all connections to friend104₂and, in response, may receive information that friend104₀is “friend” connected to friend104₂. In similar fashion, whenuser102 “likes” product116_Xoffered by a system's online catalog, the system may have received access fromuser102 to request their connections. The system may then request all connections touser102 and, in response, may receive information thatuser102 is “friend” connected to friends104₀,104₁, and104_N. The system may then identify that friend104₀who is “friend” connected touser102 is the same friend104₀who is friend connected to friend104₂, thus connecting the assets discovered through the two requests. While such embodiments may only allow discovery of assets directly connected to known assets, the predominance of social media may allow for a detailed andinformative graph101 to be discovered by a system in a relatively short time span.

Referring now toFIG. 1b, the same third party socialmedia platform graph101 is shown coupled to anasset graph180.Asset graph180 may be a graph of assets in the asset ontology ofFIG. 1.Line150 may represent a firewall separatingasset graph180 hosted on an internal computing device from one or more third party social media platforms which may, for example, be hosted with publicly accessible APIs.Asset graph180 may include assets from a highly organized asset ontology and structured connections between those assets. Initially, the structured connections may include connections between similar assets and between allied assets and include metadata describing each connection. A system may then build outasset graph180 to include additional assets and asset connections received from one or more third party social media platforms. The connections between the built out assets may include limited metadata including known information about the connection such as the third party social media platform that the connection as received from (e.g., FACEBOOK™, LINKEDIN™, TWITTER™, etc.). The system may also infer additional connections between any assets in the graph. Inferred connections may include metadata describing the inferred connection and metadata indicating that the connection is inferred. The indication that the connection is inferred may be useful for forming future inferences when an inferred connection and another connection appear to conflict or not synchronize with each other.

FIG. 1cis a conceptual diagram of a built outasset graph181 including assets from theasset ontology103 ofFIG. 1aorganized in an asset graph and built out to include assets and connections from third party social media platform'sgraph101. Assets in theasset graph181 may be connected by highly structured connections (i.e., connections having metadata indicating information about the connection). For example, each of assets IPOD™ shuffle116₀/136,IPOD™ nano134, and CLIP™116_X/144 may be connected by their common attribute value of having a “micro” form factor. In other words, a connection may be formed between each of these assets and metadata associated with the connection may indicate that the assets are connected because they each have a “micro” form factor. Likewise,IPOD™ touch132 andFUZE™142 may be connected by their common attribute value of having a “large” form factor.

For the ease of illustration, many additional structured connections between assets based on the asset ontology ofFIG. 1aare omitted fromFIG. 1c. Additional connections, for example, may be connections between each asset in assetgroup APPLE™130 and connections between each asset in assetgroup SANDISK™140 with each of the connections having metadata indicating the assets manufacturer or connections between all assets in both the assetgroup APPLE™130 and the assetgroup SANDISK™140 with metadata indicating the assets are MP3 players. Further,IPOD™ touch132 and IPOD™ shuffle116₀/136 may be connected by their common attribute value of touch screens,IPOD™ touch132, IPOD™ shuffle116₀/136, andFUZE™142 may be connected by their common attribute value of color screens, and so on.

In building out theasset graph181, a computing device may determine assets in a third party social media platform'sgraph101 that correspond to assets in the asset graph180 (shown inFIG. 1b) and associate the assets together as a single asset. For example,FIG. 1ashowed that friend104₂was the same asreviewer160_Xin the asset ontology, soFIG. 1cshowsasset graph181 built out to identify a single asset as both friend104₂andreviewer160_X. In this fashion, both the highly structured connections of the product ontology and the loosely structured connections of the social media platform may connect the common asset to other assets. For example, reviewer104₂/160_Xmay be connected toIPOD™ touch132 based on a negative review in which the reviewer criticized the large form factor of theIPOD™ touch132. Likewise, reviewer104₂/160_Xmay be connected to friend104₀via a “friend” connection.

This also illustrates that building out theasset graph181 may include adding one or more additional assets. For example,asset graph181 initially did not includeuser102 or friends104₀,104₁, and104_N. During the build out process, one or more assets in a third party social media platform'sgraph101 may be added toasset graph181. Additionally, other products “liked” ingraph101, photos “tagged” ingraph101, or any other assets ingraph101 connected either directly or via a chain of connected assets to an asset inasset graph181 may be added toasset graph181. Each asset fromgraph101 added toasset graph181 may additionally include the social connection and indicate the type of social connection (e.g., like, friend, tag, etc.).

Built outasset graph181 ofFIG. 1cincludes assets corresponding to people and products, however other asset graphs may include a much broader array of assets as well as a broader array of connection types. For example, other assets may include articles, streaming digital content, downloadable digital content, user generated content, music, or any other type of asset. Additionally, whileasset graph181 is shown in a two-dimensional space inFIG. 1c, asset connections may be relational rather than spatial, and an asset graph may be portrayed in any of many fashions. The portrayal ofFIG. 1cprovides only an example of how some assets may be connected and metadata that may be associated with some connections.

The embodiment shown inFIG. 1cincludes first class assets such asIPOD™ touch132 that may include multiple other assets or products. For example,IPOD™ touch132 may include the IPOD™ touch2010 with 8 gigabytes (“GB”) of memory, the IPOD™ touch2010 64 GB of memory, the IPOD™ touch third generation 8 GB, and others. In this fashion, a greater number of connections between assets may be usefully drawn from third party social media platforms. For example, one user may like an IPOD™ touch2010 with 8 GB for the same reasons a second user likes the IPOD™ touch2010 with 64 GB but the second user may simply have more finances. The connection between each user and the first class assetIPOD™ touch132 may include metadata indicating which specific model of the asset the user “liked” or was otherwise connected to in a third party social media platform. Thus, an asset graph may both provide a greater number of connections to make inferences about first class assets and fewer but more specific connections to make inferences about more specific assets.

Of course, the span for features or models that may be included in a single first class asset may be configured based on the purpose of the asset graph. For example, if the asset graph is designed to serve as a back end of a product catalog configured to offer products to a highly sophisticated audience, fewer models may be grouped in a single first class asset. In still other embodiments, an asset graph may not include first class assets, but rather include an asset for each individual model. While products are used by way of example to explain first class assets versus other assets, first class assets may be used with any asset types. For example, a first class asset for a downloadable or streamable movie may include a theatrical version, an extended directors cut version, and a censored version.

In other embodiments, a first class asset may include an asset graph of second class assets. Second class assets may be any assets more specifically grouped and may themselves still be broader than specific models or versions. For example, in an embodiment a first class asset may be an IPOD™, the asset may include an asset graph of second class assets including plural IPOD™ models, each second class asset may include an asset graph of third class assets including optional memory configurations, and each third class asset may include an asset graph of fourth class assets including plural colors. In such embodiments, the first class asset may be connected with other first class assets and the connections may have metadata relating to the connection (e.g., a user “liked” an article about IPODs™). The same connection may more specifically connect an asset to a second class asset within the first class asset if the connection related to a specific model and the connection may have metadata relating to the connection (e.g., a user “liked” an IPOD™ touch). Connections may connect any class of asset to any other class of asset and metadata associated with the connection may provide information relating to the connection.

In this fashion, lower tier assets may inherit all connections of higher tier assets. In this fashion, many more connections provided by one or more third party social media platforms may be made to a first tired asset, thus providing more data to assist in inferring connections between first tier assets.

FIG. 2 is a flow chart showing anexemplary method200 for building out an asset graph to include information from one or more third party social media platforms. Instep205, an asset graph may be built. For example, a computing device may automatically generate a highly structured asset graph based on a structured asset ontology. Alternatively, one or more users may manually enter assets and connections into an asset graph.

Instep210, one or more assets in the asset graph may be identified as having a corresponding asset in a third party social media platform's social graph. For example, a system may detect that a user selects (e.g., clicks) a FACEBOOK™ “like” button in an online product catalog and thereby identify both the logged in user (i.e., a first asset) and the “liked” asset (i.e., a second asset) in the FACEBOOK™ graph. Alternatively, a user may login to a system and consent to the system requesting information from the user's social graph on one or more third party social media platform. A user may choose to expressly consent, for example, so that they may receive more relevant advertisements, personalized search results, and other content tailored to the specific user.

Once one or more assets are identified instep210, a computing system may request from a third party social media platform's social graph assets and connections connected to one or more of the identified assets. For example, if the third party social media platform is FACEBOOK™, a computing system may request all of a user's connections of a certain type, for example “friends” or “likes”. Instep220, a computing system may then receive information relating to assets and/or connections from a third party social media platform connected to the identified assets.

Atstep225, a computing system may build out an asset graph to include any received assets and received connections. This step may include identifying received assets that correspond to assets already in the asset graph and associating the received assets with the assets already in the asset graph. This step may also include adding new assets to the asset graph and, if their connections connect the assets to any assets already in the asset graph, connecting the new assets to existing assets. Each connection may include metadata identifying the type of social connection, such as a “like” connection or a “friend” connection for assets received from the FACEBOOK™ graph.

One or more of the steps ofmethod200 may be performed multiple times to crawl through one or more third party social media platforms to build out asset graphs. Additionally, multiple instances of one or more step may be performed in parallel to more quickly progress through a third party social media platform to build out an asset graph. Alternatively, a third party social media platform may only be interrogated once or a predetermined number of times for connections directly related to an asset already existing in the asset graph without crawling further to determine other assets and connections.

Method

200 is divided into the finite steps shown inFIG. 2 for ease of illustration only, and any or all of the steps may be combined or further subdivided. Each of the steps may be performed by one or more modules executed by a computing device, for example the computing device ofFIG. 3.

Themethod200 ofFIG. 2 may build out theasset graph181 ofFIG. 1cin the following exemplary fashion. Instep205, a computing device may build an asset graph including plural assets, connections between the plural assets, and metadata associated with each connection. The metadata may provide information such as the type of connection, the strength of the connection, the source of the connection (e.g., whether the connection is inferred or not), and the like. For example, an asset graph may be built from the highly structuredasset ontology103 shown inFIG. 1a.

Instep210, a computing device may identify one or more assets in a social graph corresponding to assets in an asset graph. For example, auser102 logged into FACEBOOK™ may be viewing an online product catalog and click a “like” button when they are viewing a product116_Xtheuser102 likes. A computing device may thus identify both the user102 (e.g., based on their FACEBOOK™ login or other unique identification information) and identify the product116_Xthat the user “liked”. The product116_Xmay be identified as an asset in the asset graph (e.g., CLIP™144), thus the asset in the asset graph may be associated with the product in the FACEBOOK™ graph. Additionally, theuser102 may be added to theasset graph181, a connection may be added between CLIP™116_X/144 anduser102, and metadata may be associated with the connection indicating that the connection is a FACEBOOK™ “like” connection. Alternatively, a computing device may searchgraph181 to determine whetheruser102 already exists as an asset in the asset graph and add the user as an asset if the user did not exist or connect the existing user to CLIP™116_X/144 if the user already existed in the asset graph.

Instep215, a computing device may request from the FACEBOOK™ graph API assets and connections connected to the identified assets. For example, a request may be made to the FACEBOOK™ graph API for friends ofuser102 ifuser102 is logged in by making a request to <https://graph.facebook.com/me/friends?access_token= . . . >. In response, atstep220, a computing device may receive from the FACEBOOK™ graph API all of the “friends” in the FACEBOOK™ graph connected touser102. A computing device may then build out the asset graph instep225 by determining whether to add the “friends” as assets toasset graph181 or, if the returned “friends” already exist inasset graph181, whether to create connections between the assets. Any connections created may include the limited data about the connection provided by the social network (e.g., whether they are a “friend” in FACEBOOK™, whether they are a “classmate”, “friend”, “colleague”, etc. on LINKEDIN™, and the like).

Steps

210 through225 may be repeated one or more times to more fully build out theasset graph181. Additionally, multiple instances ofmethod200 may be run in parallel to request connections made to plural assets simultaneously. For example, a computing device may make substantially simultaneous requests to the FACEBOOK™ graph API for friend connections to friends104₀,104₁, and104_Nin the general format <https://graphlacebook.com/ID/CONNECTION_TYPE?ACCESS_TOKEN= . . . > with the ID being the ID of the asset (e.g., the user, the product liked, the photo tag, etc.), the CONNECTION_TYPE being the type of connection (e.g., friend, like, photo tag, etc.), and the ACCESS_TOKEN identifying an access token if a user has authorized access to their non-publicly accessible information.

These embodiments may be implemented with software executed on hardware, for example, functional software modules executed on computing devices such ascomputing device310 ofFIG. 3. Embodiments may, for example, execute modules corresponding tomethod200. Of course, steps may be performed by more than one module, a single module may perform more than one step, or any other logical division of steps ofmethod200 may be used to implement the data flow as software executed on one or more computing devices.

Computing device

310 has one or more processing device311 designed to process instructions, for example computer readable instructions (i.e., code) stored on astorage device313. By processing instructions, processing device311 may perform the steps set forth inmethod200.Storage device313 may be any type of storage device (e.g., an optical storage device, a magnetic storage device, a solid state storage device, etc.), for example a non-transitory storage device. Alternatively, instructions may be stored in remote storage devices, for example storage devices accessed over a network or the internet.Computing device310 additionally hasmemory312, aninput controller316, and anoutput controller315. Abus314 operatively couples components ofcomputing device310, including processor311,memory312,storage device313,input controller316,output controller315, and any other devices (e.g., network controllers, sound controllers, etc.).Output controller315 may be operatively coupled (e.g., via a wired or wireless connection) to a display device320 (e.g., a monitor, television, mobile device screen, touch-display, etc.) in such a fashion thatoutput controller315 can transform the display on display device320 (e.g., in response to modules executed).Input controller316 may be operatively coupled (e.g., via a wired or wireless connection) to input device330 (e.g., mouse, keyboard, touch-pad, scroll-ball, touch-display, etc.) in such a fashion that input can be received from a user (e.g., a user may navigate an online product catalog or click a “like” button relating to a product).

Of course,FIG. 3 illustratescomputing device310,display device320, andinput device330 as separate devices for ease of identification only.Computing device310,display device320, andinput device330 may be separate devices (e.g., a personal computer connected by wires to a monitor and mouse), may be integrated in a single device (e.g., a mobile device with a touch-display, such as a smartphone or a tablet), or any combination of devices (e.g., a computing device operatively coupled to a touch-screen display device, a plurality of computing devices attached to a single display device and input device, etc.).Computing device310 may be one or more servers, for example a farm of networked servers, a clustered server environment, or a cloud network of computing devices, configured to serve web pages and/or ad content.

Embodiments have been disclosed herein. However, various modifications can be made without departing from the scope of the embodiments as defined by the appended claims and legal equivalents.

Claims

1. A system for organizing assets, said system comprising:

one or more storage devices storing a data set of plural internal assets, said data set having an graph structure organizing said internal assets with one or more connections between internal assets and metadata associated with each connection; and

a processor configured to:

retrieve from one or more third party data sets one or more external assets and connections between the external assets;

determine relationships between the external assets and the internal assets; and

create at least one internal connection and metadata associated with the connection indicating the type of the connection.

2. The system ofclaim 1, wherein the external assets are one or more of people and objects relating to a social media platform.

3. The system ofclaim 2, wherein the internal assets are one or more of products, articles, blog posts, tweets, streaming content, downloadable content, websites, users, and creatives.

4. The system ofclaim 3, wherein said processor identifies some of the external assets as identical to the internal assets, and wherein said processor determines relationships between the internal assets and the external assets by identifying connections between an external asset related to an internal asset and one or more other external assets.

5. The system ofclaim 1, wherein said processor creates an inferred connection between two assets and creates metadata associated with the inferred connection indicating details of the inferred connection.

6. The system ofclaim 1, wherein the processor identifies a user and recommends an asset to the user.

7. A computer-implemented method for providing a graph of assets by one or more computing devices, the method comprising:

building, by at least one of the computing devices, an asset graph from a data set of assets, the asset graph including plural assets, at least one connection connecting each asset to one or more other asset in the graph, and metadata associated with each connection storing details relating to the connection;

identifying, by at least one of the computing devices, an asset in the asset graph that corresponds to an asset in a third party social media platform's asset graph;

requesting, by at least one of the computing devices, information relating to connections to the identified asset;

receiving, by at least one of the computing devices, the requested information relating to connections to the identified asset; and

building out, by at least one of the computing devices, the asset graph according to the received information.

8. The computer-implemented method ofclaim 7, wherein the external assets are one or more of people and objects relating to a social media platform.

9. The computer-implemented method ofclaim 8, wherein the internal assets are one or more of products, articles, blog posts, tweets, streaming content, downloadable content, websites, users, and creatives.

10. The computer-implemented method ofclaim 9, further comprising:

identifying some of the external assets as identical to the internal assets, and

determining relationships between the internal assets and the external assets by identifying connections between an external asset related to an internal asset and one or more other external assets.

11. The computer-implemented method ofclaim 7, further comprising creating an inferred connection between two assets and creating metadata associated with the inferred connection indicating details of the inferred connection.

12. The computer-implemented method ofclaim 7, further comprising identifying a user and recommending an asset to the user.

13. Computer-readable code stored on a non-transitory computer-readable medium that, when executed by a mobile device, performs the method:

14. The computer-readable code ofclaim 13, wherein the external assets are one or more of people and objects relating to a social media platform.

15. The computer-readable code ofclaim 14, wherein the internal assets are one or more of products, articles, blog posts, tweets, streaming content, downloadable content, websites, users, and creatives.

16. The computer-readable code ofclaim 15, the method further comprising:

17. The computer-readable code ofclaim 13, the method further comprising creating an inferred connection between two assets and creating metadata associated with the inferred connection indicating details of the inferred connection.

18. The computer-readable code ofclaim 13, the method further comprising identifying a user and recommending an asset to the user.