Facebook, Inc.

Ming Hua

Avichal Garg

Hong Yan

Dan Ionut Fechete

Yang Qin

Michael Steven Vernal

Saurabh Prafulla Chakradeo

Francis Luu

Wayne Kao

Bo Zhang

Ryan Allen Stout

Yu Huang

Alex Himel

Lu D Chen

Xinpan Xiao

Dharmesh A Bhatt

Rose Yao

Yun-Fang Juan

Aleksandar Ilic

Alon Michael Shalita

Andrew Rothbart

Ashish Wahi

Igor Kabiljo

Online systems, for example, social networking systems evaluate expressions based on features describing relations between entities represented in the online system. These expressions are represented using an expression language. The expression language allows features to be specified as functions of attributes from user accounts. The expressions support use of variables to represent computations, for example, sub-expressions. The expressions are dynamic, since expressions can be specified and executed at call time. The same set of expressions is used many times, e.g., to compute the same function for multiple feature sets, for example, user accounts. Expressions are preferably represented using postfix representation. However some expressions, for example, expressions using variables are represented as trees. To optimize the expressions at runtime, the expressions are cached using a representation determined to be efficient for executing the expression. The cached representation of the expression is applied to multiple feature sets, for example, user accounts.

Dynamic expressions for representing features in an online system

In one embodiment, a method includes receiving, from a client system of a first user of the communication system, an input from the first user to access a card-stack interface, generating a card cluster comprising a plurality of cards, each card comprising a suggested query referencing a query-domain and one or more query-filters, wherein each query-filter references one or more objects associated with the communication system, and wherein each card in the card cluster is ranked within the card cluster based on a predicted click-thru rate (CTR) for the card based on one or more user-engagement factors, and sending, to the client system in response to the input from the first user, the card-stack interface for display to the first user, wherein the card-stack interface comprises the card cluster, the cards of the card cluster being ordered based on the rankings associated with the cards.

Ranking recommended search queries

A social networking system provides a personalized set of bookmarks to a user based on the user's interactions with the bookmarks and/or the items associated with the bookmarks. The personalized set of bookmarks is grouped by category, and the categories are ordered in accordance with rankings for the categories. The rankings for the categories are determined based on the highest ranked item from each category of item, and the ranking of the items is determined based on the user's interactions with the bookmarks and/or associated items.

Personalized bookmarks for social networking system actions based on user activity

In one embodiment, a method includes receiving, from a client system of a first user of a communication network, an input from the first user to access a card-stack interface, generating one or more card clusters from a plurality of cards, each card comprising a query referencing a query-domain associated with the communication network and zero or more query-filters for the query-domain, wherein each query-filter references one or more objects of the communication network, each card cluster comprising one or more cards from the plurality of cards, the cards being formed into card clusters based on a card-affinity between the cards, and sending, to the client system in response to the input from the first user, the card-stack interface for display to the first user, wherein the card-stack interface comprises one or more of the card clusters.

Grouping recommended search queries in card clusters

In one embodiment, a method includes receiving, from a client device of a user of an online social network, an input from the user to access a card-stack interface, generating a plurality of cards, where each card comprises a suggested query referencing a query-domain associated with the online social network and one or more query-filters, and where each query-filter references one or more objects of the online social network, each card further comprising one or more search results corresponding to the suggested query, and each search result referencing an object of the online social network matching the suggested query of the card; and sending, to the client device in response to the input from the first user, the card-stack interface for display to the first user, wherein the card-stack interface comprises one or more of the generated cards.

Generating card stacks with queries on online social networks

A user's connections in a social networking system are grouped into a number of clusters based on a measure of the connections' relationships, or affinity, to each other. The affinities among the connections are based on the connections' own relationships and indicate a likelihood that the connections are in the same social circles. The clusters are formed based on the affinities among the user's connections, where the clusters tend to have connections that have relatively high affinities with the other connections the same cluster as compared to the connections who are not in the same cluster. An iterative hierarchical clustering algorithm may be used to collapse the connections into clusters based on affinities between pairs of the connections.

Clustering a user's connections in a social networking system

Online systems, for example, social networking systems store features describing relations between entities represented in the online system. The information describing the features is represented as a graph. The online system maintains a cumulative feature graph and an incremental feature graph. Feature values based on recent user actions are stored in the incremental graph and feature values based on previous actions are stored in the cumulative graph. Periodically, the information stored in the incremental feature graph is merged with the information stored in the cumulative feature graph. The incremental graph is marked as inactive during the merge and information based on new user actions is stored in an active incremental feature graph. If a request for feature information is received, the feature information obtained from the cumulative feature graph, inactive incremental feature graph and the active incremental feature graph are combined to determine the feature information.

Updating features based on user actions in online systems

In one embodiment, a method includes accessing a social graph that includes a number of nodes and edges connecting the nodes. Each of the edges between two of the nodes representing a single degree of separation between them. The nodes include a first node corresponding to a first user associated with an online social network and a plurality of second nodes that each correspond to a concept or a second user associated with the online social network. The method also includes generating a card cluster including a number of cards. Each card includes a suggested query referencing a query-domain associated with the online social network and zero or more query-filters. Each query-filter references one or more nodes of the plurality of nodes or one or more edges of the plurality of edges.

Ranking recommended search queries on online social networks

In one embodiment, one or more server computing devices receive, from a client computing device, a request for first information associated with a first node of a graph. The one or more server computing devices determine whether the first node is associated with a cluster of nodes. A cluster of nodes includes one or more concept nodes of the graph that are related to each other. When the first node is associated with a cluster of nodes, the one or more server computing devices access the cluster of nodes that the first node is associated with, obtain second information from one or more of the nodes in the cluster of nodes that the first node is associated with, and provide the second information for rendering by the client computing device.

Node properties in a social-networking system

In one embodiment, a method includes accessing a social graph that includes a number of nodes and edges connecting the nodes. Each of the edges between two of the nodes representing a single degree of separation between them. The nodes include a first node corresponding to a first user associated with an online social network and a plurality of second nodes that each correspond to a concept or a second user associated with the online social network. The method also includes generating a number of cards. Each card includes a suggested query referencing a query-domain associated with the online social network and zero or more query-filters for the query-domain. Each query-filter references one or more nodes of the plurality of nodes or one or more edges of the plurality of edges.

Grouping recommended search queries on online social networks

Exemplary methods, apparatuses, and systems receive a candidate object with which a user can interact within a network service. For each of a first plurality of objects with which the user has had a positive interaction, a first value representing a commonality between the candidate object and each of the first plurality of objects is determined. For each of a second plurality of objects with which a user has had a negative interaction, a second value representing a commonality between the candidate object and each of the second plurality of objects is determined. An aggregate positive distance is determined using a plurality of the first values. An aggregate negative distance is determined using a plurality of the second values. The candidate object is displayed or not displayed to the user as a recommendation based upon a difference between the aggregate positive distance and the aggregate negative distance.

Object recommendation based upon similarity distances

Online systems generate predictors for predicting actions of users of the online system. The online system receives requests to generate predictor models for predicting whether a user is likely to take an action of a particular action type. The request specifies the type of action and criteria for identifying a successful instance of the action type and a failure instance of the action type. The online system collects data including successful and failure instances of the action type. The online system generates one or more predictors of different types using the generated data. The online system evaluates and compares the performance of the different predictors generated and selects a predictor based on the performance. The online system returns a handle to access the generated predictor to the requester of the predictor.

Growing community of inventors

Mountain View, CA, United States of America

Ming Hua