Yahoo! Inc.

Facebook, Inc.

Google Inc.

Erik Nathan Vee

Jayavel Shanmugasundaram

Olivier Chatot

Shanmugasundaram Ravikumar

Utkarsh Srivastava

Keith Lanka Peiris

William Russell Maschmeyer

Thomas Scott Whitnah

Joshua Ruizhi Wang

Manish Deepak Purohit

Alexander Langenfeld

Zoya Svitkina

Minos N Garofalakis

Sergei Vassilvitskii

Peiji Chen

Sihem Amer Yahia

Jason Zien

Thomas S Whitnah

Marcus Felipe Fontoura

Jianchang Mao

Brian Frank Cooper

Qi He

John Anthony Tomlin

Alexander Fabrikant

Adam Silberstein

Jian Yang

Tasos Anastasakos

Ramana Yerneni

Raghu Ramakrishnan

Satyen Chandrakant Kale

Dongni Chen

Yuan Zhuge

Datong Chen

Suhas Sadanandan

Ashwin Kumar V Machanavajjhala

Jagadish Prasad Samantarai

Long Ji Lin

Oliver Kennedy

Srihari Venkatesan

Prashanth Bhat

Zoe Abrams

Rajagopal Baskaran

Amirali Abdullah

Chyr-Chong (Joseph) Ting

Srinath Mandalapu

Chandrashekhar Nagarajan

Sumanth Jagannath

Javavel Shanmugasundaram

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scheduling operations represented on a computation graph. One of the methods receiving, by a computation graph system, a request to generate a schedule for processing a computation graph, obtaining data representing the computation graph generating a separator of the computation graph; and generating the schedule to perform the operations represented in the computation graph, wherein generating the schedule comprises: initializing the schedule with zero nodes; for each node in the separator: determining whether the node has any predecessor nodes in the computation graph, when the node has any predecessor nodes, adding the predecessor nodes to the schedule, and adding the node in the schedule, and adding to the schedule each node in each subgraph that is not a predecessor to any node in the separator on the computation graph.

Scheduling operations on a computation graph

In one embodiment, a method includes receiving an unstructured text query inputted by a first user, identifying one or more objects associated with the online social network matching at least a portion of the unstructured text query, accessing a context-free grammar model comprising a plurality of grammars, generating one or more structured queries, each structured query corresponding to a selected grammar of a context-free grammar model, wherein each structured query is based on a natural-language string generated by the selected grammar, each structured query comprising at least one query token corresponding to each of the identified object, and sending one or more of the structured queries as suggested queries for display to the first user in response to the unstructured text query inputted by the first user.

Natural-language rendering of structured search queries

In one embodiment, a method includes receiving a query input from a client system comprising one or more n-grams, sending instructions for presenting one or more suggested modifications for the query input, each suggested modification comprising references to one or more objects associated with the online social network, receiving an indication of a selection of one of the suggested modifications, parsing the query input and the selected suggested modification using a context-free grammar model to generate an executable query command, and sending instructions to the client system for presenting one or more search results corresponding to the query command.

Modifying structured search queries on online social networks

Computer-implemented methods and systems of determining semantic place data include receiving a plurality of location data reports from a plurality of mobile devices, partitioning them into localized segments, and estimating a geographic region bucket for each segment. For clustering canopies of localized segments identified as satisfying a potential geographic overlap characterization, an overlap score is calculated that correlates the overlap among actual geographic regions covered by movement of the mobile devices generating the localized segments in that given clustering canopy. A data structure that provides a hierarchical clustering configuration of the localized segments in each geographic region bucket is generated from the determined overlap scores. Additional semantic data for nodes in the data structure can also be provided.

Analyzing semantic places and related data from a plurality of location data reports

In one embodiment, a method includes receiving, from a client system of a first user of an online social network, a first structured query comprising a natural-language string generated by a grammar model and references to one or more objects associated with the online social network, generating one or more query modifications for the first structured query, each query modification comprising references to one or more modifying objects associated with the online social network, and sending, to the client system of the first user, one or more of the query modifications as suggested modifications to the first structured query for display to the first user, each query modification being selectable by the first user to modify the natural-language string of the first structured query to further comprise references to the one or more of the modifying objects referenced in the selected query modification.

In one embodiment, a method includes accessing a social graph that includes a plurality of nodes and edges, receiving an unstructured text query, identifying nodes and edges that correspond to portions of the text query, accessing a context-free grammar model, identifying query tokens from the grammar model that correspond to the identified nodes and edges, selecting grammars having query tokens that corresponding to each of the identified nodes and edges, and generating structured queries based on the selected grammars, where the structure queries are based on strings generated by the grammars.

In one embodiment, a method includes accessing a social graph that includes a plurality of nodes and edges, receiving a structured query that includes references to selected nodes and edges, and generating one or more query modification for the structured query, where each query modification includes references to modified nodes or modified edges from the plurality of nodes and edges.

Methods and systems are disclosed which allow shifting inventory to fulfill guaranteed delivery advertisement contracts. Inventory may be allocated from a supply of unallocated inventory to one or more advertisers in accordance with guaranteed delivery agreements. Inventory may be reserved for the one or more advertisers from the remaining supply of unallocated inventory. Inventory may then be allocated to an additional advertiser by using unallocated inventory or shifted inventory, or a combination of unallocated inventory and shifted inventory. The shifted inventory is shifted out of the allocation for the first advertiser and the shifted inventory is replaced by the reserve inventory for the respective advertiser.

Methods and systems for ad placement planning

A computer implemented system produces an optimization graph in an online advertising system. The system presents at a computer, a supply of advertising inventory and advertising demand contracts, where each advertising inventory is presented as a supply node and each advertising demand contract is presented as a demand node. The system processes the supply of advertising inventory and the advertising demand contracts. For example, the system partitions the supply nodes into supply partitions and determines whether each supply node is eligible for each demand node. In addition, the system determines the number of eligible supply nodes to be drawn from each supply partition for each demand node.

Producing optimization graphs in online advertising systems

In a large-scale transaction such as the bulk loading of new records into an ordered, distributed database, a transaction limit such as an insert limit may be chosen, partitions on overfull storage servers may be designated to be moved to underfull storage servers, and the move assignments may be based, at least in part on the degree to which a storage server is underfull and the move and insertion costs of the partitions to be moved.

System and/or method for bulk loading of records into an ordered distributed database

Techniques are described herein for using an impression-trend technique to provide a display advertising supply forecast. A display advertising supply forecast is an estimate of a number of impressions, which are to occur in a future time period, that have specified attribute values. For example, the specified attribute values may be descriptive of impressions with respect to which an advertiser wishes to place ads. An impression-trend technique is a forecasting technique that uses trends regarding past impressions to forecast a number of future impressions that have specified attribute values. The past impressions include attribute values that are related to the specified attribute values.

Impression-trend techniques for providing a display advertising supply forecast

An advertisement impression distribution system includes a data processing system operable to generate an allocation plan for serving advertisement impressions. The allocation plan allocates a first portion of advertisement impressions to satisfy guaranteed demand and a second portion of advertisement impressions to satisfy nonguaranteed demand. The data processing system includes an optimizer configured to establish a relationship between the first portion of advertisement impressions and the second portion of advertisement impressions. The relationship defines a range of proportions of allocation of the first portion and the second portion. The optimizer generates a solution maximizing guaranteed demand fairness, non-guaranteed demand revenue, and click value. The solution identifies a determined proportion of the first portion of advertisement impressions to serve and a determined proportion of the second portion of advertisement impressions to serve. The data processing system outputs the allocation plan to control serving of the advertisement impressions.

System for display advertising optimization using click or conversion performance

An item of inventory is described as a Boolean expression, which is converted into a multi-level, alternating AND/OR impression tree representation with leaf nodes representing conjuncts. Processing the conjuncts of the tree through a contract index results in retrieving a set of candidate contracts that match at least some but not necessarily all impression tree leaf node predicates. Next, an AND/OR contract tree representation is constructed with each contract tree leaf node having a label representing a projection onto a discrete set of ordered symbols. Contracts with projections that cover the entire range of discrete set of ordered symbols are deemed to satisfy the item of inventory. Implementation of the contract index includes retrieval techniques to support multi-valued predicates as well as confidence threshold functions using a multi-level tree representation of multi-valued predicates.

System and method for automatic matching of contracts using a fixed-length predicate representation

Growing community of inventors

San Mateo, CA, United States of America

Erik Nathan Vee