Facebook, Inc.

Shriram Subramanian

Bryan Kauder

Jonathan Michael Arfa

Nikhil Girish Nawathe

Alyx Catherine Stevens

In one embodiment, a method includes receiving a request to identify a word representing a target concept that is in a first relationship with a particular concept such that the first relationship is analogous to a second relationship in which a first reference concept is with a second reference concept, accessing a table of word vector relationships, looking up a particular word vector, a first reference word vector, and a second reference word vector, determining an imaginary vector such that a first vector from the first reference word vector to the second reference word vector is equal to a second vector from the particular word vector to the imaginary vector, selecting a target word vector closest to the imaginary vector, identifying a target n-gram corresponding to the target word vector, and sending a response message comprising the target n-gram.

Consumer insights analysis using word embeddings

In one embodiment, a method includes receiving a request to generate k keywords each of which is semantically related to a particular subject, where the request includes an input n-gram representing the particular subject, accessing a table of word vector relationships, where the table includes a plurality of unique n-grams and their corresponding word vectors, and wherein each of the word vectors represents a semantic context of a corresponding n-gram as a point in a d-dimensional embedding space, looking up, using the table, a first word vector corresponding to the input n-gram, selecting k word vectors closest to the first word vector in the embedding space using the table and based on a similarity metric, identifying, for each of the selected word vectors, a corresponding n-gram by looking up the selected word vector in the table, and sending a response message including the identified n-grams.

In one embodiment, a method includes receiving a request to generate a visualization of public sentiments regarding a particular subject by a plurality of clusters, where each cluster includes a plurality of words semantically close to each other, constructing a first corpus of text by collecting text containing the input n-gram from a plurality of user-created content objects in the online social network, identifying a list of unique n-grams appearing in the first corpus of text, generating a table comprising unique n-grams in the list and their corresponding word vectors using a word embedding model, classifying word vectors in the table into a plurality of clusters based on semantic similarities of the word vectors, and sending, as a response to the request, instructions to display n-grams in the table in a two-dimensional display space, where n-grams corresponding to word vectors that belong to a cluster are displayed together.

In one embodiment, a method includes receiving a request to generate k words that each approximates a representation of a relationship between two concepts, where the request includes two input n-grams that each represent one of the two concepts, accessing a table of word vector relationships, where the table includes a plurality of unique n-grams and their corresponding word vectors, looking up word vectors corresponding to each of the two input n-grams using the table, calculating an average vector of the word vectors corresponding to the two input n-grams, selecting, using the table and based on a similarity metric, k word vectors closest to the average vector in the embedding space, identifying, for each of the selected word vectors, a corresponding n-gram by looking up the selected word vector in the table, and sending a response message, the response message comprising the identified n-grams.

In one embodiment, a method includes receiving a request to generate a two-dimensional visualization of public sentiments regarding a particular subject, where the request includes an input n-gram representing the particular subject, constructing a first corpus of text by collecting text containing the input n-gram from a plurality of user-created content objects in the online social network, identifying a list of unique n-grams appearing in the first corpus of text, generating a table comprising unique n-grams in the list and their corresponding word vectors using a word embedding model, condensing the d-dimensional word vectors in the table into a two-dimensional word vectors; and sending, as a response to the request, instructions to display n-grams in the table on a two-dimensional display space, where each n-gram is placed at a location of the corresponding condensed word vector.

Growing community of inventors

New York, NY, United States of America

Shriram Subramanian