Facebook, Inc.

Crimson Hexagon, Inc.

Meta Platforms, Inc.

Amaç Herdagdelen

Rousseau Newaz Kazi

Brett Matthew Westervelt

Mark Andrew Rich

Christina Joan Sauper

Maykel Andreas Louisa Jozef Anna Loomans

Soorya Vamsi Mohan Tanikella

Adam Eugene Bussing

Shuyi Zheng

Bradley Ray Green

Thomas Frederick Dimson

Bogdan State

Aditya Pal

Danilo Torres De Sa Resende

Christopher Bingham

Aykut Firat

Ehud Weinsberg

Bai Xiao

Fei Huang

Gary King

Deepayan Chakrabarti

Sourav Chatterji

Stanislav Funiak

Daniel Matthew Merl

Mitchell Brooks

Sumit Taank

Seyoung Park

Systems, methods, and non-transitory computer-readable media can acquire a set of labels associated with a set of content items. Each label in the set of labels can be associated with at least one content item in the set of content items. It can be determined that at least two labels, out of the set of labels, are related. The at least two labels can be determined to be related based on at least one of a co-occurrence metric associated with the at least two labels or a topic similarity metric associated with the at least two labels. One label can be selected, out of the at least two labels, as being representative of the at least two labels.

Systems and methods for identifying and grouping related content labels

Systems, methods, and non-transitory computer-readable media can determine one or more respective topics of interest for at least some users of a social networking system. At least some of the topics can be propagated to at least a first user, wherein the propagated topics were determined to be of interest to users that follow the first user in the social networking system. At least one topic from the propagated topics for which the first user is a topical authority is determined.

Systems and methods for recommendation of topical authorities

In one embodiment, a method includes accessing a plurality of communications, each communication being associated with a particular content item and including a text of the communication; extracting, for each of the communications, n-grams from the text of the communication; identifying mention-terms from the extracted n-grams, each mention-term being a noun-phrase; calculating a term-score for each mention-term based on a frequency of occurrence of the mention-term in the communications; and generating a mentions-module including mentions, each mention including a mention-term having a term-score greater than a threshold term-score and text from communications comprising the mention-term.

Mentions-modules on online social networks

In one embodiment, a method includes receiving, from a client system of a first user, a request associated with a particular content item; identifying communications authored by one or more users, each identified communication being associated with the particular content item; generating one or more search-results modules related to the particular content item, each search-results module being of a particular module type, wherein each search-results module includes information from a subset of the identified communications, the information corresponding to the particular module type of the search-results module, and wherein a number of communications in the subset of the identified communications including each search-results module is greater than a module-specific threshold number of communications; and sending, to the client system, a search-results interface comprising one or more of the search-results modules.

Search-results interfaces for content-item-specific modules on online social networks

In one embodiment, a method includes accessing a plurality of communications, each communication being associated with a particular content item and including a text of the communication; calculating, for each of the communications, sentiment-scores corresponding to sentiments, wherein each sentiment-score is based on a degree to which n-grams of the text of the communication match sentiment-words associated with the sentiments; determining, for each of the communications, an overall sentiment for the communication based on the calculated sentiment-scores for the communication; calculating sentiment levels for the particular content item corresponding sentiments, each sentiment level being based on a total number of communications determined to have the overall sentiment of the sentiment level; and generating a sentiments-module including sentiment-representations corresponding to overall sentiments having sentiment levels greater than a threshold sentiment level.

Sentiment-modules on online social networks

A system for identifying language(s) for content items is disclosed. The system can identify different languages for content item words segments by identifying segment languages that maximize a probability across the segments. The probability can be a combination of: an author's likelihood for the language identified for the first word; a combination of transition frequencies for selected languages identified for words, the transition frequencies indicating likelihoods that a transition occurred to the selected language from the previous word's language; and a combination of observation probabilities indicating, for a given word in the content item, a likelihood the given word is in the identified language. For an in-vocabulary word, the observation probabilities can be based on learned probability for that word. For an out-of-vocabulary word, the probability can be computed by breaking the word into overlapping n-grams and computing combined learned probabilities that each n-gram is in the given language.

Identifying multiple languages in a content item

In one embodiment, a method includes accessing a plurality of communications, each communication being associated with a particular content item and including a text of the communication; extracting, for each of the communications, quotations from the text of the communication; determining, for each extracted quotation, partitions of the quotation; grouping the extracted quotations into clusters based on a respective degree of similarity among their respective partitions; calculating a cluster-score for each cluster based on a frequency of occurrence of partitions of quotations in the cluster in the communications; and generating a quotations-module comprising representative quotations, each representative quotation being a quotation from a cluster having a cluster-score greater than a threshold cluster-score.

Quotations-modules on online social networks

Technology for media item and user language classification is disclosed. Media item classification may use models for associating language identifiers or probability distributions for multiple languages with linguistic content. User language classification may define user language models for attributing to users indications of languages they speak read, and/or write. The text classifications and user classifications may interact because the probability that given text is in a particular language may depend on a determined likelihood the user who produced the text speaks that language, or conversely, a user interacting with text in a particular language may increase the likelihood they understand that language. Some embodiments use language-tagged social media content to train n-gram classifiers for use with other social media content.

Classifying languages for objects and entities

Systems and methods are provided for classifying text based on language using one or more computer servers and storage devices. A computer-implemented method includes receiving a training set of elements, each element in the training set being assigned to one of a plurality of categories and having one of a plurality of content profiles associated therewith; receiving a population set of elements, each element in the population set having one of the plurality of content profiles associated therewith; and calculating using at least one of a stacked regression algorithm, a bias formula algorithm, a noise elimination algorithm, and an ensemble method consisting of a plurality of algorithmic methods the results of which are averaged, based on the content profiles associated with and the categories assigned to elements in the training set and the content profiles associated with the elements of the population set, a distribution of elements of the population set over the categories.

Systems and methods for calculating category proportions

Systems and methods are provided for classifying text based on language using one or more computer servers and storage devices. In general, the systems and methods can include a language classification module for classifying text of an input data set using the output of a training module. In an exemplary embodiment, a bootstrapping step feeds the output of the language classification module back into the training module to increase the accuracy of the language classification module. By iterating the language classification and training modules with input data having certain features, a user can tailor the language classification module for use with text having those or similar features.

Systems and methods for language classification

Growing community of inventors

Mountain View, CA, United States of America

Amaç Herdagdelen