Abstract: Methods for predicting network access probability of data files accessible over a computer network are provided. In one aspect, a method includes generating a primary data vector for a media file based on a stored data representation of the file, and providing the data vector for the file to an algorithm that uses past interaction information for at least one other media file from a collection of media files having a degree of similarity with the media file above a threshold similarity value. The method also includes receiving, as an output of the algorithm, a marketability score for the media file, the score indicative of a likelihood that a user will download the media file. Systems and machine-readable media are also provided.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for relevance feedback with faceted search interface. A system provides a first search query to an image search engine for initiating an image search, where the first search query indicates a first search term. The system obtains image metadata from an image collection, and obtains a listing of query refinement terms associated with the first search query from the query refinement data structure. The system determines a listing of search results containing images with one or more facets associated with the listing of query refinement terms. The system then receives user input indicating selection of at least one of the one or more facets of an image in the listing of search results. The system provides a second search query using the first search term and a second search term indicating the selected one or more facets.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for clustering search results based on image composition. A system may, for each image in a set of images, determine a compositional vector representing one or more objects and corresponding locations within the image using a trained computer-operated convolutional neural network. The system may provide each image through a clustering algorithm to produce one or more clusters based on compositional similarity. The system may provide images from the set of images clustered by composition, in which the images include a different listing of images for each of the one or more clusters. The system may provide a prioritized listing of images responsive to a user search query, in which the prioritized listing of images includes a different listing of images for each cluster of compositional similarity based on the metadata of each image associated with the cluster.

Abstract: Methods for identifying visual portions of visual media files that are responsive to at least one visual media input file submitted as a search query are provided. In one aspect, a method includes generating an index of visual portions of a plurality of visual media files from a collection of media files, and receiving from a user a search query, for the collection, including at least one visual media input file. The method also includes analyzing the index to identify at least one responsive visual media file from the collection that includes a visual portion associated with a visual similarity score, to the at least one visual media input file, which exceeds a similarity threshold value, and providing, in response to the search query, an identifier of the at least one responsive visual media file for display as responsive to the search query. Systems and machine-readable media are also provided.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for authentic content search using style classifications. A system may be a search engine that uses a set of style classifiers to detect one or more styles associated with an image and a logistic regression model to determine a level of authenticity for the image based on the associated styles. Training images are fed to train a series of neural networks that output a set of style classifiers. An image is processed through the style classifiers to determine respective probabilities for each style classification. The results from the set of style classifiers are then input to the logistic regression model to determine an authenticity score for the image. For example, the authenticity score shows how authentic is an image (e.g., a score of 1.0 refers to 100% authenticity, whereas a score of 0.0 represents a non-authentic image).

Abstract: Methods, systems, and machine-readable media for identifying synthetic media file examples to train a supervised machine learned ranking algorithm to rank relevance of media files to a search query are provided. In one aspect, a method includes identifying a search query for a search engine for a collection of media files from previously received search queries, and selecting at least one training media file from the collection as a synthetic negative example for the search query. The method also includes providing a training set to a supervised machine learned ranking algorithm. The training set includes an identification of the search query, a copy of the training media file, and a first indicator that the training media file is a synthetic negative example for the search query. The method further includes providing, to the algorithm, the search query and the collection, and receiving, from the algorithm, a ranking of the collection.

Abstract: Computer image clustering systems and methods for conducting effective media searches by grouping multimedia documents tagged by keywords into a hierarchy of images configured to: (1) maintain a first database, (2) maintain an initial occurrence matrix, (3) maintain an occurrence matrix, (4) maintain a media file activation score for each media file in the first database, (5) generate a log version of the occurrence matrix, (6) maintain an inverse media file frequency value for each descriptive term in the first database, (7) generate a descriptive term frequency matrix and generate a list of document vectors in multidimensional space (list), and (8) organize and process each media file in the list into a high activation score category and a low activation score category.

Abstract: Methods for identifying visual portions of visual media files responsive to search queries are provided. In one aspect, a method includes identifying, for each of a plurality of visual media files from a collection of media files, visual portions of the plurality of visual media files responsive to a group of search queries, and providing, in response to a search query from the group of search queries, an identifier of a visual portion of at least one of the plurality of visual media files for display as responsive to the search query. Systems and machine-readable media are also provided.

Abstract: Methods for predicting network access probability of data files accessible over a computer network are provided. In one aspect, a method includes generating a primary data vector for a media file based on a stored data representation of the file, and providing the data vector for the file to an algorithm that uses past interaction information for at least one other media file from a collection of media files having a degree of similarity with the media file above a threshold similarity value. The method also includes receiving, as an output of the algorithm, a marketability score for the media file, the score indicative of a likelihood that a user will download the media file. Systems and machine-readable media are also provided.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for composition aware image querying. A system may receive user input identifying a search query for content from a client device, where the user input indicates one or more queries assigned to one or more regions of a search input page. The system may generate a query vector for each query using a computer-operated neural language model. The system may compare the query vector to an indexed vector for each region of an image. The system may determine a listing of composition-specific images from a collection of images based on the comparison. The system may determine a ranking for each image in the listing of composition-specific images, and provide search results responsive to the search query to the client device. The search results may include a prioritized listing of the composition-specific images based on the determined ranking.

Abstract: A method for receiving an input user query from a user in any of multiple languages and creating an input user query vector for the input user query is provided. The input user query vector has a length equal to a pre-selected length determined by a dimension of a multimodal space. The method includes associating an image vector to the input user query vector in the multimodal space, the image vector having the pre-selected length, identifying, from an image database, an image associated with the image vector, and providing the image for display to the user in a results panel of a user interface. A system configured to perform the above method is also provided.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for identification of a salient portion of an image. A system may receive user input identifying a search query for content from a client device. The system may determine a listing of images responsive to the search query from an image collection. The system may obtain one or more image crops for at least one image of the listing of images based on a saliency map of the at least one image. In one or more implementations, each of the one or more image crops indicates a salient region of a corresponding image. The system may provide a set of search results responsive to the search query to the client device. In one or more implementations, the set of search results includes the obtained one or more image crops in a prioritized listing of image crops.

Abstract: Methods for identifying visual portions of visual media files that are responsive to at least one visual media input file submitted as a search query are provided. In one aspect, a method includes generating an index of visual portions of a plurality of visual media files from a collection of media files, and receiving from a user a search query, for the collection, including at least one visual media input file. The method also includes analyzing the index to identify at least one responsive visual media file from the collection that includes a visual portion associated with a visual similarity score, to the at least one visual media input file, which exceeds a similarity threshold value, and providing, in response to the search query, an identifier of the at least one responsive visual media file for display as responsive to the search query. Systems and machine-readable media are also provided.

Abstract: A method for receiving an input user query from a user in any of multiple languages and creating an input user query vector for the input user query is provided. The input user query vector has a length equal to a pre-selected length determined by a dimension of a multimodal space. The method includes associating an image vector to the input user query vector in the multimodal space, the image vector having the pre-selected length, identifying, from an image database, an image associated with the image vector, and providing the image for display to the user in a results panel of a user interface. A system configured to perform the above method is also provided.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for clustering search results based on image composition. A system may, for each image in a set of images, determine a compositional vector representing one or more objects and corresponding locations within the image using a trained computer-operated convolutional neural network. The system may provide each image through a clustering algorithm to produce one or more clusters based on compositional similarity. The system may provide images from the set of images clustered by composition, in which the images include a different listing of images for each of the one or more clusters. The system may provide a prioritized listing of images responsive to a user search query, in which the prioritized listing of images includes a different listing of images for each cluster of compositional similarity based on the metadata of each image associated with the cluster.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for identification of a salient portion of an image. A system may receive user input identifying a search query for content from a client device. The system may determine a listing of images responsive to the search query from an image collection. The system may obtain one or more image crops for at least one image of the listing of images based on a saliency map of the at least one image. In one or more implementations, each of the one or more image crops indicates a salient region of a corresponding image. The system may provide a set of search results responsive to the search query to the client device. In one or more implementations, the set of search results includes the obtained one or more image crops in a prioritized listing of image crops.

Abstract: Computer image clustering systems and methods for conducting effective media searches by grouping multimedia documents tagged by keywords into a hierarchy of images configured to: (1) maintain a first database, (2) maintain an initial occurrence matrix, (3) maintain an occurrence matrix, (4) maintain a media file activation score for each media file in the first database, (5) generate a log version of the occurrence matrix, (6) maintain an inverse media file frequency value for each descriptive term in the first database, (7) generate a descriptive term frequency matrix and generate a list of document vectors in multidimensional space (list), and (8) organize and process each media file in the list into a high activation score category and a low activation score category.

Abstract: Methods for predicting network access probability of data files accessible over a computer network are provided. In one aspect, a method includes generating a primary data vector for a media file based on a stored data representation of the file, and providing the data vector for the file to an algorithm that uses past interaction information for at least one other media file from a collection of media files having a degree of similarity with the media file above a threshold similarity value. The method also includes receiving, as an output of the algorithm, a marketability score for the media file, the score indicative of a likelihood that a user will download the media file. Systems and machine-readable media are also provided.

Abstract: Methods for analyzing data files to identify similar files to group for display within a limited visual space of a graphical user interface are provided. In one aspect, a method includes receiving a search query for a collection of media files, and identifying a subset of the media files from the collection that is responsive to the search query. The method also includes grouping the subset of the media files into a plurality of groups based on their visual similarity, wherein the visual similarity of each media file in the subset of media files is determined using an image vector corresponding to each media file, and providing the subset of the media files for display in their respective groups. Systems and machine-readable media are also provided.

Abstract: Methods, systems, and machine-readable media for identifying synthetic media file examples to train a supervised machine learned ranking algorithm to rank relevance of media files to a search query are provided. In one aspect, a method includes identifying a search query for a search engine for a collection of media files from previously received search queries, and selecting at least one training media file from the collection as a synthetic negative example for the search query. The method also includes providing a training set to a supervised machine learned ranking algorithm. The training set includes an identification of the search query, a copy of the training media file, and a first indicator that the training media file is a synthetic negative example for the search query. The method further includes providing, to the algorithm, the search query and the collection, and receiving, from the algorithm, a ranking of the collection.