US 2026/0277994 A1 · MEDIA UNIT RETRIEVAL AND RELATED PROCESSES

Abstract

Media unit retrieval methods, systems and computer program products are provided that allow a user to search for an item by iteratively presenting media units such as images representing items to the user and receiving user input consisting of selections of the presented media units (including possibly the empty selection). Features, or attributes, a user is interested in, for example semantic features, are inferred from the interaction and media units are retrieved for presentation based on similarity with user-selected media units, through sampling of a probability distribution describing the intent or interests, or combinations of approaches. Accordingly, the user-experience is akin to a conversation about what the user is looking for. Retrieval may be based on both selected and unselected media units and the selection may comprise making a selection with a single action. Further, a database of media units can capture similarity relationships for efficient media unit retrieval.

Claims26

Dependent claims

A method for decreasing an amount of time necessary for an electronic system to select relevant media units or identifiers thereof, the method comprising:

inputting a first one or more media units into a semantic model comprising an artificial neural network;
calculating a first vector comprising a first one or more values in an attribute space, the one or more values derived from neural activations of at least the artificial neural network presented with an image associated with at least one of the first one or more media units;
receiving a user input relating to a second one or more media units;
calculating a second vector μsing an intent model and the user input, the second vector comprising a second one or more values representing an estimate of the user's interests in the attribute space;
computing a measure of similarity between the first vector and the second vector;
and
selecting, using the measure of similarity, one or more relevant media units from the first one or more media units, or identifiers thereof.

4 more elements

A system for decreasing an amount of time necessary to select relevant media units or identifiers thereof, the system comprising:

a database comprising a first one or more media units;
one or more servers configured to access the first one or more media units from the database, wherein the one or more servers are further configured to
implement a semantic model comprising an artificial neural network, the semantic model configured to receive a first input and calculate a first vector, wherein the first input comprises the first one or more media units, wherein the first vector comprises a first one or more values in an attribute space, the one or more values derived from neural activations of at least the artificial neural network presented with an image associated with at least one of the first one or more media units;
implement an intent model configured to receive a second input and calculate a second vector, wherein the second input comprises a user input relating to a second one or more media units, wherein the second vector comprises a second one or more values representing an estimate of the user's interests in the attribute space;
and
compute a measure of similarity between the first vector and the second vector and select, using the measure of similarity, one or more relevant media units from the database, or identifiers thereof.

3 more elements

Dependencies are inferred from the claim text.

Media unit retrieval and related processes

Life of this patent

Media unit retrieval and related processes

Life of this patent

Abstract

Claims26