Back to results
US 11,960,531 B2Granted
Content analysis
AssigneeQubeeo Ltd.
InventorJanet Bastiman
Life of this patent
Estimating the term…
Abstract
In one aspect a computer-implemented method of processing a content item to extract information from the content item comprises steps of: receiving the content item at a processing stage; and determining whether the content item satisfies a predetermined content condition, by i) providing content of the content item to a plurality of content analysers, each of which applies machine learning content analysis thereto, in order to make an independent determination of whether that predetermined content condition is satisfied, and provides a confidence score for that independent determination, and ii) making an overall determination of whether the content item satisfies that predetermined content condition based on the confidence scores provided by the content analysers for their respective independent determinations.
Claims14
A computer-implemented method of processing a content item to extract information from the content item, the method comprising the following steps:
- receiving the content item at a processing stage;
- determining whether the content item satisfies a predetermined content condition, by i) providing content of the content item to a plurality of independently configured content analysers, each of which applies machine learning content analysis thereto, in order to make an independent determination of whether that predetermined content condition is satisfied, and provides a confidence score for that independent determination, and ii) making an overall determination of whether the content item satisfies that predetermined content condition based on the confidence scores provided by the independently configured content analysers for their respective independent determinations;
- wherein an overall determination is made that the predetermined content condition is satisfied if a combined confidence condition is satisfied by two or more of the confidence scores in combination, or an individual confidence condition is satisfied by one of the two or more confidence scores individually;
- using at least one higher-level content analyser to determine that the content item belongs to a predetermined content category, the predetermined content condition being one of multiple content conditions associated with that content category, wherein the content category is defined such that any content item that satisfies any one of the associated content conditions belongs to that content category, wherein the determining step is performed in response to determining that the content item contains belongs to that content category;
- and
- wherein the content item is a visual content item and the predetermined content condition is that the content item contains a predetermined visual structure element, wherein the machine learning content analysis applied by each of the content analysers comprises image recognition applied to image data of the content item, so as to make an independent determination of whether that predetermined visual structure element is present, and wherein the content category corresponds to a category of visual structure and each of the content conditions corresponds to a different visual structure element in that category of visual structure.
3 more elements
Executable instructions stored on non-transitory media, the executable instructions configured, when executed on one or more processors of a content analysing system having a plurality of independently configured content analysers, to implement operations comprising:
- receiving a content item to be analysed;
- accessing a configuration file, which comprises content analysis settings in a defined hierarchy;
- selecting at least a first of the content analysers based on a first set of the content analysis settings;
- using the selected first content analyser to analyse the content item;
- if the results of the analysis using the first content analyser satisfy a predetermined condition, selecting a second set of one or more of the independently configured content analysers based on a second set of the content analysis settings, which is a child of the first set of content analysis settings within the defined hierarchy, and using the selected second set of independently configured content analysers to analyse the content item;
- wherein the results of the analysis by the first content analyser comprise a first set of classification results for a first set of content classes, and the results of the analysis by the second set of independently configured content analysers comprise a second set of classification results for a second set of content classes;
- wherein the predetermined condition is a trigger condition associated with a first content class of the first set of content classes in the configuration file, the second set of independently configured content analysers being used to analyse the content item if the classification result for the first content class satisfies the trigger condition associated with the first confidence class;
- and
- wherein the classification result for the first content class is stored in an output tree;
- wherein if the trigger condition is satisfied by the confidence score for the first content class such that the second set of content analysers is to analyse the content item, at least one classification result of the second set of classification results is stored in the output tree, as a child of the classification result for the first content class.
7 more elements
A content processing system comprising:
- an input configured to receive a content item for processing;
- and
- one or more processors configured to implement operations comprising:
- determining whether the content item satisfies a predetermined content condition, by i) providing content of the content item to a plurality of independently configured content analysers, each of which applies machine learning content analysis thereto, in order to make an independent determination of whether that predetermined content condition is satisfied, and provides a confidence score for that independent determination, and ii) making an overall determination of whether the content item satisfies that predetermined content condition based on the confidence scores provided by the independently configured content analysers for their respective independent determinations;
- wherein an overall determination is made that the predetermined content condition is satisfied if a combined confidence condition is satisfied by two or more of the confidence scores in combination, or an individual confidence condition is satisfied by one of the two or more confidence scores individually;
- using at least one higher-level content analyser to determine that the content item belongs to a predetermined content category, the predetermined content condition being one of multiple content conditions associated with that content category, wherein the content category is defined such that any content item that satisfies any one of the associated content conditions belongs to that content category, wherein the determining step is performed in response to determining that the content item contains belongs to that content category;
- and
- wherein the content item is a visual content item and the predetermined content condition is that the content item contains a predetermined visual structure element, wherein the machine learning content analysis applied by each of the content analysers comprises image recognition applied to image data of the content item, so as to make an independent determination of whether that predetermined visual structure element is present, and wherein the content category corresponds to a category of visual structure and each of the content conditions corresponds to a different visual structure element in that category of visual structure.
5 more elements
Dependencies are inferred from the claim text.