Back to results
US 11,893,065 B2Granted
Document analysis architecture
AssigneeAon Risk Services, Inc. of Maryland
InventorsSamuel Cameron Fleming, John E. Bradley, III, Lewis C. Lee, Jared Dirk Sol, Scott Buzan, Timothy Seegan
Life of this patent
Estimating the term…
Abstract
Systems and methods for generation and use of document analysis architectures are disclosed. A model builder component may be utilized to receiving user input data for labeling a set of documents as in class or out of class. That user input data may be utilized to train one or more classification models, which may then be utilized to predict classification of other documents. Trained models may be incorporated into a model taxonomy for searching and use by other users for document analysis purposes.
Claims20
A system, comprising:
- one or more processors;
- and
- non-transitory computer-readable media storing computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
- storing a taxonomy of machine-learning classification models, individual ones of the machine-learning classification models configured to receive documents and predict a classification of documents in a document set utilizing predictive analytical techniques, the individual ones of the machine-learning classification models trained based at least in part on a document dataset indicated to be in class from first user input data, the documents comprising patents and patent applications, wherein the individual ones of the machine-learning classification models differ from the documents that the individual ones of the machine-learning classification models are configured to receive and analyze;
- generating a user interface configured to accept second user input data representing a search query, the search query including keywords from the second user input data;
- receiving the second user input data representing the search query;
- generating, utilizing the keywords, a first dataset based at least in part on the search query;
- determining, utilizing the first dataset, that at least one machine-learning classification model of the taxonomy of machine-learning classification models was trained by a second dataset similar to the first dataset;
- generating, based at least in part on the determining, a recommended machine-learning classification model of the machine-learning classification models determined to be most related to the search query;
- and
- based at least in part on the determining, causing display, via the user interface, of search results for the search query, the search results displaying a visual representation of the taxonomy of machine-learning models along with an emphasized portion of the visual representation associated with the at least one machine-learning classification model of the taxonomy of machine-learning models trained by the second dataset, the search results also indicating the recommended machine-learning classification model.
8 more elements
A method, comprising:
- storing a taxonomy of machine-learning models trained to predict a classification of documents in a document set utilizing predictive analytical techniques, individual ones of the machine-learning models trained based at least in part on a document dataset indicated to be in class at least in part from first user input data, wherein the individual ones of the machine-learning models differ from the documents that the individual ones of the machine-learning models are configured to receive and analyze;
- generating a user interface configured to accept second user input data representing a search query, the search query including keywords from the second user input data;
- receiving the second user input data representing a search query, the search query including keywords from the first user input data;
- generating, utilizing the keywords, a first dataset based at least in part on the search query;
- determining, utilizing the first dataset, that at least one machine-learning model of the taxonomy of machine-learning models was trained by a second dataset similar to the first dataset;
- generating based at least in part on the determining, a recommended machine-learning classification model of the machine-learning classifications determined to be most related to the search query;
- and
- based at least in part on the determining, causing display of search results for the search query, the search results displaying a visual representation of the taxonomy along with an emphasized portion of the visual representation associated with the at least one machine-learning classification model of the taxonomy of machine-learning models trained by the second dataset, the search results also indicating the recommended machine-learning classification model.
5 more elements
A system comprising:
- one or more processors;
- and
- non-transitory computer-readable media storing computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
- storing a taxonomy of machine-learning models trained to predict a classification of documents in a document set utilizing predictive analytical techniques, individual ones of the machine-learning models trained based at least in part on a document dataset indicated to be in class at least in part from first user input data, wherein the individual ones of the machine-learning models differ from the documents that the models are configured to receive and analyze;
- generating a user interface configured to accept second user input data representing a search query, the search query including keywords from the second user input data;
- receiving second user input data representing a search query, the search query including keywords from the first user input data;
- generating, utilizing the keywords, a first dataset based at least in part on the search query;
- determining, utilizing the first dataset, that at least one machine-learning model of the taxonomy of machine-learning models was trained by a second dataset similar to the first dataset;
- generating, based at least in part on the determining, a recommended machine-learning classification model of the machine-learning models determined to be most related to the search query;
- and
- based at least in part on the determining, causing display of search results for the search query, the search results displaying a visual representation of the taxonomy along with an emphasized portion of the visual representation associated with the at least one machine-learning classification model of the taxonomy of machine-learning models trained by the second dataset, the search results also indicating the recommended machine-learning classification model.
8 more elements
Dependencies are inferred from the claim text.