US 10,540,381 B1Granted

Techniques and components to find new instances of text documents and identify known response templates

AssigneeCapital One Services, LLC

InventorsJoerg Rings, William Thomas Romano, Andre Gatorano

Life of this patent

Estimating the term…

Loading patent…

Abstract

Disclosed are examples of a system, techniques and non-transitory computer-readable storage medium that enable the selection of a response document template for generating an appropriate response to notification document. Respective components may generate an electronic document from the notification document. A common theme between two or more of the electronic documents may be identified. The electronic documents identified as having the common theme may be grouped in a notification document corpus. The notification document corpus may be analyzed with reference to received notification documents stored in a document repository. Based on results of the analysis, a template of a notification response document appropriate for responding to a specific notification document received by the enterprise may be selected. The selected response document template may be forwarded to a client device for further processing.

Claims17

Dependent claims

A method, comprising:

recognizing alphanumeric characters in a notification document that was received by an enterprise, wherein the alphanumeric characters are recognized using an optical character recognition algorithm;
identifying, by a processor, patterns of alphanumeric characters in the recognized alphanumeric characters;
storing the identified patterns as a collection of alphanumeric character patterns in a memory;
arranging, in clusters by a clustering algorithm, classified notification documents of a corpus of classified notification documents, wherein:
each respective classified notification document in the corpus is assigned a common theme and includes a group of alphanumeric character patterns,
each cluster of the clusters within the corpus indicates a separate common theme,
membership in each cluster is based on an intrinsic similarity value between pairs of classified notification documents within the cluster,
the intrinsic similarity value is a mean value of a cosine similarity between the pairs of classified notification documents within the cluster, and
the classified notification documents in a respective cluster of the corpus are grouped based on the intrinsic similarity value;
accessing, by the processor, the corpus of classified notification documents;
iteratively comparing, by the processor, the collection of alphanumeric character patterns to the group of alphanumeric character patterns in each respective classified notification document in the corpus;
based on a result of each comparison of the collection of alphanumeric character patterns in the received notification document to the group of alphanumeric character patterns in each respective classified notification document of the corpus, determining a respective cosine similarity rating between the received notification document and each respective classified notification document in the corpus, wherein:
each respective cosine similarity rating indicates a degree of similarity between each respective classified notification document in the corpus and the collection of alphanumeric characters patterns from the received notification document;
identifying at least one classified notification document of the corpus having a cosine similarity rating that exceeds a predetermined similarity threshold as matching the received notification document;
and
assigning, by the processor, a common theme to the received notification document, wherein the assigned common theme is a same common theme as assigned to the identified at least one classified notification document.

13 more elements

A system comprising:

a memory storing a document repository including a plurality of electronic documents related to a plurality of notification documents received by an enterprise and a plurality of document templates including notification response documents to be sent in response to at least one of the notification documents received by the enterprise;
a document text delivery component coupled to the document repository, and operable to access the document repository;
a model text component coupled to the document text delivery component;
and
a document analytics component coupled to the model text component;
wherein:
the document text delivery component is operable to deliver the plurality of electronic documents to the model text component for processing,
the model text component is operable to:
recognize alphanumeric characters in a notification document that was received by an enterprise, wherein the alphanumeric characters are recognized using an optical character recognition algorithm;
identify patterns of alphanumeric characters in the recognized alphanumeric characters;
store the identified patterns as a collection of alphanumeric character patterns in a memory;
arrange classified notification documents in the corpus in clusters within the corpus by a clustering algorithm, wherein:
each respective classified notification document in the corpus is assigned a common theme and includes a group of alphanumeric character patterns,
each cluster of the clusters within the corpus indicates a separate common theme,
membership in each cluster is based on an intrinsic similarity value between pairs of classified notification documents within the cluster, and
the intrinsic similarity value is a mean value of a cosine similarity between the pairs of classified notification documents within the cluster;
the classified notification documents in a respective cluster of the corpus are grouped based on the intrinsic similarity value;
access the corpus of classified notification documents;
iteratively compare the collection of alphanumeric character patterns to the group of alphanumeric character patterns in each respective classified notification document in the corpus;
based on a result of each comparison of the collection of alphanumeric character patterns in the received notification document to the group of alphanumeric character patterns in each respective classified notification document of the corpus, determine a respective cosine similarity rating between the received notification document and each respective classified notification document in the corpus, wherein:
each respective cosine similarity rating indicates a degree of similarity between each respective classified notification document in the corpus and the collection of alphanumeric characters patterns from the received notification document;
identify at least one classified notification document of the corpus having a cosine similarity rating that exceeds a predetermined similarity threshold as matching the received notification document;
and
assign a common theme to the received notification document, wherein the assigned common theme is a same common theme as assigned to the identified at least one classified notification document.

21 more elements

A non-transitory computer-readable storage medium storing computer-readable programming code executable by a processor, wherein execution of the computer-readable programming code causes the processor to:

recognize alphanumeric characters in a notification document that was received by an enterprise, wherein the alphanumeric characters are recognized using an optical character recognition algorithm;
identify patterns of alphanumeric characters in the recognized alphanumeric characters;
store the identified patterns as a collection of alphanumeric character patterns in a memory;
arrange classified notification documents in the corpus in clusters within the corpus by a clustering algorithm, wherein:
each respective classified notification document in the corpus is assigned a common theme and includes a group of alphanumeric character patterns,
each cluster of the clusters within the corpus indicates a separate common theme,
membership in each cluster is based on an intrinsic similarity value between pairs of classified notification documents within the cluster,
the intrinsic similarity value is a mean value of a cosine similarity between the pairs of classified notification documents within the cluster, and
the classified notification documents in a respective cluster of the corpus are grouped based on the intrinsic similarity value;
access the corpus of classified notification documents;
iteratively compare the collection of alphanumeric character patterns to the group of alphanumeric character patterns in each respective classified notification document in the corpus;
based on a result of each comparison of the collection of alphanumeric character patterns in the received notification document to the group of alphanumeric character patterns in each respective classified notification document of the corpus, determine a respective cosine similarity rating between the received notification document and each respective classified notification document in the corpus, wherein:
each respective cosine similarity rating indicates a degree of similarity between each respective classified notification document in the corpus and the collection of alphanumeric characters patterns from the received notification document;
identify at least one classified notification document of the corpus having a cosine similarity rating that exceeds a predetermined similarity threshold as matching the received notification document;
and
assign a common theme to the received notification document, wherein the assigned common theme is a same common theme as assigned to the identified at least one classified notification document.

13 more elements

Dependencies are inferred from the claim text.