# US 2021/0065041 A1 · Analyzing documents using machine learning

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0065041 A1
- Document ID: US20210065041A1
- Status: Published application
- Kind: A1
- Application number: 16/552,471
- Filed: 2019-08-27
- Published: 2021-03-04
- Earliest claimed: 2019-08-27
- Assignees: Bank of America Corporation
- Applicants: Bank of America Corporation
- Inventors: Kishore Gopalan; Satish Chaduvu; Thomas J. Kuchcicki
- Web page: https://patentgrep.com/patent/US20210065041A1

## CPC

- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06F 16/353: into predefined classes
- G06F 40/103: Formatting, i.e. changing of presentation of documents (automatic justification G06F40/189; automatic line break hyphenation G06F40/191)
- G06F 40/117: Tagging; Marking up (details of markup languages G06F40/143); Designating a block; Setting of attributes (style sheets, e.g. eXtensible Stylesheet Language Transformation [XSLT], G06F40/154)
- G06F 40/30: Semantic analysis
- G06N 3/0499: Feedforward networks
- G06N 3/09: Supervised learning
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06F 18/22: Matching criteria, e.g. proximity measures
- G06F 18/24: Classification techniques
- G06F 18/241: relating to the classification model, e.g. parametric or non-parametric approaches
- G06F 40/10: Text processing (natural language analysis G06F40/20; semantic analysis G06F40/30; processing or translation of natural language G06F40/40)
- G06F 40/106: Display of layout of documents; Previewing
- G06N 3/045: Combinations of networks
- G06N 20/00: Machine learning
- G06V 30/414: Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A document analysis device that includes a memory operable to store a machine learning model configured to receive a sentence as an input and to output a classification identifier that is associated with a sentence type for the received sentence. The device further includes an artificial intelligence (AI) processing engine configured to receive a document comprising text, to sentences within the document, and to classify the sentences using the machine learning model. The AI processing engine is further configured to identify tagging rules for the document and to annotate one or more sentences from the document with a sentence type that matches a sentence type that is identified by the tagging rules for the document.

## Claims

### Claim 1

A document analysis device, comprising:
a memory operable to store:
a machine learning model configured to:
receive a sentence as an input; and
output a classification identifier for the received sentence, wherein the classification identifier identifies a sentence type; and
an artificial intelligence (AI) processing engine implemented by a processor operably coupled to the memory, configured to:
receive a document comprising text;
identify a plurality of sentences within the document;
classify the plurality of sentences using the machine learning model, wherein classifying the plurality of sentences associates each sentence with a classification identifier that identifies a sentence type;
identify tagging rules for the document, wherein the tagging rules identify one or more sentence types;
identify one or more sentences from the plurality of sentences with a sentence type that matches a sentence type identified by the tagging rules; and
annotate the one or more identified sentences within the document, wherein annotating the one or more identified sentences changes a format of the one or more identified sentences.

### Claim 2 (depends on 1)

The device of claim 1, wherein:
the AI processing engine is configured to:
receive a user input that modifies an annotated sentence;
add the modified annotated sentence to a set of sentences for training the machine learning model; and
retrain the machine learning model using the set of sentences for training the machine learning model.

### Claim 3 (depends on 1)

The device of claim 1, wherein the AI processing engine is configured to:
compute similarity scores between an annotated sentence and a plurality of previously classified sentences, wherein a similarity score is a numeric value that represents how similar a pair of sentences are to each other based on the text within the pair of sentences;
identify a sentence from the plurality of previously classified sentences that corresponds with a similarity score that exceeds a similarity score threshold value, wherein the similarity score threshold value indicates a minimum similarity score for a pair of sentences to be considered alternatives of each other; and
output the identified sentence as an alternative sentence for the annotated sentence.

### Claim 4 (depends on 3)

The device of claim 3, wherein the plurality of previously classified sentences are each associated with a sentence type that is different from a sentence type associated with the annotated sentence.

### Claim 5 (depends on 1)

The device of claim 1, wherein classifying the plurality of sentences is based at least in part on verb tenses used in the plurality of sentences.

### Claim 6 (depends on 1)

The device of claim 1, wherein annotating the one or more identified sentences within the document comprises identifying a sentence type for the one or more identified sentences.

### Claim 7 (depends on 1)

The device of claim 1, wherein the AI processing engine is further configured to:
display the text from the document in a first application window; and
display the one or more annotated sentences in a second application window, wherein the second application window is different from the first application window.

### Claim 8

A document analysis method, comprising:
receiving a document comprising text;
identifying a plurality of sentences within the document;
classifying the plurality of sentences using a machine learning model, wherein:
the machine learning model configured to:
receive a sentence as an input; and
output a classification identifier for the received sentence, wherein the classification identifier identifies a sentence type; and
classifying the plurality of sentences associates each sentence with a classification identifier that identifies a sentence type;
identifying tagging rules for the document, wherein the tagging rules identify one or more sentence types;
identifying one or more sentences from the plurality of sentences with a sentence type that matches a sentence type identified by the tagging rules; and
annotating the one or more identified sentences within the document, wherein annotating the one or more identified sentences changes a format of the one or more identified sentences.

### Claim 9 (depends on 8)

The method of claim 8, further comprising:
receiving a user input that modifies an annotated sentence;
adding the modified annotated sentence to a set of sentences for training the machine learning model; and
retraining the machine learning model using the set of sentences for training the machine learning model.

### Claim 10 (depends on 8)

The method of claim 8, further comprising:
computing similarity scores between an annotated sentence and a plurality of previously classified sentences;
identifying a sentence from the plurality of previously classified sentences that corresponds with a similarity score that exceeds a similarity score threshold value, wherein the similarity score threshold value indicates a minimum similarity score for a pair of sentences to be considered alternatives of each other; and
outputting the identified sentence as an alternative sentence for the annotated sentence.

### Claim 11 (depends on 10)

The method of claim 10, wherein the plurality of previously classified sentences are each associated with a sentence type that is different from a sentence type associated with the annotated sentence.

### Claim 12 (depends on 8)

The method of claim 8, wherein classifying the plurality of sentences is based at least in part on verb tenses used in the plurality of sentences.

### Claim 13 (depends on 8)

The method of claim 8, wherein annotating the one or more identified sentences within the document comprises identifying a sentence type for the one or more identified sentences.

### Claim 14 (depends on 8)

The method of claim 8, wherein further comprising:
displaying the text from the document in a first application window; and
displaying the one or more annotated sentences in a second application window, wherein the second application window is different from the first application window.

### Claim 15

A computer program comprising executable instructions stored in a non-transitory computer readable medium that when executed by a processor causes the processor to:
receive a document comprising text;
identify a plurality of sentences within the document;
classify the plurality of sentences using a machine learning model, wherein:
the machine learning model configured to:
receive a sentence as an input; and
output a classification identifier for the received sentence, wherein the classification identifier identifies a sentence type; and
classifying the plurality of sentences associates each sentence with a classification identifier that identifies a sentence type;
identify tagging rules for the document, wherein the tagging rules identify one or more sentence types;
identify one or more sentences from the plurality of sentences with a sentence type that matches a sentence type identified by the tagging rules; and
annotate the one or more identified sentences within the document, wherein annotating the one or more sentences changes a format of the one or more identified sentences.

### Claim 16 (depends on 15)

The computer program of claim 15, further comprising instructions that when executed by the processor causes the processor to:
receive a user input that modifies an annotated sentence;
add the modified annotated sentence to a set of sentences for training the machine learning model; and
retrain the machine learning model using the set of sentences for training the machine learning model.

### Claim 17 (depends on 15)

The computer program of claim 15, further comprising instructions that when executed by the processor causes the processor to:
compute similarity scores between an annotated sentence and a plurality of previously classified sentences, wherein a similarity score is a numeric value that represents how similar a pair of sentences are to each other based on the text within the pair of sentences;
identify a sentence from the plurality of previously classified sentences that corresponds with a similarity score that exceeds a similarity score threshold value, wherein the similarity score threshold value indicates a minimum similarity score value for a pair of sentences to be considered alternatives of each other; and
output the identified sentence as an alternative sentence for the annotated sentence.

### Claim 18 (depends on 17)

The computer program of claim 17, wherein the plurality of previously classified sentences are each associated with a sentence type that is different from a sentence type associated with the annotated sentence.

### Claim 19 (depends on 15)

The computer program of claim 15, wherein classifying the plurality of sentences is based at least in part on verb tenses used in the plurality of sentences.

### Claim 20 (depends on 15)

The computer program of claim 15, wherein annotating the one or more identified sentences within the document comprises identifying a sentence type for the one or more identified sentences.
