# US 2020/0394567 A1 · Target document template generation

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

- Number: US 2020/0394567 A1
- Document ID: US20200394567A1
- Status: Published application
- Kind: A1
- Application number: 16/442,335
- Filed: 2019-06-14
- Published: 2020-12-17
- Earliest claimed: 2019-06-14
- Assignees: The Toronto-Dominion Bank
- Applicants: The Toronto-Dominion Bank
- Inventors: Mok Choe; Thomas M. Kerigan; Salvatore Aspro; Evgenia Syromyatnikova
- Web page: https://patentgrep.com/patent/US20200394567A1

## CPC

- G06N 20/20: Ensemble learning
- G06F 16/93: Document management systems
- G06F 40/186: Templates
- G06F 40/279: Recognition of textual entities
- G06F 40/56: Natural language generation
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06Q 10/10: Office automation; Time management
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

To automatically generate a project document, a server in a computing environment receives input documents associated with a project, and extracts a set of features from each input document. The server determines a frequency of the words in each input document and stores the frequencies in relation to the words in the sets of words. The server than applies a document type machine-learned model to a set of words for each input document to infer a document type. The document machine-learned model may be trained using a bag-of-words representation. The server then applies a architecture pattern machine-learned model the set of input documents to determine a target architecture pattern. The server automatically generates a project document for the project based on the document types and inferred architecture pattern.

## Claims

### Claim 1

A computer system comprising:
a computer processor; and
a non-transitory computer-readable storage medium storage having instructions that when executed by the computer processor perform actions comprising:
receiving a set of input documents, the set of input documents describing a project;
extracting a set of features associated with each input document;
applying, for each input document, a document type machine-learned model to the set of features associated with the input document to infer a document type of the input document;
applying an architecture pattern machine-learned model to a concatenated set of features for the set of input documents to determine a target architecture pattern for a project document associated with the project; and
generating the project document for the project based on the document types for the set of input documents and the target architecture pattern.

### Claim 2 (depends on 1)

The computer system of claim 1, the actions further comprising:
evaluating the project document against previous project documents or input documents from the set of input documents using cosine similarity and term frequency-inverse document frequency.

### Claim 3 (depends on 1)

The computer system of claim 1, wherein the document type machine-learned model and the architecture pattern machine-learned model include random forest classifiers.

### Claim 4 (depends on 1)

The computer system of claim 1, wherein the architecture pattern machine-learned model includes a plurality of random forest classifiers.

### Claim 5 (depends on 1)

The computer system of claim 1, wherein the document type machine-learned model and the architecture pattern machine-learned model are configured to receive a bag-of-words representation as the set of features for the input document.

### Claim 6 (depends on 1)

The computer system of claim 1, wherein the set of features for each input document are a subset of words based on frequency of occurrence within the input document.

### Claim 7

A non-transitory computer-readable storage medium comprising instructions executable by a processor, the instructions comprising steps for the processor to:
receive a set of input documents, the set of input documents describing a project;
extract a set of features associated with each input document;
apply, for each input document, a document type machine-learned model to the set of features associated with the input document to infer a document type of the input document;
apply an architecture pattern machine-learned model to a concatenated set of features for the set of input documents to determine a target architecture pattern for a project document associated with the project; and
generate the project document for the project based on the document types for the set of input documents and the target architecture pattern.

### Claim 8 (depends on 7)

The non-transitory computer-readable storage medium of claim 7, the instructions further comprising:
evaluating the project document against previous project documents or input documents from the set of input documents using cosine similarity and term frequency-inverse document frequency.

### Claim 9 (depends on 7)

The non-transitory computer-readable storage medium of claim 7, wherein the document machine-learned model and the architecture pattern machine-learned model include random forest classifiers.

### Claim 10 (depends on 7)

The non-transitory computer-readable storage medium of claim 7, wherein the architecture pattern machine-learned model includes a plurality of random forest classifiers.

### Claim 11 (depends on 7)

The non-transitory computer-readable storage medium of claim 7, wherein the document type machine-learned model and the architecture pattern machine-learned model are configured to receive a bag-of-words representation as the set of features for the input documents.

### Claim 12 (depends on 7)

The non-transitory computer-readable storage medium of claim 7, wherein the set of features for each input document are a subset of words based on frequency of occurrence within the input document.

### Claim 13

A computer-implemented method for automated document generation, the method comprising:
receiving a set of input documents, the set of input documents describing a project;
extracting a set of features associated with each input document;
applying, for each input document, a document type machine-learned model to the set of features associated with the input document to infer a document type of the input document;
applying an architecture pattern machine-learned model to a concatenated set of features for the set of input documents to determine a target architecture pattern for a project document associated with the project; and
generating the project document for the project based on the document types for the set of input documents and the target architecture pattern.

### Claim 14 (depends on 13)

The computer-implemented method of claim 13, further comprising:
evaluating the project document against previous project documents or input documents from the set of input documents using cosine similarity and term frequency-inverse document frequency.

### Claim 15 (depends on 13)

The computer-implemented method of claim 13, wherein the document type machine-learned model and the architecture pattern machine-learned model are random forest classifiers.

### Claim 16 (depends on 13)

The computer-implemented method of claim 13, wherein the architecture pattern machine-learned model is a plurality of random forest classifiers.

### Claim 17 (depends on 13)

The computer-implemented method of claim 13, wherein the document type machine-learned model and the architecture pattern machine-learned model are configured to receive a bag-of-words representation as the set of features for the input documents.

### Claim 18 (depends on 13)

The computer-implemented method of claim 13, wherein the set of features for each input document are a subset of words based on frequency of occurrence within the input document.
