# US 2022/0092443 A1 · Automated knowledge base

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

- Number: US 2022/0092443 A1
- Document ID: US20220092443A1
- Status: Published application
- Kind: A1
- Application number: 17/440,472
- Filed: 2021-05-05
- Published: 2022-03-24
- Earliest claimed: 2020-05-06
- Assignees: Morgan Stanley Services Group Inc.
- Applicants: Morgan Stanley Services Group Inc.
- Inventors: Nicolas SEYOT; Richard J. HEISE; Ziad GEMAYEL; Mohamed MOUINE
- Web page: https://patentgrep.com/patent/US20220092443A1

## CPC

- G06F 8/30: Creation or generation of source code
- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06F 40/205: Parsing
- G06F 40/295: Named entity recognition
- G06F 40/30: Semantic analysis
- G06F 40/40: Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Domain specific knowledge base (KB) contains all concepts from domain and the semantic relations between concepts. The concepts and the semantic relations are extracted from an existing corpus of content for the domain. The World Wide Web Consortium (W3C) standard SKOS (Simple Knowledge Organization System) can be used and two types of semantic relations can be captured: hierarchal and associative. A Natural Language Processing (NLP) software engine can parse the input text to create a semantic knowledge graph, which is then mapped to a SKOS knowledge model. During the linguistic understanding of the text, relevant domain concepts are identified and connected by semantic links. Concepts automatically identified as most important in this domain can be promoted to another layer, referred to as “Topics.”

## Claims

### Claim 1

A computer system for generating an automated knowledge base, the computer system comprising:
a processor core;
a memory in communication with the processor core, wherein the memory stores software that when executed by the processor core, causes the processor core to:
extract semantic triples from text in a set of domain-specific electronic documents, wherein the extracted triples are represented as subject, predicate and object, such that the subject and the object have a semantic relationship;
generate an ontology from the exacted semantic triples; and
generate a Simple Knowledge Organization System knowledge base (SKOS KB) from the ontology.

### Claim 2 (depends on 1)

The computer system of claim 1, wherein the memory further stores software that when executed by the processor core causes the processor core to identify topics in the SKOS KB, wherein identifying the topics comprises:
for each concept in the SKOS KB, constructing a computational cluster graph around a centroid that is the concept, wherein the computational cluster graph comprises, as vertices, other concepts in the SKOS KB that are related and narrower than the centroid;
computing a score for each of the computational cluster graphs according to a scoring algorithm;
identifying as topic candidates one or more concepts whose corresponding computational cluster graph has a score greater than a threshold value; and
validating the topic candidates to identify one or more topics.

### Claim 3 (depends on 2)

The computer system of claim 2, wherein the computational cluster graph comprises relations between each of the vertices in the computational cluster graph.

### Claim 4 (depends on 2)

The computer system of claim 2, wherein the threshold value is an average of the scores for the computational cluster graphs.

### Claim 5 (depends on 2)

The computer system of claim 2, wherein the memory further stores software that when executed by the processor core causes the processor core to validate a topic candidate as a topic by excluding each identified topic candidate that meets at least one of the following conditions:
the identified topic candidate has at most one child in the SKOS KB; and
the identified topic candidate has at most one broader concept.

### Claim 6 (depends on 5)

The computer system of claim 5, wherein the memory further stores software that when executed by the processor core causes the processor core to discover relations between the concepts to validate topic candidates using a deep learning algorithm.

### Claim 7

A method for generating an automated knowledge base, the method comprising:
extracting, by a computer system, semantic triples from text in a set of domain-specific electronic documents, wherein the extracted triples are represented as subject, predicate and object, such that the subject and the object have a semantic relationship;
generating, by the computer system, an ontology from the exacted semantic triples; and
generating, by the computer system, a Simple Knowledge Organization System knowledge base (SKOS KB) from the ontology

### Claim 8 (depends on 7)

The method of claim 7, wherein identifying topics in the SKOS KB comprises:
for each concept in the SKOS KB, constructing a computational cluster graph around a centroid that is the concept, wherein the computational cluster graph comprises, as vertices, other concepts in the SKOS KB that are related and narrower than the centroid;
computing a score for each of the computational cluster graphs according to a scoring algorithm;
identifying as topic candidates one or more concepts whose corresponding computational cluster graph has a score greater than a threshold value; and
validating the topic candidates to identify one or more topics.

### Claim 9 (depends on 8)

The method of claim 8, wherein the computational cluster graph comprises relations between each of the vertices in the computational cluster graph.

### Claim 10 (depends on 8)

The method of claim 8, wherein the threshold value is an average of the scores for the computational cluster graphs.

### Claim 11 (depends on 8)

The method of claim 8, wherein validating a topic candidate as a topic comprises excluding each identified topic candidate that meets at least one of the following conditions:
the identified topic candidate has at most one child in the SKOS KB; and
the identified topic candidate has at most one broader concept.

### Claim 12 (depends on 11)

The method of claim 11, further comprising discovering, by the computer system, relations between the concepts to validate topic candidates using a deep learning algorithm.
