# US 2022/0277146 A1 · Determining topics and action items from conversations

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

- Number: US 2022/0277146 A1
- Document ID: US20220277146A1
- Status: Published application
- Kind: A1
- Application number: 17/682,682
- Filed: 2022-02-28
- Published: 2022-09-01
- Earliest claimed: 2020-09-01
- Assignees: Rammer Technologies, Inc.
- Applicants: Rammer Technologies, Inc.
- Inventors: Toshish Arun Jawale; Ansup Babu; Sekhar Vallath
- Web page: https://patentgrep.com/patent/US20220277146A1

## CPC

- G06F 40/35: Discourse or dialogue representation
- G06F 3/0481: based on specific properties of the displayed interaction object or a metaphor-based environment, e.g. interaction with desktop elements like windows or icons, or assisted by a cursor's changing behaviour or appearance
- G06F 3/04842: Selection of displayed objects or displayed text elements (G06F3/0482 takes precedence)
- G06F 40/166: Editing, e.g. inserting or deleting
- G06F 40/253: Grammatical analysis; Style critique
- G06F 40/268: Morphological analysis
- G06F 40/279: Recognition of textual entities
- G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking
- G06N 5/04: Inference or reasoning models
- G06N 20/00: Machine learning
- G06F 40/211: Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G10L 15/1815: {Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning}
- G10L 15/1822: {Parsing for meaning understanding}
- G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L 15/30: Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

Embodiments are directed to organizing conversation information. Two or more machine learning (ML) models and a plurality of sentences provided from a conversation may be employed to generate insight scores for each sentence such that each insight score correlates to a probability that its sentence includes one or more of an action or a question. In response to one or more sentences having insight scores that exceed a threshold value an information score and a definiteness score may be determined for the one or more sentences. And one or more insights associated with the conversation may be generated based on the one or more sentences. A report may be generated that associates the one or more insights with one or more portions of the conversation that include the one or more sentences that are associated with the insights.

## Claims

### Claim 1

A method for organizing conversation information over a network using one or more network computers that include one or more processors that execute instructions, wherein execution of the instructions arranges the one or more processors to enable the performance of actions, comprising:
employing two or more machine learning (ML) insight models to generate an insight score for each sentence in a conversation;
in response to one or more sentences associated with one or more insight scores that exceed a threshold value, performing further actions, including:
employing one or more ML information models to determine an information score for each of the one or more sentences;
determining a definiteness score for each of the one or more sentences based on decision-making words included in the one or more sentences, wherein each sentence that includes an amount of words associated with conclusive decision-making words are scored higher than each other sentence that includes another amount of words associated with the conclusive decision-making words that is less than the amount of words associated with the conclusive decision-making words;
generating one or more insights associated with the conversation based on the information score associated with each of the one or more sentences, and the definiteness score associated with each of the one or more sentences; and
generating a report that associates the one or more insights with one or more portions of the conversation that include the one or more sentences that are associated with the insights.

### Claim 2 (depends on 1)

The method of claim 1, wherein each insight score further comprises:
employing two or more partial insight scores that each correspond to an ML insight model, and wherein each insight score correlates to a probability that its sentence includes one or more of an action or a question.

### Claim 3 (depends on 1)

The method of claim 1, wherein the one or more ML information models further comprise:
determining the information score for each of the one or more sentences based on one or more of a number of included object words, a number of included subject words, a number of included insight words, or one or more included grammatical features.

### Claim 4 (depends on 1)

The method of claim 1, wherein the one or more ML information models further comprise:
providing for one or more of a data structure that include heuristics, a classifier, a rule, or an instruction; and
modifying the one or more ML information models based on one or more of a rule or an input.

### Claim 5 (depends on 1)

The method of claim 1, further comprising:
generating a graph model based on the conversation, wherein the graph model is employed to generate two or more partitions for the conversation;
generating one or more sentence models for the conversation based on one or more of a key word or a phrase;
generating a conversation model based on the one or more sentence models and the two or more partitions; and
employing the conversation model to determine one or more of a dominant message chunk of the conversation or a context change of the conversation.

### Claim 6 (depends on 1)

The method of claim 1, further comprising:
generating a preliminary graph model for the conversation;
generating one or more partitions for the preliminary graph model based on one or more greedy heuristics;
categorizing each of the one or more partitions as strong or weak based on one or more characteristics of each word and its associated relationships and interactions that are a member of each partition.

### Claim 7 (depends on 1)

The method of claim 1, further comprising:
in response to identifying one or more new words in the conversation, performing actions including:
adding a new node to a graph model of the conversation;
generating one or more new edges to the new node based on one or more relationships or interactions associated with the new word; and
modifying graph model based on the one or more relationships or interactions.

### Claim 8 (depends on 1)

The method of claim 1, further comprising:
in response to identifying a word in the conversation, performing actions including:
determining one or more characteristics of a word in the conversation;
determining one or more relationships or interactions of the word with one or more other words in the conversation;
associating the one or more characteristics with a word node in a graph model for the conversation, wherein the one or more relationships or interactions of the word are associated with one or more edges of the word node in the graph model.

### Claim 9 (depends on 1)

The method of claim 1, further comprising:
determining one or more phrasal scores for one or more portions of a sentence in the conversation;
generating one or more sub-graphs of the one or more words in the one or more portions of the sentence;
modifying the one or more phrasal scores based on one or more of word repetition, word degree or sentence length in the one or more sub-graphs;
providing a sentence keyword matrix for analysis of the conversation based on the one or more phrasal scores.

### Claim 10 (depends on 1)

The method of claim 1, wherein generating the report, further comprises:
automatically generating one or more of a task, an appointment, a reminder, or a notification based on the one or more insights;
providing one or more of the task, the appointment, the reminder, or the notification to one or more internal service or external service.

### Claim 11

A system for organizing conversation information over a network, comprising:
a network computer, comprising:
a memory that stores at least instructions; and
one or more processors that execute instructions, wherein execution of the instructions arranges the one or more processors to enable the performance of actions, comprising:
employing two or more machine learning (ML) insight models to generate an insight score for each sentence in a conversation;
in response to one or more sentences associated with one or more insight scores that exceed a threshold value, performing further actions, including:
employing one or more ML information models to determine an information score for each of the one or more sentences;
determining a definiteness score for each of the one or more sentences based on decision-making words included in the one or more sentences, wherein each sentence that includes an amount of words associated with conclusive decision-making words are scored higher than each other sentence that includes another amount of words associated with the conclusive decision-making words that is less than the amount of words associated with the conclusive decision-making words;
generating one or more insights associated with the conversation based on the information score associated with each of the one or more sentences, and the definiteness score associated with each of the one or more sentences; and
generating a report that associates the one or more insights with one or more portions of the conversation that include the one or more sentences that are associated with the insights; and
a client computer, comprising:
another memory that stores at least instructions; and
one or more processors that execute instructions, wherein execution of the instructions arranges the one or more processors to enable the performance of actions, including:
providing the plurality of sentences to the network computer; and
displaying the report.

### Claim 12 (depends on 11)

The system of claim 11, wherein each insight score further comprises:
employing two or more partial insight scores that each correspond to an ML insight model, and wherein each insight score correlates to a probability that its sentence includes one or more of an action or a question.

### Claim 13 (depends on 11)

The system of claim 11, wherein the one or more ML information models further comprise:
determining the information score for each of the one or more sentences based on one or more of a number of included object words, a number of included subject words, a number of included insight words, or one or more included grammatical features.

### Claim 14 (depends on 11)

The system of claim 11, wherein the one or more ML information models further comprise:
providing for one or more of a data structure that include heuristics, a classifier, a rule, or an instruction; and
modifying the one or more ML information models based on one or more of a rule or an input.

### Claim 15 (depends on 11)

The system of claim 11, further comprising:
generating a graph model based on the conversation, wherein the graph model is employed to generate two or more partitions for the conversation;
generating one or more sentence models for the conversation based on one or more of a key word or a phrase;
generating a conversation model based on the one or more sentence models and the two or more partitions; and
employing the conversation model to determine one or more of a dominant message chunk of the conversation or a context change of the conversation.

### Claim 16 (depends on 11)

The system of claim 11, further comprising:
generating a preliminary graph model for the conversation;
generating one or more partitions for the preliminary graph model based on one or more greedy heuristics;
categorizing each of the one or more partitions as strong or weak based on one or more characteristics of each word and its associated relationships and interactions that are a member of each partition.

### Claim 17 (depends on 11)

The system of claim 11, further comprising:
in response to identifying one or more new words in the conversation, performing actions including:
adding a new node to a graph model of the conversation;
generating one or more new edges to the new node based on one or more relationships or interactions associated with the new word; and
modifying graph model based on the one or more relationships or interactions.

### Claim 18 (depends on 11)

The system of claim 11, further comprising:
in response to identifying a word in the conversation, performing actions including:
determining one or more characteristics of a word in the conversation;
determining one or more relationships or interactions of the word with one or more other words in the conversation;
associating the one or more characteristics with a word node in a graph model for the conversation, wherein the one or more relationships or interactions of the word are associated with one or more edges of the word node in the graph model.

### Claim 19 (depends on 11)

The system of claim 11, further comprising:
determining one or more phrasal scores for one or more portions of a sentence in the conversation;
generating one or more sub-graphs of the one or more words in the one or more portions of the sentence;
modifying the one or more phrasal scores based on one or more of word repetition, word degree or sentence length in the one or more sub-graphs;
providing a sentence keyword matrix for analysis of the conversation based on the one or more phrasal scores.

### Claim 20 (depends on 11)

The system of claim 11, wherein generating the report, further comprises:
automatically generating one or more of a task, an appointment, a reminder, or a notification based on the one or more insights;
providing one or more of the task, the appointment, the reminder, or the notification to one or more internal service or external service.

### Claim 21

A network computer that is arranged to organize conversation information over a network, comprising:
a memory that stores at least instructions; and
one or more processors that execute instructions, wherein execution of the instructions arranges the one or more processors to enable the performance of actions, comprising:
employing two or more machine learning (ML) insight models to generate an insight score for each sentence in a conversation;
in response to one or more sentences associated with one or more insight scores that exceed a threshold value, performing further actions, including:
employing one or more ML information models to determine an information score for each of the one or more sentences;
determining a definiteness score for each of the one or more sentences based on decision-making words included in the one or more sentences, wherein each sentence that includes an amount of words associated with conclusive decision-making words are scored higher than each other sentence that includes another amount of words associated with the conclusive decision-making words that is less than the amount of words associated with the conclusive decision-making words;
generating one or more insights associated with the conversation based on the information score associated with each of the one or more sentences, and the definiteness score associated with each of the one or more sentences; and
generating a report that associates the one or more insights with one or more portions of the conversation that include the one or more sentences that are associated with the insights.

### Claim 22 (depends on 21)

The network computer of claim 21, further comprising:
generating a graph model based on the conversation, wherein the graph model is employed to generate two or more partitions for the conversation;
generating one or more sentence models for the conversation based on one or more of a key word or a phrase;
generating a conversation model based on the one or more sentence models and the two or more partitions; and
employing the conversation model to determine one or more of a dominant message chunk of the conversation or a context change of the conversation.

### Claim 23 (depends on 21)

The network of claim 21, further comprising:
generating a preliminary graph model for the conversation;
generating one or more partitions for the preliminary graph model based on one or more greedy heuristics;
categorizing each of the one or more partitions as strong or weak based on one or more characteristics of each word and its associated relationships and interactions that are a member of each partition.

### Claim 24 (depends on 21)

The network computer of claim 21, further comprising:
in response to identifying one or more new words in the conversation, performing actions including:
adding a new node to a graph model of the conversation;
generating one or more new edges to the new node based on one or more relationships or interactions associated with the new word; and
modifying graph model based on the one or more relationships or interactions.

### Claim 25 (depends on 21)

The network computer of claim 21, further comprising:
determining one or more phrasal scores for one or more portions of a sentence in the conversation;
generating one or more sub-graphs of the one or more words in the one or more portions of the sentence;
modifying the one or more phrasal scores based on one or more of word repetition, word degree or sentence length in the one or more sub-graphs;
providing a sentence keyword matrix for analysis of the conversation based on the one or more phrasal scores.

### Claim 26

A processor readable non-transitory storage media that includes instructions for organizing conversation information over a network, wherein execution of the instructions by one or more hardware processors enable the performance of actions, comprising:
employing two or more machine learning (ML) insight models to generate an insight score for each sentence in a conversation;
in response to one or more sentences associated with one or more insight scores that exceed a threshold value, performing further actions, including:
employing one or more ML information models to determine an information score for each of the one or more sentences;
determining a definiteness score for each of the one or more sentences based on decision-making words included in the one or more sentences, wherein each sentence that includes an amount of words associated with conclusive decision-making words are scored higher than each other sentence that includes another amount of words associated with the conclusive decision-making words that is less than the amount of words associated with the conclusive decision-making words;
generating one or more insights associated with the conversation based on the information score associated with each of the one or more sentences, and the definiteness score associated with each of the one or more sentences; and
generating a report that associates the one or more insights with one or more portions of the conversation that include the one or more sentences that are associated with the insights.

### Claim 27 (depends on 26)

The processor readable non-transitory storage media of claim 26, further comprising:
generating a graph model based on the conversation, wherein the graph model is employed to generate two or more partitions for the conversation;
generating one or more sentence models for the conversation based on one or more of a key word or a phrase;
generating a conversation model based on the one or more sentence models and the two or more partitions; and
employing the conversation model to determine one or more of a dominant message chunk of the conversation or a context change of the conversation.

### Claim 28 (depends on 26)

The processor readable non-transitory storage media of claim 26, further comprising:
generating a preliminary graph model for the conversation;
generating one or more partitions for the preliminary graph model based on one or more greedy heuristics;
categorizing each of the one or more partitions as strong or weak based on one or more characteristics of each word and its associated relationships and interactions that are a member of each partition.

### Claim 29 (depends on 26)

The processor readable non-transitory storage media of claim 26, further comprising:
in response to identifying one or more new words in the conversation, performing actions including:
adding a new node to a graph model of the conversation;
generating one or more new edges to the new node based on one or more relationships or interactions associated with the new word; and
modifying graph model based on the one or more relationships or interactions.

### Claim 30 (depends on 26)

The processor readable non-transitory storage media of claim 26, further comprising:
determining one or more phrasal scores for one or more portions of a sentence in the conversation;
generating one or more sub-graphs of the one or more words in the one or more portions of the sentence;
modifying the one or more phrasal scores based on one or more of word repetition, word degree or sentence length in the one or more sub-graphs;
providing a sentence keyword matrix for analysis of the conversation based on the one or more phrasal scores.
