Text Input Classification for Structured Relation Generation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current information extraction methods from unstructured text, such as cluttered message inboxes, fail to effectively classify and utilize relevant information in real-time, leading to missed offers and inefficient user experiences due to the inability to establish relational associations between different categories of text inputs.

Innovation Solution

A method and apparatus that classify text inputs into categories, extract entities, and generate structured relation information by associating entities from different text inputs, using keyword detection, word embeddings, and mapping tables, to provide users with relevant insights and recommendations in an intuitive display format.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If text inputs are stored in unstructured format in message inbox, then large volume of data can be stored, but information extraction and classification cannot be performed effectively

Engineering Contradiction:

Improvevolume of text data storedVSAvoidinformation extraction efficiency

Core Design Contradiction:

Quantity of substanceVSProductivity

Solution Approach 1:

The patent segments unstructured text inputs into distinct categories (offers, reminders, transactions, greetings, information sharing, requests, acknowledgments) using classification models. This segmentation enables effective information extraction by organizing the large volume of stored text data into manageable, meaningful groups that can be processed efficiently.

Inventive Principle:

Principle #1Segmentation

2Quantity of substance

If multiple messages are stored in cluttered inbox, then abundant data is available, but difficulty to identify required information increases

Engineering Contradiction:

Improvenumber of messages storedVSAvoiddifficulty to identify required information

Core Design Contradiction:

Quantity of substanceVSDifficulty of detecting and measuring

Solution Approach 1:

The classification system segments messages into distinct categories based on their content and purpose, making it easy for users to identify and access required information by category rather than searching through a cluttered inbox.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary classification layer that mediates between the raw unstructured messages and the user's information needs. This intermediary system automatically organizes messages and can establish relational associations between related messages from different categories, facilitating easier information retrieval.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Quantity of substance

If offers are mixed with personal messages, then inbox capacity is utilized, but user awareness of offers decreases

Engineering Contradiction:

Improveinbox capacity utilizationVSAvoidoffer visibility to user

Core Design Contradiction:

Quantity of substanceVSLoss of information

Solution Approach 1:

The classification model segments offers into a distinct category separate from personal messages, ensuring that offers are not lost or overlooked amidst other communications. This segmentation maintains inbox capacity utilization while preventing information loss regarding offers.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The classification system acts as an intermediary that identifies and flags offer-related messages, potentially establishing relational associations between offers and relevant personal messages or transactions, thereby ensuring user awareness of offers while maintaining efficient inbox storage.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Measurement precision

If explicit capture and cross-verification of offer information is required, then information accuracy is ensured, but process time increases

Engineering Contradiction:

Improveinformation accuracyVSAvoidtime for capture and verification

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The classification system performs preliminary action by automatically categorizing messages and extracting relevant information during the initial processing stage. This preliminary classification and information extraction reduces the need for explicit user capture and cross-verification, maintaining information accuracy while significantly reducing the time required.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The classification system serves as an intermediary that pre-processes and validates information, establishing relational associations between related messages. This intermediary processing ensures information accuracy through automated analysis while eliminating the need for time-consuming manual capture and verification steps.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11741095B2Method and apparatus for generating structured relation information based on a text input

Publication Date: 2023.08.29 SAMSUNG ELECTRONICS CO LTD

AI summary

A method of generating structured relation information in an electronic device may include: classifying a first text input received by at least one application installed on the electronic device, into at least one category; extracting, from the first text input, a first entity representing a context of the first text input; generating structured relation information by associating the first entity extracted from the first text input, with a second entity extracted from at least one second text input stored in the electronic device or a server; and displaying the structured relation information on the electronic device.