AI Chatting Model Using Attention Mechanism for User-Specific Responses

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current intelligent chatting systems have a limited number of chatting rounds due to their inability to effectively incorporate user-specific features and interests into conversation responses.

Innovation Solution

A method using a preset encoding-decoding model with an attention model to convert user input and tag data into a target word sequence, incorporating user-specific features to generate more relevant and engaging responses.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If traditional chatting models are used, then the system is simple to implement, but the number of chatting rounds is limited and user-specific interests cannot be incorporated

Engineering Contradiction:
Improveability to incorporate user-specific featuresVSAvoidmodel complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges the encoding model and decoding model into a unified neural network architecture, where the encoding model processes input sentences and the decoding model generates responses, with both models working together to incorporate user tags into the chatting process. This combination enables the system to adapt to user-specific features while maintaining a coherent architectural framework.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces a new dimension to traditional chatting models by integrating user tag information alongside input sentences. The encoding model processes both the input sentence and user tags simultaneously, creating a multi-dimensional input space that allows the system to consider both contextual information and user-specific preferences in generating responses.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Reliability

If user tags are incorporated into the chatting model, then the relevance of responses improves, but the computational complexity increases

Engineering Contradiction:
Improverelevance of responsesVSAvoidcomputational energy
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The encoding model performs preliminary processing of both input sentences and user tags before the decoding stage. By pre-processing and encoding the user tag information alongside the input sentence, the system prepares relevant features in advance, which reduces the computational burden during the response generation phase and improves the efficiency of incorporating user-specific information.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If the chatting model focuses on context only, then the implementation is straightforward, but the engagement and number of chat rounds decrease

Engineering Contradiction:
Improvenumber of chat roundsVSAvoidmodel architecture complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent implements a dynamic chatting model where the decoding model adaptively generates responses based on both the encoded context and user tags. The model dynamically adjusts its output based on the interplay between contextual information and user-specific features, enabling more engaging and personalized conversations that can sustain more chat rounds compared to static context-only models.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS10762305B2Method for generating chatting data based on artificial intelligence, computer device and computer-readable storage medium
Publication Date: 2020.09.01 BEIJING BAIDU NETCOM SCI & TECH CO LTD
  • US10762305B2 patent drawing
  • US10762305B2 patent drawing
  • US10762305B2 patent drawing

AI summary

Embodiments of the present disclosure relate to a method for generating chatting data based on AI, a computer device and a computer-readable storage medium. The method includes: converting chatting data inputted by a user into an input word sequence; converting a tag of the user into a tag word sequence; based on a preset encoding-decoding model with an attention model, predicting according to the input word sequence and the tag word sequence to obtain a target word sequence; and converting the target word sequence into reply data of the chatting data.