Corpus Processing for Chatbot Response Quality

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Open-domain chatbots face challenges in providing positive and active responses due to the need for large candidate corpora, which can include negative emotions like depression and anger, potentially causing user distress.

Innovation Solution

A corpus processing method that utilizes an emotion recognition model to identify and remove negative emotion corpora from candidate corpora before generating a reply, ensuring only positive and active responses are provided to users.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a large number of candidate corpora are used for reply, then the chatbot can provide more diverse responses, but negative emotion corpora may cause user distress

Engineering Contradiction:

Improveresponse diversityVSAvoiduser distress from negative emotions

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

The patent extracts and removes negative emotion corpora from the candidate corpus set. An emotion recognition model identifies corpora containing negative emotions (depression, anger, etc.) and excludes them from the reply candidates, allowing the system to maintain response diversity while eliminating harmful content that causes user distress

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The emotion recognition model serves as an intermediary between the candidate corpora and the reply selection process. It evaluates the emotional content of each corpus and provides guidance on which corpora to include or exclude, mediating between the need for diverse responses and the need to avoid negative emotions

Inventive Principle:

Principle #24Intermediary (Mediator)

2Object-affected harmful factors

If candidate corpora are filtered to remove negative emotions, then user experience is improved, but the number of available reply options decreases

Engineering Contradiction:

Improvenegative emotion contentVSAvoidnumber of candidate corpora

Core Design Contradiction:

Object-affected harmful factorsVSQuantity of substance

Solution Approach 1:

The patent converts the potential harm of having too many candidate corpora (risk of including negative emotions) into a benefit by using emotion recognition to identify and remove only the harmful portions. This selective filtering maintains a sufficient number of positive reply options while eliminating the harmful negative emotion content

Inventive Principle:

Principle #22Blessing in disguise (Convert harm into benefit)

Data Source

PatentUS11475069B2Corpus processing method, apparatus and storage medium

Publication Date: 2022.10.18 BEIJING XIAOMI PINECONE ELECTRONICS CO LTD

AI summary

The present disclosure relates to a corpus processing method, a corpus processing apparatus and a storage medium. The corpus processing method can include obtaining a message input by a user, retrieving a reply message matching the message input by the user from a plurality of candidate corpora, in which the plurality of the candidate corpora includes candidate corpora obtained after removing a negative emotion corpus, and sending the reply message.