EP-4738181-A1 - SESSION INFORMATION PROCESSING METHOD, APPARATUS AND SYSTEM, AND STATEMENT GENERATION METHOD AND APPARATUS

EP4738181A1EP 4738181 A1EP4738181 A1EP 4738181A1EP-4738181-A1

Abstract

The present application provides a session information processing method, apparatus and system, and a statement generation method and apparatus. The session information processing method comprises: receiving a first input of a user, wherein the first input comprises first session information; in response to the first input, carrying out matching on the first session information in a first database; when the matching is successful, inputting, into a first model, second session information and first reply information that correspond to the first session information, and the first model outputs second reply information based on the second session information and the first reply information; and displaying the second reply information, wherein the first database is a professional knowledge base in the professional field to which the first session information belongs, and the first model is an intelligent dialogue model.

Inventors

SUN, Yuwen
QIAN, Zhida
PENG, Shufang

Assignees

Guangdong Midea Kitchen Appliances Manufacturing Co., Ltd.

Dates

Publication Date: 20260506
Application Date: 20231110

Claims (20)

A method for processing conversation information, comprising: receiving a first input of a user, wherein the first input comprises first conversation information; in response to the first input, performing matching of the first conversation information in a first database; when the matching is successful, inputting second conversation information and first reply information that correspond to the first conversation information into a first model, wherein the first model outputs second reply information based on the second conversation information and the first reply information; and displaying the second reply information; wherein the first database is a professional knowledge base corresponding to a professional field to which the first conversation information belongs, and the first model is an intelligent dialogue model.
The method according to claim 1, wherein, after the performing the matching of the first conversation information in the first database, the method further comprises: in a case that the matching in the first database fails, inputting the first conversation information into the first model, wherein the first model outputs third reply information based on the first conversation information; and displaying the third reply information.
The method according to claim 1, wherein, the first database comprises a plurality of conversation information entries, and the performing the matching of the first conversation information in the first database comprises: encoding the first conversation information; performing similarity matching between the encoded first conversation information and the plurality of conversation information entries in the first database; determining that the first conversation information successfully matches the first database in a case that conversation information semantically identical to the first conversation information exists in the first database; and determining that the first conversation information fails to be matched with the first database in a case that any conversation information in the first database is semantically different from the first conversation information.
The method according to claim 3, wherein, before the encoding the first conversation information, the method further comprises: performing preprocessing on the first conversation information to obtain preprocessed first conversation information; wherein the preprocessing comprises at least one of the following: extracting a useful feature from the first conversation information, and filtering noise information from the first conversation information.
The method according to claim 3, wherein, before the encoding the first conversation information, the method further comprises: in a case that a language type of the first conversation information is inconsistent with that of the first database, inputting the first conversation information into the first model, wherein the first model then converts the language type of the first conversation information into the language type of the first database; and wherein the displaying the second reply information comprises: displaying the second reply information based on the language type of the first conversation information.
The method according to any one of claims 3 to 5, wherein, the encoding comprises at least one of the following: vector encoding processing and cosine encoding.
The method according to any one of claims 3 to 5, wherein, a matching algorithm for similarity matching comprises at least one of the following: a similarity matching algorithm based on Euclidean distance, a similarity matching algorithm based on cosine distance, and a similarity matching algorithm based on a deep learning model.
A conversation information processing apparatus, comprising: a receiving unit configured to receive a first input of a user, wherein the first input comprises first conversation information; a processing unit configured to perform matching of the first conversation information in a first database in response to the first input; and in a case of a successful matching, inputting second conversation information and first reply information corresponding to the first conversation information into a first model, wherein the first model outputs second reply information based on the second conversation information and the first reply information; and a display unit configured to display the second reply information; wherein the first database is a professional knowledge base corresponding to a professional field to which the first conversation information belongs, and the first model is an intelligent dialogue model.
A conversation information processing system, comprising: a first database in which a plurality of sets of conversation information and reply information are stored; a first model, which is an intelligent dialogue model; and a processor configured to perform steps of a method for processing conversation information according to any one of claims 1 to 7.
A sentence generation method, comprising: in response to receiving a first question sentence, identifying intent information in the first question sentence; determining a target question sentence matching the first question sentence based on the intent information; and generating an answer sentence based on a target corpus selected from at least two corpora, wherein the target corpus matches a sentence category of the target question sentence.
The sentence generation method according to claim 10, wherein, the determining the target question sentence matching the first question sentence based on the intent information comprises: obtaining at least two second question sentences based on the intent information, wherein the at least two second question sentences are associated with the intent information; and selecting the target question sentence from the at least two second question sentences.
The sentence generation method according to claim 11, wherein, the selecting the target question sentence from the at least two second question sentences comprises: obtaining a semantic similarity between the first question sentence and each of the at least two second question sentences; and selecting the target question sentence from the at least two second question sentences based on the semantic similarity.
The sentence generation method according to claim 12, wherein, the obtaining the semantic similarity between the first question sentence and each of the at least two second question sentences comprises: obtaining a first semantic vector of the first question sentence and a second semantic vector of each second question sentence; and calculating the semantic similarity based on the first semantic vector and the second semantic vector.
The sentence generation method according to claim 12, wherein, the obtaining the semantic similarity between the first question sentence and each of the at least two second question sentences comprises: inputting the first question sentence and each second question sentence into a pre-trained model to determine the semantic similarity.
The sentence generation method according to any one of claims 10 to 14, wherein, before the generating the answer sentence based on the target corpus selected from the at least two corpora, the sentence generation method comprises: selecting the target corpus from the at least two corpora based on identification information in the target question sentence, wherein the identification information corresponds to the sentence category; wherein, the sentence category comprise at least one of the following: a preset question category and a structured question category.
The sentence generation method according to claim 15, wherein, the sentence category is the structured question category; the generating the answer sentence based on the target corpus selected from the at least two corpora comprises: determining a target query sentence based on the target question sentence; querying a target corpus data in the target corpus using the target query sentence; and generating the answer sentence based on the target corpus data.
The sentence generation method according to claim 16, wherein, the determining the target query sentence based on the target question sentence comprises: extracting noun words and remaining words from the target question sentence; determining a matching query sentence template based on the remaining words; and inputting the noun words into the query sentence template to generate the target query sentence.
The sentence generation method according to claim 15, wherein, the sentence category is the preset question category; the generating the answer sentence based on the target corpus selected from the at least two corpora comprises: retrieving the answer sentence based on the target question sentence and a preset mapping relationship, wherein the preset mapping relationship represents a correspondence between the target question sentence and the answer sentence in the target corpus.
The sentence generation method according to any one of claims 10 to 14, wherein, before the identifying the intention information in the first question sentence, the sentence generation method further comprises: in response to receiving the first question sentence, updating the first question sentence based on a word in a preset vocabulary bank;.
The sentence generation method according to claim 19, wherein, the updating the first question sentence based on the word in the preset vocabulary bank in response to receiving the first question sentence comprises: extracting entity nouns from the first question sentence; searching for corresponding abstract nouns in the preset vocabulary bank based on the entity nouns; and replacing the entity nouns in the first question sentence with the abstract nouns to update the first question sentence.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to Chinese Patent Application No. 202310795163.2 filed with the China National Intellectual Property Administration on June 30, 2023 and entitled "STATEMENT GENERATION METHOD AND DEVICE, COOKING EQUIPMENT AND COMPUTER-READABLE STORAGE MEDIUM", the entire contents of which are incorporated herein by reference. This application claims priority to Chinese Patent Application No. 202310891118.7 filed with the China National Intellectual Property Administration on July 19, 2023 and entitled "SESSION INFORMATION PROCESSING METHOD, DEVICE AND SYSTEM, ELECTRONIC EQUIPMENT AND STORAGE MEDIUM", the entire contents of which are incorporated herein by reference. FIELD The present application relates to the technical field of intelligent question-answering, and specifically relates to a method for processing conversation information, apparatus and system, and a sentence generation method and apparatus. BACKGROUND Currently, for professional knowledge question-answering in specific fields, retrieval-based knowledge matching is generally established, comprising but not limited to, retrieval methods such as question-answer pairs and knowledge graph retrieval. However, when question-answering is performed through the above methods, generated answers are often fixed and highly professional, making them inconvenient to be understood. Furthermore, the above retrieval methods have limitations, which may easily lead to failure in matching the question or the answer, to reduce the accuracy and humanization of question-answering results. SUMMARY The present application aims to solve at least one of the technical problems existing in the prior art or related art. Thus, a first aspect of the present application provides a method for processing conversation information. A second aspect of the present application provides a conversation information processing apparatus. A third aspect of the present application provides a conversation information processing system. A fourth aspect of the present application provides a sentence generation method. A fifth aspect of the present application provides a sentence generation apparatus. A sixth aspect of the present application provides another sentence generation apparatus. A seventh aspect of the present application provides an electronic device. An eighth aspect of the present application provides a computer-readable storage medium. A ninth aspect of the present application provides a cooking device. In view of this, according to the first aspect of the present application, a method for processing conversation information is provided, comprising: receiving a first input of a user, wherein the first input comprises first conversation information; in response to the first input, performing matching of the first conversation information in a first database; in the case that the matching is successful, inputting second conversation information and first reply information that correspond to the first conversation information into a first model, wherein the first model outputs second reply information based on the second conversation information and the first reply information; and displaying the second reply information, wherein the first database is a professional knowledge base corresponding to a professional field to which the first conversation information belongs, and the first model is an intelligent dialogue model. The method for processing conversation information provided by the present application is configured to answer conversation information raised by users in professional fields such as a cooking field and feed back accurate and easy-to-understand reply information to users, and can improve the correctness of the results of intelligent question-answering and enhance the humanized characteristics of intelligent question-answering. Furthermore, the method for processing conversation information provided by the present application can be specifically executed by an electronic device, and the electronic device has data analyzing and data processing capabilities. Specifically, in the method for processing conversation information provided by the present application, in the case that a user asks a question corresponding to a professional field such as the cooking field through an electronic device, the user can directly input first conversation information through the first input for the electronic device. The electronic device receives and responds to the first input of the user, and compares and matches the first conversation information input by the user with a professional knowledge base (i.e., a first database) corresponding to the professional field to which the first conversation information belongs. Wherein, the first database stores a plurality of sets of conversation information and reply information in the professional field to which the first conversation information belongs. On the above basis, in the c