CN-122019725-A - Intelligent customer service reply generation method and device, electronic equipment, medium and product

CN122019725ACN 122019725 ACN122019725 ACN 122019725ACN-122019725-A

Abstract

The disclosure provides an intelligent customer service reply generation method, relates to the technical field of artificial intelligence, and particularly relates to the technical fields of intelligent customer service, natural language processing, data mining and information retrieval. The method comprises the steps of obtaining a current query requirement and a preset dialog sample, wherein the preset dialog sample comprises a dialog record which is extracted from a history dialog log and is successfully converted, determining a target dialog sample from the preset dialog sample based on semantic matching degree between the current query requirement and the history query requirement in the preset dialog sample and reply validity of customer service replies in the preset dialog sample, and adding the target dialog sample as a reference example into a prompt word template used by intelligent customer service to obtain an enhanced prompt word for providing replies for the current query requirement based on the enhanced prompt word. The method and the device improve the matching degree and interaction efficiency between the intelligent customer service reply and the user query requirement, and improve the user experience.

Inventors

HAN XINYING

Assignees

北京百度网讯科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260209

Claims (13)

1. An intelligent customer service reply generation method, the method comprising: obtaining a current query requirement and a preset dialogue sample, wherein the preset dialogue sample comprises dialogue records which are extracted from a history dialogue log and are successfully converted; Determining a target dialogue sample from the preset dialogue sample based on semantic matching degree between the current query requirement and the historical query requirement in the preset dialogue sample and reply validity of customer service reply in the preset dialogue sample; And adding the target dialogue sample as a reference example into a prompt word template used by intelligent customer service to obtain an enhanced prompt word, and providing a reply for the current query requirement based on the enhanced prompt word.
2. The method of claim 1, the method further comprising: performing conversion verification on the history dialogue log based on dialogue texts, interaction events and conversion prediction labels in the history dialogue log; If the conversion verification indicates that the conversion is successful, taking the conversation record extracted from the historical conversation log as the preset conversation sample; And determining the reply validity of the customer service reply in the preset dialogue sample based on the dialogue round, the text length of the customer service reply and the reply content in the dialogue record.
3. The method of claim 2, wherein the determining the reply validity of the customer service reply in the predetermined dialogue sample based on the dialogue turn, the text length of the customer service reply, and the reply content in the dialogue record comprises: processing the dialogue turns in the dialogue record by using a turn attenuation function to obtain the conversion efficiency of the customer service reply, wherein the turn attenuation function defines a negative correlation between the conversion efficiency and the dialogue turns; determining the information density of the customer service reply based on the position relationship between the text length of the customer service reply and a preset length interval; determining a keyword type included in the reply content, and determining the content coverage of the customer service reply based on the keyword type; And determining reply validity of the customer service reply based on the conversion efficiency, the information density and the content coverage.
4. A method according to claim 3, wherein said determining a keyword type included in the reply content and determining a content coverage of the customer service reply based on the keyword type comprises: Obtaining a pre-constructed keyword library, and matching keywords of the reply content with the keyword library, wherein the keyword library comprises at least one keyword of a guide conversion class, a service description class and a emotion pacifying class; If at least one keyword under a certain category is matched in the reply content, acquiring a preset score corresponding to the keyword; And determining the content coverage of the customer service replies based on preset scores corresponding to various keywords matched with the reply content.
5. A method according to claim 3, wherein the determining the information density of the customer service reply based on the positional relationship between the text length of the customer service reply and the preset length interval comprises: Determining the deviation degree between the text length and an interval endpoint based on the position relation between the text length of the customer service reply and a preset length interval; determining a decay fraction of the initial density score based on the degree of deviation; and processing the initial density score by adopting the attenuation proportion, and determining the information density of the customer service reply based on the obtained processing result.
6. The method of claim 2, wherein the performing conversion verification on the historical dialog log based on dialog text, interaction events, and conversion prediction tags in the historical dialog log comprises: performing text analysis on the dialogue text to determine whether key conversion elements are included in the dialogue text; identifying whether a control triggering operation acting on the conversion guide control is included in the interaction event; If the dialogue text comprises key conversion elements, the interaction event comprises control triggering operation acting on a guiding control, and the conversion prediction label is any item of deep conversion, the history dialogue log is determined to be successful in conversion; Wherein the conversion prediction tags are output by a pre-trained conversion prediction model based on user behavior and dialog context.
7. The method of claim 1, wherein the adding the target dialog sample as a reference example to a prompt word template for intelligent customer service use results in an enhanced prompt word for providing a reply to the current query demand based on the enhanced prompt word, comprising: inputting the target dialogue sample into a prompt word constructor; filling the target dialogue sample into an example placeholder in the prompt word template through the prompt word construction function; And outputting an enhanced prompt word taking the target dialogue sample as a reference example through the prompt word constructor so as to provide a reply for the current query requirement based on the enhanced prompt word.
8. The method of claim 1, wherein the determining a target dialog sample from the predetermined dialog sample based on a semantic match between the current query requirement and historical query requirements in the predetermined dialog sample and a reply validity of a customer service reply in the predetermined dialog sample comprises: selecting a candidate dialog sample from the predetermined dialog samples based on the intent similarity in the semantic matching degree, wherein the intent similarity is obtained by matching the query intent of the current query requirement with the intent label of the predetermined dialog sample; Sorting the candidate dialogue samples based on reply validity of customer service replies in the preset dialogue samples and text similarity in the semantic matching degree, wherein the text similarity is obtained by text matching of current query demands and historical query demands; and selecting a preset number of candidate dialogue samples as the target dialogue samples based on the obtained sequencing result.
9. The method of claim 8, the method further comprising: Based on at least two preset index dimensions, respectively acquiring business index data corresponding to a first reply strategy and a second reply strategy; For each preset index dimension, respectively comparing the service index data of the second reply strategy with the service index data of the first reply strategy, and judging whether parameter adjustment conditions are met or not based on the obtained comparison results; if the parameter adjustment condition is met, adjusting the preset quantity in the first reply strategy and parameters used in the reply validity evaluation process; The first reply strategy is consistent with user traffic distribution served by the second reply strategy, the first reply strategy is used for providing replies based on the enhanced prompt words, the second reply strategy is used for providing replies based on the reference prompt words, the reference prompt words are identical to prompt word templates used by the enhanced prompt words, and reference examples formed by the target dialogue samples are not included in the reference prompt words.
10. An intelligent customer service reply generation device, the device comprising: the system comprises a query demand acquisition module, a query demand analysis module and a query demand analysis module, wherein the query demand acquisition module is used for acquiring a current query demand and a preset dialogue sample, and the preset dialogue sample comprises dialogue records which are extracted from a history dialogue log and are successfully converted; A target sample determining module, configured to determine a target dialogue sample from the predetermined dialogue sample based on a semantic matching degree between the current query requirement and a historical query requirement in the predetermined dialogue sample, and a reply validity of a customer service reply in the predetermined dialogue sample; and the reference example adding module is used for adding the target dialogue sample as a reference example into a prompt word template used by intelligent customer service to obtain an enhanced prompt word, so as to provide a reply for the current query requirement based on the enhanced prompt word.
11. An electronic device, comprising: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the intelligent customer service reply generation method of any one of claims 1-9.
12. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the intelligent customer service reply generation method of any one of claims 1-9.
13. A computer program product comprising a computer program which, when executed by a processor, implements the intelligent customer service reply generation method of any one of claims 1-9.

Description

Intelligent customer service reply generation method and device, electronic equipment, medium and product Technical Field Relates to the technical field of artificial intelligence, in particular to the technical fields of intelligent customer service, natural language processing, data mining and information retrieval. In particular to an intelligent customer service reply generation method, an intelligent customer service reply generation device, electronic equipment, media and products. Background The intelligent customer service is an automatic customer service system based on an artificial intelligence technology, and can interact with a user through natural language to answer questions or guide to complete a business process. The core capability of the intelligent customer service response system depends on the used response mode to a great extent, the efficient response mode can effectively improve the matching degree between the intelligent customer service response and the user query requirement, and user experience is improved, so that continuous optimization of the conversation operation has obvious necessity. Disclosure of Invention The disclosure provides an intelligent customer service reply generation method, an intelligent customer service reply generation device, electronic equipment, media and products. According to an aspect of the present disclosure, there is provided an intelligent customer service reply generation method, including: obtaining a current query requirement and a preset dialogue sample, wherein the preset dialogue sample comprises dialogue records which are extracted from a history dialogue log and are successfully converted; Determining a target dialogue sample from the preset dialogue sample based on semantic matching degree between the current query requirement and the historical query requirement in the preset dialogue sample and reply validity of customer service reply in the preset dialogue sample; And adding the target dialogue sample as a reference example into a prompt word template used by intelligent customer service to obtain an enhanced prompt word, and providing a reply for the current query requirement based on the enhanced prompt word. According to another aspect of the present disclosure, there is provided an intelligent customer service reply generation apparatus, the apparatus including: the system comprises a query demand acquisition module, a query demand analysis module and a query demand analysis module, wherein the query demand acquisition module is used for acquiring a current query demand and a preset dialogue sample, and the preset dialogue sample comprises dialogue records which are extracted from a history dialogue log and are successfully converted; A target sample determining module, configured to determine a target dialogue sample from the predetermined dialogue sample based on a semantic matching degree between the current query requirement and a historical query requirement in the predetermined dialogue sample, and a reply validity of a customer service reply in the predetermined dialogue sample; and the reference example adding module is used for adding the target dialogue sample as a reference example into a prompt word template used by intelligent customer service to obtain an enhanced prompt word, so as to provide a reply for the current query requirement based on the enhanced prompt word. According to still another aspect of the present disclosure, there is provided an electronic device including: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the intelligent customer service reply generation method of any of the embodiments of the present disclosure. According to yet another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the intelligent customer service reply generation method according to any one of the embodiments of the present disclosure. According to yet another aspect of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the intelligent customer service reply generation method of any embodiment of the present disclosure. The method and the device improve the matching degree and interaction efficiency between the intelligent customer service reply and the user query requirement, and improve the user experience. It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification. Drawings The drawings are fo