Information query method and device, equipment, medium and product

Abstract

The invention discloses an information query method and device, equipment, a medium and a product, and relates to the technical field of digital employees, and the method comprises the steps: carrying out the first similarity matching of a target query statement and a preset question corpus in a universal domain question and answer knowledge base, according to the first similarity matching result, determining whether information recall can be carried out in the universal domain question and answer knowledge base; performing second similarity matching on the target query statement and a preset question corpus in a specific domain question and answer knowledge base under the condition that information recall cannot be performed in the general domain question and answer knowledge base, and determining whether information recall can be performed in the specific domain question and answer knowledge base or not according to a second similarity matching result; and when it is determined that information recall can be performed in the specific domain question and answer knowledge base, determining a target answer corpus from the specific domain question and answer knowledge base, and displaying the target answer corpus as target recall information to the target user. According to the invention, the effect of improving the information query precision is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of digital employees, and particularly to a method, device, equipment, medium and product for querying information. Background Art

[0002] Intelligent Customer Service is an automated service system built based on technologies such as natural language processing (NLP), machine learning, and knowledge graphs. Its core goal is to solve user consultation problems through human-computer interaction, reduce enterprise service costs, and improve response efficiency.

[0003] Currently, when the current intelligent customer service queries information for users, it usually relies on a general domain Q&A knowledge base for information query. However, due to the large number of domains involved in the general domain Q&A knowledge base, the knowledge in each domain is prone to mutual influence during information query, resulting in insufficient accuracy of information query. Summary of the Invention

[0004] The present invention provides a method, device, equipment, medium and product for querying information to solve the problem of insufficient accuracy of information query existing in the current situation of only relying on a general domain Q&A knowledge base for information query.

[0005] According to one aspect of the present invention, there is provided a method for querying information, the method comprising:

[0006] Performing a first similarity match between a target query statement sent by a target user and preset question corpus in a general domain Q&A knowledge base, and determining whether information can be recalled from the general domain Q&A knowledge base according to the first similarity match result;

[0007] In the case where it is determined that information cannot be recalled from the general domain Q&A knowledge base, performing a second similarity match between the target query statement and preset question corpus in a specific domain Q&A knowledge base, and determining whether information can be recalled from the specific domain Q&A knowledge base according to the second similarity match result; wherein, the specific domain Q&A knowledge base is determined according to a target domain associated with the target query statement;

[0008] In the case where it is determined that information can be recalled from the specific domain Q&A knowledge base, determining target answer corpus from the specific domain Q&A knowledge base, and presenting the target answer corpus as target recalled information to the target user.

[0009] According to another aspect of the present invention, there is provided a device for querying information, the device comprising:

[0010] The first similarity matching module is used to perform a first similarity matching between the target query statement sent by the target user and the preset question corpus in the general domain Q&A knowledge base, and determine whether information can be recalled from the general domain Q&A knowledge base according to the first similarity matching result;

[0011] The second similarity matching module is used to, when it is determined that information cannot be recalled from the general domain Q&A knowledge base, perform a second similarity matching between the target query statement and the preset question corpus in the specific domain Q&A knowledge base, and determine whether information can be recalled from the specific domain Q&A knowledge base according to the second similarity matching result; wherein, the specific domain Q&A knowledge base is determined according to the target domain associated with the target query statement;

[0012] The information display module is used to, when it is determined that information can be recalled from the specific domain Q&A knowledge base, determine the target answer corpus from the specific domain Q&A knowledge base, and display the target answer corpus as the target recalled information to the target user.

[0013] According to another aspect of the present invention, there is provided an electronic device, the electronic device includes:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the information query method according to any one of the present invention.

[0017] According to another aspect of the present invention, there is provided a computer-readable storage medium, the computer-readable storage medium stores computer instructions, and the computer instructions are used to implement the information query method according to any one of the present invention when executed by a processor.

[0018] According to another aspect of the present invention, there is provided a computer program product, including a computer program, and the computer program implements the information query method according to any one of the present invention when executed by a processor.

[0019] The beneficial effects of the present invention are as follows: In the first aspect, by recalling information from a specific domain Q&A knowledge base related to the target domain to which the target query statement belongs, since the specific domain Q&A knowledge base has the characteristic of domain verticalization, the knowledge of each domain is isolated, avoiding the risk of mutual influence and confusion of knowledge in each domain during information query, and improving the accuracy of information query. In the second aspect, since the general domain Q&A knowledge base usually covers high-frequency and common questions, by first attempting to recall information from the general domain Q&A knowledge base, the response rate of information query can be improved, and the query waiting time of users can be reduced.

[0020] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0022] Figure 1 It is a flowchart of a method for querying information provided in Embodiment 1 of the present invention;

[0023] Figure 2 It is a flowchart of a method for querying information provided in Embodiment 2 of the present invention;

[0024] Figure 3 It is a flowchart of a method for querying information provided in Embodiment 3 of the present invention;

[0025] Figure 4 It is a schematic structural diagram of a device for querying information provided in Embodiment 4 of the present invention;

[0026] Figure 5 It is a schematic structural diagram of an electronic device for implementing the information query method of the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] In order to enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] It should be noted that the terms "first", "second", "third", "candidate", "target", "to be consulted", etc. in the specification, claims and above-mentioned drawings of the present invention are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] Figure 1 FIG. is a flowchart of a method for querying information provided in Embodiment 1 of the present invention. This embodiment is applicable to the situation of querying information for users using a general and specific domain question and answer knowledge base. This method can be executed by a query device for information, and the query device for information can be implemented in the form of hardware and / or software, such as implemented by a personal computer. As Figure 1 shown, the method includes:

[0031] S101. Perform a first similarity match between the target query statement sent by the target user and the preset question corpus in the general domain question and answer knowledge base, and determine whether information can be recalled from the general domain question and answer knowledge base according to the first similarity match result.

[0032] Among them, the target user refers to a user with information query needs, which can be an individual user, an enterprise user, or an administrator user, etc. The specific type of the target user is not limited in this embodiment. The target query statement refers to a retrieval request expressed by the target user through a specific syntax or natural language, used to query information from a data source (such as a question and answer knowledge base, a database, etc.). The type of the target query statement can be a natural language query statement, etc.

[0033] The general domain question and answer knowledge base refers to a comprehensive knowledge base that covers a wide range of fields and is not limited to a specific industry. Its core goal is to support natural language question and answer needs in multiple scenarios through structured or semi-structured data, and it has the characteristics of breadth-first and relatively coarse knowledge granularity.

[0034] In the general domain Q&A knowledge base, preset question corpora and preset answer corpora are stored in an associated storage manner, that is, any preset question corpus is associated with at least one preset answer corpus, which is used to recall the associated at least one preset answer corpus when the target query statement hits the preset question corpus.

[0035] The preset question corpus refers to a set of questions predefined and stored in the general domain Q&A knowledge base by developers or maintainers, which is divided into standard question corpora and similar question corpora. The standard question corpus is the standardized questions predefined in the general domain Q&A knowledge base; the similar question corpus is a natural language question expressed in different ways and having the same or similar semantics as the standard question corpus.

[0036] The preset answer corpus refers to the standardized answers predefined and stored in the general domain Q&A knowledge base by developers or maintainers, and its core function is to improve the response efficiency and accuracy through standardized output.

[0037] In one implementation, the target query statement sent by the target user is obtained through an API (Application Programming Interface) or a front-end input box. Further, the target query statement is segmented, part-of-speech tagged, and stop-word filtered, and then the first sentence vector of the target query statement is generated using the BERT pre-trained model, and the second sentence vector of each preset question corpus in the general domain Q&A knowledge base is generated.

[0038] According to the first sentence vector of the target query statement and the second sentence vectors of each preset question corpus in the general domain Q&A knowledge base, a first similarity matching is performed to determine the first similarity matching result between the target query statement and each preset question corpus. Optionally, the method of the first similarity matching includes cosine similarity calculation, that is, according to the first sentence vector of the target query statement and the second sentence vectors of each preset question corpus in the general domain Q&A knowledge base, cosine similarity calculation is performed to determine the cosine similarity between the target query statement and each preset question corpus.

[0039] Further, compare the first similarity matching results between the target query statement and each preset question corpus with the first similarity threshold respectively. If there is at least one first similarity matching result between a preset question corpus and the target query statement that is greater than or equal to the first similarity threshold, it is determined that information can be recalled from the general domain Q&A knowledge base; if the first similarity matching results between each preset question corpus and the target query statement are all less than the first similarity threshold, it is determined that information cannot be recalled from the general domain Q&A knowledge base. For example, continuing with the first similarity matching being cosine similarity as an example, assume the first similarity threshold is 0.75. If there is at least one cosine similarity between a preset question corpus and the target query statement that is greater than or equal to 0.75, it is determined that information can be recalled from the general domain Q&A knowledge base; if the cosine similarities between each preset question corpus and the target query statement are all less than 0.75, it is determined that information cannot be recalled from the general domain Q&A knowledge base.

[0040] Optionally, in the case where it is determined that information can be recalled from the general domain Q&A knowledge base, determine the number of preset question corpora whose first similarity matching results are greater than or equal to the first similarity threshold. If the number is one, use the preset answer corpus corresponding to this preset question corpus as the target recalled information to be displayed to the target user; if the number is at least two, use the preset answer corpus with the highest first similarity matching result as the target recalled information to be displayed to the target user, and furthermore, use several preset answer corpora with the second highest first similarity matching results, as well as the domain information where these preset answer corpora are located, as the target recalled information to be displayed to the target user.

[0041] S102. In the case where it is determined that information cannot be recalled from the general domain Q&A knowledge base, perform a second similarity matching between the target query statement and the preset question corpora in the specific domain Q&A knowledge base, and determine whether information can be recalled from the specific domain Q&A knowledge base according to the second similarity matching results.

[0042] Among them, the specific domain Q&A knowledge base is a knowledge base constructed for a vertical domain or a professional domain, and through structured knowledge expression and semantic understanding technologies, it realizes high-precision and professional question answering. The specific domain Q&A knowledge base is determined according to the target domain associated with the target query statement. In other words, the specific domain Q&A knowledge base in this embodiment is a knowledge base constructed for the target domain. For example, assume the target domain is the "financial domain", then the specific domain Q&A knowledge base is a knowledge base constructed for the "financial domain".

[0043] In one implementation, when it is determined that information cannot be recalled from the general domain Q&A knowledge base, the target query statement is further segmented by entity words to determine at least one target entity word included in the target query statement. Then, each target entity word is matched with the domain label libraries of each predefined candidate domain, and the target domain associated with the target query statement is determined from each candidate domain according to the matching result. For example, assume that the target entity words included in the target query statement are "Entity A", "Entity B", and "Entity C", and assume that the domain label library of the "financial domain" includes "Entity A", then it is determined that the target domain associated with the target query statement is the "financial domain".

[0044] Further, according to the target domain associated with the target query statement and the candidate domains targeted by each candidate domain Q&A knowledge base, a specific domain Q&A knowledge base is determined from each candidate domain Q&A knowledge base.

[0045] Further, a BERT pre-trained model is used to generate the third sentence vectors of each predefined question corpus in the specific domain Q&A knowledge base. According to the first sentence vector of the target query statement and the third sentence vectors of each predefined question corpus in the specific domain Q&A knowledge base, a second similarity matching is performed to determine the second similarity matching result between the target query statement and each predefined question corpus. Optionally, the method for the second similarity matching includes cosine similarity calculation, that is, according to the second sentence vector of the target query statement and the third sentence vectors of each predefined question corpus in the specific domain Q&A knowledge base, a cosine similarity calculation is performed to determine the cosine similarity between the target query statement and each predefined question corpus.

[0046] Further, the second similarity matching results between the target query statement and each predefined question corpus are respectively compared with the second similarity threshold. If there is at least one predefined question corpus whose second similarity matching result with the target query statement is greater than or equal to the second similarity threshold, it is determined that information can be recalled from the specific domain Q&A knowledge base; if the second similarity matching results between each predefined question corpus and the target query statement are all less than the second similarity threshold, it is determined that information cannot be recalled from the specific domain Q&A knowledge base. For example, continuing with the example where the second similarity matching is cosine similarity, assume that the second similarity threshold is 0.9. If there is at least one predefined question corpus whose cosine similarity with the target query statement is greater than or equal to 0.9, it is determined that information can be recalled from the specific domain Q&A knowledge base; if the cosine similarity between each predefined question corpus and the target query statement is less than 0.9, it is determined that information cannot be recalled from the specific domain Q&A knowledge base.

[0047] S103. When it is determined that information can be recalled from the specific domain Q&A knowledge base, determine the target answer corpus from the specific domain Q&A knowledge base, and display the target answer corpus as the target recalled information to the target user.

[0048] In one implementation, when it is determined that information can be recalled from the specific domain Q&A knowledge base, determine the number of preset question corpora with the second similarity matching result greater than or equal to the second similarity threshold. If the number is one, display the preset answer corpus corresponding to the preset question corpus as the target recalled information to the target user; if the number is at least two, display the preset answer corpus with the highest second similarity matching result as the target recalled information to the target user, and display several preset answer corpora with the second highest second similarity matching result as the target recalled information to the target user.

[0049] In the embodiment of the present invention, the target query statement sent by the target user is first similarity matched with the preset question corpora in the general domain Q&A knowledge base, and it is determined whether information can be recalled from the general domain Q&A knowledge base according to the first similarity matching result. When it is determined that information cannot be recalled from the general domain Q&A knowledge base, the target query statement is second similarity matched with the preset question corpora in the specific domain Q&A knowledge base, and it is determined whether information can be recalled from the specific domain Q&A knowledge base according to the second similarity matching result. When it is determined that information can be recalled from the specific domain Q&A knowledge base, determine the target answer corpus from the specific domain Q&A knowledge base, and display the target answer corpus as the target recalled information to the target user. The beneficial effects are as follows:

[0050] First, by recalling information from the specific domain Q&A knowledge base related to the target domain to which the target query statement belongs, since the specific domain Q&A knowledge base has the characteristic of domain verticalization, the knowledge of each domain is isolated, avoiding the risk of mutual influence and confusion of the knowledge of each domain during information query, and improving the accuracy of information query.

[0051] Second, since the general domain Q&A knowledge base usually covers high-frequency and common questions, by first attempting to recall information from the general domain Q&A knowledge base, the response rate of information query can be improved, and the query waiting time of users can be reduced.

[0052] Embodiment 2

[0053] Figure 2 It is a flowchart of a method for querying information provided by Embodiment 2 of the present invention. This embodiment further optimizes and expands the above embodiment, and can be combined with each of the above optional implementation manners. As Figure 2 shown, the method includes:

[0054] S201. Determine the first digital employee that interacts with the target user at the current moment, and determine the type of digital employee corresponding to the first digital employee.

[0055] Among them, a digital employee is a virtual intelligent agent built based on artificial intelligence, robotic process automation (RPA), and multimodal interaction technologies. Its core goal is to automate the execution and intelligent optimization of business processes by simulating human operation and decision-making logic. It can be understood that the first digital employee refers to the digital employee that interacts with the target user at the current moment. The type of digital employee reflects the field in which the digital employee can handle information query tasks, and is divided into general-domain digital employees and specific-domain digital employees. Among them, general-domain digital employees can handle information query tasks in the general domain, are the digital employees with the strongest capabilities and the broadest knowledge, and they use complex technologies to identify user intentions; specific-domain digital employees can handle information query tasks in specific domains.

[0056] In one implementation, determine the context information of the information interaction at the current moment, determine the digital employee ID of the digital employee that interacts with the target user according to the context information, and further determine the first digital employee according to the digital employee ID. Further, obtain the employee label information corresponding to the first digital employee, and parse the type of digital employee corresponding to the first digital employee from the employee label information.

[0057] S202. When the type of digital employee is a general-domain digital employee, perform a first similarity matching between the target query statement and the preset question corpus in the general-domain question and answer knowledge base through the first task execution engine corresponding to the first digital employee, and determine whether information can be recalled from the general-domain question and answer knowledge base according to the first similarity matching result.

[0058] Among them, the task execution engine of a digital employee is the core technical module for its automated operation, responsible for converting user instructions or preset processes into executable machine operations, mainly used for parsing instructions, scheduling resources, and driving business process automation. It can be understood that the task execution engine corresponding to the first digital employee is the first task execution engine.

[0059] In one implementation, when the type of digital employee is a general-domain digital employee, perform word segmentation, part-of-speech tagging, and stop word filtering on the target query statement through the first task execution engine, and then use the BERT pre-trained model to generate the first sentence vector of the target query statement, and generate the second sentence vector of each preset question corpus in the general-domain question and answer knowledge base.

[0060] Further, the first task execution engine performs a first similarity matching based on the first sentence vector of the target query statement and the second sentence vectors of each preset question corpus in the general domain Q&A knowledge base, to determine the first similarity matching results between the target query statement and each preset question corpus. Then, the first similarity matching results between the target query statement and each preset question corpus are respectively compared with the first similarity threshold. If there is at least one first similarity matching result between a preset question corpus and the target query statement that is greater than or equal to the first similarity threshold, it is determined that information can be recalled from the general domain Q&A knowledge base; if the first similarity matching results between each preset question corpus and the target query statement are all less than the first similarity threshold, it is determined that information cannot be recalled from the general domain Q&A knowledge base.

[0061] By determining the first digital employee that interacts with the target user at the current moment and determining the digital employee type corresponding to the first digital employee, in the case where the digital employee type is a general domain digital employee, the first task execution engine corresponding to the first digital employee performs a first similarity matching between the target query statement and the preset question corpus in the general domain Q&A knowledge base. The beneficial effects are as follows:

[0062] First, by invoking the digital employee to perform the information query task, compared with the way of using a human customer service for information query, it saves labor costs, improves the response speed of information query, and can also enhance the fault tolerance ability of information query.

[0063] Second, by preferentially invoking the general domain digital employee to perform the initial matching, it avoids prematurely starting the specific domain Q&A knowledge base that consumes high computing power. This hierarchical mechanism can reduce the processing delay of high-frequency simple questions and at the same time reduce the occupancy of hardware resources such as GPUs.

[0064] S203. In the case where it is determined that information cannot be recalled from the general domain Q&A knowledge base, at least one target real employee is determined from each candidate real employee according to the target domain and the expertise domain labels associated with each candidate real employee, and the target identification information of each target real employee is displayed to the target user.

[0065] Among them, the candidate real employees refer to at least one real employee who is good at the target domain, such as a technical R & D personnel or a management personnel, etc. For example, if the target domain is the "financial domain", then the candidate real employees are technical R & D personnel or management personnel who are good at the "financial domain", etc. The target identification information refers to the unique identifier that can reflect the identity of the target real employee, including but not limited to name, position, employee number, etc.

[0066] In one implementation, in the case where it is determined that information recall cannot be performed from the general domain Q&A knowledge base, the target domain is matched with the expertise domain labels associated with each candidate real employee, and each candidate real employee whose expertise domain label matches the target domain is used as the target real employee. For example, assume the target domain is the "financial domain", the expertise domain label associated with candidate real employee 1 is the "education domain", the expertise domain label associated with candidate real employee 2 is the "technology domain", and the expertise domain label associated with candidate real employee 3 is the "financial domain", then candidate real employee 3 is used as the target real employee.

[0067] Furthermore, the target identification information of each target real employee is displayed to the target user through the front-end dialog box, and a prompt message is generated in the front-end dialog box to prompt the target user to perform a selection operation on any target identification information. Such as "We have recommended domain experts for you. You may want to ask them." etc.

[0068] By determining at least one target real employee from each candidate real employee according to the target domain and the expertise domain labels associated with each candidate real employee in the case where it is determined that information recall cannot be performed from the general domain Q&A knowledge base, and displaying the target identification information of each target real employee to the target user, it enables the target user to independently select real employees who are good at the target domain to provide feedback information in the case where information recall cannot be performed from the general domain Q&A knowledge base, ensuring the success rate of information recall and improving the user's information query experience.

[0069] S204. In response to the target user's selection operation on the target identification information, determine the real employee to be consulted from each target real employee, and determine the digital employee corresponding to the real employee to be consulted as the second digital employee.

[0070] Among them, the second digital employee is the digital employee exclusive to the real employee to be consulted, that is, the digital avatar of the real employee to be consulted. It is a virtual identity created based on artificial intelligence, virtual reality, and multi-modal technologies, and can simulate the characteristics of the real employee to be consulted and perform specific tasks.

[0071] In one implementation, the target user performs a selection operation on any target identification information through the front-end dialog box. In response to the target user's selection operation on the target identification information, the target identification information is matched with the target identification information of each target real employee to determine the real employee to be consulted from each target real employee. Furthermore, determine the digital employee corresponding to the real employee to be consulted as the second digital employee.

[0072] S205. Perform a second similarity matching between the target query statement and the preset question corpus in the exclusive Q&A knowledge base through the second task execution engine corresponding to the second digital employee.

[0073] Among them, different target real employees maintain their own specific domain Q&A knowledge bases to update the preset question corpus and preset answer corpus in the maintained specific domain Q&A knowledge bases periodically or in real time. The exclusive Q&A knowledge base is the specific domain Q&A knowledge base maintained for the real employees to be consulted.

[0074] In one implementation, the second task execution engine uses the BERT pre-trained model to generate the fourth sentence vectors of the preset question corpora in the exclusive Q&A knowledge base. According to the first sentence vector of the target query statement and the fourth sentence vectors of the preset question corpora in the exclusive Q&A knowledge base, a second similarity matching is performed to determine the second similarity matching results between the target query statement and the preset question corpora.

[0075] By responding to the selection operation of the target user on the target identification information, the real employee to be consulted is determined from each target real employee, and the digital employee corresponding to the real employee to be consulted is determined as the second digital employee; the second task execution engine corresponding to the second digital employee performs a second similarity matching between the target query statement and the preset question corpora in the exclusive Q&A knowledge base. The beneficial effects are as follows:

[0076] First, compared with the general domain Q&A knowledge base, the similarity matching based on the real employee exclusive Q&A knowledge base improves the matching accuracy of professional questions.

[0077] Second, by calling the digital employee to execute the information query task, compared with the real employee answering questions in person, it saves labor costs, improves the response speed of information query, and can also improve the fault tolerance ability of information query.

[0078] Third, the second digital employee can only access the exclusive Q&A knowledge base of the real employee to be consulted, realizing data isolation and ensuring data security.

[0079] S206. Determine whether information can be recalled from the exclusive Q&A knowledge base according to the second similarity matching results. In the case where it is determined that information cannot be recalled from the exclusive Q&A knowledge base, obtain the artificial answer corpus generated by the real employee to be consulted for the target query statement.

[0080] Among them, the artificial answer corpus refers to the answer corpus generated by the real employee to be consulted manually.

[0081] In one implementation, the second similarity matching results between the target query statement and the preset question corpora are respectively compared with the second similarity threshold. If the second similarity matching results between each preset question corpus and the target query statement are all less than the second similarity threshold, it is determined that information cannot be recalled from the specific domain Q&A knowledge base.

[0082] In the case where it is determined that information cannot be recalled from the exclusive Q&A knowledge base, a prompt message is sent to the real employee to be consulted, which is used to prompt the real employee to be consulted to send artificial answer corpus for the target query statement through the front-end dialog box. After the real employee to be consulted sends the artificial answer corpus through the front-end dialog box, the artificial answer corpus is obtained.

[0083] S207. Display the artificial answer corpus as the target recalled information to the target user.

[0084] In one implementation, the artificial answer corpus is displayed to the target user through the front-end dialog box.

[0085] By determining whether information can be recalled from the exclusive Q&A knowledge base according to the second similarity matching result, in the case where it is determined that information cannot be recalled from the exclusive Q&A knowledge base, the artificial answer corpus generated by the real employee to be consulted for the target query statement is obtained, and the artificial answer corpus is displayed to the target user as the target recalled information. The beneficial effects are as follows:

[0086] First, the artificial response process is triggered only when the recall from the exclusive Q&A knowledge base fails, avoiding the real-time call of the high-cost exclusive Q&A knowledge base to process problems and reducing the overall response latency of the system.

[0087] Second, the traditional system returns "unanswerable" when there is no matching result, while this solution uses the artificial answer corpus as a backup, reducing the session interruption rate and significantly improving the satisfaction, especially in the high-value customer service scenario.

[0088] Optionally, after displaying the artificial answer corpus as the target recalled information to the target user, it further includes:

[0089] 1) Input the first prompt text, the target query statement, and the artificial answer corpus into the first large language model through the second task execution engine.

[0090] The large language model refers to a deep learning model with a large number of parameters, which can generate natural language text or understand the meaning of language text. The large language model can handle various natural language tasks, such as text classification, question answering, dialogue, etc. Usually, the parameter scale of the large language model can even reach the level of hundreds of billions.

[0091] The first prompt text is used to prompt the first large language model to refine the text according to the target query statement and the artificial answer corpus, and generate the first supplementary question corpus and the first supplementary answer corpus respectively.

[0092] In one embodiment, the first prompt text, the target query statement, and the artificial answer corpus are input into the first large language model through the second task execution engine. The first large language model refines the target query statement according to the first prompt text to generate the first supplementary question corpus, and refines the artificial answer corpus to generate the first supplementary answer corpus.

[0093] Exemplarily, the first prompt text may include "Please refine the question corpus and answer corpus according to the following conversation", etc.

[0094] 2) Update the preset question corpus in the exclusive Q&A knowledge base according to the first supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the first supplementary answer corpus.

[0095] In one embodiment, generate Q&A corpus supplement confirmation information according to the first supplementary question corpus and the first supplementary answer corpus, and display the Q&A corpus supplement confirmation information to the real employee to be consulted. When the real employee to be consulted confirms that the first supplementary question corpus and the first supplementary answer corpus are correct, update the preset question corpus in the exclusive Q&A knowledge base according to the first supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the first supplementary answer corpus.

[0096] Input the first prompt text, the target query statement, and the artificial answer corpus into the first large language model through the second task execution engine, update the preset question corpus in the exclusive Q&A knowledge base according to the first supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the first supplementary answer corpus. The beneficial effects are as follows:

[0097] Firstly, the incremental knowledge update of the exclusive Q&A knowledge base is realized through the second task execution engine, solving the lag problem of the traditional knowledge base relying on full-volume data retraining, and ensuring the real-time nature of the knowledge update in the exclusive Q&A knowledge base.

[0098] Secondly, the update of the traditional knowledge base requires experts to manually label data and retrain the model. In this solution, the second task execution engine automatically updates the knowledge in the exclusive Q&A knowledge base according to the first supplementary question corpus and the first supplementary answer corpus, reducing manual intervention.

[0099] Embodiment III

[0100] Figure 3 The flowchart of a method for querying information provided in Embodiment III of the present invention further optimizes and expands the above embodiments and can be combined with the above various optional embodiments. As Figure 3 shown, the method includes:

[0101] S301. Input the second prompt text and the target query statement into the second large language model through the first task execution engine.

[0102] Among them, the second prompt text is used to prompt the second large language model to determine whether the query intention of the target query statement is a non-inquiry intention. Non-inquiry intentions include, but are not limited to, chatting intentions, non-help-seeking intentions, non-help-seeking-for-oneself intentions, etc. For example, when the target query statement is "Okay, thank you for answering", "Then I'll ask Teacher Chen", etc., it means that the query intention of the target query statement is a chatting intention; when the target query statement is "Hello, I'm XXX and I have completed the operation and maintenance information registration", "I see that you are the person on duty for environmental operation and maintenance today; I want to consult the system address of environmental operation and maintenance", etc., it means that the query intention of the target query statement is a non-help-seeking intention.

[0103] In one implementation, input the second prompt text and the target query statement into the second large language model through the first task execution engine. The second large language model determines whether the query intention of the target query statement is a non-inquiry intention according to the second prompt text.

[0104] S302. In the case where the second large language model determines that the query intention is not a non-inquiry intention, perform a first similarity matching between the target query statement and the preset question corpus in the general domain Q&A knowledge base.

[0105] Input the second prompt text and the target query statement into the second large language model through the first task execution engine; among them, the second prompt text is used to prompt the second large language model to determine whether the query intention of the target query statement is a non-inquiry intention. In the case where the second large language model determines that the query intention is not a non-inquiry intention, perform a first similarity matching between the target query statement and the preset question corpus in the general domain Q&A knowledge base. The beneficial effects are as follows:

[0106] First, the second prompt text guides the second large language model to identify the user's query intention, reducing meaningless retrievals from the general domain Q&A knowledge base. This mechanism can filter out non-business-related queries and reduce the server load.

[0107] Second, trigger the similarity matching when it is confirmed that it is not a non-inquiry intention, avoiding redundant calculations.

[0108] S303. Determine whether information can be recalled from the general domain Q&A knowledge base according to the first similarity matching result. In the case where it is determined that information cannot be recalled from the general domain Q&A knowledge base, perform a second similarity matching between the target query statement and the preset question corpus in the specific domain Q&A knowledge base.

[0109] S304. In the case where the second similarity matching results are all less than the similarity threshold, at least one query keyword included in the target query statement is respectively textually matched with the preset answer corpus in the specific domain Q&A knowledge base.

[0110] In one implementation, if the second similarity matching results between the target query statement and each preset question corpus in the specific domain Q&A knowledge base are all less than the similarity threshold, keyword extraction is performed on the target query statement to obtain at least one query keyword. Further, at least one query keyword is respectively textually matched with each preset answer corpus in the specific domain Q&A knowledge base.

[0111] S305. If there is at least one preset answer corpus containing any of the query keywords, it is determined that information can be recalled from the specific domain Q&A knowledge base.

[0112] In one implementation, it is determined whether there is at least one preset answer corpus containing any of the query keywords according to the text matching result. If it is determined that there is at least one preset answer corpus containing any of the query keywords, it is determined that information can be recalled from the specific domain Q&A knowledge base.

[0113] Exemplarily, assume that the query keywords included in the target query statement are "keyword 1", "keyword 2", and "keyword 3". Then, if there is at least one preset answer corpus containing "keyword 1", "keyword 2", and / or "keyword 3", it is determined that information can be recalled from the specific domain Q&A knowledge base.

[0114] S306. In the case where it is determined that information can be recalled from the specific domain Q&A knowledge base, the preset answer corpus containing any of the query keywords is used as the target answer corpus, and the target answer corpus is used as the target recalled information to be displayed to the target user.

[0115] Exemplarily, assume that preset answer corpus 1 in the specific domain Q&A knowledge base contains any of the query keywords. Then, preset answer corpus 1 is used as the target answer corpus, and preset answer corpus 1 is used as the target recalled information to be displayed to the target user.

[0116] By, in the case where the second similarity matching results are all less than the similarity threshold, respectively textually matching at least one query keyword included in the target query statement with the preset answer corpus in the specific domain Q&A knowledge base, if there is at least one of the preset answer corpora containing any of the query keywords, it is determined that information can be recalled from the specific domain Q&A knowledge base, and using the preset answer corpus containing any of the query keywords as the target answer corpus, the beneficial effects are as follows:

[0117] In the first aspect, when information cannot be accurately recalled through similarity matching, text matching based on keywords is used as a fallback strategy, which can capture the preset answer corpus directly associated with the query keywords in the Q&A knowledge base of a specific domain, avoid missed detections caused by semantic deviations, and improve the information recall rate.

[0118] In the second aspect, it can effectively handle spelling mistakes and colloquial expressions in the user input and achieve fault tolerance through text character-level matching.

[0119] Optionally, the method further includes:

[0120] A. Input the third prompt text and the target conversation information into the third large language model through the second task execution engine.

[0121] Among them, the target conversation information is the conversation information generated between the real employee to be consulted and the consulting user. For example, it can be the chat record between the real employee to be consulted and the consulting user. The consulting user refers to the user who consults information from the real employee to be consulted.

[0122] The third prompt text is used to prompt the third large language model to refine the text according to the target conversation information and generate the second supplementary question corpus and the second supplementary answer corpus.

[0123] In one implementation, the third prompt text and the target conversation information are input into the third large language model through the second task execution engine. The third large language model refines the text of the target conversation information according to the third prompt text to generate the second supplementary question corpus and the second supplementary answer corpus.

[0124] B. Update the preset question corpus in the exclusive Q&A knowledge base according to the second supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the second supplementary answer corpus.

[0125] In one implementation, generate Q&A corpus supplement confirmation information according to the second supplementary question corpus and the second supplementary answer corpus, and display the Q&A corpus supplement confirmation information to the real employee to be consulted. When the real employee to be consulted confirms that the second supplementary question corpus and the second supplementary answer corpus are correct, update the preset question corpus in the exclusive Q&A knowledge base according to the second supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the second supplementary answer corpus.

[0126] Input the third prompt text and the target session information into the third large language model through the second task execution engine; wherein, the target session information is the session information generated between the real employee to be consulted and the consulting user, and the third prompt text is used to prompt the third large language model to refine the text according to the target session information to generate the second supplementary question corpus and the second supplementary answer corpus; update the preset question corpus in the exclusive Q&A knowledge base according to the second supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the second supplementary answer corpus. The beneficial effects are as follows:

[0127] First, the continuous self-update of the exclusive Q&A knowledge base is realized. Compared with the traditional manual annotation method, the maintenance cycle of the exclusive Q&A knowledge base is shortened, effectively solving the problem of knowledge lag in the scenario of frequent business rule changes.

[0128] Second, compared with the traditional rule engine (such as regular expression matching) and statistical models, this solution refines the Q&A corpus through the context awareness ability of the large language model, which can further improve the accuracy of Q&A corpus refinement.

[0129] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by users or fully authorized by all parties, and the collection, use, and processing of relevant data comply with the relevant laws, regulations, and standards in the relevant regions.

[0130] Embodiment 4

[0131] Figure 4 FIG. is a schematic structural diagram of an information query device provided in Embodiment 4 of the present invention, which is applicable to the situation of querying information for users using a general and specific domain Q&A knowledge base, such as Figure 4 As shown, the device includes:

[0132] The first similarity matching module 41 is used to perform a first similarity matching between the target query statement sent by the target user and the preset question corpus in the general domain Q&A knowledge base, and determine whether information can be recalled from the general domain Q&A knowledge base according to the first similarity matching result;

[0133] The second similarity matching module 42 is used to perform a second similarity matching between the target query statement and the preset question corpus in the specific domain Q&A knowledge base when it is determined that information cannot be recalled from the general domain Q&A knowledge base, and determine whether information can be recalled from the specific domain Q&A knowledge base according to the second similarity matching result; wherein, the specific domain Q&A knowledge base is determined according to the target domain associated with the target query statement;

[0134] An information display module 43, configured to, when it is determined that information can be recalled from the specific domain Q&A knowledge base, determine a target answer corpus from the specific domain Q&A knowledge base, and display the target answer corpus as target recalled information to the target user.

[0135] Optionally, the first similarity matching module 41 is specifically configured to:

[0136] Determine a first digital employee that interacts with the target user at the current moment, and determine the digital employee type corresponding to the first digital employee;

[0137] When the digital employee type is a general domain digital employee, perform a first similarity matching between the target query statement and the preset question corpus in the general domain Q&A knowledge base through the first task execution engine corresponding to the first digital employee.

[0138] Optionally, the device further includes an identification information display module, specifically configured to:

[0139] Determine at least one target real employee from each of the candidate real employees according to the target domain and the expertise field tags associated with each candidate real employee; display the target identification information of each of the target real employees to the target user;

[0140] The second similarity matching module 42 is specifically configured to: in response to a selection operation of the target user on the target identification information, determine a real employee to be consulted from each of the target real employees, and determine the digital employee corresponding to the real employee to be consulted as a second digital employee;

[0141] Perform a second similarity matching between the target query statement and the preset question corpus in the exclusive Q&A knowledge base through the second task execution engine corresponding to the second digital employee; wherein, the exclusive Q&A knowledge base is the specific domain Q&A knowledge base maintained by the real employee to be consulted.

[0142] Optionally, the second similarity matching module 42 is further specifically configured to:

[0143] In a case where the second similarity matching results are all less than the similarity threshold, perform text matching between at least one query keyword included in the target query statement and the preset answer corpus in the specific domain Q&A knowledge base respectively;

[0144] If there is at least one of the preset answer corpus that contains any of the query keywords, determine that information can be recalled from the specific domain Q&A knowledge base;

[0145] The information display module 43 is specifically configured to:

[0146] Use the preset answer corpus containing any of the query keywords as the target answer corpus.

[0147] Optionally, the device further includes an artificial reply module, specifically for:

[0148] Determine whether information can be recalled from the exclusive Q&A knowledge base according to the second similarity matching result;

[0149] In the case where it is determined that information cannot be recalled from the exclusive Q&A knowledge base, obtain the artificial answer corpus generated by the real employee to be consulted for the target query statement;

[0150] Display the artificial answer corpus as the target recalled information to the target user.

[0151] Optionally, the device further includes a first Q&A corpus update module, specifically for:

[0152] Input the first prompt text, the target query statement, and the artificial answer corpus into the first large language model through the second task execution engine; wherein, the first prompt text is used to prompt the first large language model to perform text refinement according to the target query statement and the artificial answer corpus, and generate a first supplementary question corpus and a first supplementary answer corpus respectively;

[0153] Update the preset question corpus in the exclusive Q&A knowledge base according to the first supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the first supplementary answer corpus.

[0154] Optionally, the device further includes a query intention recognition module, specifically for:

[0155] Input the second prompt text and the target query statement into the second large language model through the first task execution engine; wherein, the second prompt text is used to prompt the second large language model to determine whether the query intention of the target query statement is a non-inquiry intention;

[0156] The first similarity matching module 41 is specifically for:

[0157] In the case where the second large language model determines that the query intention is not a non-inquiry intention, perform a first similarity matching between the target query statement and the preset question corpus in the general domain Q&A knowledge base.

[0158] Optionally, the device further includes a second Q&A corpus update module, specifically for:

[0159] Input the third prompt text and the target session information into the third large language model through the second task execution engine; wherein, the target session information is the session information generated between the real employee to be consulted and the consulting user, and the third prompt text is used to prompt the third large language model to refine the text according to the target session information to generate a second supplementary question corpus and a second supplementary answer corpus;

[0160] Update the preset question corpus in the exclusive Q&A knowledge base according to the second supplementary question corpus, and update the preset answer corpus in the exclusive Q&A knowledge base according to the second supplementary answer corpus.

[0161] The information query device provided by the embodiments of the present invention can execute the information query method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the method.

[0162] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0163] Embodiment Five

[0164] Figure 5 FIG. shows a schematic structural diagram of an electronic device 50 that can be used to implement the embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described herein and / or claimed.

[0165] As Figure 5 shown, the electronic device 50 includes at least one processor 51, and a memory communicatively connected to at least one processor 51, such as a read only memory (ROM) 52, a random access memory (RAM) 53, etc., wherein the memory stores a computer program executable by at least one processor, and the processor 51 can execute various appropriate actions and processes according to the computer program stored in the read only memory (ROM) 52 or the computer program loaded from the storage unit 58 into the random access memory (RAM) 53. In the RAM 53, various programs and data required for the operation of the electronic device 50 can also be stored. The processor 51, the ROM 52, and the RAM 53 are connected to each other through a bus 54. The input / output (I / O) interface 55 is also connected to the bus 54.

[0166] Multiple components in the electronic device 50 are connected to the I / O interface 55, including: an input unit 56, such as a keyboard, a mouse, etc.; an output unit 57, such as various types of displays, speakers, etc.; a storage unit 58, such as a magnetic disk, an optical disc, etc.; and a communication unit 59, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 59 allows the electronic device 50 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0167] The processor 51 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 51 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 51 executes the various methods and processes described above, such as the information query method.

[0168] In some embodiments, the information query method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as the storage unit 58. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 50 via the ROM 52 and / or the communication unit 59. When the computer program is loaded into the RAM 53 and executed by the processor 51, one or more steps of the information query method described above can be executed. Alternatively, in other embodiments, the processor 51 can be configured to execute the information query method by any other suitable means (e.g., by means of firmware).

[0169] The various embodiments of the systems and technologies described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGA), application-specific integrated circuits (ASIC), application-specific standard products (ASSP), systems-on-a-chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a dedicated or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0170] A computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, a special purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The computer programs can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0171] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0172] In order to provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0173] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0174] A computing system can include a client and a server. The client and the server are generally remote from each other and typically interact via a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0175] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is imposed herein.

[0176] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for querying information, characterized in that: The method comprises: Performing a first similarity match between the target query sentence sent by the target user and the preset question corpus in the general domain question-answering knowledge base, and determining whether information can be recalled from the general domain question-answering knowledge base according to the first similarity matching result; In the case where it is determined that information cannot be recalled from the general domain question-answering knowledge base, a second similarity matching is performed between the target query statement and a preset question corpus in the specific domain question-answering knowledge base, and it is determined whether information can be recalled from the specific domain question-answering knowledge base according to the second similarity matching result; wherein the specific domain question-answering knowledge base is determined according to the target domain associated with the target query statement; When it is determined that information can be recalled from the domain-specific question-answering knowledge base, a target answer corpus is determined from the domain-specific question-answering knowledge base, and the target answer corpus is displayed to the target user as target recall information.

2. The method according to claim 1, characterized in that The first similarity matching of the target query sentence sent by the target user with the preset question corpus in the general domain question-answering knowledge base includes: Determine a first digital employee who is currently interacting with the target user, and determine a digital employee type corresponding to the first digital employee; In the case where the digital employee type is a general-domain digital employee, a first task execution engine corresponding to the first digital employee performs a first similarity match between the target query statement and a preset question corpus in the general-domain question-and-answer knowledge base.

3. The method according to claim 1, wherein when it is determined that information cannot be recalled from the general domain question-answering knowledge base, the method further comprises: Determine at least one target real employee from the candidate real employees according to the target field and the proficiency field labels associated with the candidate real employees; Displaying the target identification information of each of the target real employees to the target user; The second similarity matching of the target query statement with the preset question corpus in the specific domain question-answering knowledge base includes: In response to the target user's selection operation on the target identification information, determining a real employee to be consulted from each of the target real employees, and determining a digital employee corresponding to the real employee to be consulted as a second digital employee; The target query statement is matched with the preset question corpus in the exclusive question and answer knowledge base by a second task execution engine corresponding to the second digital employee; wherein the exclusive question and answer knowledge base is the specific field question and answer knowledge base maintained by the real employee to be consulted.

4. The method according to claim 1, characterized in that: The determining whether information can be recalled from the specific domain question-answering knowledge base according to the second similarity matching result includes: In the case where the second similarity matching results are all less than the similarity threshold, text matching is performed on at least one query keyword included in the target query statement with a preset answer corpus in the specific field question and answer knowledge base; If there is at least one of the preset answer corpora containing any of the query keywords, it is determined that information can be recalled from the specific field question and answer knowledge base; The step of determining a target answer corpus from the domain-specific question-answering knowledge base includes: The preset answer corpus containing any of the query keywords is used as the target answer corpus.

5. The method according to claim 3, after the second task execution engine corresponding to the second digital employee performs a second similarity matching on the target query statement and the preset question corpus in the exclusive question and answer knowledge base, further comprising: Determining whether information can be recalled from the exclusive question-and-answer knowledge base according to the second similarity matching result; When it is determined that information cannot be recalled from the exclusive question-and-answer knowledge base, obtaining artificial answer corpus generated by the real employee to be consulted for the target query statement; The artificial answer corpus is displayed to the target user as target recall information.

6. The method according to claim 5, after presenting the artificial answer corpus as target recall information to the target user, further comprising: The first prompt text, the target query sentence and the manual answer corpus are input into the first large language model through the second task execution engine; wherein the first prompt text is used to prompt the first large language model to perform text extraction according to the target query sentence and the manual answer corpus to generate a first supplementary question corpus and a first supplementary answer corpus respectively; The preset question corpus in the exclusive question and answer knowledge base is updated according to the first supplementary question corpus, and the preset answer corpus in the exclusive question and answer knowledge base is updated according to the first supplementary answer corpus.

7. The method according to claim 2, before performing a first similarity matching between the target query sentence sent by the target user and the preset question corpus in the general domain question-answering knowledge base, further comprising: Inputting the second prompt text and the target query statement into the second largest language model through the first task execution engine; wherein the second prompt text is used to prompt the second largest language model to determine whether the query intent of the target query statement is a non-inquiry intent; The first similarity matching of the target query sentence sent by the target user with the preset question corpus in the general domain question-answering knowledge base includes: When the second largest language model determines that the query intent is not a non-inquiry intent, a first similarity matching is performed between the target query statement and a preset question corpus in a general domain question-answering knowledge base.

8. The method according to claim 3, further comprising: The third prompt text and the target conversation information are input into the third language model through the second task execution engine; wherein the target conversation information is the conversation information generated between the real employee to be consulted and the consulting user, and the third prompt text is used to prompt the third language model to perform text extraction according to the target conversation information to generate a second supplementary question corpus and a second supplementary answer corpus; The preset question corpus in the exclusive question and answer knowledge base is updated according to the second supplementary question corpus, and the preset answer corpus in the exclusive question and answer knowledge base is updated according to the second supplementary answer corpus.

9. An information query device, characterized in that: The device comprises: A first similarity matching module is used to perform a first similarity matching between a target query sentence sent by a target user and a preset question corpus in a general domain question-answering knowledge base, and determine whether information can be recalled from the general domain question-answering knowledge base according to the first similarity matching result; A second similarity matching module is used to perform a second similarity matching between the target query statement and a preset question corpus in the specific domain question and answer knowledge base when it is determined that information cannot be recalled from the general domain question and answer knowledge base, and determine whether information can be recalled from the specific domain question and answer knowledge base based on the second similarity matching result; wherein the specific domain question and answer knowledge base is determined based on the target domain associated with the target query statement; The information display module is used to determine the target answer corpus from the specific field question and answer knowledge base when it is determined that information can be recalled from the specific field question and answer knowledge base, and display the target answer corpus as target recall information to the target user.

10. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the information query method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute the information query method according to any one of claims 1 to 8.

12. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the information query method according to any one of claims 1 to 8.

Images