Automatic question and answer method and device in recruitment

Abstract

The embodiment of the invention discloses an automatic question answering system method and device in recruitment. The method comprises the following steps: determining recruitment states of recruiters and candidates; judging a current communication theme according to the recruitment state; response rules triggered by the communication theme in the response rule set are determined, and the response rules comprise rule conditions and rule execution; and when the value of at least one rule condition is true, executing rule execution corresponding to the rule condition, thereby realizing accurate automatic response between the recruiter and the candidate, and improving user experience.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and more particularly to an automatic question-answering system method and device for recruitment. Background Art

[0002] With the development of mobile Internet, smart recruitment has been used more and more widely. In the process of smart recruitment, it is common to use pre-trained models for semantic recognition and automatically respond to candidates. However, when automatically responding based on this method, no model can be 100% accurate. In other words, if the semantic recognition is not accurate enough, there will often be a misunderstanding of the candidate's status or the current topic, and it is easy for the two parties to chat on different channels or answer irrelevant questions, which reduces the efficiency of the recruiter and affects the image of the recruiter, resulting in a poor experience of automatic response.

[0003] Based on this, a more accurate automatic question-answering method in recruitment is needed. Summary of the invention

[0004] The embodiments of this specification provide a method, device and storage medium for automatic question answering in recruitment, which are used to solve the following technical problems: a more accurate method for automatic question answering in recruitment is needed.

[0005] To solve the above technical problems, one or more embodiments of this specification are implemented as follows:

[0006] In a first aspect, an embodiment of the present specification provides an automatic question-and-answer method in recruitment, which is applied to a recruitment system in which a set of answer rules is pre-configured, and the method includes: determining the recruitment status of the recruiter and the candidate; judging the current communication topic based on the recruitment status; determining the answer rule triggered by the communication topic in the answer rule set, wherein the answer rule includes rule conditions and rule execution; when at least one rule condition evaluates to true, executing the rule execution corresponding to the rule condition.

[0007] In a second aspect, one or more embodiments of this specification provide an electronic device, including:

[0008] at least one processor; and,

[0009] a memory communicatively connected to the at least one processor; wherein,

[0010] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to the first aspect.

[0011] In a third aspect, an embodiment of the present specification provides a non-volatile computer storage medium storing computer executable instructions. When a computer reads the computer executable instructions in the storage medium, the instructions enable one or more processors to execute the method described in the first aspect.

[0012] At least one of the above-mentioned technical solutions adopted in one or more embodiments of the present specification can achieve the following beneficial effects: by determining the recruitment status of the recruiter and the candidate; judging the current communication topic according to the recruitment status; determining the response rule triggered by the communication topic in the response rule set, wherein the response rule includes rule conditions and rule execution; when at least one rule condition is true, executing the rule execution corresponding to the rule condition, thereby realizing accurate automatic response between the recruiter and the candidate and improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0014] Figure 1 A schematic diagram of a system architecture provided in an embodiment of this specification;

[0015] Figure 2 A flowchart of an automatic question-answering method in recruitment provided by an embodiment of this specification;

[0016] Figure 3 A schematic diagram of a table of a set of dimension value sets provided in an embodiment of this specification;

[0017] Figure 4 A schematic diagram of adding a new business rule to the embodiment of this specification;

[0018] Figure 5 A schematic diagram of the execution order and mutual exclusion relationship of a rule execution provided in an embodiment of this specification;

[0019] Figure 6 A schematic diagram of an automatic response process in the whole recruitment process provided by an embodiment of this specification;

[0020] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0021] The embodiments of this specification provide a method, device and storage medium for automatic question answering in recruitment.

[0022] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0023] like Figure 1 As shown, Figure 1 A schematic diagram of a system architecture provided in an embodiment of this specification. Figure 1 The architecture includes a configuration domain, a support domain, and an operation domain. Users can configure various answering rules in advance through the configuration domain, wherein the answering rules include rule conditions and rule execution. The configured answering rules will be converted and stored in the rule library of the support domain. When the recruiter and the candidate (the candidate may be a user who has not yet established contact but may meet the recruitment conditions) communicate, automatic answers are made based on the recruitment status and contextual information of both parties, thereby avoiding ambiguity based on semantic analysis and achieving accurate answers based on answering rules.

[0024] like Figure 2 As shown, Figure 2 A flowchart of an automatic question-answering method in recruitment provided in an embodiment of the present specification is applied to a recruitment system with a set of answer rules pre-configured.

[0025] Figure 2 The process in may include the following steps:

[0026] S202: Determine the recruitment status of the recruiter and the candidate.

[0027] The recruitment status here includes the communication status between the recruiter and the candidate (whether the message has been replied), whether there is contact information, whether the person matches the position, who sent the last message, the source of the candidate, the status of the candidate being selected, etc., and can also include the context involved in multiple rounds of communication between the two parties, etc.

[0028] S204: Determine the current communication topic according to the recruitment status.

[0029] The communication topic refers to the specific business scenario or the main information involved in the conversation between the two parties, such as company situation, position situation, etc.

[0030] S206: Determine a response rule in the response rule set that is triggered by the communication topic.

[0031] The response rule includes rule conditions and rule execution. The general format of the response rule is: when {certain rule conditions are true}, {certain rule execution} is executed.

[0032] The rule conditions contain a series of expressions or a combination of some expressions. The expression is the smallest logical operation unit, which performs a certain operation and gives a specific output. The rule conditions are composed of a group of expressions that return true / false. The expression group is a combination of a group of expressions, and the combination method supports ('and', 'or'). The rule conditions contain multiple dimensions, and each dimension contains a corresponding latitude value set, that is, a set of all possible values ​​of a certain dimension. For example, invited, eliminated, etc.

[0033] Rule execution is the action triggered when a rule condition is true. Common types include business actions, system actions, etc. For example, follow-up chat, counter-question, etc.

[0034] like Figure 3 As shown, Figure 3 A schematic diagram of a table of a set of dimension value sets provided in an embodiment of this specification. In this schematic diagram, exemplary value sets are given for the value sets of multiple dimensions in the rule conditions and rule execution. Obviously, in actual applications, the specific dimensions and dimension value sets are not limited to this, and the recruiter can add them in the configuration field based on actual needs.

[0035] Rule conditions and rule execution can be done in Figure 1 In the illustrated architecture, the configuration is pre-configured by the recruiter based on the configuration domain. The configuration of the recruiter can be performed using natural language.

[0036] S208: When at least one rule condition evaluates to true, the rule corresponding to the rule condition is executed.

[0037] Specific rule execution may include proactive greetings, chatting or asking questions, proactive inquiries, or follow-up questions on the previous topic.

[0038] For example, when a suitable candidate is found, a greeting message is sent proactively; when a candidate asks a specific question about a position, a prepared answer is provided, and then a counter-question or a chat is initiated. For example, if a candidate asks about the salary, an answer such as "Does 5-8k meet your expectations?" is provided; proactive inquiries are made to candidates regarding issues that the company is concerned about, such as whether the work location matches the company; when a candidate does not respond for a long time after being asked, a personalized follow-up question is provided for the previous question, such as "Seeing that you are online, do you have any concerns about the work content?", etc., so that the company can intelligently conduct personalized text question-and-answer conversations with candidates.

[0039] In one embodiment, the response rule set is pre-configured based on the following method, specifically including: obtaining the natural language input by the recruiter, and determining the dimension value and business dimension corresponding to the natural language; generating corresponding rule conditions and rule executions according to the dimension value and business dimension; splitting the rule conditions into multiple rule condition expressions, and splitting the rule execution into multiple rule execution expressions; generating response rules corresponding to the multiple rule condition expressions, and generating execution actions corresponding to the rule execution expressions and associated with the response rules.

[0040] For example, for the natural language input by the recruiter, after determining that the dimensions include candidate status, message status, and person-job matching status, and then determining that the business dimension includes business actions, the corresponding rule conditions and rule executions can be generated according to the dimension values ​​and business dimensions, including the following:

[0041] Rule conditions:

[0042] ((Candidate status is not eliminated or invited) and (Message has been read) and (Position matching status is not potential match)) or ((Position matching status is not potential match));

[0043] Rule execution: 1) Request contact information; 2) Change candidate status; 3) Send a text reply.

[0044] Furthermore, the rule condition is split into a plurality of rule condition expressions, and the rule execution is split into a plurality of rule execution expressions.

[0045] Continuing with the previous example, the expression obtained by splitting is as follows: (

[0047] (Candidate status is not Eliminated or Invited) Row 1. Expression 1

[0048] And row 1. Expression relation 1

[0049] (Message read) Line 1. Expression 2

[0050] And row 1. Expression relationship 2

[0051] (The person-job matching status is not a potential match) Line 1. Expression 3

[0052] ) Expression group line 1

[0053] Or inter-row relationship (can only be or) (

[0055] (The person-job matching status is not a potential match) Line 2. Expression 1

[0056] ) Row 2

[0057] In the above example, for multiple expression relationships in the same row of expression (i.e. row one. expression relationship one and row one. expression relationship two), the expression relationship in the same row of expression can only be set to "and"; in addition, the inter-row relationship between expression row one and expression row two can only be set to "or". Multiple rule condition expressions containing the inter-row relationship and expression relationship are generated, and business rules are added based on the aforementioned expressions and saved.

[0058] like Figure 4 As shown, Figure 4 This is a schematic diagram of adding business rules in the embodiment of this specification. In this diagram, the primary key is added in sequence +1, and the expression name can be customized. Four operations are supported: contains, does not contain, is equal to, is not equal to, and the dimension value is associated with the business dimension field of this row, multiple options. The Enable column indicates that the expression is effective after checking it.

[0059] A rule condition and a rule execution can be associated with each other through the same rule name. For example, if a rule condition is named AR0001, and the corresponding rule execution is assigned the value AR0001 through the corresponding association field, it can be determined that the rule execution is associated with the rule condition. Generally, a rule condition can be associated with multiple rule executions, while a rule execution can only be associated with one rule condition.

[0060] In one embodiment, when generating response rules corresponding to the multiple rule condition expressions, the rule type corresponding to the response rule can be determined, and a response rule containing the rule type can be generated; wherein the rule type includes global rules, company-level rules, position-level rules and coverage types, and the coverage type is used to characterize whether the rules can overlap or coexist.

[0061] The calculation order of the three types of rules is: position type rule > company type rule > global rule. Multiple rules of each type are allowed to exist. If the hit position type rule has an override, the company type rule will not be calculated. If the hit company type rule has an override, the global rule will not be calculated.

[0062] Specifically, when there are multiple executable rule executions, the rule execution corresponding to the rule condition is executed, and the multiple executable rule executions are executed in the following logical order:

[0063] Obtain candidate ID, position ID and company ID;

[0064] Execute the rule corresponding to the position ID and determine whether there is an overriding type. If so, end and return the execution result; otherwise,

[0065] Get all the rule executions corresponding to the company logo and determine whether there is an overriding type. If so, end and return the execution result; otherwise,

[0066] Get the global rules and query the order of the global rules. At the same time, add the company-level rule order + K (for example, K = 1000, the company-level rule order + 1000), the global rule order + 2K (correspondingly, the global rule order + 2000), and create a return result for each rule;

[0067] The rules are executed in order according to the sorting, and all the execution results that meet the conditions are returned, thereby ensuring the sequential execution of the position type rules, company type rules and global rules.

[0068] In one implementation, the rule execution is split into multiple rule execution expressions, and the execution order and mutually exclusive relationship of the rule execution expressions can be determined to generate multiple rule execution expressions containing the execution order and mutually exclusive relationship. Figure 5 As shown, Figure 5 A schematic diagram of the execution order and mutual exclusion relationship of a rule execution provided in an embodiment of this specification.

[0069] In this diagram, the execution order (or priority order) is directly determined by the sorting sequence number. The smaller the sequence number, the higher the priority. The mutually exclusive relationship is determined by checking the rows and columns. That is, if multiple rules selected at the same time in a row or a column have a mutually exclusive relationship. Multiple rules in a mutually exclusive relationship will only be executed with the priority first. In other words, when there are multiple executable rules to be executed, they will be screened according to the execution order and the mutually exclusive relationship, and the rules with the first execution order will be retained and executed.

[0070] By splitting the rule conditions into multiple rule expressions and determining the logical relationships among them, and splitting the rule execution into multiple rule execution expressions and determining the execution order and mutual exclusion relationship, when multiple rules are triggered, the rule engine decides which rule to execute first, so as to achieve a hierarchical implementation of the response rules, achieve consistency in execution, and avoid invalid or even redundant responses.

[0071] In one implementation, an external big model can also be called to provide a richer content answer. Specifically, the communication topic is classified to determine the corresponding topic classification; prompt words corresponding to the topic classification are generated; and the prompt words, response rules and topic classification are sent to the big model so that the big model generates the corresponding dialogue content. After the big model generates the dialogue content, the dialogue content can be returned to the recruitment system, and the recruitment system can forward it to the candidate.

[0072] In this way, the context of communication with candidates can be pre-classified and designed. According to the classification algorithm of the chat records in the specified window, 90 categories of topics and sub-topic classifications are preset for recruitment scenarios. Users (including recruiters and candidates) and agents complete the extraction of knowledge base content under each sub-topic and generate prompt words for answers. The large model comprehensively completes the generation of dialogue content based on natural language description, constraints, and topic classification, thereby achieving richer personalized answers.

[0073] By setting the response rules in advance, the recruitment system can automatically obtain the communication status of both parties (including whether it is screening, replying or greeting, what is the main content of the current discussion, etc.), and the backend system will be responsible for evaluating the rules based on the input data and determining the corresponding actions. It will check which rules are triggered and determine the execution order based on the priority, and then be responsible for performing the corresponding actions. This may include calling external services, updating databases, or sending notifications. Figure 6 As shown, Figure 6 This is a flow chart of automatic response in the whole recruitment process provided by the embodiment of this specification. After the recruitment system obtains the chat context information, it automatically screens and determines the communication topic, and then executes reply, greeting, follow-up chat, etc. in sequence according to the topic, and finally sends out an interview invitation, realizing accurate automatic response in the whole recruitment process.

[0074] In the second aspect, Figure 7 As shown, Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification, the device includes:

[0075] at least one processor; and,

[0076] a memory communicatively connected to the at least one processor; wherein,

[0077] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to the first aspect.

[0078] In the third aspect, based on the same idea, the embodiment of this specification also provides a non-volatile computer storage medium corresponding to the above method, storing computer executable instructions. When the computer reads the computer executable instructions in the storage medium, the instructions enable one or more processors to execute the method described in the first aspect.

[0079] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0080] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0081] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0082] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. An automatic question-answering method in recruitment, applied to a recruitment system in which a set of answering rules is pre-configured, the method comprising: Determine the recruitment status of recruiters and candidates; Determine the current communication topic based on the recruitment status; Determining a response rule in the response rule set that is triggered by the communication topic, wherein the response rule includes a rule condition and a rule execution; When at least one rule condition evaluates to true, the rule execution corresponding to the rule condition is executed.

2. The method of claim 1, wherein: The response rule set is pre-configured based on the following method: Acquire the natural language input by the recruiter, and determine the dimension value and business dimension corresponding to the natural language; Generate corresponding rule conditions and rule execution according to the dimension value and business dimension; Splitting the rule condition into a plurality of rule condition expressions, and splitting the rule execution into a plurality of rule execution expressions; Response rules corresponding to the plurality of rule condition expressions are generated, and execution actions corresponding to the rule execution expressions and associated with the response rules are generated.

3. The method of claim 2, wherein: Split the rule execution into multiple rule execution expressions, including: The execution order and mutually exclusive relationship of the rule execution expressions are determined, and a plurality of rule execution expressions including the execution order and mutually exclusive relationship are generated.

4. The method of claim 3, wherein: When there are multiple executable rule executions, executing the rule execution corresponding to the rule condition includes: Screening is performed according to the execution order and the mutually exclusive relationship, and the rules with the previous execution order are retained and executed.

5. The method of claim 2, wherein: Split the rule condition into multiple rule expressions, including: Determine a logical relationship between the plurality of regular expressions, wherein the logical relationship comprises an inter-line relationship and an expression relationship; The inter-row relationship is set to be or, and the expression relationship is set to be and; A plurality of rule condition expressions including the inter-row relationships and expression relationships are generated.

6. The method of claim 2, wherein: Generating response rules corresponding to the multiple rule condition expressions, including: Determine the rule type corresponding to the response rule, and generate a response rule containing the rule type; The rule types include global rules, company-level rules, position-level rules and coverage types, and the coverage type is used to characterize whether the rules can be overlapped or coexist.

7. The method of claim 6, wherein: When there are multiple executable rule executions, executing the rule execution corresponding to the rule condition includes: The plurality of executable rules are executed in the following logical order: Obtain candidate ID, position ID and company ID; Execute the rule corresponding to the position ID and determine whether there is an overriding type. If so, end and return the execution result; otherwise, Get all the rule executions corresponding to the company logo and determine whether there is an overriding type. If so, end and return the execution result; otherwise, Get the global rules and query the order of the global rules. At the same time, the company-level rule order + K, the global rule order + 2K, create a return result for each rule, where K is a natural number greater than the number of global rules; Execute in order according to the order, and return all execution results that meet the conditions.

8. The method of claim 1, wherein: The method further comprises: Classifying the communication topics and determining corresponding topic classifications; generating prompt words corresponding to the subject classification; The prompt words, response rules and topic classification are sent to the big model so that the big model can generate corresponding dialogue content.

9. The method of claim 1, wherein: The specific operations corresponding to the rule execution include: Take the initiative to say hello, chat or ask questions, actively inquire or follow up on the previous topic.

10. An electronic device comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 9.

Images