An ai voice interactive post screening and recommendation method and system for blue-collar job seekers

Abstract

The application relates to the field of intelligent interaction, in particular to an AI voice interactive post screening and recommendation method and system for blue-collar job hunting, which acquires a voice query instruction of a user end, extracts screening conditions through voice and semantic recognition, and evaluates the demand definiteness of the user; if the demand definiteness does not satisfy the matching condition, an interaction strategy is determined according to the definiteness, supplementary information is acquired through multiple rounds of voice question and answer, a candidate post set is determined according to the supplemented conditions, time change attribute information and user information of each post are acquired, time suggestion information is generated and the candidate posts are sorted according to the time suggestion information, and finally a post recommendation list is output. The application realizes differentiated interaction guidance through demand definiteness evaluation, generates personalized time suggestions in combination with dynamic time attributes of the posts, and improves the post matching efficiency and user experience in the blue-collar job hunting scene.

Description

Technical Field

[0001] This application relates to the field of intelligent interactive technology, and in particular to an AI voice interactive job screening and recommendation method and system for blue-collar job seekers. Background Technology

[0002] Recruitment applications primarily rely on graphical user interfaces (GUIs) for interaction, requiring users to manually input keywords or navigate filter menus. While some existing technologies incorporate voice input, these typically only convert voice commands to text and perform simple keyword matching. However, in blue-collar job-seeking scenarios, user queries are often highly colloquial and fragmented, such as directly asking "Any jobs nearby?" or "Any daily pay jobs?" rather than inputting filters according to pre-defined criteria. Existing systems struggle to accurately extract complete filter conditions from such unstructured spoken expressions and lack proactive mechanisms to guide users in completing their information. Furthermore, the clarity of expression varies among users, and existing systems fail to differentiate between them and employ tailored interaction strategies. This results in ineffective guidance for users with unclear expressions and unnecessary interactive redundancy for those with clear expressions.

[0003] Secondly, existing recruitment systems typically rely on static matching criteria for recommendation ranking, rarely considering the dynamic changes in job information within the actual labor market. The recruitment status of blue-collar jobs is highly time-sensitive; positions may fill up quickly or enter different recruitment stages. Furthermore, factors such as a user's available start time and the urgency of their job search significantly impact matching effectiveness. Current technology lacks effective modeling of these dynamic time attributes, meaning recommended positions may be outdated or conflict with a user's schedule. On the other hand, information such as remaining job openings is often provided unilaterally by employers, leading to dynamic attrition, such as simultaneous recruitment through other channels or data inaccuracies. Existing systems lack verification mechanisms for this information, potentially recommending positions that are already full or contain distorted information, thus affecting the accuracy of job search decisions. Summary of the Invention

[0004] To address one or more problems in the existing technology, the main objective of this application is to provide an AI-powered voice-interactive job screening and recommendation method and system for blue-collar job seekers.

[0005] To achieve the aforementioned objectives, this application proposes an AI-powered voice-interactive job screening and recommendation method for blue-collar job seekers, the method comprising:

[0006] Obtain the voice query command input by the user;

[0007] The voice query command is subjected to voice and semantic recognition to extract filtering conditions, and the clarity of the user's needs is evaluated based on the filtering conditions.

[0008] Based on the evaluation results, analyze whether the required clarity meets the preset matching conditions;

[0009] If the required clarity does not meet the job matching conditions, the interaction strategy is determined by combining the voice query command and the required clarity, and the voice Q&A continues to interact with the user terminal to obtain supplementary information until the preset matching conditions are met.

[0010] Based on the supplementary information and screening criteria, a candidate job set is determined;

[0011] Obtain time change attribute information and user information for each candidate position; generate time suggestion information for each candidate position based on the time change attribute information and the user information; and sort the candidate position set according to the time suggestion information.

[0012] Based on the sorted set of candidate positions, a recommended list of positions is generated and output to the user terminal.

[0013] This application also provides an AI-powered voice-interactive job screening and recommendation system for blue-collar job seekers, including:

[0014] The acquisition module is used to acquire the voice query commands input by the user.

[0015] The recognition and extraction module is used to perform voice and semantic recognition on the voice query command, extract filtering conditions, and evaluate the clarity of the user's needs based on the filtering conditions;

[0016] The analysis module is used to analyze whether the required clarity meets preset matching conditions based on the evaluation results;

[0017] The judgment module is used to determine the interaction strategy by combining the voice query command and the clarity of the requirement if the required clarity does not meet the job matching conditions, and continue to perform voice Q&A to interact with the user terminal to obtain supplementary information until the preset matching conditions are met.

[0018] The determination module is used to determine the candidate job set based on the supplementary information and screening criteria;

[0019] The first generation module is used to obtain time change attribute information and user information for each candidate position, generate time suggestion information for each candidate position based on the time change attribute information and the user information, and sort the candidate position set according to the time suggestion information.

[0020] The second generation module is used to generate a job recommendation list based on the sorted set of candidate jobs and output it to the user terminal.

[0021] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described above.

[0022] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0023] This application's embodiment of the AI-powered voice-interactive job selection and recommendation method and system for blue-collar job seekers uses voice commands as the interaction entry point, eliminating the need for users to manually fill out forms or navigate filter menus, significantly lowering the operational threshold. More importantly, it determines whether the user has clearly stated key information such as job type, location, salary, and availability. Based on the clarity of the information, the interaction strategy is switched: users with clear expressions quickly enter the matching process, avoiding redundant follow-up questions; users with vague expressions or missing information are guided through targeted multi-round question-and-answer sessions until the necessary information is completed. After completing the condition completion and recalling candidate jobs, a dynamic time-dimensional adaptation analysis is introduced. By acquiring dynamic attributes such as job posting time, remaining positions, and recruitment deadline, combined with the user's available start time and time urgency, specific time suggestions are generated for each job, such as "go immediately," "contact as soon as possible," or "time mismatch." The final recommendation list presented to the user not only matches the static conditions with the user's needs but also aligns with the real-time status of the job and the user's actual available start time, solving the problems of recommending expired jobs or time mismatches. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating an embodiment of an AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to this application.

[0025] Figure 2 This is a flowchart illustrating an embodiment of an AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to this application.

[0026] Figure 3 This is a schematic block diagram of an AI voice-interactive job screening and recommendation system for blue-collar job seekers, according to an embodiment of this application.

[0027] Figure 4 This is a schematic block diagram of the structure of a computer device according to an embodiment of this application;

[0028] Figure 5 This is a comparative schematic diagram of the remaining job quota verification mechanism according to an embodiment of this application.

[0029] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0031] Reference Figure 1 This application provides an AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers, the method comprising:

[0032] S1. Obtain the voice query command input by the user;

[0033] S2. Perform speech and semantic recognition on the voice query command, extract filtering conditions, and evaluate the clarity of the user's needs based on the filtering conditions;

[0034] S3. Based on the evaluation results, analyze whether the required clarity meets the preset matching conditions;

[0035] S4. If the required clarity does not meet the job matching conditions, then the interaction strategy is determined by combining the voice query command and the required clarity, and the voice Q&A continues to interact with the user terminal to obtain supplementary information until the preset matching conditions are met.

[0036] S5. Determine the candidate job set based on the supplementary information and screening criteria;

[0037] S6. Obtain time change attribute information and user information for each candidate position; generate time suggestion information for each candidate position based on the time change attribute information and the user information; and sort the candidate position set according to the time suggestion information.

[0038] S7. Based on the sorted set of candidate positions, generate a recommended list of positions and output it to the user terminal.

[0039] As described in steps S1-S3 above, step one involves obtaining the user's voice query command. In blue-collar job-seeking scenarios, users typically want to express their job needs in the most natural and quickest way, rather than facing a bunch of forms or menus after opening an application. Voice commands are the input method closest to human daily communication habits. Users can directly say things like, "I want to find a welding job nearby" or "Are there any daily-paid moving jobs?" without having to fill in conditions such as job type, location, and salary. Step two involves two layers of processing. The first layer is the conventional speech-to-text conversion and the identification of meaningful filtering conditions from the text, such as job type, location, salary requirements, and start time. The second layer is that the system does not assume that the user has clearly stated all the conditions, but actively assesses the clarity of the user's expression. The so-called clarity of needs is a comprehensive judgment indicator. For example, if a user says, "I want to find a welding job in Futian District with a daily wage of 300 yuan," this sentence contains specific information such as job type, location, salary, and payment method, and the system can determine it as having high clarity. If a user only says they're looking for work or if there are any jobs nearby, the extracted filter conditions are few and contain vague terms like "nearby," which the system will classify as low specificity. There's also an intermediate state, such as a user saying they want to find a general laborer with a higher salary. Here, the job type is specific, but the salary is vague, and the location is missing, which falls under medium specificity. The evaluation principle compares the extracted conditions with a pre-set set of key dimensions, counting how many dimensions are covered and detecting the presence of vague terms, thus quantifying the level of specificity. The pre-set matching conditions in step three can be understood as the minimum standard at which the system believes it can directly conduct job searches without further questioning the user. Typically, when the specificity of the request reaches high specificity, meaning the user has provided sufficiently specific and complete filter conditions, the system determines it can directly enter the matching stage. If the specificity does not meet this standard, such as being medium or low specificity, then the current information is considered insufficient to provide accurate recommendations, and further interaction with the user is needed to supplement the information. The design principle here is to avoid two extremes: one is repeatedly asking follow-up questions to users who have already explained themselves clearly, creating redundant interaction; the other is directly returning a massive amount of irrelevant results to users with vague expressions, leaving them to filter through them themselves. This judgment determines whether to proceed directly to the fast track or enter the guided Q&A process.

[0040] As described in steps S4-S7 above, step four is the specific implementation of the differentiated interaction strategy. Different interaction methods will be adopted based on the clarity level assessed in step two. For users with medium clarity, i.e., the user has clearly stated some conditions but important dimensions are missing or vague terms are used, a completion mode will be adopted. For example, if a user says they want to find a general worker in Bao'an District but doesn't specify their salary requirements, the system will follow up with salary-related questions according to a preset priority, such as "What is your expected monthly salary range?" For users with low clarity, such as those who only say they want to find work, the system will adopt an exploration mode. It will first obtain the user's current location and then push the most popular job types in that area for the user to choose from, such as sorting workers, porters, and forklift drivers. Users can easily express their preferences by selecting whichever they are interested in, without having to construct complex search statements themselves. After each round of question and answer, the newly provided information will be added to the filtering conditions, and the clarity will be reassessed until the matching conditions are met. Step 5: Once the user's needs are clear enough to meet the matching criteria, all collected filtering conditions, including those extracted from the initial voice and those supplemented in subsequent rounds of question-and-answer sessions, are combined into a complete query condition. This query is then used to retrieve all positions that meet these conditions from the job database, forming a candidate job set. Step 6: Traditional recruitment recommendation systems only perform static matching, checking if job tags and user conditions match. However, blue-collar job seekers are extremely sensitive to timeliness. A job might be actively recruiting in the morning but already full by the afternoon. A user might be able to start today, or they might have to wait three days to begin work. This step introduces dynamic adaptation analysis along the time dimension. Specifically, the system obtains the time-varying attributes for each candidate job, including when the job was posted, the employer's recruitment deadline, and the number of remaining openings. Simultaneously, the system also gathers information about the user's available start time from previous voice interactions, as well as the user's expressed urgency, such as saying "the sooner the better" or "I can only start next week." Based on this information, several comparisons are performed. First, the system determines whether the user can start work before the job deadline. If not, the position is a time mismatch for the user and should be lowered in priority or given a clear warning. Second, based on how long the position has been posted and historical data of similar positions, the system determines whether the position is currently in a high-demand, saturated, or nearing completion phase, providing different suggestions for each stage. Finally, considering the user's time urgency and the remaining positions, the system assesses whether there is a shortage of openings. Based on the analysis results from these dimensions, the system generates a time-related suggestion for each position, such as suggesting immediate application, contacting for confirmation as soon as possible, scheduling an application, or suggesting alternative positions if the time is not a match. These suggestions, along with traditional matching scores, then determine the final recommended order of positions.For example, a job posting with a slightly lower match but a suggested start time might appear ahead of a job posting with a higher match but a suggested start time that doesn't match. Step seven presents the sorted job list to the user in an appropriate format. Considering the usage habits of blue-collar users, the recommendation list can include both text information and voice prompts.

[0041] As mentioned above, using voice commands as the interaction entry point eliminates the need for users to manually fill out forms or navigate filter menus, significantly lowering the operational threshold. More importantly, it determines whether the user has clearly stated key information such as job type, location, salary, and availability. Based on the clarity of the information, the interaction strategy switches: users with clear expressions quickly enter the matching process, avoiding redundant follow-up questions; users with vague expressions or missing information are guided through targeted multi-round Q&A sessions until the necessary information is provided. After completing the condition completion and recalling candidate positions, dynamic time-based adaptation analysis is introduced. By acquiring dynamic attributes such as the job posting time, remaining positions, and recruitment deadline, combined with the user's available start time and time urgency, specific time suggestions are generated for each position, such as "go immediately," "contact as soon as possible," or "time mismatch." The final recommendation list presented to the user not only matches the static conditions with the user's needs but also aligns with the real-time status of the positions and the user's actual available start time, solving the problem of recommending expired positions or time mismatches.

[0042] In one embodiment, the steps of performing speech and semantic recognition on the voice query command, extracting filtering conditions, and evaluating the clarity of the user's needs based on the filtering conditions include:

[0043] The voice query command is subjected to speech recognition and semantic parsing to extract the filtering conditions from the voice query command;

[0044] The filtering conditions are matched with a preset set of filtering dimensions to determine the number of matched filtering dimensions. The preset set of filtering dimensions includes job type, location, salary, and time.

[0045] The dimension coverage ratio is calculated based on the number of matched filtering dimensions and the total number of dimensions in the preset filtering dimension set.

[0046] If the dimension coverage ratio is greater than or equal to a preset ratio threshold, then the requirement clarity is determined to be high clarity.

[0047] If the dimension coverage ratio is less than the preset ratio threshold, then determine whether the filtering conditions contain fuzzy words;

[0048] If the filtering criteria contain vague words, then the required specificity is determined to be low.

[0049] If the filtering criteria do not contain vague words, then the required clarity is determined to be medium clarity.

[0050] As mentioned above, in step 1, the user's spoken words are recognized and converted into text. Valuable job-seeking elements are then extracted from this text. For example, if a user says they want to find a daily-paid porter in Bao'an District, the system needs to extract three conditions: job type (porter), location (Bao'an District), and daily payment method. The time dimension, such as start time, may not be included. The extraction principle utilizes entity recognition technology in natural language processing to map keywords in spoken language to predefined filtering dimensions. In step 2, the system pre-sets four core filtering dimensions: job type, location, salary, and time. These four dimensions are determined based on the information types users most frequently focus on in blue-collar job-seeking scenarios. All filtering conditions extracted in the previous step are compared one by one with these four dimensions to see if each dimension has a corresponding condition. For example, if the user mentioned job type and location but not salary and time, the number of matched dimensions is 2. In step 3, the formula for calculating the dimension coverage ratio is simple: divide the number of matched dimensions by 4. For example, if the user provides two dimensions (job type and location), the coverage ratio is 50%. If a user provides three dimensions—job type, location, and salary—the coverage rate is 75%. If the user provides all four dimensions, the coverage rate is 100%. Step 4 sets the preset coverage threshold to a configurable value, such as 75%. When a user's coverage rate reaches 75% or higher, it means the user has proactively provided information on three or four core dimensions. This implies the user's expression is quite complete, and the system can consider the user's needs to be clear. Step 5 states that when a user's coverage rate is low, such as only mentioning job type or location, with a coverage rate of only 25% or 50%, it cannot be directly judged as low clarity. This is because there is another possibility: although the user provides few dimensions, their expression is very specific. For example, a user saying "looking for a job nearby" only involves the location dimension and uses vague terms; "nearby" is a typical vague term. On the other hand, a user saying "looking for welding work" only provides one dimension, but welding is a specific job type. The key to distinguishing between these two situations is to detect whether there are vague terms in the filtering conditions, such as words like "nearby," "surrounding area," "about," or "suitable point," or generic terms like "job," "job," or "position." Step 6: When users use vague or generic terms in their expressions, it indicates that their needs are not yet clear. For example, if a user says they want to find a suitable job, "suitable" is vague, and "job" is a general term. The system cannot determine any valuable filtering criteria from this. In this case, the user doesn't need the system to ask questions about a missing dimension, but rather needs the system to provide some popular options to help them build understanding and gradually clarify what they really want. Step 7: When users provide a limited number of dimensions but their expression is clear, for example, if a user only says they are looking for welding work without specifying location, salary, or time, but the job title of welding is clear and there are no vague terms, then the user's needs are clear, just incomplete.The system knows what job the user wants, and can fill in the missing information by asking questions about the other missing dimensions in turn.

[0051] Reference Figure 2 In one embodiment, the step of determining the interaction strategy by combining the voice query command and the clarity of the demand includes:

[0052] S21. Obtain the required clarity;

[0053] S22. If the required clarity is medium clarity, the interaction strategy is determined to be the completion mode. The completion mode is used to compare the extracted filtering conditions with the preset filtering dimension set, determine the dimensions in the preset filtering dimension set that are not covered by the filtering conditions as missing dimensions, and generate a question instruction for the missing dimensions according to the preset dimension priority order.

[0054] S23. If the required clarity is low, the interaction strategy is determined to be exploration mode. The exploration mode is used to obtain the current location information of the user terminal, recall popular job information based on the current location information, generate an option list containing the popular job information, and send a selection request to the user terminal.

[0055] As described above, when the clarity of the need is determined to be medium, it means the user has provided some specific filtering criteria. For example, the user said they are looking for welding work, specifying the job but not the location, salary, or working hours. In this case, the missing information dimensions need to be identified and questions asked of the user in a logical order. Specifically, the extracted filtering criteria are compared one by one with the four preset dimensions: job type, location, salary, and time. Dimensions without corresponding conditions are marked as missing dimensions. For example, if the user only provided the job type, then the missing dimensions are location, salary, and time. The system then decides what to ask first according to a preset priority order. Typically, location has the highest priority because users usually want to work near their residence; second is salary, which is the core benefit most important to job seekers; and finally, time, such as the start date. Based on this priority, the system generates a question instruction for the missing dimension with the highest priority, such as asking which area the user prefers to work in. When the clarity of the need is determined to be low, it means the user's expression contains vague or general terms, such as the user only saying they want to find work or what jobs are available nearby. In this situation, the user's own understanding of their needs may not be clear, and if the system asks for specific locations or salaries, the user may not be able to answer. The design idea behind the exploration mode is to switch roles, from questioner to recommender. The system first obtains the user's current location, and then recalls information on recently popular job postings in the area, such as sorting clerks, porters, and forklift drivers urgently needed in nearby industrial parks. The system compiles these popular job postings into a short list of options and announces them to the user via voice. For example, the system might ask whether the user is more interested in sorting clerks, porters, or forklift drivers. The user simply selects an option from the list, and the system can infer the user's potential preferences, thus transforming vague needs into clear filtering conditions before continuing the subsequent matching process. This method reduces the difficulty of expression for the user.

[0056] In one embodiment, the completion mode, the method further includes:

[0057] Obtain the filtering criteria;

[0058] Iterate through each dimension in the preset set of filtering dimensions, match the filtering conditions with the dimensions, and if there is no information corresponding to the dimension in the filtering conditions, mark the dimension as a missing dimension.

[0059] Based on the preset dimension priority order, select the dimension with the highest current priority from the marked missing dimensions as the target dimension;

[0060] Analyze the type of the target dimension, and according to the type of the target dimension, call the corresponding question template from the preset question template library. The question template library includes job question templates, location question templates, salary question templates, and time question templates.

[0061] The invoked question template is converted into a voice question command, and the voice question command is sent to the user terminal.

[0062] As mentioned above, the first step uses the filtering criteria extracted from the voice commands as input for the current processing. These criteria may be initially stated by the user or added after several rounds of question-and-answer sessions. The second step pre-sets four core dimensions: job type, location, salary, and time. The system checks each of these four dimensions to see if the existing filtering criteria contain information corresponding to that dimension. For example, if the user says they are looking for welding work, the job type dimension has information, but the location, salary, and time dimensions do not, so these three dimensions are marked as missing dimensions. This transforms the vague "what is missing" into a clear list, allowing the system to clearly know what information it needs to obtain from the user. This ensures comprehensive information collection and avoids overlooking any important dimensions. In the third step, the importance of the four dimensions is not entirely equal. In blue-collar job-seeking scenarios, location is often the most important factor for users because commuting distance directly affects their willingness to go to work. Salary is the next most important factor, as it is a core benefit. Then comes job type, although the system usually only enters the completion mode after the user has already stated their job type. Finally, time, such as the available start date, is important. The system prioritizes the missing dimensions and selects the highest priority one as the target dimension to ask in this round. For example, if location is missing, it prioritizes location; if location is available but salary is missing, it asks about salary. The principle behind this step is to focus on only the most important missing information each time, avoiding overwhelming the user with multiple questions at once. The effect is a more natural interaction, with the user only needing to answer a simple question each time. In the fourth step, since the questioning methods differ for different dimensions—asking about location should be "Which region do you hope to work in?", asking about salary should be "What is your approximate salary expectation?", and asking about start date should be "When can you start work?"—the system pre-prepares standard question templates for each dimension, forming a template library. Once the target dimension is determined, the system directly retrieves the corresponding template from the library. Decoupling the question content from the dimension type allows the system to flexibly generate questions that conform to natural language habits without needing to dynamically construct sentences each time. The effect is that the question statements are standardized, clear, and easy for users to understand. The fifth step converts the template into a voice command and sends it. This involves synthesizing the text-based question template into speech, which is then played back through the user's device. This step uses conventional speech synthesis technology, but its effect is to allow users to interact solely through hearing without looking at the screen. This is very practical for blue-collar users who use the technology outdoors or in scenarios where their hands are not readily available.

[0063] In one embodiment, the steps of obtaining time-change attribute information and user information for each candidate position, and generating time suggestion information for each candidate position based on the time-change attribute information and the user information, include:

[0064] The posting time, remaining number of positions, and employer-set deadline for each candidate position are obtained as the time change attribute information.

[0065] The system obtains the available start date information provided by the user during the voice Q&A interaction, extracts time urgency keywords from the voice query command, and combines the available start date information and time urgency keywords to determine the user information.

[0066] Analyze the difference between the job posting time and the current time, and calculate the posting duration for each candidate position based on the difference between the job posting time and the current time;

[0067] The posting duration is compared with a preset posting duration threshold to determine the recruitment timeframe for each candidate position.

[0068] The available start date information is compared with the recruitment deadline to determine whether the user can complete the onboarding process before the recruitment deadline.

[0069] The time urgency keywords are matched with the remaining positions available to determine whether the user's time urgency matches the remaining positions available.

[0070] Based on the comparison results of the recruitment timeframe, the available start date and the recruitment deadline, and the matching results of the time urgency and the remaining number of positions, time-recommended information for each candidate position is generated.

[0071] As mentioned above, the first step is to obtain the time-related attribute information of the job postings. This involves collecting three types of data for each candidate job: the job posting time, the number of remaining positions indicated by the employer, and the employer's set deadline. The posting time calculates how long the job has been listed, the number of remaining positions reflects the current level of competition, and the deadline is the hard threshold for whether a user can meet the requirements. These three types of data together constitute a complete profile of the job in terms of time. The second step is to obtain the user's time-related information, extracting two aspects from the previous voice Q&A. First, the user's explicitly stated start date, such as answering that they can start work next Monday or can start anytime. Second, keywords indicating time urgency identified from the user's voice commands, such as expressions like "as soon as possible," "as quickly as possible," or "can start today." Combining these two pieces of information reflects the user's actual time constraints and subjective sense of urgency. The third step is to calculate the posting duration of the job. This is done by subtracting the posting time from the current moment, resulting in a time difference, i.e., how long the job has been posted. For example, if a job was posted at 10:00 AM yesterday and it is now 10:00 AM today, then the posting duration is 24 hours. This value directly relates to determining the job's timeliness. The fourth step is to determine the recruitment timeline. The calculated posting duration is compared to a preset threshold to determine whether the position is currently in a high-demand, saturated, or nearing completion phase. For example, 24 hours might be considered a high-demand period, 24 to 48 hours a saturated period, and over 48 hours a nearing completion phase. Different phases represent different levels of urgency. Positions in the high-demand period can be scheduled more readily, while those in the nearing completion phase require immediate action. A quantified timeframe is used to characterize the job's lifecycle. The fifth step is to compare the available start date with the recruitment deadline. This determines whether the user can start before the employer's deadline. For example, if a user says they can start in three days, but the deadline is tomorrow, they cannot start on time. This comparison is a hard constraint; if it's not met, the job's value to the user is significantly reduced. The sixth step is to match the user's expressed urgency with the remaining openings. If the user says "as soon as possible" and there's only one opening left, it's considered a good match, meaning the user's urgency aligns with the job's demand, and the user is advised to act immediately. If the user is not in a hurry and there are plenty of openings, or if the user is in a hurry but there are also many openings, the suitability will differ, and the corresponding suggestions will also differ. Step seven integrates the recruitment timeline, the comparison results of start dates, and the matching results between urgency and available openings to generate the final time recommendation. For example, if a position is in its final stages, the user can start before the deadline, and the user has a high urgency but limited openings, the system will recommend going immediately. Conversely, if the user cannot start before the deadline, regardless of the other two dimensions, the system will directly give a time mismatch suggestion. Through this multi-dimensional integration, each suggestion output considers both the actual status of the position and the user's personal circumstances.

[0072] For example, in actual voice conversations during blue-collar job interviews, users express their available start dates in far more complex ways than an idealized standard format. Users rarely state clearly, "I can start next Monday," but rather use natural expressions that incorporate conditions, ranges, vague terms, and even dynamic processes. Here are some typical examples: First, conditional expressions. The user says, "If accommodation can be arranged, I can go tomorrow." In this statement, "I can go tomorrow" is predicated on the availability of accommodation. If the system ignores the condition and directly extracts tomorrow as the available start date, then if the position doesn't provide accommodation, the user won't actually go, rendering the system's time comparison meaningless. Second, range-based expressions with exceptions. The user says, "Anything is fine this week, but not Saturday." This is a complex structure with excluded days within the time range. If the system simply interprets it as Monday to Sunday of this week, it will incorrectly assume Saturday is also acceptable, leading to discrepancies when comparing it with the job deadline. Third, vague expressions. The user says, "About two or three days later," "as soon as possible," or "in a while." These expressions lack specific dates, offering only semantically vague estimates. If the system forcibly maps to a specific date, for example, interpreting "approximately two or three days later" as "two days later," while the user might actually be able to start work in three days, then when the job deadline is two days later, the system will misjudge that the user has enough time, when in reality they won't. Fourthly, there are dynamic expressions. A user might say, "I'm available anytime, but I need to resign from my current job first." This implies a dynamic process; the start date depends on when the resignation procedures are completed, not a fixed point in time. Current voice recruitment systems typically assume users will directly provide precise dates or simple time periods, lacking the ability to parse such complex expressions. When encountering this type of input, the system either cannot process it, causing the interaction to interrupt, or it uses simplistic rules leading to incorrect parsing, making subsequent comparisons between the start date and the job deadline unreliable, ultimately generating misleading time suggestions.

[0073] To address this issue, this embodiment first determines whether the user's expression is conditional. If so, the condition is recorded separately as a prerequisite for employment, and basic time information is extracted. This approach allows the conditional content to be used for matching and filtering when recommending jobs. For example, if the system detects that a user requires accommodation, it prioritizes recommending jobs that provide accommodation and proactively informs the user that this condition has been met, thus enhancing the relevance and persuasiveness of the recommendations. For non-conditional information, further processing is performed based on the type of time expression. Range-based expressions are parsed into earliest and latest boundary values, forming a time window; fuzzy expressions are converted into a reasonable time range using preset mapping rules, such as approximately two or three days later corresponding to at least two days later and at most three days later; dynamic expressions identify the preceding actions and are also converted into a time range. Finally, a standardized earliest and latest start date is output. When this time window is compared with the job's deadline, if the user's time window overlaps with the deadline, the user is considered eligible to start work. If the window ends earlier than the deadline, the user is eligible to start work. If the window begins later than the deadline, the user is ineligible to start work. This time window-based comparison method is more robust and accurate than judging based on a single point in time.

[0074] Specifically, obtaining the available start date information provided by the user during the voice Q&A interaction includes:

[0075] Semantic analysis is performed on the available start time information provided by the user terminal to determine whether the available start time information is conditional information;

[0076] If the available start time information is conditional information, then extract the conditional content and basic time information from the available start time information, record the conditional content as a prerequisite for start-up, and use the basic time information as the available start time.

[0077] If the available start time information is not conditional information, then the available start time is extracted according to the time expression type of the available start time information;

[0078] The extracted start dates are analyzed into the earliest and latest start dates, which are then compared with the recruitment deadline.

[0079] As mentioned above, the available start time given by users in actual conversations is often not a simple date or time period, but rather comes with certain preconditions. For example, a user might say, "If accommodation can be arranged, I can go tomorrow," or "As long as the salary is paid daily, I can start today." The system first performs semantic analysis on the user's original statement, identifying whether it contains conditional conjunctions such as "if," "as long as," or "if only," and whether it contains a conclusive time statement. The principle behind this step is to extract the logical structure from natural language, distinguishing between preconditions and genuine time commitments. Complex time expressions are broken down into separately processable components. When the judgment result is conditional information, the user's statement is further split into two parts. One part is the conditional content, such as accommodation being arranged or the salary being paid daily; this part is recorded separately as the preconditions for starting work. The other part is the basic time information, such as "I can go tomorrow" or "I can start today," which is extracted as the available start time. The advantage of this approach is that the system uses the basic time for subsequent time comparisons, while the conditional content can be used for secondary screening when recommending positions. For example, if a job offers accommodation, the system can proactively inform the user that the job meets their prerequisites, thereby improving the relevance and persuasiveness of the recommendation. If the user's expression lacks conditional conjunctions, the system parses it according to conventional time expression types. For example, if the user says "next Monday," "this week is fine," or "about two or three days later," these are respectively precise, range-based, and fuzzy. The system calls the corresponding parsing rules to extract specific time information. This step is conventional natural language time parsing, based on a pre-set time lexicon and mapping rules. Regardless of whether the user provides a precise date, time period, or fuzzy estimate, it must ultimately be converted into two standardized values: the earliest and latest start dates. For example, if the user says "next Monday," the earliest and latest start dates are both next Monday; if the user says "this week is fine," the earliest is this Monday and the latest is this Sunday; if the user says "about two or three days later," the earliest is two days later and the latest is three days later. This standardized format allows the system to consistently compare with the job's recruitment deadline to determine if the user can start before the deadline. This unifies the ever-changing natural language time expressions into a computable data structure.

[0080] In one embodiment, the step of obtaining the remaining positions for each candidate position includes:

[0081] Get the total number of times users contacted each candidate position through the system within a preset time period, and the number of times users reported that the position was full;

[0082] Based on the total number of contacts initiated and the number of times users reported that the positions were full, the full feedback ratio is calculated, which is the number of times users reported that the positions were full divided by the total number of contacts initiated.

[0083] If the full feedback ratio is higher than the preset ratio threshold, it is determined that the position has been consumed through other channels, and the position's quota status is "shortage".

[0084] If the full feedback ratio is lower than or equal to the preset threshold, the initial remaining number of positions filled by the employer will be used as the remaining number of positions.

[0085] As mentioned above, this embodiment addresses the problem of inaccurate remaining positions for blue-collar jobs. The remaining positions employers fill in when posting job openings often don't match the actual number. This is either because other channels, such as offline recruitment and telephone applications, are simultaneously consuming these positions, causing the data to become outdated, or because employers deliberately inflate the numbers to attract more job seekers. This embodiment doesn't rely on data provided by employers; instead, it infers the true status of the available positions by analyzing the actual interactions between job seekers and the job postings. The first step involves counting how many users have contacted the job posting through the system recently, such as calling the phone number provided or submitting an application. Simultaneously, the system also tracks the feedback from these users after contacting the job. The most crucial type of feedback is whether the position is full or no longer available. This data represents genuine traces left by users during their actual job search process, which employers cannot interfere with or tamper with. The second step calculates the percentage of users who reported the position was full. Dividing the number of times the position was full by the total number of contacts yields a percentage between 0 and 1. This percentage reflects the probability that a user was rejected after contacting the job posting. For example, if 20 people contact the employer for a certain position through the system, and 6 of them report that the position is already filled, then the "filled" feedback rate is 30%. The higher this rate, the fewer actual vacancies there are for that position. The third step determines the vacancy status based on the rate by comparing the calculated "filled" feedback rate with a preset threshold. This threshold can be set according to the actual situation, such as 20% or 30%. If the rate is higher than the threshold, it means that a large number of users contacted the employer and were told that the position was full. This means that the vacancies for that position have been largely consumed through other channels, and the static remaining vacancies in the system are no longer reliable. Therefore, the vacancy status is determined to be tight. If the rate is lower than or equal to the threshold, it means that most users did not encounter a full position after contacting the employer, and the initial remaining vacancies filled in by the employer are relatively reliable. The system continues to use the data filled in by the employer. The fourth step outputs the vacancy status. Positions determined to be in tight vacancies will be marked during the subsequent time suggestion generation process. When generating time suggestions, users will be given priority to be notified that the vacancies for that position may be full, and it is recommended to verify as soon as possible to avoid users making a wasted trip.

[0086] It's worth noting that in the blue-collar labor market, the number of remaining job openings is a highly volatile and easily distorted data point. Employers often exhibit two serious problems when posting job openings. The first is the dynamic decay problem. Employers may recruit for the same position simultaneously through multiple channels, such as physical stores, direct phone recruitment, and agency referrals. When the system shows five openings, other channels may have filled their positions within minutes, but the system remains unaware. Users make their judgments based on the system's displayed ample availability, only to find the position is already full after contacting the employer, wasting time and effort. The second is the problem of false labeling. Some employers, in order to attract as many job seekers as possible, deliberately exaggerate the number of remaining openings, listing ten when there is only one vacancy, or keeping a posting for a position that has already stopped recruiting. This data misrepresentation renders the remaining openings in the system completely meaningless, misleading users with false information, frequently encountering full positions, and gradually eroding their trust in the system. Currently, voice recruitment systems typically directly accept the remaining openings filled by employers without any verification mechanism. When encountering dynamic decay or false labeling, the system still operates according to false data, pushing positions that appear to have ample openings but are actually full to users, or misjudging positions that should be marked as scarce as having ample openings.

[0087] This embodiment proposes a solution. It doesn't rely on data actively provided by employers, but rather infers the true status of available positions by analyzing the real interactions between job seekers and job postings. When a large number of users contact the system and report that the position is full, even if the employer has entered a number of available positions, it can accurately determine that the position actually has very few openings. This mechanism, based on group behavior verification, essentially uses the experiences of real job seekers as sensors to dynamically detect the actual saturation of positions. The user feedback of "full" is itself the most direct evidence; by statistically analyzing the proportion of these feedbacks, the system can effectively identify which positions have already been exhausted through other channels and which positions have more reliable data. This embodiment does not require access to any third-party data or real-time updates from employers; it only utilizes user feedback data naturally accumulated through its own operations to achieve continuous verification and dynamic correction of remaining positions. This allows this recommendation method to maintain high accuracy even in scenarios involving inflated data and multi-channel recruitment.

[0088] refer to Figure 5For example, a logistics company posted a mover position through this system, specifying 10 remaining openings. However, the company also posted the same job information through offline stores, internal employee referrals, and other online platforms. Due to the higher efficiency of offline recruitment, eight people successfully applied and joined the company within half a day, but the employer did not update the remaining openings in the system, which still showed 10 openings. At this point, 20 users viewed the job through the system's voice interaction and called the employer's provided contact number. After each call, the system proactively inquired about the contact result, such as asking via voice whether the user successfully contacted the employer or if the position was still open. Fifteen users selected "position full" or "no longer hiring" in their feedback. The system recorded these 15 responses as the number of times the position was considered full. This method calculates the full-position feedback rate by obtaining the total number of times users contacted the position through the system within a preset time period (e.g., the last 6 hours), which is 20, and the number of times users reported the position was full (15 times). The percentage of full-position feedback is calculated as 15 divided by 20, yielding 75%. The system's preset threshold is 30%, and 75% is significantly higher than this threshold. Therefore, the system determines that the position was filled through other channels, indicating a shortage of openings, and no longer accepts the employer's data of 10 openings. Subsequently, when another user searches for a mover position via voice, the system adds a prompt to the time-based suggestion information generated for that position, such as announcing via voice that the position may have already been filled through other channels and suggesting a phone call to confirm. The user, after hearing the prompt, decides to call and confirms that the position is indeed full, thus avoiding a wasted trip. In contrast, if the feedback rate for another position being full is only 10%, below the 30% threshold, the system would determine that the employer's initial remaining openings are reliable and continue to use that data for subsequent time-based suggestion generation and ranking recommendations.

[0089] Reference Figure 3 This application also provides an AI-powered voice-interactive job screening and recommendation system for blue-collar job seekers, comprising:

[0090] Module 1 is used to acquire voice query commands input by the user.

[0091] The recognition and extraction module 2 is used to perform voice and semantic recognition on the voice query command, extract filtering conditions, and evaluate the clarity of the user's needs based on the filtering conditions;

[0092] Analysis module 3 is used to analyze whether the required clarity meets preset matching conditions based on the evaluation results;

[0093] The judgment module 4 is used to determine the interaction strategy by combining the voice query command and the clarity of the requirement if the required clarity does not meet the job matching conditions, and continue to perform voice Q&A to interact with the user terminal to obtain supplementary information until the preset matching conditions are met.

[0094] Module 5 is used to determine the candidate job set based on the supplementary information and screening conditions;

[0095] The first generation module 6 is used to obtain time change attribute information and user information for each candidate position, generate time suggestion information for each candidate position based on the time change attribute information and the user information, and sort the candidate position set according to the time suggestion information.

[0096] The second generation module 7 is used to generate a job recommendation list based on the sorted candidate job set and output it to the user terminal.

[0097] As described above, it is understood that each component of the AI ​​voice-interactive job screening and recommendation system for blue-collar job seekers proposed in this application can achieve the function of any of the AI ​​voice-interactive job screening and recommendation methods for blue-collar job seekers described above, and the specific structure will not be repeated here.

[0098] Reference Figure 4 This application also provides a computer device, which may be a server, and its internal structure may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores monitoring data and other data. The network interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements an AI-powered voice-interactive job screening and recommendation method for blue-collar job seekers.

[0099] The processor described above executes the AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers, including: acquiring a voice query command input by a user; performing voice and semantic recognition on the voice query command, extracting filtering conditions, and evaluating the clarity of the user's needs based on the filtering conditions; analyzing whether the clarity of needs meets preset matching conditions based on the evaluation results; if the clarity of needs does not meet the job matching conditions, determining an interaction strategy by combining the voice query command and the clarity of needs, and continuing to interact with the user through voice question-and-answer to obtain supplementary information until the preset matching conditions are met; determining a candidate job set based on the supplementary information and filtering conditions; acquiring time-change attribute information and user information for each candidate job, generating time-suggestion information for each candidate job based on the time-change attribute information and the user information, and sorting the candidate job set based on the time-suggestion information; and generating a job recommendation list based on the sorted candidate job set and outputting it to the user.

[0100] One embodiment of this application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements an AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers, comprising the following steps: acquiring a voice query command input by a user; performing voice and semantic recognition on the voice query command, extracting filtering conditions, and evaluating the clarity of the user's needs based on the filtering conditions; analyzing whether the clarity of needs meets preset matching conditions based on the evaluation results; if the clarity of needs does not meet the job matching conditions, determining an interaction strategy by combining the voice query command and the clarity of needs, and continuing to interact with the user through voice question-and-answer to obtain supplementary information until the preset matching conditions are met; determining a candidate job set based on the supplementary information and filtering conditions; acquiring time-change attribute information and user information for each candidate job, generating time-recommendation information for each candidate job based on the time-change attribute information and the user information, and sorting the candidate job set based on the time-recommendation information; and generating a job recommendation list based on the sorted candidate job set and outputting it to the user.

[0101] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0102] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0103] The above description is only a preferred embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural changes made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. An AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers, characterized in that: The method includes: Obtain the voice query command input by the user; The voice query command is subjected to voice and semantic recognition to extract filtering conditions, and the clarity of the user's needs is evaluated based on the filtering conditions. Based on the evaluation results, analyze whether the required clarity meets the preset matching conditions; If the required clarity does not meet the job matching conditions, the interaction strategy is determined by combining the voice query command and the required clarity, and the voice Q&A continues to interact with the user terminal to obtain supplementary information until the preset matching conditions are met. Based on the supplementary information and screening criteria, a candidate job set is determined; Obtain time change attribute information and user information for each candidate position; generate time suggestion information for each candidate position based on the time change attribute information and the user information; and sort the candidate position set according to the time suggestion information. Based on the sorted set of candidate positions, a recommended list of positions is generated and output to the user terminal.

2. The AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to claim 1, characterized in that, The steps of performing speech and semantic recognition on the voice query command, extracting filtering conditions, and evaluating the clarity of the user's needs based on the filtering conditions include: The voice query command is subjected to speech recognition and semantic parsing to extract the filtering conditions from the voice query command; The filtering conditions are matched with a preset set of filtering dimensions to determine the number of matched filtering dimensions. The preset set of filtering dimensions includes job type, location, salary, and time. The dimension coverage ratio is calculated based on the number of matched filtering dimensions and the total number of dimensions in the preset filtering dimension set. If the dimension coverage ratio is greater than or equal to a preset ratio threshold, then the requirement clarity is determined to be high clarity. If the dimension coverage ratio is less than the preset ratio threshold, then determine whether the filtering conditions contain fuzzy words; If the filtering criteria contain vague words, then the required specificity is determined to be low. If the filtering criteria do not contain vague words, then the required clarity is determined to be medium clarity.

3. The AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to claim 2, characterized in that, The step of determining the interaction strategy by combining the voice query command and the clarity of the demand includes: Obtain the required clarity; If the required clarity is medium clarity, then the interaction strategy is determined to be the completion mode. The completion mode is used to compare the extracted filtering conditions with a preset set of filtering dimensions, determine the dimensions in the preset set of filtering dimensions that are not covered by the filtering conditions as missing dimensions, and generate a question instruction for the missing dimensions according to the preset dimension priority order. If the required clarity is low, the interaction strategy is determined to be exploration mode. The exploration mode is used to obtain the current location information of the user terminal, recall popular job information based on the current location information, generate an option list containing the popular job information, and send a selection request to the user terminal.

4. The AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to claim 3, characterized in that, The completion mode and method further include: Obtain the filtering criteria; Iterate through each dimension in the preset set of filtering dimensions, match the filtering conditions with the dimensions, and if there is no information corresponding to the dimension in the filtering conditions, mark the dimension as a missing dimension. Based on the preset dimension priority order, select the dimension with the highest current priority from the marked missing dimensions as the target dimension; Analyze the type of the target dimension, and according to the type of the target dimension, call the corresponding question template from the preset question template library. The question template library includes job question templates, location question templates, salary question templates, and time question templates. The invoked question template is converted into a voice question command, and the voice question command is sent to the user terminal.

5. The AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to claim 1, characterized in that, The steps of obtaining time-change attribute information and user information for each candidate position, and generating time-recommendation information for each candidate position based on the time-change attribute information and the user information, include: The posting time, remaining number of positions, and employer-set deadline for each candidate position are obtained as the time change attribute information. The system obtains the available start date information provided by the user during the voice Q&A interaction, extracts time urgency keywords from the voice query command, and combines the available start date information and time urgency keywords to determine the user information. Analyze the difference between the job posting time and the current time, and calculate the posting duration for each candidate position based on the difference between the job posting time and the current time; The posting duration is compared with a preset posting duration threshold to determine the recruitment timeframe for each candidate position. The available start date information is compared with the recruitment deadline to determine whether the user can complete the onboarding process before the recruitment deadline. The time urgency keywords are matched with the remaining positions available to determine whether the user's time urgency matches the remaining positions available. Based on the comparison results of the recruitment timeframe, the available start date and the recruitment deadline, and the matching results of the time urgency and the remaining number of positions, time-recommended information for each candidate position is generated.

6. The AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to claim 5, characterized in that, The acquisition of the available start date information provided by the user during the voice Q&A interaction specifically includes: Semantic analysis is performed on the available start time information provided by the user terminal to determine whether the available start time information is conditional information; If the available start time information is conditional information, then extract the conditional content and basic time information from the available start time information, record the conditional content as a prerequisite for start-up, and use the basic time information as the available start time. If the available start time information is not conditional information, then the available start time is extracted according to the time expression type of the available start time information; The extracted start dates are analyzed into the earliest and latest start dates, which are then compared with the recruitment deadline.

7. The AI-powered voice-interactive job selection and recommendation method for blue-collar job seekers according to claim 5, characterized in that, The steps for obtaining the remaining positions for each candidate position include: Get the total number of times users contacted each candidate position through the system within a preset time period, and the number of times users reported that the position was full; Based on the total number of contacts initiated and the number of times users reported that the positions were full, the full feedback ratio is calculated, which is the number of times users reported that the positions were full divided by the total number of contacts initiated. If the full feedback ratio is higher than the preset ratio threshold, it is determined that the position has been consumed through other channels, and the position's quota status is "shortage". If the full feedback ratio is lower than or equal to the preset threshold, the initial remaining number of positions filled by the employer will be used as the remaining number of positions.

8. An AI-powered voice-interactive job screening and recommendation system for blue-collar job seekers, characterized in that: include: The acquisition module is used to acquire the voice query commands input by the user. The recognition and extraction module is used to perform voice and semantic recognition on the voice query command, extract filtering conditions, and evaluate the clarity of the user's needs based on the filtering conditions; The analysis module is used to analyze whether the required clarity meets preset matching conditions based on the evaluation results; The judgment module is used to determine the interaction strategy by combining the voice query command and the clarity of the requirement if the required clarity does not meet the job matching conditions, and continue to perform voice Q&A to interact with the user terminal to obtain supplementary information until the preset matching conditions are met. The determination module is used to determine the candidate job set based on the supplementary information and screening criteria; The first generation module is used to obtain time change attribute information and user information for each candidate position, generate time suggestion information for each candidate position based on the time change attribute information and the user information, and sort the candidate position set according to the time suggestion information. The second generation module is used to generate a job recommendation list based on the sorted set of candidate jobs and output it to the user terminal.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images