Human resource multi-specialty intelligent question and answer method, system and device and medium

Abstract

The invention provides a human resource multi-specialty intelligent question and answer method, system and device and a medium, and the method comprises the steps: collecting human resource knowledge data, carrying out the data cleaning and disassembly and data format conversion of the human resource knowledge data, and importing the data into a knowledge base; the method comprises the following steps: collecting human resource data, preprocessing the human resource data, performing entity recognition through a natural speech processing model, and constructing an entity library related to human resources; based on the entity library, carrying out data slicing processing, and according to the type of the sliced data, storing the sliced data into a database in a structured storage and vectorized storage mode; constructing a training data set of the model, training a large language model by using the training data set, and performing model optimization processing to generate a knowledge base language model of the human resource multi-specialty intelligent question-answering system; and obtaining human resource consultation questions of the employees, performing question analysis and retrieval through the knowledge base language model, generating matched answers, and outputting the answers.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and more specifically to a method, system, device and medium for intelligent question answering of multiple disciplines in human resources. Background Art

[0002] With the rapid development of informatization and digitalization of enterprise human resource management and the increasing demand for efficient and convenient services, online consultation systems have received widespread attention as an innovative customer service tool. The traditional consultation model for human resource issues generally adopts the offline consultation method of employees. This model lacks integrity and coordination, and the daily workload of professional managers of human resources is heavy, resulting in slow information exchange and inability to respond to employee consultation needs in a timely manner.

[0003] Specifically, the current channels for employees to obtain information on professional policies and systems such as salary, benefits, and social security are limited, and can only be obtained through manual consultation with professional managers, which is inefficient and cannot respond in a timely manner. Daily policy inquiries are mainly managed offline manually, policy responses are relatively scattered, information acquisition and analysis are slow, and the degree of intelligence is not high. Efficient management and timely transmission of information are impossible, which affects the quality and efficiency of management work. It is necessary to further expand the human resources professional field channels for online intelligent inquiry digitization. Once the number of inquiries is large, the feedback speed is slow, and the work efficiency is low, it not only increases the workload of grassroots employees, but also reduces the employee service experience.

[0004] In view of the above situation, although there are some intelligent question-and-answer platforms on the market, the functions of such platforms are relatively simple and not flexible enough to meet the multi-professional needs of human resources, and the effect of use is not very obvious. In addition, in view of the uniqueness of multi-professional human resources, there are few packaged programs in such intelligent question-and-answer platforms that promote management advancement, and most of them need to be redeveloped, which is time-consuming and ineffective. Summary of the invention

[0005] In view of the above problems, the purpose of the present invention is to provide a multi-professional intelligent question-and-answer method, system, device and medium for human resources, which utilizes artificial intelligence technology, natural language processing, and vector space model technology, and provides online self-service, question-and-answer services and other services for human resources-related issues by simulating human dialogue and communication through robots, thereby effectively improving employees' business operation experience.

[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions: In a first aspect, the present invention discloses a multi-professional intelligent question-answering method for human resources, comprising: Collect human resource knowledge data, clean and disassemble it, convert its data format, and import it into the knowledge base; Collect human resources data, perform entity recognition through natural speech processing models after preprocessing, and build a human resources-related entity library; Based on the entity library, data slicing processing is performed, and according to the type of slice data, the slice data is stored in a preset database through structured storage and vectorized storage; Based on the knowledge base and preset database, a training data set for the model is constructed, and the large language model is trained using the training data set. After model tuning, a knowledge base language model for the multi-professional intelligent question-answering system for human resources is generated; Obtain employees' human resources consulting questions, parse and retrieve questions through the knowledge base language model of the human resources multi-professional intelligent question-and-answer system, generate matching answers, and output the answers.

[0007] Furthermore, the collecting of human resource knowledge data, data cleaning and decomposition and data format conversion of the human resource knowledge data, and importing the data into the knowledge base include: Connect data with human resource management systems and human resource integration platforms through APIs to obtain basic employee information, attendance records, and collect human resource documents, training materials, and company rules and regulations; Use Python and Spark tools to clean human resource knowledge data, remove data noise from human resource knowledge data, fill missing values, and correct format errors; Use Pandas and OpenRefine to convert human resource knowledge data into a unified format; Classify human resource knowledge data according to profession, and import them into the knowledge base according to different categories.

[0008] Furthermore, the human resources data is collected, pre-processed, and then entity recognition is performed through a natural speech processing model, and a human resources-related entity library is constructed, including: Collect human resources data, and perform data cleaning, deduplication, word segmentation, and part-of-speech tagging in sequence; Perform entity recognition on the processed human resource data through a natural speech processing model, identify the relationship between entities in the text, and generate recognition results; Build a human resources-related entity library to store recognition results.

[0009] Further, the data slicing process is performed based on the entity library, and according to the type of the slice data, the slice data is stored in a preset database through structured storage and vectorized storage, including: Slice the data stored in the entity library according to different professional fields, and process the data slices in parallel through a distributed computing framework to perform classified storage of the sliced ​​data; When the slice data is relational data, a relational database is used for storage; When the slice data is non-relational data, use the MongoDB database for storage; Create query indexes for relational databases and MongoDB databases by setting query conditions, and configure query optimizers for relational databases; When the slice data is text data, the text data is converted into vector data and stored in a vector database; The HNSW algorithm is used to build an index for the vector database and the ANN vector retrieval algorithm. Furthermore, the method of constructing a training data set for the model based on the knowledge base and the preset database, and using the training data set to train the large language model includes: Based on the knowledge base and preset database, collect human resources documents, training materials and company rules and regulations as training data, clean, remove duplicates, segment and tag the training data, integrate professional dictionaries in the field of human resources, and generate training data sets; According to the multi-professional application scenarios and needs of human resources, select the corresponding large language model for model training. Furthermore, the knowledge base language model of the human resources multi-professional intelligent question-answering system is generated after the model tuning process, including: Use grid search, random search or Bayesian optimization methods to search for hyperparameters, and adjust the hyperparameters of large language models for different application scenarios in multiple human resources disciplines to find the best model parameters; Integrate the large language model of each application scenario to generate the knowledge base language model of the multi-professional intelligent question-answering system for human resources; Evaluation indicators for the human resources question-answering task are set for the knowledge base language model, the model performance is evaluated using the evaluation indicators, and the model is iterated and optimized based on the evaluation results.

[0010] Furthermore, the human resources consulting questions of employees are obtained, and the questions are parsed and retrieved through the knowledge base language model of the human resources multi-professional intelligent question-answering system, matching answers are generated, and the answers are output, including: Obtain employees' human resources consulting questions and process them through NLP and CV technology; Perform word segmentation, part-of-speech tagging, and syntactic analysis on the human resources consulting questions, and perform intent recognition optimization for the preset question types in multiple professional fields of human resources to obtain the intent of the questions; According to the question intent, use the inverted index and full-text search engine to search the entire knowledge base, and combine it with the vector database to perform vector search and integrate the retrieved information; Generate answers using preset response templates based on the retrieved information; On the preset interactive interface, the answer is output through multimodal output.

[0011] In a second aspect, the present invention further discloses a human resources multi-professional intelligent question-answering system, comprising: The data collection module is used to collect human resource knowledge data, clean and disassemble the human resource knowledge data, convert the data format, and import it into the knowledge base; The data extraction module is used to collect human resources data, perform entity recognition through the natural speech processing model after preprocessing, and build a human resources related entity library; A data processing module is used to perform data slicing processing based on the entity library, and store the slice data in a preset database through structured storage and vectorized storage according to the type of slice data; The model building module is used to build a training data set for the model based on the knowledge base and the preset database, use the training data set to train the large language model, and generate a knowledge base language model for the human resources multi-professional intelligent question-answering system after model tuning; The question-and-answer service module is used to obtain employees' human resources consulting questions, parse and retrieve questions through the knowledge base language model of the human resources multi-professional intelligent question-and-answer system, generate matching answers, and output the answers.

[0012] In a third aspect, the present invention further discloses a multi-professional intelligent question-answering device for human resources, comprising: A memory device for storing a multi-professional intelligent question-answering program for human resources; A processor is used to implement the steps of the human resources multi-professional intelligent question and answer method as described in any of the above items when executing the human resources multi-professional intelligent question and answer program.

[0013] In a fourth aspect, the present invention further discloses a readable storage medium, on which a human resources multi-professional intelligent question and answer program is stored. When the human resources multi-professional intelligent question and answer program is executed by a processor, the steps of the human resources multi-professional intelligent question and answer method as described in any one of the above items are implemented.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention utilizes artificial intelligence technology, natural language processing, and vector space model technology, and uses robots to simulate human dialogue and communication to explore new business question-and-answer services for human resources. It uses instruction sets and natural semantic analysis (NLP) and other technologies to provide online self-service, question-and-answer services, etc., thereby realizing online human resources consulting services and improving employees' business operation experience.

[0015] 2. The present invention uses natural language processing (NLP) technology to enable machines to understand and generate human language. NLP technology includes word segmentation, part-of-speech tagging, syntactic analysis, semantic understanding and other aspects. These technologies work together on the questions input by users to convert them into a structured form that can be understood by computers, laying the foundation for subsequent knowledge retrieval and answer generation.

[0016] 3. The present invention converts human resource data into numerical vectors, text vectors, image vectors and other forms through vectorization technology and stores them in the database. The vector space model is used to adopt a specific storage structure and index algorithm to efficiently store and query these data, thus realizing efficient similarity search.

[0017] 4. The present invention uses machine learning (ML) and deep learning technology to learn and summarize knowledge from a large amount of data, and realizes the self-evolution and optimization of the model. By training the model, it can more accurately understand the user's intention and provide more precise answers. In addition, deep learning technology also enables the present invention to handle more complex language phenomena and semantic relationships, and improve the accuracy and intelligence level of question and answer.

[0018] 5. The present invention adopts modal interaction technology and supports multiple interaction modes such as text and image. These multimodal interaction technologies enable users to interact with the system in a more natural and convenient way, improving user experience and satisfaction.

[0019] It can be seen that compared with the prior art, the present invention has outstanding substantive features and significant progress, and the beneficial effects of its implementation are also obvious. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention 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 embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0021] Figure 1 It is a method flow chart of a specific implementation mode of the present invention.

[0022] Figure 2It is a system structure diagram of a specific implementation mode of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the scheme of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] See also Figure 1 As shown, this embodiment provides a human resources multi-professional intelligent question answering method, including the following steps: S1: Collect human resource knowledge data, clean and decompose the data, convert the data format, and import it into the knowledge base.

[0025] In the specific implementation method, firstly, data connection is carried out with human resource management system (ERP), human resource integration platform and other systems through API to ensure that key data such as basic information of employees and attendance records can be obtained; policy documents, training materials and company rules and regulations related to human resource majors are collected to ensure the pertinence and effectiveness of the data. Thus, the collection of human resource knowledge data is completed.

[0026] Then, use Python, Spark and other tools to clean the collected human resources knowledge data, including removing noise, filling missing values, correcting format errors, etc. In the data cleaning process, pay special attention to the processing of sensitive information. For example, when processing employee personal privacy data, ID card numbers, contact information, etc. must be ensured to comply with data protection regulations. Structural decomposition of human resources knowledge data, complex human resources policy documents are decomposed into structured data, such as policy terms, effective dates, applicable objects, etc., to facilitate subsequent query and analysis.

[0027] Furthermore, we will unify data standards and develop human resources data standards to ensure that data formats from different sources are unified, such as employee numbers, job titles, etc. We will use tools such as Pandas and OpenRefine to convert data formats, and support multiple data format conversions (such as CSV, JSON, Parquet) to meet standard requirements. Finally, we will merge the data to resolve format differences between data.

[0028] Finally, establish a multi-professional classification system for human resources, such as recruitment, training, performance, compensation, etc., to facilitate organization and retrieval in the knowledge base; create a unified data view, and manually import the resulting graphical interface or script file into the knowledge base; introduce a version control mechanism to record the history of each data import for easy tracing and recovery.

[0029] S2: Collect human resources data, perform entity recognition through natural speech processing models after preprocessing, and build a human resources related entity library.

[0030] In a specific implementation, human resources data is obtained from the original data source, and valuable information is screened and obtained. This process usually includes the following steps: (1) Preprocessing: Preprocessing of human resources data, including data cleaning, deduplication, word segmentation, part-of-speech tagging, etc., helps filter and eliminate data information. Identify and tag professional terms in the field of human resources, such as "salary and benefits" and "performance appraisal", to improve the accuracy of subsequent processing.

[0031] (2) Entity recognition and relationship extraction: Use models such as BERT and Spacy to perform entity recognition and identify the relationships between entities in the text as the basis for building semantic analysis. Build a human resources-related entity library, including employees, positions, departments, etc., and the relationships between them.

[0032] S3: Based on the entity library, data slicing processing is performed, and according to the type of slice data, the slice data is stored in a preset database through structured storage and vectorized storage.

[0033] In a specific implementation, data slicing is a technique for dividing a large data set into smaller, more manageable parts. It can efficiently process and utilize a large amount of raw data. This step first slices the data stored in the entity library according to different professional fields, and processes the data slices in parallel through a distributed computing framework to perform classified storage of the sliced ​​data.

[0034] It should be noted that data slicing is used in the following aspects: (1) Parallel processing: Data is sliced ​​according to different professional fields of human resources (such as recruitment, training, performance, etc.), and the data slices are distributed to multiple processing units. Distributed computing frameworks such as Hadoop and Spark are used for parallel processing, which helps the system realize parallel processing and professional analysis, and improves processing speed and efficiency.

[0035] (2) Balanced processing: Data slicing helps to balance the load between different professional units of human resources. It dynamically adjusts resource allocation according to the load of the processing unit to avoid the situation where some units are overloaded and other units are idle.

[0036] (3) Data recovery and fault tolerance: During the data slicing process, a backup is usually created for each human resources professional slice, and an automatic recovery mechanism is designed to automatically restore from the backup when a slice is damaged or lost.

[0037] When the slice data is relational data, a relational database is used for storage. Specifically, for data with clear relationships, a relational database (such as MySQL, PostgreSQL, etc.) is used for storage. A reasonable ER diagram is designed for human resources data to clarify the relationship between entities such as employees, departments, and positions, support complex queries and transaction processing, and is suitable for storing and managing structured data.

[0038] When the slice data is non-relational data, use the MongoDB database for storage. Specifically, for data that does not require a strict relational model, use a non-relational database (such as MongoDB, etc.) for storage, which has higher flexibility and scalability and is suitable for storing semi-structured or unstructured data.

[0039] Create query indexes for relational databases and MongoDB databases by setting query conditions, and configure query optimizers for relational databases. Specifically, create common query indexes for commonly used query conditions in the human resources field (such as employee name, department name, leave system, title evaluation policy, etc.). Use multi-level index structures such as B-tree and B+ tree to improve query efficiency. Use query optimizers (such as MySQL's EXPLAIN) to analyze and optimize query performance.

[0040] When the slice data is text data, the text data is converted into vector data and stored in a vector database. Specifically, vectorized storage is an effective way to process unstructured data. Converting text data into high-dimensional vector representation is semantically similar. The implementation of vectorized storage depends on vectorized algorithms and vector databases.

[0041] When selecting a vectorization algorithm, the default is to write all types of data into memory first, flush them to disk regularly, use the HNSW algorithm to build indexes and the ANN vector retrieval algorithm, and combine the two algorithms to form a hybrid retrieval to quickly capture the text semantics of the nearest neighbor in the vector space. Select a vectorization algorithm suitable for human resources text data, such as BERT or Word2Vec, to calculate the similarity between policy documents.

[0042] A dedicated vector database (milvus) is used to store and retrieve vector data, which supports efficient vector indexing and query operations and can quickly find the vector data most similar to the query vector.

[0043] Based on vectorized storage, this method designs a matching strategy for human resource policies and achieves efficient semantic retrieval and matching through hybrid retrieval. When a user enters a query, the query content is converted into a vector representation, and the vector database is searched for the most matching content to answer it, ensuring accurate retrieval of policy content related to the user's question.

[0044] S4: Based on the knowledge base and the preset database, a training data set for the model is constructed, and the large language model is trained using the training data set. After model tuning, the knowledge base language model of the multi-professional intelligent question-and-answer system for human resources is generated.

[0045] In a specific implementation, building a knowledge base language model for a multi-professional intelligent question-answering system for human resources is a complex and sophisticated process involving multiple links such as data set selection, model training, and model tuning. Through reasonable data set selection, effective model training, and meticulous model tuning, a high-performance multi-professional intelligent question-answering system for human resources is built to provide users with accurate and fast question-answering services. The specific process of this step is as follows: S401: Dataset selection. Collect the company's internal human resources policy documents, rules and regulations, training materials and human resources professional public data as important sources of training data. Supplement external data sets and combine them with public data sets in the human resources industry to improve the generalization ability of the model.

[0046] S402: Model training. First, the collected data is preprocessed by cleaning, deduplication, word segmentation, part-of-speech tagging, etc., and professional dictionaries in the field of human resources are integrated to improve the accuracy of subsequent processing. Then, according to the specific multi-professional application scenarios and needs of human resources, a suitable language model is selected for training to enhance the semantic features and contextual information of the captured text. Finally, the model is trained using the preprocessed data set, and the model is fine-tuned for specific tasks in the field of human resources (such as policy interpretation, employee consultation responses, etc.), and the model parameters are adjusted to minimize the loss function and improve the accuracy of the model. During the training process, cross-validation and other techniques are used to evaluate the performance of the model and prevent overfitting.

[0047] S403: Model tuning. First, use grid search, random search, Bayesian optimization and other methods to search for hyperparameters. Adjust the model's hyperparameters (such as training frequency, number of iterations, etc.) for different application scenarios in human resources to find the best model configuration. Then integrate multiple models for different application scenarios to improve overall accuracy. Finally, design evaluation indicators for human resources question-and-answer tasks, such as policy interpretation accuracy, consultation response satisfaction, etc., evaluate the model, calculate accuracy, failure rate and other indicators to evaluate the model's performance, and iterate and optimize the model based on the evaluation results to improve its performance.

[0048] S5: Obtain employees’ human resources consulting questions, parse and retrieve the questions through the knowledge base language model of the human resources multi-professional intelligent question-answering system, generate matching answers, and output the answers.

[0049] In a specific implementation manner, the human resources consulting question is firstly processed by word segmentation, part-of-speech tagging, and syntactic analysis, and intent recognition optimization is performed for preset question types in multiple professional fields of human resources to obtain the intent of the question; then, an inverted index and a full-text search engine are used to search the entire knowledge base according to the intent of the question, and a vector search is performed in conjunction with a vector database to integrate the retrieved information; at this time, an answer is generated based on the retrieved information using a preset reply template; and the answer is output on a preset interactive interface through a multimodal output method.

[0050] As an example, the specific implementation process of this step is as follows: First, we integrated employee consultation channels (such as email, phone, online platforms, etc.), and used NLP and CV technologies for input processing to ensure accurate reception and analysis of questions. Then, we used natural language processing technology to process questions by word segmentation, part-of-speech tagging, and syntactic analysis, supported context management, and optimized intent recognition for common types of questions in multiple professional fields of human resources (such as salary and benefits inquiries, leave process consultations, and professional title review processes, etc.) to understand the intent and key information of the questions.

[0051] At this time, based on the understanding of the question intent, the knowledge base is searched first, and the inverted index and full-text search engine (such as Elasticsearch) are used for full-library search, combined with the vector database (Milvus) for vector search to ensure accurate answers to policy-related questions.

[0052] Finally, the answers are generated based on the retrieved information, matching the question and answer format that suits the user. Templated responses are designed for common questions, and personalized responses are generated based on user information and historical conversations.

[0053] When presenting answers, it supports multi-modal output methods such as text and images, which can provide a convenient information acquisition experience. Aiming at the specific needs of multiple professional fields of human resources, a friendly interactive interface and process can be designed to improve user satisfaction.

[0054] The present invention provides a multi-professional intelligent question-and-answer method for human resources. By systematically collecting, processing and storing human resources knowledge and data, and training and optimizing a large language model, it achieves accurate analysis and rapid answers to employee consulting questions, and significantly improves the efficiency and quality of human resources consulting services.

[0055] See also Figure 2As shown, the present invention also discloses a human resources multi-professional intelligent question and answer system, including: a data collection module 1, a data extraction module 2, a data processing module 3, a model building module 4 and a question and answer service module 5. The data collection module 1 is used to collect human resource knowledge data, clean and decompose the human resource knowledge data, convert the data format, and import it into the knowledge base.

[0056] The data extraction module 2 is used to collect human resource information data, perform entity recognition through a natural speech processing model after preprocessing, and build a human resource related entity library.

[0057] The data processing module 3 is used to perform data slicing processing based on the entity library, and store the slice data into a preset database through structured storage and vectorized storage according to the type of slice data.

[0058] The model building module 4 is used to build a training data set for the model based on the knowledge base and the preset database, use the training data set to train the large language model, and generate a knowledge base language model for the human resources multi-professional intelligent question-and-answer system after model tuning.

[0059] The question-answering service module 5 is used to obtain the human resources consulting questions of employees, parse and retrieve the questions through the knowledge base language model of the human resources multi-professional intelligent question-answering system, generate matching answers, and output the answers.

[0060] The specific implementation of the human resources multi-professional intelligent question and answer system of this embodiment is basically the same as the specific implementation of the human resources multi-professional intelligent question and answer method mentioned above, and will not be repeated here.

[0061] The present invention also discloses a human resources multi-professional intelligent question and answer device, comprising a processor and a memory; wherein, when the processor executes the human resources multi-professional intelligent question and answer program stored in the memory, the steps of the human resources multi-professional intelligent question and answer method as described in any one of the above items are implemented.

[0062] Furthermore, the multi-professional intelligent question-answering device for human resources in this embodiment may also include: The input interface is used to obtain the human resources multi-professional intelligent question-and-answer program imported from the outside, and save the obtained human resources multi-professional intelligent question-and-answer program to the memory, and can also be used to obtain various instructions and parameters transmitted by external terminal devices, and transmit them to the processor, so that the processor can use the above various instructions and parameters to carry out corresponding processing. In this embodiment, the input interface can specifically include but is not limited to a USB interface, a serial interface, a voice input interface, a fingerprint input interface, a hard disk reading interface, etc.

[0063] The output interface is used to output various data generated by the processor to the terminal device connected thereto, so that other terminal devices connected to the output interface can obtain various data generated by the processor. In this embodiment, the output interface may specifically include but is not limited to a USB interface, a serial interface, etc.

[0064] The communication unit is used to establish a remote communication connection between the human resources multi-professional intelligent question-answering device and an external server so that the human resources multi-professional intelligent question-answering device can mount the image file to the external server. In this embodiment, the communication unit may specifically include but is not limited to a remote communication unit based on wireless communication technology or wired communication technology.

[0065] The keyboard is used to obtain various parameter data or instructions input by the user by tapping the keycaps in real time.

[0066] The display is used to display the relevant information of the multi-professional intelligent question and answer process of human resources in real time.

[0067] The mouse can be used to assist users in inputting data and simplify user operations.

[0068] The present invention also discloses a readable storage medium, wherein the readable storage medium includes a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable hard disk, a CD-ROM or any other form of storage medium known in the technical field. The readable storage medium stores a human resources multi-professional intelligent question-answering program, and when the human resources multi-professional intelligent question-answering program is executed by a processor, the steps of the human resources multi-professional intelligent question-answering method described in any one of the above items are implemented.

[0069] In summary, the present invention utilizes artificial intelligence technology, natural language processing, and vector space model technology to provide online self-service, question-and-answer services, and other services for human resources-related issues through robots simulating human dialogue and communication, thereby effectively improving employees' business operation experience.

[0070] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. As for the method disclosed in the embodiment, since it corresponds to the system disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0071] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0072] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0073] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0074] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each module may exist physically separately, or two or more modules may be integrated into one unit.

[0075] Similarly, each processing unit in each embodiment of the present invention may be integrated into one functional module, or each processing unit may exist physically, or two or more processing units may be integrated into one functional module.

[0076] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0077] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0078] The above is a detailed introduction to the human resources multi-professional intelligent question-and-answer method, system, device and readable storage medium provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, the present invention can also be improved and modified, and these improvements and modifications also fall within the scope of protection of the present invention.

Claims

1. A multi-professional intelligent question-answering method for human resources, characterized in that: include: Collect human resource knowledge data, clean and disassemble it, convert its data format, and import it into the knowledge base; Collect human resources data, perform entity recognition through natural speech processing models after preprocessing, and build a human resources-related entity library; Based on the entity library, data slicing processing is performed, and according to the type of slice data, the slice data is stored in a preset database through structured storage and vectorized storage; Based on the knowledge base and preset database, a training data set for the model is constructed, and the large language model is trained using the training data set. After model tuning, a knowledge base language model for the multi-professional intelligent question-answering system for human resources is generated; Obtain employees' human resources consulting questions, parse and retrieve questions through the knowledge base language model of the human resources multi-professional intelligent question-and-answer system, generate matching answers, and output the answers.

2. The human resources multi-professional intelligent question-answering method according to claim 1 is characterized in that: The collecting of human resource knowledge data, data cleaning and decomposition, data format conversion of the human resource knowledge data, and importing into the knowledge base includes: Connect data with the human resource management system and human resource integration platform through API to obtain basic employee information, attendance records, and collect human resource documents, training materials, and company rules and regulations; Use Python and Spark tools to clean human resource knowledge data, remove data noise from human resource knowledge data, fill missing values, and correct format errors; Use Pandas and OpenRefine to convert human resource knowledge data into a unified format; Classify human resource knowledge data according to profession, and import them into the knowledge base according to different categories.

3. The human resources multi-professional intelligent question-answering method according to claim 1 is characterized in that: The human resources data is collected, pre-processed, and then entity recognized through a natural speech processing model, and a human resources-related entity library is constructed, including: Collect human resources data, and perform data cleaning, deduplication, word segmentation, and part-of-speech tagging in sequence; Perform entity recognition on the processed human resource data through a natural speech processing model, identify the relationship between entities in the text, and generate recognition results; Build a human resources-related entity library to store recognition results.

4. The human resources multi-professional intelligent question-answering method according to claim 3 is characterized in that: The data slicing process is performed based on the entity library, and according to the type of the slice data, the slice data is stored in a preset database through structured storage and vectorized storage, including: Slice the data stored in the entity library according to different professional fields, and process the data slices in parallel through a distributed computing framework to perform classified storage of the sliced ​​data; When the slice data is relational data, a relational database is used for storage; When the slice data is non-relational data, use the MongoDB database for storage; Create query indexes for relational databases and MongoDB databases by setting query conditions, and configure query optimizers for relational databases; When the slice data is text data, the text data is converted into vector data and stored in a vector database; The HNSW algorithm is used to build an index for the vector database and the ANN vector retrieval algorithm.

5. The human resources multi-professional intelligent question-answering method according to claim 4 is characterized in that: The method of constructing a training data set for the model based on the knowledge base and the preset database, and using the training data set to train the large language model includes: Based on the knowledge base and preset database, collect human resources documents, training materials and company rules and regulations as training data, clean, remove duplicates, segment and tag the training data, integrate professional dictionaries in the field of human resources, and generate training data sets; According to the multi-professional application scenarios and needs of human resources, select the corresponding large language model for model training.

6. The human resources multi-professional intelligent question-answering method according to claim 5 is characterized in that: The knowledge base language model of the human resources multi-professional intelligent question-answering system is generated after the model tuning process, including: Use grid search, random search or Bayesian optimization methods to search for hyperparameters, and adjust the hyperparameters of large language models for different application scenarios in multiple human resources disciplines to find the best model parameters; Integrate the large language model of each application scenario to generate the knowledge base language model of the multi-professional intelligent question-answering system for human resources; Evaluation indicators for the human resources question-answering task are set for the knowledge base language model, the model performance is evaluated using the evaluation indicators, and the model is iterated and optimized based on the evaluation results.

7. The human resources multi-professional intelligent question-answering method according to claim 1 is characterized in that: The human resources consulting questions of employees are obtained, and the knowledge base language model of the human resources multi-professional intelligent question-answering system is used to parse and retrieve the questions, generate matching answers, and output the answers, including: Obtain employees' human resources consulting questions and process them through NLP and CV technology; Perform word segmentation, part-of-speech tagging, and syntactic analysis on the human resources consulting questions, and perform intent recognition optimization for the preset question types in multiple professional fields of human resources to obtain the intent of the questions; According to the question intent, use the inverted index and full-text search engine to search the entire knowledge base, and combine it with the vector database to perform vector search and integrate the retrieved information; Generate answers using preset response templates based on the retrieved information; On the preset interactive interface, the answer is output through multimodal output.

8. A multi-professional intelligent question-answering system for human resources, characterized in that: The system adopts the human resources multi-professional intelligent question-answering method as described in any one of claims 1 to 7; The system comprises: The data collection module is used to collect human resource knowledge data, clean and disassemble the human resource knowledge data, convert the data format, and import it into the knowledge base; The data extraction module is used to collect human resources data, perform entity recognition through the natural speech processing model after preprocessing, and build a human resources related entity library; A data processing module is used to perform data slicing processing based on the entity library, and store the slice data into a preset database through structured storage and vectorized storage according to the type of slice data; The model building module is used to build a training data set for the model based on the knowledge base and the preset database, use the training data set to train the large language model, and generate a knowledge base language model for the human resources multi-professional intelligent question-answering system after model tuning; The question-and-answer service module is used to obtain employees' human resources consulting questions, parse and retrieve questions through the knowledge base language model of the human resources multi-professional intelligent question-and-answer system, generate matching answers, and output the answers.

9. A multi-professional intelligent question-answering device for human resources, characterized in that: include: A memory device for storing a multi-professional intelligent question-answering program for human resources; A processor is used to implement the steps of the human resources multi-professional intelligent question and answer method as described in any one of claims 1 to 7 when executing the human resources multi-professional intelligent question and answer program.

10. A readable storage medium, characterized in that: The readable storage medium stores a human resources multi-professional intelligent question and answer program, and when the human resources multi-professional intelligent question and answer program is executed by the processor, the steps of the human resources multi-professional intelligent question and answer method as described in any one of claims 1 to 7 are implemented.

Images