CN117474092A - Enterprise knowledge base construction system based on AIGC - Google Patents

Abstract

The invention discloses an enterprise knowledge base construction system based on AIGC, which includes: a knowledge base, which is established within the enterprise and is open to internal employees of the enterprise and customers who cooperate with the enterprise. The knowledge base sets up login verification, and the login verification marks and distinguishes employees and customers. ; Marking model, grading different employees and customers, marking employees and customers, forming employee-specific tag codes and recording the browsing content of employees and customers; Splitting model, splitting the data and text transmitted to the knowledge base into Framework content and branch content. The present invention builds an enterprise knowledge base based on AIGC, performs the same push according to the data and text that customers and employees need to retrieve, and marks the data and text when retrieving data and text within the enterprise.

Description

An enterprise knowledge base construction system based on AIGC

Technical field

The present invention relates to the technical field of knowledge base construction based on AIGC, and is specifically an enterprise knowledge base construction system based on AIGC.

Background technique

AIGC technology can automatically generate a large amount of content to improve production efficiency; generate novel creativity and ideas by learning and imitating large amounts of data; assist the creation and design process, and provide creators with inspiration and innovative materials.

An enterprise knowledge base serves as a self-service desk for customers and employees to store, organize, and share information. Help employees complete their work more efficiently, allowing them to focus on their tasks and what is really important, without having to conduct a large and heavy search for a certain thing or task, that is, it can complete the refining of the key points of things or tasks. The construction of an enterprise knowledge base needs to be based on the actual needs and actual situation of the enterprise. The advantages of the enterprise knowledge base include: providing personalized learning experience and feedback, automatically generating a large amount of content, and generating novel ideas and innovations by learning and imitating a large amount of data. ideas etc.

Building an enterprise knowledge base based on AIGC can better serve customers and employees. When customers and employees need to know the corresponding data and related content, they only need to enter the corresponding question to generate data and content related to the issue, so that calls to customers and employees.

Chinese Patent Announcement No. CN111460078B discloses a process-based knowledge base management system for engineering supervision enterprises. This management system realizes the knowledge management of the entire enterprise process and the rapid search of relevant knowledge in the knowledge base, and effectively solves the problem of various departments of the enterprise providing knowledge to customers. Problems such as inconsistent knowledge service information, inconsistent knowledge service standards, insufficient service experience, and inability to perform rapid knowledge retrieval during service can better provide knowledge management services to enterprises.

In the process of retrieving knowledge data, due to the inconsistency of the keywords received by AIGC, when searching based on questions from customers or employees, multiple answers are often obtained. Although the retrieved content is similar in a large framework, However, there are still obvious differences in the details. It is precisely because of the differences that there are differences in the data processing process, which affects the final results. When employees and customers get different results from retrieval, it will cause confusion for customers. Doubts about corporate professionalism.

Contents of the invention

One of the purposes of the present invention is to provide an enterprise knowledge base construction system based on AIGC. The enterprise knowledge base is constructed based on AIGC, and accurate content is extracted based on the information selected by employees' search, so as to reduce the occurrence of similar searches. Data differences improve the synchronization rate of multi-terminal retrieval.

In order to achieve the above objectives, the present invention is implemented through the following technical solutions: an enterprise knowledge base construction system based on AIGC, including:

knowledge base , established within the enterprise and open to internal employees and customers who cooperate with the enterprise, a knowledge base Set up login verification to mark employees and customers;

The tagging model classifies different employees and customers, tags employees and customers, forms employee-specific tag codes, and records the browsing content of employees and customers;

Split the model and send it to the knowledge base Internal data and text are split into frame content/> and branch content/> , branch content/> For frame content/> extension, and refine keywords based on data and text, and use keywords/> Combine with each other to form related words;

The retrieval model pushes relevant data and text based on the retrieval of employees and customers. When pushing, the lowest-level employees and customers frame the content. push, employee level promotion increases branch content/> Push, the higher the employee level, the branch content/> The more you push, the more you branch out with the use of data and text/> with frame content/> conversion between.

In one or more embodiments of the invention, employees are labeled , the customer is marked as/> , and generate special tags about employees at different levels, namely/> , where,/> is the level, which is divided into: ,/> is the employee type,/> Number the employee and create a knowledge base for the employee/> When retrieving, branch content with different amounts of data/> push.

In one or more embodiments of the present invention, the above split model first performs a knowledge base Refining the content, that is, first retrieving a piece of text or data, obtaining the content described by the text or data, deleting the marginalized content based on the content, and comparing it with the content described by the text or data. For the deleted marginalized content Text or data that does not change the original content meaning of the text or data is defined as branch content/> .

In one or more embodiments of the present invention, the above retrieval model includes targeting branch content Mark the clipboard, the clipboard is generated when the employee retrieves the content, and the employee selects the frame content that needs to be used/> and branch content/> After that, the selected content will be generated in the clipboard, marked according to the number of uses of the branch content, and placed in the branch content/> The location generates a subscript, i.e./> , among which,/> To select branch content/> ,/> is the number of selections.

In one or more embodiments of the present invention, the above-mentioned employee retrieves the branch content selected in the clipboard When, you can modify the branch content/> Make modifications and modify the branch content/> is recorded and included in the branch content/> The modification position generates a correction table/> , correction table/> Display all modified content and modify the table/> In the storage state, when retrieving the contents of the clipboard, modify the table/> Expand.

In one or more embodiments of the present invention, when multiple departments in an enterprise perform a joint task, a set of employees of the departments included in the joint task is generated. , where,/> is the task number,/> Number employees in each department, and create a knowledge base for employees within the employee collection/> Inner branch content/> When retrieving, generate a mark that is different from the modified content of employees who do not belong to the employee set.

In one or more embodiments of the present invention, the above unmodified branch content Frame content based on the number of times selected/> conversion ratio:

;

in, For branch content/> Conversion ratio, calculated as a percentage. Different levels of employees correspond to different proportions of branch content/> Push, that is/> , when the employee level is/> When, branch content/> Push ratio/> is 0.

In one or more embodiments of the present invention, the above construction is for branch content The split module, the branch content/> Search to generate the content of this branch/> Detailed information, based on branch content/> Details about branch content/> Calculate the amount of differential content contained in:

;

in, For branch content/> The quantity of materials in;/> For branch content/> The number of values in , when the number of difference contents/> When, branch content/> break down.

In one or more embodiments of the present invention, the above decomposed branch content divided into variables/> and quantitative , among which, quantitative/> It is text and data with no difference content, and the variable/> In order to have replaceable text and data, quantitative analysis is performed through a language analysis model/> Carry out analysis and make quantitative judgments/> Completeness, complete quantification/> Content marked as frame extension.

In one or more embodiments of the present invention, the above keywords Relate to each other to form related words/> , where,/> is the nth keyword, and the two keywords are combined to form a related word/> , where x is the associated word number, based on employee or customer/> Use the keywords generated by the search to search for related content.

The invention provides an enterprise knowledge base construction system based on AIGC. Compared with existing technology, it has the following beneficial effects:

1. This invention builds enterprise knowledge base based on AIGC, and based on customer The data and text that employees need to retrieve are pushed in the same way. When retrieving data and text within the enterprise, the data and text are marked. When subsequent corresponding pushes are made, the retrieved data and text can be filtered to reduce data differences.

2. In the process of building the knowledge base, classify the employee levels within the enterprise, filter the data and text according to the different levels of employees, and identify low-level employees or customers When searching, push content that fits the large framework to reduce data or content deviations and reduce the problem of data differences in retrieval that lead to deviations in work results.

3. When building a knowledge base based on AIGC, keywords are refined and summarized based on the content of the knowledge base, and keyword extensions are performed on related content. When searching, it can be extended through the interweaving of different keywords. Different keywords, based on employee level and customer Push relevant data and text.

4. When constructing the content inside the knowledge base, the data and content are divided into frame content and branch content according to the data and text inside the knowledge base. When retrieving, by marking the frame content and branch content, in When using retrieval content, the retrieval content can be continuously marked and modified to match keywords based on the content used by high-level employees after retrieval and related modifications.

Description of the drawings

Figure 1 is a schematic diagram of the system of the present invention.

Detailed ways

The following will disclose multiple embodiments of the present invention in the accompanying drawings. For clarity of explanation, many practical details will be explained in the following description. However, it will be understood that these practical details should not limit the invention. That is to say, in some embodiments of the present invention, these practical details are not necessary. In addition, in order to simplify the drawings, some conventional structures and components are simply illustrated in the drawings, and the same reference numerals will be used to represent the same or similar components in all the drawings. And if possible in implementation, the features of different embodiments can be applied interactively.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have their ordinary meanings and can be understood by one familiar with the art. Furthermore, the definitions of the above-mentioned words in commonly used dictionaries should be interpreted as meanings consistent with the relevant fields of the present invention in the content of this specification. Unless specifically defined, these terms will not be interpreted as having an idealized or overly formal meaning.

Please refer to Figure 1. The present invention provides an enterprise knowledge base construction system based on AIGC. When building the knowledge base, the data and text are split for the knowledge base, and different data and text are pushed according to the levels of internal employees of the enterprise. When customers or low-level employees search through the enterprise knowledge base, content that is closer to the framework can be recommended. include:

knowledge base , established within the enterprise and open to internal employees and customers who cooperate with the enterprise, a knowledge base Set up login verification to mark employees and customers;

The tagging model classifies different employees and customers, tags employees and customers, forms employee-specific tag codes, and records the browsing content of employees and customers;

Split the model and send it to the knowledge base Internal data and text are split into frame content/> and branch content/> , branch content/> For frame content/> extension, and refine keywords based on data and text, and use keywords/> Combine with each other to form related words;

The retrieval model pushes relevant data and text based on the retrieval of employees and customers. When pushing, the lowest-level employees and customers frame the content. Push, employee level promotion increases branch content/> Push, the higher the employee level, the branch content/> The more you push, the more you branch out with the use of data and text/> with frame content/> conversion between.

In this embodiment, different levels of employees and customers are used to push different content. When low-level employees and customers perform searches, the retrieved content fits the frame content. , while the frame content/> For knowledge base/> It contains more accurate content, and there is no problem of data differences. It can be unified when junior employees and customers use these data.

As a low-level employee or customer, the data and text you need to use are shallow, and the frame content and branch content/> The difference is that the frame content/> It is the framework of data or content and is the basic component of data and content. There is no difference, and branch content/> frame content/> The extension is uncertain data and has differences between different data or contents.

In one embodiment, wherein the employee is labeled , the customer is marked as/> , and generate special tags about employees at different levels, namely/> , where,/> is the level, which is divided into: ,/> is the employee type,/> Number the employee and create a knowledge base for the employee/> When retrieving, branch content with different amounts of data/> push. /> The levels decrease in sequence.

In this embodiment, employees are graded and the framework content can be targeted and branch content push, and for frame content/> and branch content/> In terms of frame content/> The accuracy is higher and there is no differential information. Therefore, when pushing relevant content, employees with higher levels have higher content discrimination ability and the ability to distinguish the authenticity of the pushed content.

Higher-level employees have access to more branch content , branch content/> and frame content/> Make mark distinctions so that employees can be reminded of push content when they check it, so that employees can view branch content/> Make a distinction.

In one embodiment, the split model first performs the knowledge base Refining the content, that is, first retrieving a piece of text or data, obtaining the content described by the text or data, deleting the marginalized content based on the content, and comparing it with the content described by the text or data. For the deleted marginalized content Text or data that does not change the original content meaning of the text or data is defined as branch content/> .

In this embodiment, the knowledge base Split the internal text or data into frame content/> and branch content/> , branch content/> After being deleted, the general direction of the text or data will not be changed, but the frame content/> After being deleted, there will be problems of incoherence. Therefore, after deleting marginalized content, deviations in the meaning of the original content will be defined as frame content/> , which does not deviate from the meaning of the original content, is defined as branch content/> .

Based on this, branch content and frame content/> division, able to analyze data and text, and frame content/> Changes in will lead to changes in the original meaning, so for lower-level employees and customers, the content of the framework/> The promotion can be more precise and there will be no differentiated content.

In one embodiment, the retrieval model includes targeting branch content Mark the clipboard, the employee generates the clipboard when retrieving content, and the employee selects the frame content that needs to be used/> and branch content/> After that, the selected content will be generated in the clipboard, marked according to the number of uses of the branch content, and placed in the branch content/> The location generates a subscript, i.e./> , where,/> To select branch content/> ,/> is the number of selections.

In this embodiment, by detecting branch content The number of selections can branch content/> tag, branch content/> The higher the number of selections, the more relevant the branch content/> The higher the relevant keyword search push, the more employees unlock branch content with different amounts of data as the level increases/> Branch content with a higher number of selections will be removed first/> .

Among them, the branch content The number of selections will be displayed directly on the push interface after employees retrieve it, so that employees can refer to it when using this part of the content, so as to serve as a reminder so that employees can use this branch of content. to identify the authenticity.

In one embodiment, the employee retrieves the selected branch content in the clipboard When, you can modify the branch content Make modifications and modify the branch content/> is recorded and included in the branch content/> The modification position generates a correction table/> , correction table/> Display all modified content and modify the table/> In the storage state, when retrieving the contents of the clipboard, modify the table/> Expand.

In this example, branch content Different employees will modify some of the values or content, and the enterprise knowledge base/> Branch content included in/> Employees who have reached a certain level in all departments can perform searches. When conducting joint tasks among multiple departments, one of the employees will search for branch content/> After modifications are made, they will be recorded in the knowledge base/> to avoid differences in the content used between multiple parts.

After generating the revision table, it can be recorded in the knowledge base Within, the selected branch content/> When a revision table is included, an inverted triangle mark will be generated below the modified content. Selecting the inverted triangle will expand all modification records for the modified content, so that employees from multiple departments can make corresponding references when retrieving and retrieving content to avoid content. difference.

In one embodiment, when multiple departments within an enterprise perform a joint task, a set of employees of the departments included in the joint task is generated. , where,/> is the task number,/> Number employees in each department, and create a knowledge base for employees within the employee collection/> Inner branch content/> When retrieving, generate a mark that is different from the modified content of employees who do not belong to the employee set.

In this embodiment, the branch content retrieved within the same employee set After being modified by the employees in the employee set, marks of different colors are generated. According to the different color marks, the employees in the employee set can be branched into other data/> When retrieving, make corresponding content modifications without having to modify branch content through communication between departments/> to modify.

And in the branch content When the number of calls is huge, the modified branch content/> When selected, the modified branch content on the clipboard/> It will automatically be modified to the content modified by employees in the same employee set, avoiding branching content/> In case of omissions during retrieval, the synchronization of data used between multiple departments is further ensured.

In one embodiment, the unmodified branch content Frame content based on the number of times selected/> conversion ratio:

;

in, For branch content/> Conversion ratio, calculated as a percentage. Different levels of employees correspond to different proportions of branch content/> Push, that is/> , when the employee level is/> When, branch content/> Push ratio/> is 0.

In this embodiment, the calculation for branch content to frame content/> conversion ratio, in branch content/> After multiple selections without modification, the content of the branch/> Will gradually transform into frame content/> , meaning the branch content/> There is no content difference, therefore, the employee level is/> The frame content that can be retrieved at the time/> Will be gradually added, and after review by senior staff, branch content/> It will gradually be opened to lower levels.

Add high-level staff for branch content audits can further facilitate lower-level employees or customers Perform corresponding searches and open branch content/> When employees search, they will push branch content that accounts for a certain percentage of the total pushed content/> , so that branch content can be gradually branched/> of openness.

In one embodiment, building for branch content The split module, the branch content/> Search to generate the content of this branch/> Detailed information, based on branch content/> Details about branch content/> Calculate the amount of differential content contained in:

;

in, For branch content/> The quantity of materials in;/> For branch content/> The number of values in , when the number of difference contents/> When, branch content/> break down.

In this embodiment, calculating the amount of differential content can be done for branch content Carry out further classification to be able to select and push based on the number of different content, at lower level employees or customers/> Ability to push differential content count/> when performing deeper searches Branch content/> , and there will be no difference in content that will affect the final presentation result.

In one embodiment, the decomposed branch content divided into variables/> And quantitative/> , among which, quantitative/> It is text and data with no difference content, and the variable/> In order to have replaceable text and data, quantitative analysis is performed through a language analysis model/> Carry out analysis and make quantitative judgments/> Completeness, complete quantification/> Content marked as frame extension.

In this embodiment, the calibration frame extension content can further branch content Distinguishing between low-level employees or customers/> During further retrieval, the frame content/> can be modified by pushing the frame extension content. Provide further explanation or expansion, so that the content can be interpreted and pushed at a deeper level, and more complete creative content can be generated.

In one embodiment, the keyword Relate to each other to form related words/> , where,/> is the nth keyword, and the two keywords are combined to form a related word/> , where x is the associated word number, based on employee or customer/> Use the keywords generated by the search to search for related content.

In this example, by placing employees or customers The keywords that need to be searched are combined to form related words, and related content is combined and pushed based on the related words.

Example:

When an employee or customer of a company logs into the enterprise knowledge base to search for relevant content, the proportion of frame content and branch content that employees or customers at different levels can retrieve is as shown in the following table:

Among them, the unmodified branch content Frame content based on the number of times selected/> The conversion ratio is shown in the table below:

branch content When the conversion ratio reaches 100%, it is converted into frame content/> .

To sum up, the technical solutions disclosed in the above embodiments of the present invention have at least the following advantages:

1. This invention develops enterprise knowledge base based on AIGC built and based on the customer/> The data and text that employees need to retrieve are pushed in the same way. When retrieving data and text within the enterprise, the data and text are marked. When subsequent corresponding pushes are made, the retrieved data and text can be filtered to reduce data differences.

2. Working on knowledge base During the construction process, the employee levels within the enterprise are classified, and the data and text are filtered according to the different levels of employees. When low-level employees or customers/> When searching, push content that fits the large framework to reduce data or content deviations and reduce the problem of data differences in retrieval that lead to deviations in work results.

3. Knowledge base based on AIGC When building, according to the knowledge base /> Refining and summarizing the keywords of the content, and extending the keywords to the related content. When searching, different keywords can be extended through the interweaving of different keywords, and then based on employee levels and customers/> Push relevant data and text.

4. In the knowledge base When building internal content, it is based on the knowledge base/> Internal data and text are divided into frame contents/> and branch content/> , when retrieving, by querying the frame content/> and branch content/> When using the search content, the search content can be continuously marked and modified, so that keywords can be matched based on the content used by high-level employees after the search and related modifications.

Although the present invention has been disclosed in conjunction with the above embodiments, it is not intended to limit the present invention. Anyone familiar with the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the present invention is The scope of protection shall be defined by the appended claims.

Claims (10)

1. An AIGC-based enterprise knowledge base construction system, comprising:

knowledge baseEstablishing a knowledge base for opening enterprise-oriented staff and clients cooperated with the enterprise in the enterprise>Setting up login verification, and marking and distinguishing staff and clients by the login verification;

the marking model is used for grading different staff and clients, marking the staff and the clients to form staff exclusive marking codes and recording browsing contents of the staff and the clients;

splitting the model and conveying the model to a knowledge baseInternal data and text splitting into frame contents>Branch content->Branch content->For frame content->Is used for extracting keywords according to the data and the characters and passing through the keywordsCombining each other to form a related word;

search model based on staff and guestsUser searching to push related data and text, and during pushing, the lowest staff and clients perform frame contentIs a push of staff level promotion increase +.>The higher the employee level the branch content +.>The more push, the more branching content is along with the use of data and text>And frame content->And between the transitions.

2. The AIGC based enterprise knowledge base building system of claim 1, wherein staff members are labeled asThe customer is marked +.>And generates special marks concerning staff of different levels, i.e. +.>Wherein->The level is divided into levels according to employee time and capability: />，/>For employee type->Numbering staff, carrying out knowledge base on staff +.>Branching content of different data amount according to level during searching>Is pushed by the push button.

3. The AIGC-based enterprise knowledge base construction system of claim 2, wherein the splitting model first performs knowledge baseThe extraction of the content, namely, firstly, searching a section of text or data to obtain the content described by the text or data, carrying out the deletion of the bordered content according to the content, comparing with the content described by the text or data, and defining the text or data which does not change the original content meaning of the text or data as branch content for the deleted bordered content>。

4. The AIGC based enterprise knowledge base building system of claim 3, wherein the retrieval model comprises a search model for branch contentClipboard for marking, which is generated by staff during searching content, and which selects frame content to be used +.>And branch content->Then, the selected content is generated in the clipboard, marked according to the number of times of use of the branch content, and added in the branch content->Position-generating corner marks, i.e->Wherein->For selecting branch content->，/>For the number of selections.

5. The AIGC based enterprise knowledge base building system of claim 4, wherein staff members pick up selected branch content in the clipboardIn this case, the branching content can be added>Modified branch content +.>Recorded and in branch content +.>Generates a correction table +.>Correction of the form->Displaying all modified contents and correcting the table +.>In order to take in the state, when the contents of the shear plate are called, the correction table +.>And (5) unfolding.

6. The AIGC based enterprise knowledge base building system of claim 5, wherein when multiple departments within an enterprise perform a joint task, a staff set of departments included by the joint task is generatedWherein->For task number->Staff numbering for each department, staff within staff set carrying out knowledge base +.>Content of inner branch->Upon invocation, a tag is generated that is distinct from the employee modified content that does not belong to the employee collection.

7. The AIGC based enterprise knowledge base building system of claim 6, wherein unmodified branch contentFrame content according to the selected times>Conversion ratio of (2):

；

wherein,for branching content->Conversion proportion, calculated by percentage, the staff different levels correspond to branch content of different proportions respectively +.>Push, i.e.)>When employee level is +.>Branch content->Push ratio->Is 0.

8. The AIGC based enterprise knowledge base building system of claim 7, wherein building is directed to branch contentSplit module of (1) for branching content->Retrieving to generate the branch content->According to the branching content +.>Is about the branch content>The number of the contained difference contents is calculated:

；

wherein,for branching content->The amount of material in (a); />For branching content->The number of values in (1) when the number of contents is different +.>When it is, branch content is performed>And (5) decomposing.

9. The AIGC based enterprise knowledge base building system of claim 8, wherein the decomposed branch contentDivided into variables->Quantification +.>Wherein, quantitatively->For text and data without differential content, and the variable +.>For the purpose of having alternative text and data, the quantification is made by means of a linguistic analysis model>Analyzing, judging and quantifyingIs the whole quantification->Marked as frame extension content.

10. The AIGC based enterprise knowledge base building system of claim 1, wherein keywordsInterrelating to form a related word->Wherein->For the nth keyword, two keywords are combined to form a related wordWherein x is the number of the associated word according to employee or client +.>The keyword generated by the search is searched for the relevant content.