CN118036613A

CN118036613A - Author management method, system and storage medium based on network platform

Info

Publication number: CN118036613A
Application number: CN202410186003.2A
Authority: CN
Inventors: 王斌斌
Original assignee: Wuxian Shenghuo Beijing Information Technology Co ltd
Current assignee: Wuxian Shenghuo Beijing Information Technology Co ltd
Priority date: 2024-02-20
Filing date: 2024-02-20
Publication date: 2024-05-14

Abstract

The embodiment of the disclosure provides an author management method, an author management system and a storage medium based on a network platform, wherein the method comprises the following steps: receiving a request parameter, wherein the request parameter comprises an author I D and a text of a desired operation, and the text of the desired operation comprises at least one operation sentence; sending the expected operation text to a rule engine; receiving a rule matching result returned by the rule engine; and judging whether all the operation sentences in the expected operation texts are matched with the local rules or not, and if so, executing all the operation sentences in the expected operation texts on all authors according to authors I D. According to the embodiment of the disclosure, through the rule engine, the automatic matching of the operation sentences managed by the author is performed, so that the automatic management of the author is realized, the labor cost is reduced, and the management efficiency is improved.

Description

Author management method, system and storage medium based on network platform

Technical Field

The embodiment of the disclosure relates to the technical field of author management, in particular to an author management method, an author management system, an electronic device and a computer readable storage medium based on a network platform.

Background

With the continuous development of artificial intelligence technology, the large computer model can play different roles in various aspects, and the effects of improving efficiency and being easy to operate can be achieved on business by integrating the large model.

The author management center of the traditional content community platform can only identify and manage authors through simple keyword matching, and the method has the problems of low accuracy, low management efficiency and the like. In addition, it is also difficult for existing author management centers to manage multidimensional information of authors who post content, such as interests of the authors, areas of locale, social networks, etc.

The semantic similarity model technology is a natural language processing technology based on deep learning, can realize tasks such as text classification, similarity calculation and the like through understanding language semantics, and has been widely applied in a plurality of fields such as machine translation, text abstract, emotion analysis and the like.

The management method in the prior art comprises the following steps: 1. operators need to open an author management background, and then manually click on an author to edit a popup window for operation. 2. And the operators give a list of the authors of the developers, and the developers perform batch author data operation through the interface.

The prior art has the following defects:

(1) Long operation flow and low efficiency: operators need to edit one by one through webpage operation.

(2) The manpower cost is high: part of operators cannot be independently completed and need to rely on developers.

(3) The learning cost is high: for new users, platform usage training is required.

Disclosure of Invention

An object of an embodiment of the present disclosure is to provide a web platform-based author management method, system, electronic device, and computer-readable storage medium, so as to solve the foregoing problems in the prior art.

In order to achieve the above objective, the technical solution adopted in the embodiments of the present disclosure is as follows:

in one aspect, an embodiment of the present disclosure provides a method for managing authors based on a network platform, where the method includes: receiving a request parameter, wherein the request parameter comprises at least one author ID and at least one text of a desired operation, and the text of the desired operation comprises at least one management operation statement;

the expected operation text is sent to a rule engine, wherein the rule engine is used for matching the expected operation text with a preset local rule;

receiving a rule matching result returned by the rule engine;

And judging whether all the management operation sentences in the expected operation texts are matched with the local rules or not, and if so, executing all the management operation sentences in the expected operation texts on all authors according to the author ID.

Illustratively, the determining whether all management operation sentences in each of the expected operation texts match the local rule further includes:

if not, extracting the management operation statement which is not matched with the upper rule;

Packaging preset basic operation texts and management operation sentences which are not matched with the rules into corresponding request parameters respectively, and sending the request parameters to a pre-trained semantic similarity model, wherein the semantic similarity model carries out similarity matching calculation on the management operation sentences which are not matched with the rules and each sentence in the preset basic operation texts according to the guidance of the request parameters, and the basic operation texts comprise various grades of authors, various classifications and opening or closing of authors and are divided into operation sentences;

Receiving similarity results of management operation sentences which are returned by the semantic similarity model and are not matched with the upper rules and each basic operation sentence in the basic operation text;

screening all basic operation sentences with the similarity to each management operation sentence which is not matched with the upper rule being greater than or equal to a preset threshold value, selecting the basic operation sentence with the maximum similarity to the corresponding management operation sentence from all basic operation sentences, and respectively executing all management operation sentences which are matched with the upper rule and the selected expected operation sentences for all authors according to the author ID.

Illustratively, after the performing, on all authors, all the management operation statements matching the upper rule and the selected basic operation statement, the method further includes: and returning an execution result and notifying an operator of the execution result.

And receiving records formed by unexecuted management operation sentences submitted by the operation and corresponding basic operation sentences, and storing the records into a database.

Illustratively, after storing the record in the database, the method further comprises:

The method comprises the steps that at intervals of preset time, a model trainer inquires whether an unprocessed record of operation feedback exists in the database within the preset time, and if so, the model trainer receives the unprocessed record of operation feedback;

adding unexecuted management operation sentences into a corresponding basic operation sentence training set, retraining a semantic similarity model, and generating a new semantic similarity model;

and returning the new semantic similarity model to the model trainer.

Illustratively, after the returning the new semantic similarity model to the model trainer, the method further comprises:

Receiving a notification that the model trainer completes the training of the semantic similarity model;

Switching to a new semantic similarity model.

Illustratively, the semantic similarity model is based on the method comprising: the BERT model is constructed and formed.

Illustratively, after receiving the request parameter, the method further comprises:

Sending the expected operation text to a text cleaning tool for local text cleaning;

Receiving cleaned expected operation text returned by the text cleaning tool;

the cleaning process comprises the following steps: and removing useless character strings and useless sentences in the expected operation text.

Another aspect of the embodiments of the present disclosure proposes an author management system based on a network platform, which is characterized in that the system includes:

a receiving request module, configured to receive a request parameter, where the request parameter includes at least one author ID and at least one text of a desired operation, and the text of the desired operation includes at least one management operation statement;

the first sending module is used for sending the expected operation text to a rule engine, wherein the rule engine is used for matching the expected operation text with a preset local rule;

The receiving matching result module is used for receiving a rule matching result returned by the rule engine;

And the judging module is used for judging whether all the management operation sentences in the expected operation texts are matched with the local rules or not, and if so, executing all the management operation sentences in the expected operation texts on all authors according to the author ID.

Another aspect of the disclosed embodiments proposes an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, said processor implementing the method as described above when executing said program.

Another aspect of the disclosed embodiments proposes a computer readable storage medium, on which a computer program is stored, which program, when being executed by a processor, implements a method as described above.

The beneficial effects of the embodiment of the disclosure are that:

According to the embodiment of the disclosure, the rule engine is added, and the function of executing the author operation according to the input content is added, so that the labor cost is reduced, the use efficiency is improved, meanwhile, the learning cost is also reduced, and operators only need to know the operation type and concept of the author management platform.

Drawings

FIG. 1 is a flow chart of an author management method based on a network platform according to an embodiment of the disclosure;

FIG. 2 is a schematic diagram of a method employing a semantically similar model in accordance with an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of an author management system architecture based on a network platform according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of an author management system process based on a network platform using an embodiment of the present disclosure.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present disclosure more apparent, the embodiments of the present disclosure will be further described in detail below with reference to the accompanying drawings. It should be understood that the detailed description is merely illustrative of the disclosed embodiments and is not intended to limit the disclosed embodiments.

The terms involved in the embodiments of the present disclosure are explained as follows:

Semantic similarity model: is a computer model for evaluating similarity between texts, and herein mainly uses the BERT model.

Author management center: is a platform for unifying author-related operations.

Operators: refers to a large number of people who use the author management center.

Author binding group: grouping according to the type of store of the author, such as cartoon group, star chasing group, cotton doll group.

The authors rated: indicating the level of the group in which the author is located.

The authors are divided into: whether the author participates in the content creation incentive program, and if so, monthly traffic reddening can be performed.

As shown in fig. 1, an aspect of an embodiment of the present disclosure proposes a method for managing authors based on a network platform, where the method includes:

Step S110, receiving request parameters, wherein the request parameters comprise at least one author ID and at least one text of a desired operation, and the text of the desired operation comprises at least one management operation statement.

The embodiment of the disclosure can be applied to network platforms such as community platforms for publishing contents, management of network platform authors can be realized by acquiring the author ID of the network platform and texts of expected operations, and multiple request parameters can be received at one time in the management method of the disclosure, for example, at least one author ID is received in any batch, and at least one expected operation text is received at the same time, wherein the author ID and the text contents of the expected operations are provided by operators, so that management operation sentences can be used for carrying out operations such as various grouping, various grading, opening or closing division and the like on the authors so as to realize the management of the authors. The author ID is a unique identification for identifying the author of the platform content distribution, and when a specific operation sentence is executed, the corresponding author is determined to be managed according to the author ID. The content that authors post on the platform is not limited to text, pictures, video, etc.

As two examples, as in example 1, the operator fills in the text input box with a value of 1 at the receiving end of the author management center, and operates the text input box to fill in the text operation that is expected to be executed as follows: "modify authors grouped into caricature groups, modify authors rated — class B. The text of the desired operation may also have emoticons, as in example 2, the text of the desired operation is "modify authors grouping into animation groups, set authors ranking to S level, [ emotions ]". Here, the author ID may be filled in according to actual situations, and the text of the desired operation is filled in according to actual situations.

and step S111, sending the expected operation text to a text cleaning tool to clean the local text.

Specifically, in step S111, the text cleansing tool is a set of code tools written in Java, so that redundant characters in the text of the desired operation can be removed. For example, the original text content is "modify-to-animation group", and is changed into "modify-to-animation group" through the cleaning tool.

The cleaning process comprises the following steps: and removing useless character strings and useless sentences in the expected operation text. According to the above example, as example 1, the text operation that is desired to be performed is: the "modification authors are grouped into cartoon groups, the modification authors are rated-B-class", and the modification authors are grouped into cartoon groups after the text cleaning tool cleans out redundant characters, and the modification authors are rated B-class ". As in example 2, the text of the desired operation is "modify author group to animation group, set author rank to S level, [ expression ]", change to "modify author group to animation group, set author rank to S level after the text washing tool washes out useless labels.

And step S112, receiving cleaned expected operation text returned by the text cleaning tool.

The author management center in the embodiment of the disclosure receives the cleaned expected operation text.

And step S120, the expected operation text is sent to a rule engine, wherein the rule engine is used for matching the expected operation text with a preset local rule.

Step S120 of the disclosed embodiment firstly invokes a rule engine interface, and sends the expected operation text of step 110 or the expected operation text cleaned in step 112 to a rule engine to match with a preset local rule. The rule engine is a database pre-stored by the developer in the author management center. The local rule can be set according to the actual situation. The rule engine includes a regular matching rule. Regular matching rules include a number of matching rules for various groupings of authors, various hierarchies, and on or off divisions. According to the above example, for example, the regular matching rule includes: "modify authors group to% s" and so on, where% s is placeholder, can match all characters, can match to the desired operation sentence of example 1 "modify authors group to caricature group".

And step S130, receiving a rule matching result returned by the rule engine.

The matching result content in the embodiment of the disclosure includes hit items, namely, matching correct and miss items, namely, not matching correct. According to the above example, as in example 1, the expected operation statement with the correct matching result is "the modification authors are grouped into the cartoon group", and the expected operation statement with the correct non-matching result is "the modification authors are ranked as B-level"; the miss items are "modify authors group to animation group" and "set authors to rank to S-level" as in example 2. The hit item is that the sentence in the expected operation text is matched with the content in the matching rule, and the miss item is that the sentence in the expected operation text is not matched with the content in the matching rule.

And step 140, judging whether all operation sentences in the expected operation texts are matched with the local rules, and if so, executing all management operation sentences in the expected operation texts on all authors according to the author ID.

Embodiments of the present disclosure reduce manual matching and manipulation processes by using a rules engine to match desired manipulation text to local rules. After matching with the local rules of the rule engine, the executive program can execute corresponding management operation according to the matching result, so as to realize the management of the author. For example, any one lot receives at least one author ID and at least one desired operation text, and if some or all of the management operation sentences in the desired operation text match the local rule, each management operation sentence matching the rule is executed for all authors of the lot.

Illustratively, as shown in fig. 2, the determining whether all the operation sentences in each of the expected operation texts match the local rule further includes:

and step S150, if not, filtering the management operation sentences matched with the upper rules, and extracting the management operation sentences not matched with the upper rules.

In the embodiment of the disclosure, the hit statement is filtered, and the operation statement of the missed item is extracted. According to the above example, as in example 1, the expected operation statement that does not match correctly is "modify author rank to B-rank"; the miss items in example 2 are "modify authors group to animation group" and "set authors to rank to S-level".

Step S160, respectively packaging preset basic operation texts and management operation sentences which are not matched with the rules into corresponding request parameters, and sending the request parameters to a pre-trained semantic similarity model, wherein the semantic similarity model carries out similarity matching calculation on the operation sentences which are not matched with the rules and each sentence in the preset basic operation texts according to guidance of the request parameters, and the basic operation texts comprise various grades of authors, various classifications and opening or closing of authors and are divided into operation sentences.

Step S160 of the embodiment of the present disclosure packages the missed item sentence in step S150 and the sentence list in the preset basic operation text into corresponding request parameters, and sends the request parameters to the pre-trained semantic similarity model, so as to request the semantic similarity model to calculate the similarity. The basic operation text is a database preset by a developer and stored in an author management center. The request parameters are mainly prompt words, and the missed item sentences and the preset basic operation text sentences are respectively filled into preset prompt word templates to generate corresponding prompt words. The basic operation text includes: the author groups, and opens or closes the author groups into management operation sentences, etc., the basic operation text can be set according to practical situations, for example, each sentence of the basic operation text list can include: "modify authors into starburst", "modify authors into comic groups", "modify authors' rank into B-rank", "set authors into cartoon groups", "turn on authors into" turn off authors into "and" set authors rank into S-rank ", etc., which are not listed here. According to the above example, as in example 1, the expected operation statement that does not match correctly is "modify author rank to B-rank"; and packaging the "modification author grade B" and all sentences in the basic operation text, wherein the "modification author grade B" is the modification author grade B "," the modification author preference is the cartoon "," the modification author preference is the star chasing "," the modification author grade D "," the author grouping is modified to the cartoon group "," the opening authors are divided "," the closing authors are divided "and" the setting author grade S ", together into corresponding request parameters.

Request parameters: the method comprises the steps that a text to be matched is in grade B, the matched text is in grade B, the author preference is animation, the author preference is star tracking, the author authoring grade is grade D, the authors are grouped and modified into animation groups, the authors are opened, the authors are closed and the authors are set to be classified into grade S, and the similarity of each sentence in the text to be matched and the matched text is calculated.

The request parameters are sent to a pre-trained semantic similarity model.

The missed sentences are picked up as "modify author group to animation group, set author rank to S level" in the text input by the user in example 2 through step 150, and "modify author group to animation group" and "set author rank to S level" are generated by segmentation, and parameters are composed with basic operation text data respectively, a parameter is requested: the method comprises the steps of modifying an author group into a cartoon group, matching the text= [ "modifying the author creation level to be level B", "modifying the author preference to be a cartoon", "modifying the author preference to be a star-chasing", "modifying the author creation level to be level D", "modifying the author group into the cartoon group", "opening the author to be divided into", "closing the author to be divided into", "setting the author to be classified into level S" ], and calculating the similarity of each sentence in the text to be matched and the matched text.

Request two parameters: the method comprises the steps of setting an author grade to be S-level, matching the text to be matched to be B-level, modifying the author preference to be cartoon, modifying the author preference to be star-following, modifying the author authoring grade to be D-level, modifying the author group to be cartoon group, opening the author to be divided into closing the author to be divided into S-level, setting the author grade to be S-level, and calculating the similarity of each sentence in the text to be matched and the matched text.

And respectively sending the first request parameter and the second request parameter to a pre-trained semantic similarity model. And the semantic similarity model performs similarity analysis according to each request parameter, and calculates the similarity between the text to be matched, namely, the missed operation sentence and each basic operation sentence of the sentence list in the basic operation text.

The semantic similarity model is based on the following steps: the BERT model is constructed and formed.

And step S170, receiving a similarity result of each expected operation statement which is returned by the semantic similarity model and does not match with the upper rule and each basic operation statement in the basic operation text.

In the embodiment of the disclosure, according to the above examples, the similarity results of "modify author rank B" in example 1 and each sentence "modify author authoring rank B", "modify author preference cartoon", "modify author preference star", "modify author authoring rank D", "modify authors into cartoon group", "open author into", "close author into" and "set author rank S" are [0.99,0.55,0.43,0.90,0.45,0.02,0.01,0.42].

The return results of the four operation sentences after the "modify author group to animation group" and the basic operation text in the request one in example 2 are [0.99,0.001,0.005,0.004], which means that the similarity between the "modify author group to animation group" and the "modify author group to animation group" is 0.99, the similarity between the "open author group" and the "close author group" is 0.001, the similarity between the "close author group" and the "set author rank to S level" is 0.004, the similarity between the "modify author group to animation group" and the four operation sentences before the basic operation text is lower, and the similarity results are omitted here. In the second request, "set author to rank at S level" and return results of four operation sentences after the basic operation text are [0.004,0.001,0.005,0.99], specific numerical values are only illustrative, and are not repeated here, the "set author to rank at S level" has lower similarity with the first four operation sentences of the basic operation text, and the similarity results are omitted here.

Step S180, screening all basic operation sentences with similarity greater than or equal to a preset threshold value with management expected operation sentences which are not matched with the upper rules, selecting the basic operation sentence with the maximum similarity with the corresponding management operation sentence from all basic operation sentences, and respectively executing all management operation sentences matched with the upper rules and the selected basic operation sentences for all authors according to the author ID.

The preset threshold value in the embodiment of the disclosure may be set according to actual conditions or experience, all basic operation sentences with similarity greater than or equal to the preset threshold value are filtered, the basic operation sentence with the greatest similarity is selected as the sentence to be executed, and the preset threshold value may be set according to actual conditions. The author management center receives the management operation sentences which are matched correctly and the selected basic operation sentences, and executes the management operation sentences one by one; the method of the present disclosure determines the operation statement to be executed and automatically executes the operation statement. By adding the semantic similarity model, the function of executing the author operation according to the input content is added, so that the labor cost is reduced, the use efficiency is improved, meanwhile, the learning cost is also reduced, and operators only need to know the operation type and concept of the author management platform.

According to the above example, as in example 1, if the preset threshold is 0.6, two of the preset thresholds are satisfied, but the highest level is the "modified author creation level is the B level", and the sentence is taken as the execution sentence.

In example 2, if the preset threshold is 0.95, request one, only "author group is set as animation group" remain in step S170. The basic operation sentence screened in step S180 is executed. The modification of the author group to the animation group is performed. Request two, the execution sets the author hierarchy to S level.

The author management method of the embodiment of the disclosure greatly reduces labor cost by using a rule engine and a semantic similarity model, liberates both hands of a developer, and achieves the effects of reducing cost and enhancing efficiency; the learning cost is low, operators do not need to know platform interaction, and can operate by hands only by knowing the names and concepts of the functional points.

Illustratively, after the performing, on all authors, all the management operation statements matching the upper rule and the selected basic operation statement, the method further includes:

Step S190, returning an execution result and notifying an operator of the execution result.

After the execution of step S190 is finished, the operator is notified of the execution result by means of a mobile phone short message or an office software message. If the unexecuted sentence exists, the operator submits the operation sentence which is the most similar to the unexecuted management operation sentence in the basic operation text and the unexecuted management operation sentence to the author management platform. Operators can reject wrong operation sentences and do not feed back.

Step 200, receiving records formed by unexecuted management operation sentences submitted by the operation and corresponding basic operation sentences, and storing the records into a database.

Step S210, at intervals of preset time, the model trainer inquires whether the database has records of unprocessed operation feedback in the preset time, and if so, the model trainer receives record data of unprocessed operation feedback.

In step S200, a record formed by the unexecuted management operation statement submitted by the operation and the basic operation statement with the highest similarity is received. The model trainer receives an operation feedback record which is not processed in the preset time of the database, wherein the unprocessed operation feedback record is data which is not subjected to model training again, and error management operation sentences are removed from the data of the unexecuted management operation sentences, namely the error management operation sentences cannot be trained, and only sentences of management operation which should be executed are fed back, but expected operation sentences which are not executed are fed back.

And step S220, adding the unexecuted management operation sentences into the corresponding basic operation sentence training set, retraining the semantic similarity model, and generating a new semantic similarity model.

In the embodiment of the disclosure, the corpus fed back by the operation is added into the training set of the semantic similarity model, and is added in an increment mode. If the semantic similarity model cannot identify that 'set users are grouped into cartoons', the 'set users are grouped into cartoons' are added into a basic operation sentence through operator feedback into a training set of 'modify authors into cartoons', and the similarity of the training set sample and the basic operation sentence 'modify authors into cartoons' is a preset similarity value, so that retraining is performed. Semantic similarity model training is a person skilled in the art will appreciate that it is not described in detail here.

Step S230, returning the new semantic similarity model to the model trainer.

step S240, receiving a notification that the model trainer completes training of the semantic similarity model;

step S250, switching to a new semantic similarity model.

After the training is successful in the embodiment of the disclosure, the model trainer informs an author management center through interface call, and the new model B is trained and can be switched at any time. The author management center switches to a new semantic model B for operation. And the automatic iteration semantic similarity model is realized, no perception is generated by a user, and the automatic switching of the new model and the old model is performed.

As shown in fig. 3, another aspect of the embodiments of the present disclosure provides an author management system based on a network platform, the system including:

A receiving request module 110, configured to receive a request parameter, where the request parameter includes at least one author ID and at least one text content of a desired operation, and the text content of the desired operation includes at least one management operation statement;

The first sending module 120 is configured to send the expected operation text to a rule engine, where the rule engine is configured to match the expected operation text with a preset local rule;

a receiving matching result module 130, configured to receive a rule matching result returned by the rule engine;

And the judging module 140 is configured to judge whether all the management operation sentences in each desired operation text match the local rule, if yes, execute all the management operation sentences in each desired operation text for all authors according to the author ID.

Illustratively, the system further comprises:

a second sending module 111, configured to send the desired operation text to a text cleansing tool for performing local text cleansing;

The cleansing module 112 removes useless character strings and useless sentences in the desired operation text.

And the cleaning result receiving module 113 is used for receiving cleaned expected operation text returned by the text cleaning tool.

The judging module is further configured to judge whether all management operation sentences in each of the expected operation texts match with a local rule; if not, filtering the management operation sentences matched with the upper rules, and extracting the management operation sentences not matched with the upper rules;

the system further comprises:

the third sending module 150 is configured to package a preset basic operation text and each management operation sentence not matched with the rule into corresponding request parameters, and send the request parameters to a pre-trained semantic similarity model, where the semantic similarity model performs similarity matching calculation on the management operation sentence not matched with the rule and each sentence in the preset basic operation text according to guidance of the request parameters, and the basic operation text includes various classification of authors, various classification, and opening or closing of authors into operation sentences;

The similarity receiving result module 160 is configured to receive a similarity result between each management operation sentence which is returned by the semantic similarity model and each basic operation sentence in the basic operation text, where the management operation sentence is not matched with the rule;

The screening module 170 is configured to screen all basic operation sentences having a similarity greater than or equal to a preset threshold value with each management operation sentence not matching the upper rule, select a basic operation sentence having a maximum similarity of the corresponding management operation sentence from all basic operation sentences, and execute all management operation sentences matching the upper rule and the selected basic operation sentences for all authors respectively according to the author ID.

Fig. 4 is an operation process of an author management center adopting an embodiment of the present disclosure, and by adding a semantic similarity model, the embodiment of the present disclosure adds a function of executing author operations according to input content, thereby reducing labor cost, improving use efficiency, and reducing learning cost, and an operator only needs to know operation types and concepts of an author management platform. The text cleaning tool and the rule engine belong to an author management center platform, namely the system of the disclosure, the semantic similarity model and model training are external calls, and the database belongs to the author management center.

The intelligent author management system based on the semantic similarity model is realized based on an author management center, wherein the author management center is constructed by adopting the following technology, 1. The back-end technology: java Spring Boot framework. Spring Boot is a fast developing framework, provides many ready-made components and libraries, and can quickly construct Web applications. It also provides a number of useful features such as auto-configuration, dependent injection, etc. 2. Front-end technology: vue.js framework. Vue.js is a popular JavaScript framework for building single page Web applications. It provides a number of useful features such as componentization, responsive data binding, etc. 3. Database: mySQL database. MySQL is a popular relational database that can store and manage data for applications. 4. The construction tool comprises: maven and npm. Maven is a Java project management tool for building and managing Java projects. npm is a JavaScript package manager for managing dependent items of the vue.js item. 5. Safety: spring Security framework. Spring Security is a powerful Security framework for protecting Web applications from various attacks. API design: RESTful API. The RESTful API is an API design style based on the HTTP protocol, which can make the application program more flexible, extensible and easy to maintain. 7. Unit test: JUnit and Vue Test Utils. JUnit is a Java unit test framework for testing Java code. The Vue Test Utils is a vue.js Test tool for testing vue.js components. 8. Deployment: docker and Kubernetes. Dock is a containerization platform that packages applications into containers and runs in different environments. Kubernetes is a container orchestration tool that can manage and expand containerized applications.

Another aspect of the disclosed embodiments provides an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the method as described above when executing the program.

Another aspect of the disclosed embodiments provides a computer readable storage medium having stored thereon a computer program which when executed by a processor implements a method as described above.

The foregoing is merely a preferred implementation of the embodiments of the disclosure, and it should be noted that, for a person skilled in the art, several improvements and modifications may be made without departing from the principles of the embodiments of the disclosure, which should also be considered as protective scope of the embodiments of the disclosure.

Claims

1. An author management method based on a network platform, the method comprising:

Receiving a request parameter, wherein the request parameter comprises at least one author ID and at least one text of a desired operation, and the text of the desired operation comprises at least one management operation statement;

receiving a rule matching result returned by the rule engine;

2. The method of claim 1, wherein determining whether all management operation sentences in each of the desired operation texts match local rules further comprises:

Screening all basic operation sentences with the similarity to each management operation sentence which is not matched with the upper rule being greater than or equal to a preset threshold value, selecting the basic operation sentence with the maximum similarity to the corresponding management operation sentence from all basic operation sentences, and respectively executing all management operation sentences which are matched with the upper rule and the selected basic operation sentences for all authors according to the author ID.

3. The method according to claim 1 or 2, wherein after the performing of all management operation sentences and the selected basic operation sentences of the matching upper rules for all authors, respectively, the method further comprises: returning an execution result and notifying an operator of the execution result;

4. A method according to claim 3, wherein after storing the record in the database, the method further comprises:

and returning the new semantic similarity model to the model trainer.

5. The method of claim 4, wherein after the returning the new semantic similarity model to the model trainer, the method further comprises:

Switching to a new semantic similarity model.

6. The method according to any one of claims 1-5, wherein the semantic similarity model is based on a method comprising: the BERT model is constructed and formed.

7. The method according to any one of claims 1 to 5, wherein after receiving the request parameter, the method further comprises:

Receiving cleaned expected operation text returned by the text cleaning tool;

8. An author management system based on a network platform, the system comprising:

9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1-7 when the program is executed by the processor.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any one of claims 1-7.