Intelligent demand processing method and device

Abstract

The embodiment of the invention relates to an intelligent demand processing method and device. The method comprises the following steps: selecting a large language model; constructing a priority prediction model and training the priority prediction model; updating a demand database based on the two types of models and the product feedback information of the customer; updating a demand database based on the two types of models and acquired batch product demand information; performing batch priority attenuation on unprocessed demands of the demand database regularly; the high-priority demand list is updated once when the demand database is updated once; when a demand list browsing instruction sent by a product manager is received, the high-priority demand list is sent back to the product manager; when a demand project configuration instruction sent by a product manager is received, the high-priority demand list, the demand database and the project configuration list are updated based on instruction data, demand change notification is carried out based on an updating result, and the high-priority demand list is sent back to the current product manager. The human cost can be reduced, and the analysis efficiency and the demand response speed can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an intelligent demand processing method and device. Background Art

[0002] The demand management work of general software development companies is usually completed manually by product managers of each product line. However, we have found through practice that this demand management method that relies entirely on manual processing is not suitable for IoT application developers. Most of the business products provided by IoT application developers require high business customization and high-frequency function iteration for different regions and different customers; in this case, the number and complexity of demands they face will inevitably be much greater than that of general software development companies. At this time, if they continue to use traditional manual demand management methods, it will not only incur high labor costs, but also lead to problems such as inefficient demand analysis and untimely demand response. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent demand processing method, device, electronic device and computer-readable storage medium in view of the defects of the prior art. The present invention pre-uses a type of public large language model (Large Language Model, LLM) as the first large language model and uses it to identify demand events, and builds an intelligent model for predicting demand priority according to the input demand feature vector based on a type of machine learning model as the first priority prediction model and trains it; then, every time a product feedback message sent by a customer is received, the first large language model and the first priority prediction model are used to identify the demand information of the product feedback information received at that time, and the demand database is updated in time based on the identification result; and after a batch of product demand information is collected through a type of public information collection channel at regular intervals, the first large language model and the first priority prediction model are used to identify the demand information of the product feedback information received at that time, and the demand database is updated in time based on the identification result; The test model identifies the demand information of the batch product demand information collected at that time and updates the demand database in time based on the identification results; and regularly performs batch priority decay processing on the unprocessed demands in the demand database; and updates the high-priority demand list every time the demand database is updated; and upon receiving the demand list browsing instruction sent by the product manager, pushes the latest high-priority demand list to it; and upon receiving the demand project configuration instruction sent by the product manager, updates the high-priority demand list, demand database and project configuration list based on the instruction data and notifies the demand changes in time based on the latest data, and pushes the latest high-priority demand list to the current product manager. Through the present invention, on the one hand, two types of artificial intelligence (AI) models can be used to achieve the purpose of improving demand analysis efficiency, improving demand analysis accuracy, improving demand response speed, and reducing labor costs; on the other hand, the priority decay mechanism can be used to achieve the purpose of dynamically optimizing demand priorities.

[0004] To achieve the above object, a first aspect of an embodiment of the present invention provides an intelligent demand processing method, the method comprising:

[0005] A class of public large language models is used as the corresponding first large language model; and based on a class of machine learning models, an intelligent model for predicting demand priority according to an input demand feature vector is constructed as the corresponding first priority prediction model; and the first priority prediction model is trained based on a preset first data set; the machine learning model includes at least a decision tree model and a random forest model;

[0006] Whenever a product feedback message is received from a customer, the preset demand database is updated based on the first language model, the first priority prediction model and the first product feedback message received at that time;

[0007] After collecting batches of product demand information through a type of public information collection channel at regular intervals, the demand database is updated based on the first language model, the first priority prediction model, and the first batch of product demand information collected at that time; the public information collection channel includes at least a market research channel and a social media information collection channel; the first batch of product demand information includes multiple first product demand information;

[0008] Regularly performing batch priority decay processing on the unprocessed demands in the demand database;

[0009] Whenever the demand database is updated, the preset high-priority demand list is updated based on the latest demand database;

[0010] Every time a first demand management instruction is received from a product manager, the corresponding first instruction type and first instruction data are extracted from the first demand management instruction received at that time; and the first instruction type is identified; if the first instruction type is a demand list browsing instruction, the latest high-priority demand list is sent back to the current product manager; if the first instruction type is a demand project configuration instruction, the high-priority demand list, the demand database and the preset project configuration list are updated based on the first instruction data, and demand change notifications are made based on the update results, and the latest high-priority demand list is sent back to the current product manager.

[0011] Preferably, the disclosed large language model at least includes GPT series models, BERT series models, BART series models, and Wenxin series models, and the disclosed large language model has completed model pre-training and customized fine-tuning; all natural language processing tasks supported by the disclosed large language model include demand event recognition tasks, and the natural language processing instruction template corresponding to the demand event recognition task is a demand event recognition instruction template; the demand event recognition instruction template is a formatting template; the demand event recognition instruction template is used to use the text information input by the template as the corresponding current target text, and inform the large language model through a narrative natural language to perform corresponding demand event extraction processing on the current target text, and require that each extracted demand event be assembled in the text format of a demand event binary consisting of a demand subject and a demand content, and require that all the obtained demand event binary tuples constitute the corresponding model to generate text output;

[0012] The first priority prediction model is used to predict the demand priority according to the input demand feature vector and output the corresponding priority prediction vector; the demand feature vector includes a demand importance level, a demand urgency level and a demand feasibility level; the priority prediction vector includes a plurality of priority prediction probabilities; each of the priority prediction probabilities corresponds to a priority level; each demand importance level, demand urgency level, demand feasibility level and priority level is a positive integer value level;

[0013] The first data set includes a plurality of first data records; the first data record includes a first training model vector and a first label vector; the first training model vector includes a first training importance level, a first training urgency level, and a first training feasibility level; the first label vector includes a plurality of first label prediction probabilities; each of the first label prediction probabilities corresponds to a priority level;

[0014] The demand database includes a plurality of first demand data records; the first demand data records include a first identification field, a first time field, a first source field, a first subject field, a first content field, a first importance field, a first urgency field, a first feasibility field, a first priority field and a first status field; the first source field includes at least customer, market research, and social media; the first status field includes processed and unprocessed;

[0015] The high priority requirement list includes a plurality of second requirement data records; the second requirement data records include a second identification field, a second subject field, a second content field, a second priority field, a first project identification field and a first development assessment field;

[0016] The project configuration list includes multiple first project configuration records; the first project configuration record includes a first project identification field, a first project name field, a first project cost field, a first project start time field, a first project end time field and a first project notification interface field; the first project notification interface field is composed of one or more first personnel notification interfaces; each of the first personnel notification interfaces corresponds to a project-related personnel; the interface types of the first personnel notification interfaces include at least an email notification interface, a SMS notification interface, a phone notification interface, an office system notification interface and an instant messaging platform notification interface; the instant messaging platform notification interface includes at least a WeChat platform notification interface and a DingTalk office platform notification interface;

[0017] The first requirement management instruction includes the first instruction type and the first instruction data; the first instruction type includes at least the requirement list browsing instruction and the requirement project configuration instruction; when the first instruction type is the requirement list browsing instruction, the corresponding first instruction data is empty; when the first instruction type is the requirement project configuration instruction, the corresponding first instruction data includes one or more first requirement configuration data, and the first requirement configuration data includes a first requirement identifier and a first project identifier.

[0018] Preferably, the training of the first priority prediction model based on a preset first data set specifically includes:

[0019] Step 31, based on a preset first segmentation ratio, the first data set is divided into two sub-data sets, which are recorded as a corresponding first training set and a first evaluation set;

[0020] Wherein, both the first training set and the first evaluation set are composed of a plurality of the first data records, and a ratio of the total number of records in the first training set to the total number of records in the first evaluation set satisfies the first segmentation ratio;

[0021] Step 32, initializing the first iteration counter to 0;

[0022] Step 33, perform a round of traversal on all the first data records of the first training set; and in this round of traversal, record the first data record currently traversed as the corresponding current training record; and input the first training model vector of the current training record into the first priority prediction model to perform demand priority prediction and use the prediction vector output by the model as the corresponding first prediction vector; and form a corresponding first prediction-label pair by the first prediction vector and the first label vector of the current training record; and at the end of this round of traversal, bring all the first prediction-label pairs obtained into the preset first model loss function; and perform a round of optimization on the model parameters of the first priority prediction model in the direction of minimizing the first model loss function;

[0023] Wherein, the first model loss function at least includes a regularized maximum likelihood function;

[0024] Step 34, adding 1 to the first iteration counter; and identifying whether the first iteration counter after adding 1 exceeds a preset first iteration number threshold; if so, go to step 35; if not, return to step 33;

[0025] Step 35, perform a round of traversal on all the first data records of the first evaluation set; and in this round of traversal, record the first data record currently traversed as the corresponding current evaluation record; and input the first training model vector of the current evaluation record into the first priority prediction model to perform demand priority prediction and use the prediction vector output by the model as the corresponding second prediction vector; and form a corresponding second prediction-label pair by the second prediction vector and the first label vector of the current training record; and at the end of this round of traversal, perform corresponding accuracy, precision, recall and F1 score calculations based on all the obtained second prediction-label pairs to obtain the corresponding first accuracy, first precision, first recall and first F1 score;

[0026] Step 36, identifying the first accuracy, the first precision, the first recall and the first F1 score; if the first accuracy does not meet the preset first accuracy range or the first accuracy does not meet the preset first accuracy range or the first recall does not meet the preset first recall range or the first F1 score does not meet the preset first F1 score range, returning to step 32 to continue training; if the first accuracy meets the first accuracy range and the first accuracy meets the first accuracy range and the first recall meets the first recall range and the first F1 score meets the first F1 score range, confirming that the model training is completed.

[0027] Preferably, updating the preset demand database based on the first language model, the first priority prediction model and the first product feedback information received at that time specifically includes:

[0028] Step 41, inputting the first product feedback information into the demand event recognition instruction template to assemble the instruction text to obtain the corresponding first instruction text; inputting the first instruction text into the first large language model to perform demand event recognition processing to obtain the corresponding first generated text; and extracting each demand event binary in the first generated text as the corresponding first demand event binary;

[0029] Wherein, the first demand event binary group includes a first demand subject and a first demand content;

[0030] Step 42, assigning a corresponding first demand source to each first demand event tuple, and setting each first demand source as a customer; and forming a corresponding first demand information record from each first demand event tuple and the corresponding first demand source; and forming a corresponding first demand information record set from all the obtained first demand information records;

[0031] Step 43, sending the first demand information record set to a preset demand three-property analysis interface; and receiving the first demand analysis record set sent back by the information three-property analysis interface;

[0032] The three-characteristics analysis interface of demand is a preset processing interface for performing a level analysis on the importance, urgency and feasibility of demand information; the level of each importance, urgency and feasibility returned by the three-characteristics analysis interface of demand is a positive integer value level;

[0033] The first demand analysis record set includes a plurality of first demand analysis records; the first demand analysis records correspond one to one with the first demand information records; the first demand analysis records include a first importance level, a first urgency level and a first feasibility level;

[0034] Step 44, the first importance level, the first urgency level and the first feasibility level of each of the first demand analysis records in the first demand analysis record set form a corresponding first demand feature vector; the first demand feature vector is input into the first priority prediction model to perform demand priority prediction to obtain a corresponding first priority prediction vector; and the priority level corresponding to the largest first priority prediction probability in the first priority prediction vector is used as the corresponding first priority level;

[0035] Wherein, the first priority prediction vector includes a plurality of the first priority prediction probabilities; each of the first priority prediction probabilities corresponds to a priority level;

[0036] Step 45, perform a round of traversal of all the first demand information records in the first demand information record set; and in this round of traversal, take the first demand information record currently traversed as the corresponding current demand information record, and take the first demand subject, the first demand content, the first demand source, the first importance level, the first urgency level, the first feasibility level and the first priority level corresponding to the current demand information record as the corresponding current demand subject, current demand content, current demand source, current importance level, current urgency level, current feasibility level and current priority level, and the current time as the corresponding current update time; and calculate the text similarity between the current demand subject and each of the first subject fields of the demand database to obtain the corresponding first subject similarity, and calculate the text similarity between the current demand content and each of the first content fields of the demand database to obtain the corresponding first content similarity, and compare the first subject similarity with the corresponding The product of the first content similarities is used as the corresponding first demand similarity; and the largest first demand similarity is used as the corresponding maximum similarity; and whether the maximum similarity exceeds a preset first similarity threshold is identified; if the maximum similarity exceeds the first similarity threshold, the first demand data record corresponding to the maximum similarity in the demand database is recorded as the corresponding current demand data record, and when the first status field of the current demand data record is unprocessed and the first priority field is lower than the current priority level, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field and the first priority field of the current demand data record are updated to the corresponding current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level and the current priority level;If the maximum similarity does not exceed the first similarity threshold, a unique identifier is assigned as the corresponding current demand identifier, and a corresponding current demand state is set to unprocessed, and a first demand data record is added to the demand database as the corresponding current demand data record, and the first identifier field, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field, the first priority field and the first status field of the current demand data record are set to the corresponding current demand identifier, the current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level, the current priority level and the current demand state. ;

[0037] Preferably, updating the demand database based on the first language model, the first priority prediction model and the first batch of product demand information collected at the time specifically includes:

[0038] Step 51, input each of the first product demand information of the first batch of product demand information into the demand event recognition instruction template to assemble the instruction text to obtain the corresponding second instruction text; input the second instruction text into the first language model to perform demand event recognition processing to obtain the corresponding second generated text; and extract each demand event binary group in the second generated text as the corresponding second demand event binary group;

[0039] Wherein, the second demand event tuple includes a second demand subject and a second demand content;

[0040] Step 52, assign a corresponding second demand source to each second demand event binary; and use the public information collection channel corresponding to the first batch of product demand information as the corresponding current collection channel; and identify the current collection channel; if the current collection channel is a market research channel, all second demand sources are set to market research; if the current collection channel is a social media information collection channel, all second demand sources are set to social media; and each second demand event binary and the corresponding second demand source form a corresponding second demand information record; and all the obtained second demand information records form a corresponding second demand information record set;

[0041] Step 53, sending the second demand information record set to the demand three-property analysis interface; and receiving the second demand analysis record set sent back by the information three-property analysis interface;

[0042] Wherein, the second demand analysis record set includes a plurality of second demand analysis records; the second demand analysis records correspond one-to-one to the second demand information records; the second demand analysis records include a second importance level, a second urgency level and a second feasibility level;

[0043] Step 54, the second importance level, the second urgency level and the second feasibility level of each second demand analysis record in the second demand analysis record set form a corresponding second demand feature vector; the second demand feature vector is input into the first priority prediction model to perform demand priority prediction to obtain a corresponding second priority prediction vector; and the priority level corresponding to the maximum second priority prediction probability in the second priority prediction vector is used as the corresponding second priority level;

[0044] The second priority prediction vector includes a plurality of second priority prediction probabilities; each second priority prediction probability corresponds to a priority level;

[0045] Step 55, perform a round of traversal of all the second demand information records in the second demand information record set; and in this round of traversal, take the second demand information record currently traversed as the corresponding current demand information record, and take the second demand subject, the second demand content, the second demand source, the second importance level, the second urgency level, the second feasibility level and the second priority level corresponding to the current demand information record as the corresponding current demand subject, current demand content, current demand source, current importance level, current urgency level, current feasibility level and current priority level, and the current time as the corresponding current update time; and calculate the text similarity between the current demand subject and each of the first subject fields of the demand database to obtain the corresponding second subject similarity, and calculate the text similarity between the current demand content and each of the first content fields of the demand database to obtain the corresponding second content similarity, and compare each of the second subject similarities with the corresponding The product of the second content similarities is used as the corresponding second demand similarity; and the largest second demand similarity is used as the corresponding maximum similarity; and whether the maximum similarity exceeds the first similarity threshold is identified; if the maximum similarity exceeds the first similarity threshold, the first demand data record corresponding to the maximum similarity in the demand database is recorded as the corresponding current demand data record, and when the first status field of the current demand data record is unprocessed and the first priority field is lower than the current priority level, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field and the first priority field of the current demand data record are updated to the corresponding current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level and the current priority level;If the maximum similarity does not exceed the first similarity threshold, a unique identifier is assigned as the corresponding current demand identifier, and a corresponding current demand state is set to unprocessed, and a first demand data record is added to the demand database as the corresponding current demand data record, and the first identifier field, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field, the first priority field and the first status field of the current demand data record are set to the corresponding current demand identifier, the current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level, the current priority level and the current demand state. ;

[0046] Preferably, the periodic batch priority attenuation processing of the unprocessed demands in the demand database specifically includes:

[0047] A round of traversal is performed regularly on all the first demand data records in the demand database whose first status fields are unprocessed according to the preset first time frequency; and in this round of traversal, the first demand data record currently traversed is used as the corresponding current record; and the time difference between the current time and the first time field of the current record is calculated and the calculation result is used as the corresponding first duration; and based on the first duration, a preset first correspondence table is queried, and the first priority attenuation factor field of the first correspondence record whose first duration range field in the first correspondence table satisfies the first duration is extracted as the corresponding current attenuation factor; and the first priority field of the current record is extracted as the corresponding pre-attenuation priority; and based on the current attenuation factor and the pre-attenuation priority, the corresponding post-attenuation priority is calculated = pre-attenuation priority - current attenuation factor; and the first time field and the first priority field of the current record are reset to the corresponding current time and the post-attenuation priority;

[0048] Among them, the first correspondence table is a data table used to reflect the correspondence between duration range and priority attenuation factor; the first correspondence table includes multiple first correspondence records; the first correspondence record includes the first duration range field and the first priority attenuation factor field; the first priority attenuation factor field is an integer value, the minimum value is 0, and the maximum value is the positive integer value corresponding to the maximum priority level.

[0049] Preferably, updating the preset high-priority demand list based on the latest demand database specifically includes:

[0050] Step 71, delete the second demand data record whose first item identification field is empty in the high priority demand list; and record the first demand data record whose first status field is unprocessed and whose first priority field is higher than the preset first priority threshold in the demand database as the corresponding first high priority record; and form a corresponding first high priority record set from all the obtained first high priority records;

[0051] Step 72, sending the first high priority record set to a preset demand development evaluation interface; and receiving the first demand evaluation record set sent back by the demand development evaluation interface;

[0052] The demand development evaluation interface is a preset processing interface for evaluating the development time and development cost of demand information;

[0053] The first demand assessment record set includes a plurality of first demand assessment records; the first demand assessment records correspond to the first high priority records one by one; the first demand assessment records include a first demand development time and a first demand development cost;

[0054] Step 73, perform a round of traversal of all the first high priority records in the first high priority record set; and in this round of traversal, take the first high priority record currently traversed as the corresponding current high priority record; and form a corresponding first development evaluation information by the first requirement development time and the first requirement development cost corresponding to the current high priority record; and add a second requirement data record to the high priority requirement list as the corresponding current newly added record; and set the second identification field, the second subject field, the second content field, the second priority field and the first development evaluation field of the currently newly added record to the first identification field, the first subject field, the first content field, the first priority field and the first development evaluation information corresponding to the current high priority record; and set the first project identification field of the currently newly added record to empty; and at the end of this round of traversal, sort all the second requirement data records in the high priority requirement list in descending order according to the second priority field.

[0055] Preferably, updating the high-priority requirement list, the requirement database, and the preset project configuration list based on the first instruction data and notifying the requirement change based on the update result specifically includes:

[0056] Extract one or more of the first demand configuration data from the first instruction data; perform a round of traversal on all the obtained first demand configuration data; and in the traversal process, use the first demand configuration data currently traversed as the corresponding current demand configuration data; use the first demand identifier and the first project identifier of the current demand configuration data as the corresponding current demand identifier and current project identifier; use the first demand data record whose first identifier field in the demand database matches the current demand identifier as the corresponding first record; use the second demand data record whose second identifier field in the high priority demand list matches the current demand identifier as the corresponding second record; use the first project configuration record whose first project identifier field in the project configuration list matches the current project identifier as the corresponding third record; update the first time field of the first record to the current time, and update the first status field of the first record to processed; and update the first project of the second record The identification field is updated to the corresponding current project identification, and the demand development duration and demand development cost information of the first development evaluation field of the second record are extracted as the corresponding current demand development duration and current demand development cost; and the first project cost field and the first project end time field of the third record are extracted as the corresponding first cost and first end time; and the corresponding second end time is calculated based on the current demand development duration and the first end time = the first end time + the current demand development duration; and the corresponding second cost is calculated based on the current demand development cost and the first cost = the first cost + the current demand development cost; and the first project cost field and the first project end time field of the third record are reset to the corresponding second cost and the second end time; and the reset first record and the third record form a corresponding first notification record; and the first notification record is sent to all the first personnel notification interfaces of the first project notification interface field of the third record.

[0057] The second aspect of the embodiment of the present invention provides a device for implementing the intelligent demand processing method described in the first aspect, the device comprising: an intelligent model preparation module, a first demand update module, a second demand update module, a demand priority attenuation module, a high priority demand update module, and a product manager interaction module;

[0058] The intelligent model preparation module is used to use a class of public large language models as the corresponding first large language model; and based on a class of machine learning models, an intelligent model for predicting demand priority according to the input demand feature vector is constructed as the corresponding first priority prediction model; and the first priority prediction model is trained based on a preset first data set; the machine learning model includes at least a decision tree model and a random forest model;

[0059] The first demand updating module is used to update the preset demand database based on the first language model, the first priority prediction model and the first product feedback information received each time a product feedback information sent by a customer is received;

[0060] The second demand update module is used to update the demand database once after collecting batch product demand information once through a type of public information collection channel at regular intervals, based on the first language model, the first priority prediction model and the first batch of product demand information collected at that time; the public information collection channel includes at least a market research channel and a social media information collection channel; the first batch of product demand information includes multiple first product demand information;

[0061] The demand priority decay module is used to periodically perform batch priority decay processing on the unprocessed demands in the demand database;

[0062] The high priority requirement updating module is used to update the preset high priority requirement list based on the latest requirement database every time the requirement database is updated;

[0063] The product manager interaction module is used to extract the corresponding first instruction type and first instruction data from the first demand management instruction received at that time each time a first demand management instruction sent by the product manager is received; and identify the first instruction type; if the first instruction type is a demand list browsing instruction, the latest high-priority demand list is sent back to the current product manager; if the first instruction type is a demand project configuration instruction, the high-priority demand list, the demand database and the preset project configuration list are updated based on the first instruction data, and demand change notification is made based on the update result, and the latest high-priority demand list is sent back to the current product manager.

[0064] A third aspect of an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a transceiver; the processor is used to couple with the memory, read and execute instructions in the memory, so as to implement the method steps described in the first aspect above;

[0065] The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

[0066] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions. When the computer instructions are executed by a computer, the computer executes the instructions of the method described in the first aspect above.

[0067] The embodiments of the present invention provide a method, device, electronic device and computer-readable storage medium for intelligent demand processing. As can be seen from the above content, the embodiments of the present invention pre-use a type of public large language model as the first large language model and use it to identify demand events, and build an intelligent model for predicting demand priority according to the input demand feature vector based on a type of machine learning model as the first priority prediction model and train it; then, every time a product feedback message sent by a customer is received, the first large language model and the first priority prediction model are used to identify the demand information of the product feedback information received at that time and the demand database is updated in time based on the identification result; and after a batch of product demand information is collected through a type of public information collection channel at regular intervals, the first large language model and the first priority prediction model are used to identify the demand information of the product feedback information received at that time and the demand database is updated in time based on the identification result. The priority prediction model identifies the demand information of the batch product demand information collected at that time and updates the demand database in time based on the identification results; and regularly performs batch priority decay processing on the unprocessed demands in the demand database; and updates the high-priority demand list every time the demand database is updated; and upon receiving the demand list browsing instruction sent by the product manager, pushes the latest high-priority demand list to it; and upon receiving the demand project configuration instruction sent by the product manager, updates the high-priority demand list, demand database and project configuration list based on the instruction data and notifies the demand changes in time based on the latest data, and pushes the latest high-priority demand list to the current product manager. On the one hand, the embodiment of the present invention utilizes two types of artificial intelligence models to improve the efficiency of demand analysis, improve the accuracy of demand analysis, improve the speed of demand response, and reduce labor costs; on the other hand, it utilizes the priority decay mechanism to realize the dynamic optimization function of demand priority. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A schematic diagram of an intelligent demand processing method provided in Embodiment 1 of the present invention;

[0069] Figure 2 A module structure diagram of an intelligent demand processing device provided in Embodiment 2 of the present invention;

[0070] Figure 3 A schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all 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.

[0072] Embodiment 1 of the present invention provides an intelligent demand processing method, such as Figure 1 As shown in the schematic diagram of an intelligent demand processing method provided in the first embodiment of the present invention, the method mainly includes the following steps:

[0073] Step 1: taking a class of public large language models as the corresponding first large language model; and constructing an intelligent model for predicting demand priority according to an input demand feature vector based on a class of machine learning models as the corresponding first priority prediction model; and training the first priority prediction model based on a preset first data set;

[0074] Specifically, it includes: step 11, taking a class of public large language models as the corresponding first large language model;

[0075] Here, the disclosed large language model mentioned in the embodiment of the present invention should at least include GPT series models, BERT series models, BART series models, Wenxin series models, etc.; this disclosed large language model requires that model pre-training and customized fine tuning have been completed; all natural language processing tasks supported by this disclosed large language model should include event recognition / event extraction (EE) tasks, and the required event recognition task is a type of conventional EE task;

[0076] The natural language processing instruction template corresponding to the demand event recognition task of the embodiment of the present invention is called a demand event recognition instruction template; this demand event recognition instruction template is a formatting template; the demand event recognition instruction template is used to use the text information input by the template as the corresponding current target text, and inform the large language model through a narrative natural language to perform corresponding demand event extraction processing on the current target text, and requires that each extracted demand event be assembled in the text format of a demand event binary composed of a demand subject and a demand content, and requires that the corresponding model generated text output is composed of all the obtained demand event binary tuples;

[0077] For example, the demand event identification instruction template is:

[0078] "Current target text = <text information entered in the template>;

[0079] Please extract demand events from the current target text, with the following requirements: 1) assemble each extracted demand event into a text format of a demand event tuple consisting of a demand subject and a demand content; 2) generate text output from the corresponding model composed of all the demand event tuples obtained;

[0080] Step 12, based on a class of machine learning models, an intelligent model for predicting demand priority according to the input demand feature vector is constructed as the corresponding first priority prediction model;

[0081] Here, the machine learning model mentioned in the embodiment of the present invention includes at least a decision tree model and a random forest model; the first priority prediction model of the embodiment of the present invention is used to predict the demand priority according to the input demand feature vector and output the corresponding priority prediction vector; wherein the demand feature vector includes the demand importance level, the demand urgency level and the demand feasibility level; the priority prediction vector includes multiple priority prediction probabilities; each priority prediction probability corresponds to a priority level; it should be noted that each demand importance level, demand urgency level, demand feasibility level and priority level here is a positive integer value level;

[0082] Step 13, training a first priority prediction model based on a preset first data set;

[0083] Wherein, the first data set includes a plurality of first data records; the first data record includes a first training model vector and a first label vector; the first training model vector includes a first training importance level, a first training urgency level, and a first training feasibility level; the first label vector includes a plurality of first label prediction probabilities; each first label prediction probability corresponds to a priority level;

[0084] Specifically, it includes: step 131, dividing the first data set into two sub-data sets based on a preset first division ratio and recording them as a corresponding first training set and a first evaluation set;

[0085] The first segmentation ratio is a preset ratio parameter, such as 8:2; the first training set and the first evaluation set are both composed of a plurality of first data records, and the ratio of the total number of records in the first training set to the total number of records in the first evaluation set satisfies the first segmentation ratio;

[0086] Step 132, initializing the first iteration counter to 0;

[0087] Step 133, perform a round of traversal on all first data records of the first training set; and in this round of traversal, record the first data record currently traversed as the corresponding current training record; and input the first training model vector of the current training record into the first priority prediction model to perform demand priority prediction and use the prediction vector output by the model as the corresponding first prediction vector; and form a corresponding first prediction-label pair by the first prediction vector and the first label vector of the current training record; and at the end of this round of traversal, bring all the obtained first prediction-label pairs into the preset first model loss function; and perform a round of optimization on the model parameters of the first priority prediction model in the direction of minimizing the first model loss function;

[0088] Wherein, the first model loss function at least includes a regularized maximum likelihood function;

[0089] Step 134, adding 1 to the first iteration counter; and identifying whether the first iteration counter after adding 1 exceeds the preset first iteration number threshold; if so, go to step 135; if not, return to step 133;

[0090] Step 135, perform a round of traversal on all first data records of the first evaluation set; and in this round of traversal, record the first data record currently traversed as the corresponding current evaluation record; and input the first training model vector of the current evaluation record into the first priority prediction model to perform demand priority prediction and use the prediction vector output by the model as the corresponding second prediction vector; and form a corresponding second prediction-label pair by the second prediction vector and the first label vector of the current training record; and at the end of this round of traversal, perform corresponding accuracy, precision, recall and F1 score calculations based on all the obtained second prediction-label pairs to obtain the corresponding first accuracy, first precision, first recall and first F1 score;

[0091] Step 136, identifying the first accuracy, the first precision, the first recall rate and the first F1 score; if the first accuracy does not satisfy the preset first accuracy range or the first accuracy does not satisfy the preset first accuracy range or the first recall rate does not satisfy the preset first recall rate range or the first F1 score does not satisfy the preset first F1 score range, returning to step 132 to continue training; if the first accuracy satisfies the first accuracy range and the first accuracy satisfies the first accuracy range and the first recall rate satisfies the first recall rate range and the first F1 score satisfies the first F1 score range, confirming that the model training is completed.

[0092] Here, the first accuracy range, the first precision range, the first recall range and the first F1 score range of the embodiment of the present invention are four preset value ranges.

[0093] Step 2: when each product feedback message sent by a customer is received, the preset demand database is updated based on the first language model, the first priority prediction model and the first product feedback message received at that time;

[0094] The demand database includes a plurality of first demand data records; the first demand data record includes a first identification field, a first time field, a first source field, a first subject field, a first content field, a first importance field, a first urgency field, a first feasibility field, a first priority field and a first status field; the first source field includes at least customer, market research, and social media; the first status field includes processed and unprocessed;

[0095] Specifically, it includes: step 21, inputting the first product feedback information into the demand event recognition instruction template to assemble the instruction text to obtain the corresponding first instruction text; inputting the first instruction text into the first large language model to perform demand event recognition processing to obtain the corresponding first generated text; and extracting each demand event binary in the first generated text as the corresponding first demand event binary;

[0096] The first demand event tuple includes the first demand subject and the first demand content;

[0097] Step 22, assigning a corresponding first demand source to each first demand event tuple, and setting each first demand source as a customer; and forming a corresponding first demand information record from each first demand event tuple and the corresponding first demand source; and forming a corresponding first demand information record set from all the obtained first demand information records;

[0098] Step 23, sending the first demand information record set to a preset demand three-property analysis interface; and receiving the first demand analysis record set sent back by the information three-property analysis interface;

[0099] Here, the demand three-characteristics analysis interface of the embodiment of the present invention is a preset processing interface for performing grade analysis on the importance, urgency and feasibility of demand information; the processing structure can be a manual processing interface or a machine processing interface; the importance, urgency and feasibility grades returned by the demand three-characteristics analysis interface are all positive integer values;

[0100] The first requirement analysis record set returned by the requirement three-property analysis interface includes a plurality of first requirement analysis records; wherein the first requirement analysis records correspond to the first requirement information records one by one; the first requirement analysis records include a first importance level, a first urgency level, and a first feasibility level;

[0101] Step 24, the first importance level, the first urgency level and the first feasibility level of each first demand analysis record of the first demand analysis record set are used to form a corresponding first demand feature vector; the first demand feature vector is input into the first priority prediction model to perform demand priority prediction to obtain a corresponding first priority prediction vector; and the priority level corresponding to the largest first priority prediction probability in the first priority prediction vector is used as the corresponding first priority level;

[0102] Wherein, the first priority prediction vector includes a plurality of first priority prediction probabilities; each first priority prediction probability corresponds to a priority level;

[0103] Step 25, perform a round of traversal of all first demand information records in the first demand information record set; and in this round of traversal, take the currently traversed first demand information record as the corresponding current demand information record, and take the first demand subject, first demand content, first demand source, first importance level, first urgency level, first feasibility level and first priority level corresponding to the current demand information record as the corresponding current demand subject, current demand content, current demand source, current importance level, current urgency level, current feasibility level and current priority level, and the current time as the corresponding current update time; and The text similarity with each first subject field of the demand database is calculated to obtain the corresponding first subject similarity, and the text similarity between the current demand content and each first content field of the demand database is calculated to obtain the corresponding first content similarity, and the product of each first subject similarity and its corresponding first content similarity is used as the corresponding first demand similarity; and the maximum first demand similarity is used as the corresponding maximum similarity; and whether the maximum similarity exceeds the preset first similarity threshold is identified; if the maximum similarity exceeds the first similarity threshold, the first demand data record corresponding to the maximum similarity in the demand database is recorded as the corresponding The current demand data record, and when the first status field of the current demand data record is unprocessed and the first priority field is lower than the current priority level, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field and the first priority field of the current demand data record are updated to the corresponding current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level and the current priority level; if the maximum similarity does not exceed the first similarity threshold, a unique identifier is assigned as the corresponding current demand identifier, and a corresponding current demand status is set to unprocessed, and a first demand data record is added to the demand database as the corresponding current demand data record, and the first identifier field, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field, the first priority field and the first status field of the current demand data record are set to the corresponding current demand identifier, the current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level, the current priority level and the current demand status.

[0104] Here, the first similarity threshold is a preset threshold parameter.

[0105] Step 3, after collecting batch product demand information through a public information collection channel at regular intervals, update the demand database based on the first language model, the first priority prediction model and the first batch of product demand information collected at that time;

[0106] Among them, the public information collection channels include at least market research channels and social media information collection channels; the first batch of product demand information includes multiple first product demand information;

[0107] Specifically, it includes: step 31, inputting each first product demand information of the first batch of product demand information into the demand event recognition instruction template to assemble the instruction text to obtain the corresponding second instruction text; inputting the second instruction text into the first language model to perform demand event recognition processing to obtain the corresponding second generated text; and extracting each demand event binary group in the second generated text as the corresponding second demand event binary group;

[0108] The second demand event dyad includes the second demand subject and the second demand content;

[0109] Step 32, assign a corresponding second demand source to each second demand event binary; and use the public information collection channel corresponding to the first batch of product demand information as the corresponding current collection channel; and identify the current collection channel; if the current collection channel is a market research channel, all second demand sources are set to market research; if the current collection channel is a social media information collection channel, all second demand sources are set to social media; and each second demand event binary and the corresponding second demand source form a corresponding second demand information record; and all the obtained second demand information records form a corresponding second demand information record set;

[0110] Step 33, sending the second demand information record set to the demand three-property analysis interface; and receiving the second demand analysis record set sent back by the information three-property analysis interface;

[0111] Wherein, the second demand analysis record set includes a plurality of second demand analysis records; the second demand analysis records correspond one to one with the second demand information records; the second demand analysis records include a second importance level, a second urgency level and a second feasibility level;

[0112] Step 34, the second importance level, the second urgency level and the second feasibility level of each second demand analysis record of the second demand analysis record set are used to form a corresponding second demand feature vector; the second demand feature vector is input into the first priority prediction model to perform demand priority prediction to obtain a corresponding second priority prediction vector; and the priority level corresponding to the largest second priority prediction probability in the second priority prediction vector is used as the corresponding second priority level;

[0113] Wherein, the second priority prediction vector includes a plurality of second priority prediction probabilities; each second priority prediction probability corresponds to a priority level;

[0114] Step 35, perform a round of traversal of all second demand information records in the second demand information record set; and in this round of traversal, take the currently traversed second demand information record as the corresponding current demand information record, and take the second demand subject, second demand content, second demand source, second importance level, second urgency level, second feasibility level and second priority level corresponding to the current demand information record as the corresponding current demand subject, current demand content, current demand source, current importance level, current urgency level, current feasibility level and current priority level, and the current time as the corresponding current update time; and The text similarity between the topic and each first topic field of the demand database is calculated to obtain the corresponding second topic similarity, and the text similarity between the current demand content and each first content field of the demand database is calculated to obtain the corresponding second content similarity, and the product of each second topic similarity and its corresponding second content similarity is used as the corresponding second demand similarity; and the maximum second demand similarity is used as the corresponding maximum similarity; and whether the maximum similarity exceeds the first similarity threshold is identified; if the maximum similarity exceeds the first similarity threshold, the first demand data record corresponding to the maximum similarity in the demand database is recorded as the corresponding current The previous demand data record is processed, and when the first status field of the current demand data record is unprocessed and the first priority field is lower than the current priority level, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field and the first priority field of the current demand data record are updated to the corresponding current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level and the current priority level; if the maximum similarity does not exceed the first similarity threshold, a unique identifier is assigned as the corresponding current demand identifier, and a corresponding current demand status is set to unprocessed, and a first demand data record is added to the demand database as the corresponding current demand data record, and the first identification field, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field, the first priority field and the first status field of the current demand data record are set to the corresponding current demand identifier, the current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level, the current priority level and the current demand status.

[0115] Step 4, regularly perform batch priority decay processing on the unprocessed demands in the demand database;

[0116] Specifically, it includes: regularly traversing all the first demand data records in the demand database whose first status fields are unprocessed according to the preset first time frequency; and in this round of traversal, taking the first demand data record currently traversed as the corresponding current record; and calculating the time difference between the current time and the first time field of the current record and taking the calculation result as the corresponding first duration; and querying the preset first correspondence table based on the first duration, extracting the first priority attenuation factor field of the first correspondence record whose first duration range field in the first correspondence table satisfies the first duration as the corresponding current attenuation factor; and extracting the first priority field of the current record as the corresponding pre-attenuation priority; and calculating the corresponding post-attenuation priority based on the current attenuation factor and the pre-attenuation priority = pre-attenuation priority - current attenuation factor; and resetting the first time field and the first priority field of the current record to the corresponding current time and post-attenuation priority;

[0117] Among them, the first time frequency is a pre-set time frequency; the first correspondence table is a data table used to reflect the correspondence between time range and priority attenuation factor; the first correspondence table includes multiple first correspondence records; the first correspondence record includes a first time range field and a first priority attenuation factor field; the first priority attenuation factor field is an integer value, the minimum value is 0, and the maximum value is a positive integer value corresponding to the maximum priority level.

[0118] Step 5: Whenever the demand database is updated, the preset high-priority demand list is updated based on the latest demand database;

[0119] The high-priority requirement list includes a plurality of second requirement data records; the second requirement data record includes a second identification field, a second subject field, a second content field, a second priority field, a first project identification field, and a first development assessment field;

[0120] Specifically, it includes: step 51, deleting the second demand data record whose first item identification field is empty in the high priority demand list; and recording the first demand data record whose first status field in the demand database is unprocessed and whose first priority field is higher than the preset first priority threshold as the corresponding first high priority record; and forming the corresponding first high priority record set from all the obtained first high priority records;

[0121] Here, the first priority threshold is a preset threshold parameter;

[0122] Step 52, sending the first high priority record set to a preset demand development evaluation interface; and receiving the first demand evaluation record set sent back by the demand development evaluation interface;

[0123] Here, the demand development evaluation interface of the embodiment of the present invention is a preset processing interface for evaluating the development time and development cost of demand information; the processing interface can be a manual processing interface or a machine processing interface;

[0124] The first demand assessment record set returned by the demand development assessment interface includes a plurality of first demand assessment records; wherein the first demand assessment records correspond to the first high priority records one by one; the first demand assessment records include the first demand development time and the first demand development cost;

[0125] Step 53, perform a round of traversal of all first high priority records in the first high priority record set; and in this round of traversal, use the currently traversed first high priority record as the corresponding current high priority record; and form a corresponding first development evaluation information from the first requirement development time and the first requirement development cost corresponding to the current high priority record; and add a second requirement data record to the high priority requirement list as the corresponding current newly added record; and set the second identification field, second subject field, second content field, second priority field and first development evaluation field of the currently newly added record to the first identification field, first subject field, first content field, first priority field and first development evaluation information corresponding to the current high priority record; and set the first project identification field of the currently newly added record to empty; and at the end of this round of traversal, sort all second requirement data records in the high priority requirement list in descending order according to the second priority field.

[0126] Step 6, when each first demand management instruction sent by the product manager is received, the corresponding first instruction type and first instruction data are extracted from the first demand management instruction received at that time; and the first instruction type is identified; if the first instruction type is a demand list browsing instruction, the latest high-priority demand list is sent back to the current product manager; if the first instruction type is a demand project configuration instruction, the high-priority demand list, the demand database and the preset project configuration list are updated based on the first instruction data, and a demand change notification is issued based on the update result, and the latest high-priority demand list is sent back to the current product manager;

[0127] Specifically, it includes: step 61, when each first demand management instruction sent by a product manager is received, extracting the corresponding first instruction type and first instruction data from the first demand management instruction received at that time;

[0128] Here, the first demand management instruction of the embodiment of the present invention includes a first instruction type and a first instruction data; wherein the first instruction type includes at least a demand list browsing instruction and a demand project configuration instruction; when the first instruction type is a demand list browsing instruction, the corresponding first instruction data is empty; when the first instruction type is a demand project configuration instruction, the corresponding first instruction data includes one or more first demand configuration data, and the first demand configuration data includes a first demand identifier and a first project identifier;

[0129] Step 62, identifying the first instruction type;

[0130] Step 63: If the first instruction type is a requirement list browsing instruction, the latest high-priority requirement list is sent back to the current product manager;

[0131] Step 64: if the first instruction type is a requirement project configuration instruction, then based on the first instruction data, the high-priority requirement list, the requirement database, and the preset project configuration list are updated, and a requirement change notification is issued based on the update result, and the latest high-priority requirement list is sent back to the current product manager;

[0132] Among them, the project configuration list includes multiple first project configuration records; the first project configuration record includes a first project identification field, a first project name field, a first project cost field, a first project start time field, a first project end time field and a first project notification interface field; the first project notification interface field is composed of one or more first personnel notification interfaces; each first personnel notification interface corresponds to a project-related personnel; the interface types of the first personnel notification interface include at least an email notification interface, a SMS notification interface, a phone notification interface, an office system notification interface and an instant messaging platform notification interface; the instant messaging platform notification interface includes at least a WeChat platform notification interface and a DingTalk office platform notification interface;

[0133] In the current step 64, the specific processing steps of updating the high-priority requirement list, the requirement database and the preset project configuration list based on the first instruction data and notifying the requirement change based on the update result include: extracting one or more first requirement configuration data from the first instruction data; and performing a round of traversal of all the obtained first requirement configuration data; and in the traversal process, taking the currently traversed first requirement configuration data as the corresponding current requirement configuration data; and taking the first requirement identifier and the first project identifier of the current requirement configuration data as the corresponding current requirement identifier and the current project identifier; and taking the first requirement data record whose first identifier field in the requirement database matches the current requirement identifier as the corresponding first record; and taking the second requirement data record whose second identifier field in the high-priority requirement list matches the current requirement identifier as the corresponding second record; and taking the first project configuration record whose first project identifier field in the project configuration list matches the current project identifier as the corresponding third record; and updating the first time field of the first record to the current time , and update the first status field of the first record to processed; and update the first project identification field of the second record to the corresponding current project identification, and extract the demand development time and demand development cost information of the first development evaluation field of the second record as the corresponding current demand development time and current demand development cost; and extract the first project cost field and the first project end time field of the third record as the corresponding first cost and first end time; and calculate the corresponding second end time = first end time + current demand development time based on the current demand development time and the first end time; and calculate the corresponding second cost = first cost + current demand development cost based on the current demand development cost and the first cost; and reset the first project cost field and the first project end time field of the third record to the corresponding second cost and second end time; and form a corresponding first notification record by the reset first record and the third record; and send the first notification record to all first personnel notification interfaces of the first project notification interface field of the third record.

[0134] Figure 2 This is a module structure diagram of an intelligent demand processing device provided in the second embodiment of the present invention. The device is a terminal device or server that implements the aforementioned method embodiment, and can also be a device that enables the aforementioned terminal device or server to implement the aforementioned method embodiment. For example, the device can be a device or chip system of the aforementioned terminal device or server. Figure 2 As shown, the device includes: an intelligent model preparation module 201, a first demand update module 202, a second demand update module 203, a demand priority decay module 204, a high priority demand update module 205 and a product manager interaction module 206.

[0135] The intelligent model preparation module 201 is used to use a class of public large language models as the corresponding first large language model; and based on a class of machine learning models, build an intelligent model for demand priority prediction according to the input demand feature vector as the corresponding first priority prediction model; and train the first priority prediction model based on a preset first data set; the machine learning model includes at least a decision tree model and a random forest model.

[0136] The first demand updating module 202 is used to update the preset demand database based on the first language model, the first priority prediction model and the first product feedback information received each time a product feedback information sent by a customer is received.

[0137] The second demand update module 203 is used to update the demand database based on the first language model, the first priority prediction model and the first batch of product demand information collected at that time after collecting batch product demand information through a type of public information collection channel at regular intervals; the public information collection channel includes at least market research channels and social media information collection channels; the first batch of product demand information includes multiple first product demand information.

[0138] The demand priority decay module 204 is used to periodically perform batch priority decay processing on the unprocessed demands in the demand database.

[0139] The high priority requirement updating module 205 is used to update the preset high priority requirement list based on the latest requirement database every time the requirement database is updated.

[0140] The product manager interaction module 206 is used to extract the corresponding first instruction type and first instruction data from the first demand management instruction received each time a first demand management instruction sent by the product manager is received; and identify the first instruction type; if the first instruction type is a demand list browsing instruction, the latest high-priority demand list is sent back to the current product manager; if the first instruction type is a demand project configuration instruction, the high-priority demand list, the demand database and the preset project configuration list are updated based on the first instruction data, and demand change notifications are made based on the update results, and the latest high-priority demand list is sent back to the current product manager.

[0141] An intelligent demand processing device provided in an embodiment of the present invention can execute the method steps in the above method embodiment. Its implementation principle and technical effects are similar and will not be repeated here.

[0142] It should be noted that it should be understood that the division of the various modules of the above device is only a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or they can be physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be all implemented in the form of hardware; some modules can also be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. For example, the intelligent model preparation module can be a separately established processing element, or it can be integrated in a chip of the above device. In addition, it can also be stored in the memory of the above device in the form of program code, and called and executed by a processing element of the above device. The function of the above-mentioned module is determined. The implementation of other modules is similar. In addition, all or part of these modules can be integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each module above can be completed by an integrated logic circuit of hardware in the processor element or instructions in the form of software.

[0143] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASIC), or one or more digital signal processors (DSP), or one or more field programmable gate arrays (FPGA). For another example, when a module above is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0144] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the above method embodiments are generated. The above computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The above-mentioned computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the above-mentioned computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, Bluetooth, microwave, etc.) methods. The above-mentioned computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The above-mentioned available medium can be a magnetic medium (such as a floppy disk, a hard disk, a tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state disk (SSD)), etc.

[0145] Figure 3 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. The electronic device may be a terminal device or a server for implementing the method of the aforementioned embodiment, or may be a terminal device or a server for implementing the method of the aforementioned embodiment connected to the aforementioned terminal device or server. Figure 3 As shown, the electronic device may include: a processor 301 (such as a CPU), a memory 302, and a transceiver 303; the transceiver 303 is coupled to the processor 301, and the processor 301 controls the transceiver 303. Various instructions may be stored in the memory 302 to complete various processing functions and implement the processing steps described in the aforementioned embodiment method. Preferably, the electronic device involved in the embodiment of the present invention also includes: a power supply 304, a system bus 305 and a communication port 306. The system bus 305 is used to realize the communication connection between components. The above-mentioned communication port 306 is used for connecting and communicating between the electronic device and other peripherals.

[0146] exist Figure 3The system bus 305 mentioned in the figure can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface is used to realize the communication between the database access device and other devices (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM), and may also include non-volatile memory (Non-Volatile Memory), such as at least one disk storage.

[0147] The above-mentioned processor can be a general-purpose processor, including a central processing unit CPU, a network processor (Network Processor, NP), a graphics processing unit (Graphics Processing Unit, GPU), etc.; it can also be a digital signal processor DSP, an application-specific integrated circuit ASIC, a field programmable gate array FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0148] It should be noted that an embodiment of the present invention further provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the method and processing process provided in the above embodiments.

[0149] The embodiments of the present invention provide a method, device, electronic device and computer-readable storage medium for intelligent demand processing. As can be seen from the above content, the embodiments of the present invention pre-use a type of public large language model as the first large language model and use it to identify demand events, and build an intelligent model for predicting demand priority according to the input demand feature vector based on a type of machine learning model as the first priority prediction model and train it; then, every time a product feedback message sent by a customer is received, the first large language model and the first priority prediction model are used to identify the demand information of the product feedback information received at that time and the demand database is updated in time based on the identification result; and after a batch of product demand information is collected through a type of public information collection channel at regular intervals, the first large language model and the first priority prediction model are used to identify the demand information of the product feedback information received at that time and the demand database is updated in time based on the identification result. The priority prediction model identifies the demand information of the batch product demand information collected at that time and updates the demand database in time based on the identification results; and regularly performs batch priority decay processing on the unprocessed demands in the demand database; and updates the high-priority demand list every time the demand database is updated; and upon receiving the demand list browsing instruction sent by the product manager, pushes the latest high-priority demand list to it; and upon receiving the demand project configuration instruction sent by the product manager, updates the high-priority demand list, demand database and project configuration list based on the instruction data and notifies the demand changes in time based on the latest data, and pushes the latest high-priority demand list to the current product manager. On the one hand, the embodiment of the present invention utilizes two types of artificial intelligence models to improve the efficiency of demand analysis, improve the accuracy of demand analysis, improve the speed of demand response, and reduce labor costs; on the other hand, it utilizes the priority decay mechanism to realize the dynamic optimization function of demand priority.

[0150] The professionals should further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by 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 terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can 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.

[0151] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented 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.

[0152] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent demand processing method, characterized in that: The method comprises: A class of public large language models is used as the corresponding first large language model; and based on a class of machine learning models, an intelligent model for predicting demand priority according to an input demand feature vector is constructed as the corresponding first priority prediction model; and the first priority prediction model is trained based on a preset first data set; the machine learning model includes at least a decision tree model and a random forest model; Whenever a product feedback message is received from a customer, the preset demand database is updated based on the first language model, the first priority prediction model and the first product feedback message received at that time; After collecting batches of product demand information through a type of public information collection channel at regular intervals, the demand database is updated based on the first language model, the first priority prediction model, and the first batch of product demand information collected at that time; the public information collection channel includes at least a market research channel and a social media information collection channel; the first batch of product demand information includes multiple first product demand information; Regularly performing batch priority decay processing on the unprocessed demands in the demand database; Whenever the demand database is updated, the preset high-priority demand list is updated based on the latest demand database; Every time a first demand management instruction is received from a product manager, the corresponding first instruction type and first instruction data are extracted from the first demand management instruction received at that time; and the first instruction type is identified; if the first instruction type is a demand list browsing instruction, the latest high-priority demand list is sent back to the current product manager; if the first instruction type is a demand project configuration instruction, the high-priority demand list, the demand database and the preset project configuration list are updated based on the first instruction data, and demand change notifications are made based on the update results, and the latest high-priority demand list is sent back to the current product manager.

2. The intelligent demand processing method according to claim 1, characterized in that: The disclosed large language model at least includes GPT series models, BERT series models, BART series models, and Wenxin series models. The disclosed large language model has completed model pre-training and customized fine-tuning; all natural language processing tasks supported by the disclosed large language model include demand event recognition tasks, and the natural language processing instruction template corresponding to the demand event recognition task is a demand event recognition instruction template; The demand event identification instruction template is a formatting template; The demand event identification instruction template is used to use the text information input by the template as the corresponding current target text, and inform the large language model through a narrative natural language to perform corresponding demand event extraction processing on the current target text, and requires that each extracted demand event be assembled in the text format of a demand event binary composed of a demand subject and a demand content, and requires that the corresponding model generated text output is composed of all the obtained demand event binary tuples; The first priority prediction model is used to predict the demand priority according to the input demand feature vector and output the corresponding priority prediction vector; the demand feature vector includes a demand importance level, a demand urgency level and a demand feasibility level; the priority prediction vector includes a plurality of priority prediction probabilities; each of the priority prediction probabilities corresponds to a priority level; each demand importance level, demand urgency level, demand feasibility level and priority level is a positive integer value level; The first data set includes a plurality of first data records; the first data record includes a first training model vector and a first label vector; the first training model vector includes a first training importance level, a first training urgency level, and a first training feasibility level; the first label vector includes a plurality of first label prediction probabilities; each of the first label prediction probabilities corresponds to a priority level; The demand database includes a plurality of first demand data records; the first demand data records include a first identification field, a first time field, a first source field, a first subject field, a first content field, a first importance field, a first urgency field, a first feasibility field, a first priority field and a first status field; the first source field includes at least customer, market research, and social media; the first status field includes processed and unprocessed; The high priority requirement list includes a plurality of second requirement data records; the second requirement data records include a second identification field, a second subject field, a second content field, a second priority field, a first project identification field and a first development assessment field; The project configuration list includes a plurality of first project configuration records; the first project configuration record includes a first project identification field, a first project name field, a first project cost field, a first project start time field, a first project end time field and a first project notification interface field; the first project notification interface field is composed of one or more first personnel notification interfaces; each of the first personnel notification interfaces corresponds to a project-related personnel; the interface types of the first personnel notification interfaces include at least an email notification interface, a text message notification interface, a telephone notification interface, an office system notification interface and an instant messaging platform notification interface; The instant messaging platform notification interface at least includes a WeChat platform notification interface and a DingTalk office platform notification interface; The first requirement management instruction includes the first instruction type and the first instruction data; the first instruction type includes at least the requirement list browsing instruction and the requirement project configuration instruction; when the first instruction type is the requirement list browsing instruction, the corresponding first instruction data is empty; when the first instruction type is the requirement project configuration instruction, the corresponding first instruction data includes one or more first requirement configuration data, and the first requirement configuration data includes a first requirement identifier and a first project identifier.

3. The intelligent demand processing method according to claim 2, characterized in that: The training of the first priority prediction model based on the preset first data set specifically includes: Step 31, based on a preset first segmentation ratio, the first data set is divided into two sub-data sets which are recorded as a corresponding first training set and a first evaluation set; Wherein, both the first training set and the first evaluation set are composed of a plurality of the first data records, and a ratio of the total number of records in the first training set to the total number of records in the first evaluation set satisfies the first segmentation ratio; Step 32, initializing the first iteration counter to 0; Step 33, perform a round of traversal on all the first data records of the first training set; and in this round of traversal, record the first data record currently traversed as the corresponding current training record; and input the first training model vector of the current training record into the first priority prediction model to perform demand priority prediction and use the prediction vector output by the model as the corresponding first prediction vector; and form a corresponding first prediction-label pair by the first prediction vector and the first label vector of the current training record; and at the end of this round of traversal, bring all the first prediction-label pairs obtained into the preset first model loss function; and perform a round of optimization on the model parameters of the first priority prediction model in the direction of minimizing the first model loss function; Wherein, the first model loss function at least includes a regularized maximum likelihood function; Step 34, adding 1 to the first iteration counter; and identifying whether the first iteration counter after adding 1 exceeds a preset first iteration number threshold; if so, go to step 35; if not, return to step 33; Step 35, perform a round of traversal on all the first data records of the first evaluation set; and in this round of traversal, record the first data record currently traversed as the corresponding current evaluation record; and input the first training model vector of the current evaluation record into the first priority prediction model to perform demand priority prediction and use the prediction vector output by the model as the corresponding second prediction vector; and form a corresponding second prediction-label pair by the second prediction vector and the first label vector of the current training record; and at the end of this round of traversal, perform corresponding accuracy, precision, recall and F1 score calculations based on all the obtained second prediction-label pairs to obtain the corresponding first accuracy, first precision, first recall and first F1 score; Step 36, identifying the first accuracy, the first precision, the first recall and the first F1 score; if the first accuracy does not meet the preset first accuracy range or the first accuracy does not meet the preset first accuracy range or the first recall does not meet the preset first recall range or the first F1 score does not meet the preset first F1 score range, returning to step 32 to continue training; if the first accuracy meets the first accuracy range and the first accuracy meets the first accuracy range and the first recall meets the first recall range and the first F1 score meets the first F1 score range, confirming that the model training is completed.

4. The intelligent demand processing method according to claim 2, characterized in that: The updating of data in the preset demand database based on the first language model, the first priority prediction model and the first product feedback information received at the time specifically includes: Step 41, inputting the first product feedback information into the demand event recognition instruction template to assemble the instruction text to obtain the corresponding first instruction text; inputting the first instruction text into the first large language model to perform demand event recognition processing to obtain the corresponding first generated text; and extracting each demand event binary in the first generated text as the corresponding first demand event binary; Wherein, the first demand event binary group includes a first demand subject and a first demand content; Step 42, assigning a corresponding first demand source to each first demand event tuple, and setting each first demand source as a customer; and forming a corresponding first demand information record from each first demand event tuple and the corresponding first demand source; and forming a corresponding first demand information record set from all the obtained first demand information records; Step 43, sending the first demand information record set to a preset demand three-property analysis interface; and receiving the first demand analysis record set sent back by the information three-property analysis interface; The three-characteristics analysis interface of demand is a preset processing interface for performing a level analysis on the importance, urgency and feasibility of demand information; the level of each importance, urgency and feasibility returned by the three-characteristics analysis interface of demand is a positive integer value level; The first demand analysis record set includes a plurality of first demand analysis records; the first demand analysis records correspond one to one with the first demand information records; the first demand analysis records include a first importance level, a first urgency level and a first feasibility level; Step 44, the first importance level, the first urgency level and the first feasibility level of each of the first demand analysis records in the first demand analysis record set form a corresponding first demand feature vector; the first demand feature vector is input into the first priority prediction model to perform demand priority prediction to obtain a corresponding first priority prediction vector; and the priority level corresponding to the largest first priority prediction probability in the first priority prediction vector is used as the corresponding first priority level; Wherein, the first priority prediction vector includes a plurality of the first priority prediction probabilities; each of the first priority prediction probabilities corresponds to a priority level; Step 45, perform a round of traversal of all the first demand information records in the first demand information record set; and in this round of traversal, take the first demand information record currently traversed as the corresponding current demand information record, and take the first demand subject, the first demand content, the first demand source, the first importance level, the first urgency level, the first feasibility level and the first priority level corresponding to the current demand information record as the corresponding current demand subject, current demand content, current demand source, current importance level, current urgency level, current feasibility level and current priority level, and the current time as the corresponding current update time; and calculate the text similarity between the current demand subject and each of the first subject fields of the demand database to obtain the corresponding first subject similarity, and calculate the text similarity between the current demand content and each of the first content fields of the demand database to obtain the corresponding first content similarity, and compare the first subject similarity with the corresponding The product of the first content similarities is used as the corresponding first demand similarity; and the largest first demand similarity is used as the corresponding maximum similarity; and whether the maximum similarity exceeds a preset first similarity threshold is identified; if the maximum similarity exceeds the first similarity threshold, the first demand data record corresponding to the maximum similarity in the demand database is recorded as the corresponding current demand data record, and when the first status field of the current demand data record is unprocessed and the first priority field is lower than the current priority level, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field and the first priority field of the current demand data record are updated to the corresponding current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level and the current priority level;If the maximum similarity does not exceed the first similarity threshold, a unique identifier is assigned as the corresponding current demand identifier, and a corresponding current demand state is set to unprocessed, and a first demand data record is added to the demand database as the corresponding current demand data record, and the first identifier field, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field, the first priority field and the first status field of the current demand data record are set to the corresponding current demand identifier, the current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level, the current priority level and the current demand state. ; 5. The intelligent demand processing method according to claim 4, characterized in that: The updating of the demand database based on the first language model, the first priority prediction model and the first batch of product demand information collected at the time specifically includes: Step 51, input each of the first product demand information of the first batch of product demand information into the demand event recognition instruction template to assemble the instruction text to obtain the corresponding second instruction text; input the second instruction text into the first language model to perform demand event recognition processing to obtain the corresponding second generated text; and extract each demand event binary group in the second generated text as the corresponding second demand event binary group; Wherein, the second demand event tuple includes a second demand subject and a second demand content; Step 52, assign a corresponding second demand source to each second demand event binary; and use the public information collection channel corresponding to the first batch of product demand information as the corresponding current collection channel; and identify the current collection channel; if the current collection channel is a market research channel, all second demand sources are set to market research; if the current collection channel is a social media information collection channel, all second demand sources are set to social media; and each second demand event binary and the corresponding second demand source form a corresponding second demand information record; and all the obtained second demand information records form a corresponding second demand information record set; Step 53, sending the second demand information record set to the demand three-property analysis interface; and receiving the second demand analysis record set sent back by the information three-property analysis interface; Wherein, the second demand analysis record set includes a plurality of second demand analysis records; the second demand analysis records correspond one-to-one to the second demand information records; the second demand analysis records include a second importance level, a second urgency level and a second feasibility level; Step 54, the second importance level, the second urgency level and the second feasibility level of each second demand analysis record in the second demand analysis record set form a corresponding second demand feature vector; the second demand feature vector is input into the first priority prediction model to perform demand priority prediction to obtain a corresponding second priority prediction vector; and the priority level corresponding to the maximum second priority prediction probability in the second priority prediction vector is used as the corresponding second priority level; The second priority prediction vector includes a plurality of second priority prediction probabilities; each second priority prediction probability corresponds to a priority level; Step 55, perform a round of traversal of all the second demand information records in the second demand information record set; and in this round of traversal, take the second demand information record currently traversed as the corresponding current demand information record, and take the second demand subject, the second demand content, the second demand source, the second importance level, the second urgency level, the second feasibility level and the second priority level corresponding to the current demand information record as the corresponding current demand subject, current demand content, current demand source, current importance level, current urgency level, current feasibility level and current priority level, and the current time as the corresponding current update time; and calculate the text similarity between the current demand subject and each of the first subject fields of the demand database to obtain the corresponding second subject similarity, and calculate the text similarity between the current demand content and each of the first content fields of the demand database to obtain the corresponding second content similarity, and compare each of the second subject similarities with the corresponding The product of the second content similarities is used as the corresponding second demand similarity; and the largest second demand similarity is used as the corresponding maximum similarity; and whether the maximum similarity exceeds the first similarity threshold is identified; if the maximum similarity exceeds the first similarity threshold, the first demand data record corresponding to the maximum similarity in the demand database is recorded as the corresponding current demand data record, and when the first status field of the current demand data record is unprocessed and the first priority field is lower than the current priority level, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field and the first priority field of the current demand data record are updated to the corresponding current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level and the current priority level;If the maximum similarity does not exceed the first similarity threshold, a unique identifier is assigned as the corresponding current demand identifier, and a corresponding current demand state is set to unprocessed, and a first demand data record is added to the demand database as the corresponding current demand data record, and the first identifier field, the first time field, the first source field, the first subject field, the first content field, the first importance field, the first urgency field, the first feasibility field, the first priority field and the first status field of the current demand data record are set to the corresponding current demand identifier, the current update time, the current demand source, the current demand subject, the current demand content, the current importance level, the current urgency level, the current feasibility level, the current priority level and the current demand state. ; 6. The intelligent demand processing method according to claim 2, characterized in that: The periodically performing batch priority attenuation processing on the unprocessed demands in the demand database specifically includes: A round of traversal is performed regularly on all the first demand data records in the demand database whose first status fields are unprocessed according to the preset first time frequency; and in this round of traversal, the first demand data record currently traversed is used as the corresponding current record; and the time difference between the current time and the first time field of the current record is calculated and the calculation result is used as the corresponding first duration; and based on the first duration, a preset first correspondence table is queried, and the first priority attenuation factor field of the first correspondence record whose first duration range field in the first correspondence table satisfies the first duration is extracted as the corresponding current attenuation factor; and the first priority field of the current record is extracted as the corresponding pre-attenuation priority; and based on the current attenuation factor and the pre-attenuation priority, the corresponding post-attenuation priority is calculated = pre-attenuation priority - current attenuation factor; and the first time field and the first priority field of the current record are reset to the corresponding current time and the post-attenuation priority; Among them, the first correspondence table is a data table used to reflect the correspondence between duration range and priority attenuation factor; the first correspondence table includes multiple first correspondence records; the first correspondence record includes the first duration range field and the first priority attenuation factor field; the first priority attenuation factor field is an integer value, the minimum value is 0, and the maximum value is the positive integer value corresponding to the maximum priority level.

7. The intelligent demand processing method according to claim 2, characterized in that: The updating of the preset high-priority demand list based on the latest demand database specifically includes: Step 71, delete the second demand data record whose first item identification field is empty in the high priority demand list; and record the first demand data record whose first status field is unprocessed and whose first priority field is higher than the preset first priority threshold in the demand database as the corresponding first high priority record; and form a corresponding first high priority record set from all the obtained first high priority records; Step 72, sending the first high priority record set to a preset demand development evaluation interface; and receiving the first demand evaluation record set sent back by the demand development evaluation interface; The demand development evaluation interface is a preset processing interface for evaluating the development time and development cost of demand information; The first demand assessment record set includes a plurality of first demand assessment records; the first demand assessment records correspond to the first high priority records one by one; the first demand assessment records include a first demand development time and a first demand development cost; Step 73, perform a round of traversal of all the first high priority records in the first high priority record set; and in this round of traversal, take the first high priority record currently traversed as the corresponding current high priority record; and form a corresponding first development evaluation information by the first requirement development time and the first requirement development cost corresponding to the current high priority record; and add a second requirement data record to the high priority requirement list as the corresponding current newly added record; and set the second identification field, the second subject field, the second content field, the second priority field and the first development evaluation field of the currently newly added record to the first identification field, the first subject field, the first content field, the first priority field and the first development evaluation information corresponding to the current high priority record; and set the first project identification field of the currently newly added record to empty; and at the end of this round of traversal, sort all the second requirement data records in the high priority requirement list in descending order according to the second priority field.

8. The intelligent demand processing method according to claim 2, characterized in that: The updating of the high-priority requirement list, the requirement database, and the preset project configuration list based on the first instruction data and notifying the requirement change based on the update result specifically includes: Extract one or more of the first demand configuration data from the first instruction data; perform a round of traversal on all the obtained first demand configuration data; and in the traversal process, use the first demand configuration data currently traversed as the corresponding current demand configuration data; use the first demand identifier and the first project identifier of the current demand configuration data as the corresponding current demand identifier and current project identifier; use the first demand data record whose first identifier field in the demand database matches the current demand identifier as the corresponding first record; use the second demand data record whose second identifier field in the high priority demand list matches the current demand identifier as the corresponding second record; use the first project configuration record whose first project identifier field in the project configuration list matches the current project identifier as the corresponding third record; update the first time field of the first record to the current time, and update the first status field of the first record to processed; and update the first project of the second record The identification field is updated to the corresponding current project identification, and the demand development duration and demand development cost information of the first development evaluation field of the second record are extracted as the corresponding current demand development duration and current demand development cost; and the first project cost field and the first project end time field of the third record are extracted as the corresponding first cost and first end time; and the corresponding second end time is calculated based on the current demand development duration and the first end time = the first end time + the current demand development duration; and the corresponding second cost is calculated based on the current demand development cost and the first cost = the first cost + the current demand development cost; and the first project cost field and the first project end time field of the third record are reset to the corresponding second cost and the second end time; and the reset first record and the third record form a corresponding first notification record; and the first notification record is sent to all the first personnel notification interfaces of the first project notification interface field of the third record.

9. A device for executing the intelligent demand processing method according to any one of claims 1 to 8, characterized in that: The device comprises: an intelligent model preparation module, a first demand update module, a second demand update module, a demand priority attenuation module, a high priority demand update module, and a product manager interaction module; The intelligent model preparation module is used to use a class of public large language models as the corresponding first large language model; and based on a class of machine learning models, an intelligent model for predicting demand priority according to the input demand feature vector is constructed as the corresponding first priority prediction model; and the first priority prediction model is trained based on a preset first data set; the machine learning model includes at least a decision tree model and a random forest model; The first demand updating module is used to update the preset demand database based on the first language model, the first priority prediction model and the first product feedback information received each time a product feedback information sent by a customer is received; The second demand update module is used to update the demand database once after collecting batch product demand information once through a type of public information collection channel at regular intervals, based on the first language model, the first priority prediction model and the first batch of product demand information collected at that time; the public information collection channel includes at least a market research channel and a social media information collection channel; the first batch of product demand information includes multiple first product demand information; The demand priority decay module is used to periodically perform batch priority decay processing on the unprocessed demands in the demand database; The high priority requirement updating module is used to update the preset high priority requirement list based on the latest requirement database every time the requirement database is updated; The product manager interaction module is used to extract the corresponding first instruction type and first instruction data from the first demand management instruction received at that time each time a first demand management instruction sent by the product manager is received; and identify the first instruction type; if the first instruction type is a demand list browsing instruction, the latest high-priority demand list is sent back to the current product manager; if the first instruction type is a demand project configuration instruction, the high-priority demand list, the demand database and the preset project configuration list are updated based on the first instruction data, and demand change notification is made based on the update result, and the latest high-priority demand list is sent back to the current product manager.

10. An electronic device, characterized in that: include: memory, processors, and transceivers; The processor is used to couple with the memory, read and execute instructions in the memory, so as to implement the method according to any one of claims 1 to 8; The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a computer, the computer is enabled to execute the method according to any one of claims 1 to 8.

Images