Method and device for SKU efficient ex-warehouse in SaaS commodity system

Abstract

The invention provides an SKU efficient ex-warehouse method and device in a SaaS commodity system, and belongs to the technical field of data processing, and the method comprises the steps: importing an SKU inventory exchange code; acquiring a user request based on the token bucket; requesting a transaction to lock the SKU according to a user request; selecting an unused redemption code according to the SKU inventory redemption code and a user request based on the ZPOPMIN command; if selection succeeds, unused redemption code logic is processed, and the lock is released; and if the redemption code is used up, reminding that there is no available redemption code. According to the SKU efficient ex-warehouse method and device in the SaaS commodity system, based on the token bucket, the user request is acquired, the commodity requested to be exchanged by the user is locked, based on the ZPOPMIN command, the unused exchange code is selected, the unused exchange code logic is processed, and the corresponding SKU is accessed based on the distributed lock to complete ex-warehouse, so that distributed transaction processing is more efficient, and ex-warehouse efficiency is higher.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method and device for efficiently shipping SKUs in a SaaS commodity system. Background Art

[0002] SKU outbound refers to the process of taking a specific stock keeping unit (SKU) out of the warehouse and preparing it for shipment to customers in inventory management. In existing trading systems, SKU outbound usually adopts the method of database global lock, selects an unused redemption code from the SKU inventory through the selectlimit 1 statement, and then processes it. The database global lock seriously affects the concurrent performance of the system, especially during high-traffic promotional activities. In addition, frequent database query operations also increase the database load, greatly reducing the efficiency of SKU outbound.

[0003] In view of this, there is an urgent need for a method and device for efficiently issuing SKUs in a SaaS product system to at least solve the above-mentioned deficiencies. Summary of the invention

[0004] One of the purposes of the present invention is to provide a method and device for efficient SKU outbound delivery in a SaaS commodity system, wherein the SKU inventory redemption code is stored in a Redis ZSet queue, and at the same time, based on a token bucket, a user request is obtained, and the commodity redeemed by the user is first locked through a transaction. After locking, based on a ZPOPMIN command, an unused redemption code is selected according to the SKU inventory redemption code and the user request, the logic of the unused redemption code is processed, and the corresponding SKU is accessed based on a distributed lock to complete the outbound delivery, thereby improving the efficiency of distributed transaction processing and the outbound delivery efficiency.

[0005] An embodiment of the present invention provides a method for efficiently issuing SKUs from a SaaS commodity system, comprising: Import SKU inventory redemption codes; Based on the token bucket, obtain user requests; Based on the user's request, request the transaction to lock the SKU; After locking the SKU, based on the ZPOPMIN command, select the unused redemption code according to the SKU inventory redemption code and user request; If the selection is successful, process the unused redemption code logic and release the lock; If the redemption codes are used up, a reminder will be given that there are no available redemption codes.

[0006] Preferably, obtaining a user request based on a token bucket includes: When a user makes a request, try to get a token from the token bucket; If the acquisition attempt is successful, the request sent by the user is regarded as the user request; If the attempt to obtain fails, an error message is returned.

[0007] An embodiment of the present invention provides a method for efficiently issuing SKUs from a SaaS commodity system, further comprising: According to the preset rate, the token generator periodically adds tokens to the token bucket; Get the maximum capacity of the token bucket. When the bucket is full, discard the excess tokens.

[0008] Preferably, based on the ZPOPMIN command, selecting an unused redemption code according to the SKU inventory redemption code and the user request includes: Get the target score of the SKU inventory redemption code in the Redis ordered set; Based on the ZPOPMIN command, the SKU inventory redemption codes are popped up in order from low to high according to the target scores and used as the codes to be selected; If there is no character match between the code to be selected and the historically used redemption code in the preset historically used redemption code library, the corresponding code to be selected will be used as the unused redemption code to be selected.

[0009] Preferably, obtaining the target score of the SKU inventory redemption code in the Redis ordered set includes: Get the product information of the target product corresponding to the SKU inventory redemption code; Extract product information features based on product information; According to the template position of the template corresponding to the preset feature description factor to construct the feature type of the product information feature, a first product information feature description factor is constructed; Obtain the historical request records of the requesting person; Determine the redemption code score of the historically requested product based on the historical request records; Constructing a second product information feature description factor based on the same construction principle as the historical request record and the first product information feature description factor; According to the historical request records, obtain the conversation records of the requesting personnel during the request process; Parse recorded conversations and adjust redemption code scores to obtain training scores; The second product information feature description factor is used as a preset neural network model input, and the training score is used as the neural network model output to train the scoring model; The first product information feature description factor is input into the scoring model to obtain the target score.

[0010] Preferably, parsing the recorded conversation and adjusting the redemption code score to obtain the training score includes: Parse recorded conversations and extract conversation semantic sequences; According to the conversation semantics in the conversation semantic sequence, identify the description semantics of the historical request product redemption code; Obtaining the target semantics of the historical request commodity redemption code description semantics within a preset sequence range in the conversation semantic sequence, and determining the recognition evaluation of the historical request commodity redemption code corresponding to the historical request commodity redemption code description semantics; Adjust the redemption code score based on the recognition evaluation and obtain the training score.

[0011] Preferably, adjusting the redemption code score according to the recognition evaluation to obtain the training score includes: Sort the historical requested product redemption codes in ascending order according to the redemption code scores to obtain a historical requested product redemption code sequence; Get the evaluation characteristics of historically requested product redemption codes; Determine an evaluation member group consisting of adjacent historical request commodity redemption codes in a historical request commodity redemption code sequence; If the inter-group evaluation characteristics of the evaluation member groups meet the preset characteristic conditions, the redemption scores of the historical requested commodity redemption codes corresponding to the evaluation member groups are exchanged to complete the adjustment; Among them, the preset characteristic conditions are: The historical requested commodity redemption code that is relatively front in the historical requested commodity redemption code sequence in the evaluation member group is used as the first target historical redemption code, and the evaluation feature of the first target historical redemption code is a feature representation of fast pop-up; The historical requested commodity redemption code that is relatively later in the historical requested commodity redemption code sequence in the evaluation member group is used as the second target historical redemption code, and the evaluation feature of the second target historical redemption code is a feature representation of slow pop-up; The speed level corresponding to the evaluation feature of any member of the evaluation member group is less than or equal to a preset level threshold.

[0012] Preferably, obtaining the target score of the SKU inventory redemption code in the Redis ordered set also includes: Filter the target historical request records based on the conversation records; Parse the target historical request record and read the system historical SKU sequence corresponding to the target historical request; Based on a preset SKU information feature extraction template and according to the system historical SKUs in the system historical SKU sequence, extract a first SKU information feature set; Based on a preset relationship feature extraction template and according to a target sequence position of the system historical SKU corresponding to the first SKU information feature set in the system historical SKU sequence, extracting a sorting relationship feature of the system historical SKU corresponding to the first SKU information feature set in the system historical SKU sequence; Determine the ranking score corresponding to the ranking relationship feature according to the ranking relationship feature and the preset ranking score comparison table; Taking the first SKU information feature set as input and the ranking score as output, training the sequence ranking score determination model; The second SKU information feature set of the SKU inventory redemption code in the Redis ordered set extracted based on the SKU information feature extraction template is input into the sequence sorting score determination model to obtain the target score.

[0013] An embodiment of the present invention provides a method for efficiently issuing SKUs from a SaaS commodity system, further comprising: If the redemption code is used up, obtain the future inbound logistics information of the SaaS product system; According to the future incoming logistics information, obtain the expected incoming time of future inventory goods; Get future inventory information; Get the preset target construction rules for SKU inventory redemption code; Generate future redemption codes based on target building rules, future inventory product information, and the expected entry time of future inventory products; Pre-issuing future redemption codes to corresponding users who are prompted to use up redemption codes. When issuing, dynamically display the estimated time for future inventory items to enter the warehouse based on preset display rules; Obtain user feedback on future redemption codes, and determine whether to issue future redemption codes based on the feedback; When it is determined that a future redemption code is issued and the future product corresponding to the future redemption code is put into storage, a remote reminder is given to the corresponding user.

[0014] An embodiment of the present invention provides a device for efficiently shipping SKUs in a SaaS commodity system, comprising: SKU stock redemption code import subsystem, used to import SKU stock redemption codes; User request acquisition subsystem, used to obtain user requests based on token bucket; The SKU locking subsystem is used to request transactions to lock SKUs based on user requests; The unused redemption code selection subsystem is used to select unused redemption codes based on the SKU inventory redemption codes and user requests based on the ZPOPMIN command after the SKU is locked; The outbound subsystem is used to process the logic of unused redemption codes and release the lock if the selection is successful; The reminder subsystem is used to remind you that there are no available redemption codes if the redemption codes are used up.

[0015] Preferably, the user request acquisition subsystem acquires the user request based on the token bucket and performs the following operations: When a user makes a request, try to get a token from the token bucket; If the acquisition attempt is successful, the request sent by the user is regarded as the user request; If the attempt to obtain fails, an error message is returned.

[0016] An embodiment of the present invention provides a device for efficiently removing SKUs from a SaaS commodity system, which further performs the following operations: According to the preset rate, the token generator periodically adds tokens to the token bucket; Get the maximum capacity of the token bucket. When the bucket is full, discard the excess tokens.

[0017] The beneficial effects of the present invention are: The present invention stores the SKU inventory redemption code in the Redis ZSet queue. At the same time, based on the token bucket, the user request is obtained, and the commodity requested to be redeemed by the user is first locked through a transaction. After locking, based on the ZPOPMIN command, an unused redemption code is selected according to the SKU inventory redemption code and the user request, the unused redemption code logic is processed, and the corresponding SKU is accessed based on the distributed lock to complete the outbound delivery, thereby improving the efficiency of distributed transaction processing and the outbound delivery efficiency.

[0018] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in the present application documents.

[0019] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 It is a schematic diagram of a method for efficiently shipping SKUs in a SaaS commodity system in an embodiment of the present invention; Figure 2 It is a schematic diagram of a device for efficiently shipping SKUs in a SaaS commodity system in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0022] The embodiment of the present invention provides a method for efficiently removing SKUs from a SaaS commodity system. Figure 1 As shown, including: Import SKU inventory redemption codes. The SKU (Stock Keeping Unit) inventory redemption codes are stored in the Redis ZSet queue and imported by the operator. The SKU inventory redemption code is a string of characters or codes used to redeem inventory products in the SaaS product system. The Redis ZSet queue is used to store SKU inventory redemption codes. Each redemption code is an element with a unique score. Based on the token bucket, a user request is obtained; wherein the user request is: the user's request information for the product redemption code in the SaaS product system; According to the user's request, a transaction is requested to lock the SKU; wherein the transaction is: a database operation to lock the SKU of the requested redemption product corresponding to the user's request; After locking the SKU, based on the ZPOPMIN command, select the unused redemption code according to the SKU inventory redemption code and user request; If the selection is successful, the logic of unused redemption codes is processed and the lock is released; the logic of unused redemption codes is: the management process of unused redemption codes obtained but not used by users, such as: validity period management; the lock is released: based on the distributed lock, the other locked SKU inventory redemption codes except the unused redemption codes are unlocked; If the redemption codes are used up, a reminder will be given that there are no available redemption codes; Among them, based on the token bucket, obtaining the user request includes: When a user requests, try to obtain a token from the token bucket; the token bucket is a container for storing tokens; the token is an access token used to call the trading system API; If the acquisition attempt is successful, the request sent by the user is regarded as the user request; If the attempt to obtain fails, an error message is returned; The token generator periodically adds tokens to the token bucket according to a preset rate; wherein the token generator periodically adds tokens to the token bucket according to a preset rate, and the token generation rate can be configured according to business requirements, for example: generating a preset number of tokens per second; Get the maximum capacity of the token bucket. When the bucket is full, discard the excess tokens. Based on the ZPOPMIN command, an unused redemption code is selected according to the SKU inventory redemption code and the user request, including: adding the SKU inventory redemption code as a member to the ordered set of Redis, storing the SKU inventory redemption code in the ordered set in ascending order of its score, executing the ZPOPMIN command, popping up the SKU inventory redemption codes in sequence from low to high in the ordered set, comparing the popped-up SKU inventory redemption code with the user request, if the popped-up SKU inventory redemption code is not used and the product corresponding to the popped-up SKU inventory redemption code matches the product requested by the user, then the corresponding SKU inventory redemption code is used as the selected unused redemption code.

[0023] The working principle and beneficial effects of the above technical solution are: The present invention stores the SKU inventory redemption code in the Redis ZSet queue. At the same time, based on the token bucket, the user request is obtained, and the commodity requested to be redeemed by the user is first locked through a transaction. After locking, based on the ZPOPMIN command, an unused redemption code is selected according to the SKU inventory redemption code and the user request, the unused redemption code logic is processed, and the corresponding SKU is accessed based on the distributed lock to complete the outbound delivery, thereby improving the efficiency of distributed transaction processing and the outbound delivery efficiency.

[0024] In one embodiment, based on the ZPOPMIN command, selecting an unused redemption code according to the SKU inventory redemption code and the user request includes: Get the target score of the SKU stock redemption code in the Redis ordered set; the target score can be preset manually or determined by the redemption priority of the SKU stock redemption code. For example, the longer the time represented by the timestamp is from the current time, the higher the redemption priority and the lower the target score. Based on the ZPOPMIN command, the SKU inventory redemption codes are popped up in order from low to high according to the target scores and used as the codes to be selected; If there is no character match between the code to be selected and the historical redemption code in the preset historical redemption code library, the corresponding code to be selected is used as the unused redemption code to be selected. The preset historical redemption code library is constructed according to the historical request record of the requested redemption code.

[0025] The working principle and beneficial effects of the above technical solution are: The present invention obtains the target points of SKU inventory redemption codes in the Redis ordered set, introduces the ZPOPMIN command, and efficiently pops up an unused redemption code through the ZPOPMIN command, thereby improving the query efficiency of the unused redemption code and further improving the outbound management speed.

[0026] In one embodiment, obtaining a target score of a SKU inventory redemption code in a Redis ordered set includes: Get the product information of the target product corresponding to the SKU inventory redemption code; the product information includes: product name, product description, price, discount and sales volume, etc.; Extract product information features based on product information; wherein the product information features are obtained based on information extraction templates preset for feature types, and product information features include: product types, product discounts, and product sales. The information extraction template is, for example, a template for extracting product discounts based on product information; Constructing a first commodity information feature description factor according to a template position of a preset feature description factor construction template corresponding to a feature type of the commodity information feature; wherein the preset feature description factor construction template is: a vector of a feature type of a vector element at each vector element position is preset; Get the historical request records of the requester; the historical request records are: the records of the user's historical requests for redemption codes, and the historical request records include: historically requested products, historically requested product information, and the scores of the historical SKU inventory redemption codes corresponding to the historical products in the stored Redis ordered set; Determine the redemption code score of the historically requested product based on the historical request records; According to the same construction principle of the historical request record and the first product information feature description factor, a second product information feature description factor is constructed; wherein the second product information feature description factor is: a product information feature description vector of the historical request product; According to the historical request records, the conversation records of the requester during the request process are obtained; wherein the conversation records are: the semantics of the requester's evaluation of the request process, which is obtained by collecting and semantically analyzing the sound of a preset sound receiving device of the user request device (for example: the coupon collection area interface of the supermarket self-service checkout device), for example: "Coupons that I don't need always pop up". Parse the recorded conversation and adjust the redemption code score to obtain the training score; wherein, the redemption code score is adjusted by: performing a rationality analysis on the scores of the historical SKU inventory redemption codes in the stored Redis ordered set to obtain the analysis results and adjust them, and the rationality analysis is based on the sorting basis of the Redis ordered set; The second product information feature description factor is used as a preset neural network model input, and the training score is used as the neural network model output to train the scoring model; The first product information feature description factor is input into the scoring model to obtain the target score.

[0027] The working principle and beneficial effects of the above technical solution are: The present invention obtains the product information of the target product that is exchanged corresponding to the SKU inventory redemption code, extracts the product information features based on the product information, sets the product information features at the template position of the feature description factor construction template according to the feature type, and obtains the first product information feature description factor; obtains the historical request record of the requester, and parses the historical request record to determine the redemption code score of the historically requested product. According to the same construction principle of the historical request record and the first product information feature description factor, the second product information feature description factor is constructed; according to the historical request record, the conversation record of the requester in the request process is obtained, the conversation record is parsed, the evaluation semantics of the requester in the request process is obtained, and whether the redemption code score needs to be adjusted is judged according to the evaluation semantics. When a negative evaluation of the request process appears in the evaluation semantics, the score of the historical SKU inventory redemption code in the stored Redis ordered set is rationally analyzed to obtain the analysis result and adjust it. After the adjustment, the reorganized training score is obtained; the second product information feature description factor is used as the preset neural network model input and the training score is used as the output of the neural network model to train the scoring model, and the first product information feature description factor is input into the scoring model to obtain the target score, thereby improving the quality of the training data; further, the accuracy of the target score is improved, and the efficiency of selecting unused redemption codes in subsequent sequences is also higher.

[0028] In one embodiment, parsing the recorded conversation and adjusting the redemption code score to obtain the training score includes: Parse the recorded conversation and extract the conversation semantic sequence; where parsing the recorded conversation refers to using natural language processing technology to analyze and understand the content in the conversation record; the conversation semantic sequence is a sequence formed by arranging the semantic content of each sentence in the conversation in the order of appearance; According to the conversation semantics in the conversation semantic sequence, the description semantics of the historical requested commodity redemption code is identified; wherein the description semantics of the historical requested commodity redemption code is: the description semantics of the historical requested commodity redemption code in the process of requesting the redemption code by the requester, such as: "coupon for expiring cereal"; Obtain the target semantics of the historical request commodity redemption code description semantics within the preset sequence range in the conversation semantic sequence, and determine the recognition evaluation of the historical request commodity redemption code corresponding to the historical request commodity redemption code description semantics; wherein the preset sequence range is manually set, such as: the range of the first 8 sequence positions and the last 8 sequence positions of the sequence position of the historical request commodity redemption code description semantics in the conversation semantic sequence; the recognition evaluation is the evaluation result of the target semantic input semantic understanding model input, such as: "jump out speed is a bit slow", "jump out speed is too slow" and "jump out speed is very timely", etc.; Adjust the redemption code score based on the recognition evaluation and obtain the training score.

[0029] The working principle and beneficial effects of the above technical solution are: When users make requests in the past, they will also give feedback while operating, such as: "Why is the redemption code for a certain product so slow?" This means that when the redemption code of this product is stored in the Redis ZSet queue, the score assigned as the basis for popping up is unreasonable, which is not suitable for the training of the subsequent scoring model and needs to be adjusted. The present invention uses natural language processing technology to analyze and record conversations to obtain conversation semantic sequences, identify the description semantics of historical request commodity redemption codes in the sequence, and the description semantics of historical request commodity redemption codes are the description semantics of the historical request commodity redemption codes of the requesting personnel in the process of requesting redemption codes. The sequence range of the description semantics of historical request commodity redemption codes in the conversation semantic sequence is introduced, and the specific description content semantics of the historical request commodity redemption codes of the requesting personnel are obtained in a targeted manner, that is, the target semantics, and the target semantics are input into the semantic understanding model to obtain the recognition evaluation of the requesting personnel, and the redemption code score is adjusted based on the recognition evaluation. When the recognition evaluation indicates that the redemption code recognition speed is slow, the redemption code score of the corresponding historical request commodity redemption code is lowered, and when the recognition evaluation indicates that the redemption code recognition speed is fast, the redemption code score of the corresponding historical request commodity redemption code is increased to obtain the adjusted training score, thereby improving the rationality and accuracy of the training score adjustment.

[0030] In one embodiment, adjusting the redemption code score according to the recognition evaluation to obtain the training score includes: Sort the historical requested commodity redemption codes in ascending order according to the redemption code scores to obtain a historical requested commodity redemption code sequence; wherein the historical requested commodity redemption code sequence is: the result obtained by sorting the historical requested commodity redemption codes in ascending order according to their redemption code scores; Obtaining evaluation features of historically requested product redemption codes; wherein the evaluation features include: whether the code pops up quickly or slowly, and the speed level, and the rules for determining the speed level based on the recognition evaluation are manually preset; Determine an evaluation member group consisting of adjacent historical request commodity redemption codes in a historical request commodity redemption code sequence; If the inter-group evaluation characteristics of the evaluation member groups meet the preset characteristic conditions, the redemption scores of the historical requested commodity redemption codes corresponding to the evaluation member groups are exchanged to complete the adjustment; Among them, the preset characteristic conditions are: The historical requested commodity redemption code that is relatively front in the historical requested commodity redemption code sequence in the evaluation member group is used as the first target historical redemption code, and the evaluation feature of the first target historical redemption code is a feature representation of fast pop-up; The historical requested commodity redemption code that is relatively later in the historical requested commodity redemption code sequence in the evaluation member group is used as the second target historical redemption code, and the evaluation feature of the second target historical redemption code is a feature representation of slow pop-up; The speed level corresponding to the evaluation feature of any member of the evaluation member group is less than or equal to a preset level threshold, wherein the preset level threshold is manually preset.

[0031] The working principle and beneficial effects of the above technical solution are: When adjusting the redemption code score, the present invention sorts the historical requested commodity redemption codes in ascending order according to the redemption code scores to obtain a historical requested commodity redemption code sequence; obtains evaluation features of the historical requested commodity redemption codes, the evaluation features including whether the evaluation pops up quickly or slowly, and a specific speed level; Generally, for commodity redemption codes with standard scores determined based on unified scores, when the evaluation characteristics are inappropriate, commodity redemption codes with adjacent scores are more likely to have score deviations. Therefore, the evaluation member group consisting of adjacent historical request commodity redemption codes is extracted. When the inter-group conditions of the evaluation member group (the evaluation characteristics corresponding to the members) meet the characteristic conditions, the redemption scores of the corresponding historical request commodity redemption codes are exchanged to complete the adjustment; Among them, the preset characteristic conditions are specifically: the pop-up speed of the members in the evaluation member group who are relatively early in the historical request product redemption code sequence is fast, and the pop-up speed of the members in the evaluation member group who are relatively late in the historical request product redemption code sequence is slow, and their speed level is less than the preset level threshold, which realizes the rapid local adjustment of inappropriate scoring and improves the quality of the scoring model training data.

[0032] In one embodiment, obtaining a target score of a SKU inventory redemption code in a Redis ordered set further includes: Filter the target historical request records according to the conversation records; wherein the target historical request records are: historical request records that do not contain pop-up objections to product redemption codes as determined by parsing the conversation records; Parse the target historical request record and read the system historical SKU sequence corresponding to the target historical request; wherein the target historical request is: the redemption code request in the target historical request record; the system historical SKU sequence is: the sequence of SKU inventory redemption codes of the products locked by the target historical request sorted from small to large according to the scores; Based on a preset SKU information feature extraction template and according to the system historical SKU in the system historical SKU sequence, a first SKU information feature set is extracted; wherein the preset SKU information feature extraction template is: a template for extracting information of the product corresponding to the SKU serial code by comparing the SKU serial code, such as: production date, size, color and model; Based on a preset relationship feature extraction template and according to the target sequence position of the system historical SKU corresponding to the first SKU information feature set in the system historical SKU sequence, a sorting relationship feature of the system historical SKU corresponding to the first SKU information feature set in the system historical SKU sequence is extracted; wherein the relationship feature extraction template is a template for extracting the relationship between the first SKU information feature set corresponding to the system historical SKU and the target sequence position of the system historical SKU in the system historical SKU sequence, and the sorting relationship feature is, for example: for system historical SKUs representing commodities of the same size, color and model, the corresponding system historical SKU with a production date farther away from the current one has a target sequence position closer to the front; Determine the ranking score corresponding to the ranking relationship feature according to the ranking relationship feature and a preset ranking score comparison table; wherein the ranking score comparison table stores manually preset pre-selected ranking relationship features and corresponding ranking scores; The first SKU information feature set is used as input and the ranking score is used as output to train a sequence ranking score determination model; wherein the sequence ranking score determination model is an AI model that determines the ranking score (target score) of the SKU in the Redis ordered set with a subsequent specified pop-up order according to the SKU information features. The smaller the ranking score, the closer it is to the front in the Redis ordered set and the more limited it is to pop up; The second SKU information feature set of the SKU inventory redemption code in the Redis ordered set extracted based on the SKU information feature extraction template is input into the sequence sorting score determination model to obtain the target score.

[0033] The working principle and beneficial effects of the above technical solution are: The present invention introduces a target historical request record in which there is no pop-up objection to the commodity redemption code, determines the system historical SKU sequence corresponding to the target historical request, introduces a SKU information feature extraction template, extracts the first SKU information feature set in the system historical SKU in the system historical SKU sequence; introduces the first SKU information feature set, extracts the correlation relationship (sorting relationship feature) between the first SKU information feature set and its corresponding system historical SKU in the system historical SKU sequence; refines the sorting score according to the sorting relationship feature and a preset sorting score comparison table, takes the first SKU information feature set as input and the sorting score as output, trains a sequence sorting score determination model, and then inputs the second SKU information feature set of the SKU inventory redemption code into the sequence sorting score determination model to obtain a target score, thereby further improving the determination accuracy of the target score.

[0034] An embodiment of the present invention provides a method for efficiently issuing SKUs from a SaaS commodity system, further comprising: If the redemption code is used up, the future incoming logistics information of the SaaS product system is obtained; the future incoming logistics information is: the driving information of the logistics vehicle that is going to enter the goods in the future, which is obtained through the logistics vehicle node; According to the future incoming logistics information, the estimated incoming time of the future inventory goods is obtained; among which, the estimated incoming time of the future inventory goods is predicted based on the distance between the current logistics vehicle position and the warehouse position determined by analyzing the future incoming logistics information; Obtaining future inventory product information; wherein the future inventory product information is: product information of products to be put into storage in the future; Obtain the preset target building rules for the SKU inventory redemption code; wherein the target building rules are set by the system administrator; Generate a future redemption code according to the target construction rule, the future inventory product information and the expected entry time of the future inventory product; when generating the future redemption code, first generate the redemption code according to the target construction rule and the future inventory product information, and then mark the expected entry time of the future inventory product in the blank area of ​​the future redemption code display area; Pre-issuing future redemption codes to corresponding users who are prompted to have used up their redemption codes, and dynamically displaying the estimated time for future inventory items to enter the warehouse based on preset display rules when issuing them; wherein the preset display rules are, for example: flashing red characters; Obtaining user feedback information on the future redemption code, and determining whether to issue the future redemption code based on the feedback information; wherein the feedback information is feedback on whether the user receives the future redemption code, such as: clicking a confirmation area on the capacitive screen; When it is determined that a future redemption code is issued and the future product corresponding to the future redemption code is put into storage, a remote reminder is given to the corresponding user. The remote reminder can be pushed through the user service system bound to the user.

[0035] The working principle and beneficial effects of the above technical solution are: When the redemption code is exhausted, the present invention introduces the future warehousing logistics information of the SaaS product system and predicts the expected warehousing time of future inventory products; generates a redemption code according to the preset target construction rule of the SKU inventory redemption code and the future inventory product information, and then marks the expected warehousing time of the future inventory products in the blank area of ​​the future redemption code display area to obtain the future redemption code, pre-issues the future redemption code to the corresponding user who is prompted that the redemption code is exhausted, and dynamically displays the expected warehousing time of the future inventory products, when the user receives the future redemption code, the future product corresponding to the future redemption code is put into storage, and the product arrival notification is pushed through the user service system, thereby realizing the advance reservation of future inventory products and improving the user experience.

[0036] The embodiment of the present invention provides a device for efficiently removing SKUs from a SaaS commodity system. Figure 2 As shown, including: SKU stock redemption code import subsystem 1, used to import SKU stock redemption codes; User request acquisition subsystem 2, used for acquiring user requests based on a token bucket; SKU locking subsystem 3, used to request a transaction to lock the SKU according to a user request; The unused redemption code selection subsystem 4 is used to select an unused redemption code according to the SKU inventory redemption code and the user request based on the ZPOPMIN command after the SKU is locked; Outbound subsystem 5, for processing the logic of unused redemption codes and releasing the lock if the selection is successful; The reminder subsystem 6 is used to remind that there are no available redemption codes if the redemption codes are used up.

[0037] In one embodiment, the user request acquisition subsystem acquires the user request based on the token bucket and performs the following operations: When a user makes a request, try to get a token from the token bucket; If the acquisition attempt is successful, the request sent by the user is regarded as the user request; If the attempt to obtain fails, an error message is returned.

[0038] The embodiment of the present invention provides a device for efficiently sending out SKUs from a SaaS commodity system, and further performs the following operations: According to the preset rate, the token generator periodically adds tokens to the token bucket; Get the maximum capacity of the token bucket. When the bucket is full, discard the excess tokens.

[0039] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for efficiently removing SKUs from a SaaS product system, characterized in that: include: Import SKU inventory redemption codes; Based on the token bucket, obtain user requests; Based on the user's request, request the transaction to lock the SKU; After locking the SKU, based on the ZPOPMIN command, select the unused redemption code according to the SKU inventory redemption code and user request; If the selection is successful, process the unused redemption code logic and release the lock; If the redemption codes are used up, a reminder will be given that there are no available redemption codes.

2. A method for efficient SKU outbound delivery in a SaaS commodity system as claimed in claim 1, characterized in that: Based on the token bucket, obtain the user request, including: When a user makes a request, try to get a token from the token bucket; If the acquisition attempt is successful, the request sent by the user is regarded as the user request; If the attempt to obtain fails, an error message is returned.

3. A method for efficient SKU outbound delivery in a SaaS commodity system as claimed in claim 2, characterized in that: Also includes: According to the preset rate, the token generator periodically adds tokens to the token bucket; Get the maximum capacity of the token bucket. When the bucket is full, discard the excess tokens.

4. A method for efficient SKU outbound delivery in a SaaS commodity system as claimed in claim 1, characterized in that: Based on the ZPOPMIN command, select unused redemption codes according to the SKU inventory redemption codes and user requests, including: Get the target score of the SKU inventory redemption code in the Redis ordered set; Based on the ZPOPMIN command, the SKU inventory redemption codes are popped up in order from low to high according to the target scores and used as the codes to be selected; If there is no character match between the code to be selected and the historically used redemption code in the preset historically used redemption code library, the corresponding code to be selected will be used as the unused redemption code to be selected.

5. A method for efficient SKU outbound delivery in a SaaS commodity system as claimed in claim 4, characterized in that: Get the target score of the SKU inventory redemption code in the Redis ordered set, including: Get the product information of the target product corresponding to the SKU inventory redemption code; Extract product information features based on product information; According to the template position of the template corresponding to the preset feature description factor to construct the feature type of the product information feature, a first product information feature description factor is constructed; Obtain the historical request records of the requesting person; Determine the redemption code score of the historically requested product based on the historical request records; Constructing a second product information feature description factor based on the same construction principle as the historical request record and the first product information feature description factor; According to the historical request records, obtain the conversation records of the requesting personnel during the request process; Parse recorded conversations and adjust redemption code scores to obtain training scores; The second product information feature description factor is used as a preset neural network model input, and the training score is used as the neural network model output to train the scoring model; The first product information feature description factor is input into the scoring model to obtain the target score.

6. A method for efficient SKU outbound delivery in a SaaS commodity system as claimed in claim 5, characterized in that: Parse recorded conversations and adjust redemption code scores to obtain training scores, including: Parse recorded conversations and extract conversation semantic sequences; According to the conversation semantics in the conversation semantic sequence, identify the description semantics of the historical request product redemption code; Obtaining the target semantics of the historical request commodity redemption code description semantics within a preset sequence range in the conversation semantic sequence, and determining the recognition evaluation of the historical request commodity redemption code corresponding to the historical request commodity redemption code description semantics; Adjust the redemption code score based on the recognition evaluation and obtain the training score.

7. A method for efficient SKU outbound delivery in a SaaS commodity system as claimed in claim 6, characterized in that: Adjust the redemption code score based on the recognition evaluation and obtain the training score, including: Sort the historical requested product redemption codes in ascending order according to the redemption code scores to obtain a historical requested product redemption code sequence; Get the evaluation characteristics of historically requested product redemption codes; Determine an evaluation member group consisting of adjacent historical request commodity redemption codes in a historical request commodity redemption code sequence; If the inter-group evaluation characteristics of the evaluation member groups meet the preset characteristic conditions, the redemption scores of the historical requested commodity redemption codes corresponding to the evaluation member groups are exchanged to complete the adjustment; Among them, the preset characteristic conditions are: The historical requested commodity redemption code that is relatively front in the historical requested commodity redemption code sequence in the evaluation member group is used as the first target historical redemption code, and the evaluation feature of the first target historical redemption code is a feature representation of fast pop-up; The historical requested commodity redemption code that is relatively later in the historical requested commodity redemption code sequence in the evaluation member group is used as the second target historical redemption code, and the evaluation feature of the second target historical redemption code is a feature representation of slow pop-up; The speed level corresponding to the evaluation feature of any member of the evaluation member group is less than or equal to a preset level threshold.

8. A device for efficiently removing SKUs from warehouses in a SaaS commodity system, characterized in that: include: SKU stock redemption code import subsystem, used to import SKU stock redemption codes; User request acquisition subsystem, used to obtain user requests based on token bucket; The SKU locking subsystem is used to request transactions to lock SKUs based on user requests; The unused redemption code selection subsystem is used to select unused redemption codes based on the SKU inventory redemption codes and user requests based on the ZPOPMIN command after the SKU is locked; The outbound subsystem is used to process the logic of unused redemption codes and release the lock if the selection is successful; The reminder subsystem is used to remind you that there are no available redemption codes if the redemption codes are used up.

9. In the device for efficient SKU outbound delivery in a SaaS commodity system as claimed in claim 8, the user request acquisition subsystem acquires the user request based on the token bucket and performs the following operations: When a user makes a request, try to get a token from the token bucket; If the acquisition attempt is successful, the request sent by the user is regarded as the user request; If the attempt to obtain fails, an error message is returned.

10. The device for efficiently removing SKUs from a SaaS commodity system according to claim 9 further performs the following operations: According to the preset rate, the token generator periodically adds tokens to the token bucket; Get the maximum capacity of the token bucket. When the bucket is full, discard the excess tokens.

Images