A large model-based commodity automatic pricing system and method

Abstract

The application discloses a large model-based commodity automatic pricing system and method. Through the first intelligent agent module, the text content in the pricing demand list is subjected to semantic extraction by using a large language model to obtain structured commodity identification information and quantity information. Then, the second intelligent agent module combines a commodity data source and a preset pricing strategy to determine the pricing result of each commodity item to be priced, so that the corresponding, pricing and strategy adjustment of each commodity item in the pricing demand list are completed by the modules in cooperation, thereby reducing the processing amount of the artificial corresponding item by item. Meanwhile, the pricing generation module generates a pricing list based on the pricing result, so that the field organization and expression of the pricing list are relatively consistent, and the efficiency and consistency of the pricing processing are improved.

Description

Technical Field

[0001] This application relates to the fields of natural language processing and data processing technology, specifically to an automatic commodity pricing system and method based on a large model. Background Technology

[0002] In business scenarios such as product sales and procurement, business partners often need to generate corresponding quotations based on the product requirements submitted by customers. These customer-submitted product requirements are typically presented in various formats, with inconsistent descriptions of product names, models, specifications, quantities, etc., requiring verification against the company's own product records and pricing information.

[0003] The process of generating a quotation involves matching each item described in the requirements with its corresponding product record, determining the price of each item based on its price and the relevant pricing rules, and then compiling these into a quotation. Because the number of product items in a requirement can be large, and the descriptions of products differ across requirements, the processing volume is significant, and ensuring the consistency and accuracy of the results is difficult, thus limiting overall processing efficiency.

[0004] Therefore, a solution is needed to improve the efficiency and consistency of commodity quotation processing. Summary of the Invention

[0005] To address this issue, this application provides an automatic commodity pricing system and method based on a large model, which solves the problems of large processing volume, insufficient processing efficiency, and inconsistency in existing pricing methods.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] In a first aspect, this application provides an automatic commodity pricing system based on a large model, the system comprising: an information input module, a first intelligent agent module, a second intelligent agent module, and a pricing generation module;

[0008] The information input module is used to obtain a list of quotation requirements;

[0009] The first intelligent agent module is used to obtain the text content in the quotation demand list, and to extract the semantics of the text content through a large language model to obtain structured demand data including the product identification information and demand quantity information of each product item to be quoted.

[0010] The second intelligent agent module is used to determine the product price information corresponding to each product item to be quoted from the product data source based on the structured demand data, and adjust the product price information corresponding to each product item to be quoted based on the preset quoting strategy to obtain the quoting result corresponding to each product item to be quoted.

[0011] The quotation generation module is used to generate a quotation sheet corresponding to the quotation requirement list based on the quotation results corresponding to each of the items to be quoted.

[0012] Secondly, this application provides a method for automatic commodity pricing based on the system, the method comprising:

[0013] Obtain a list of quote requirements;

[0014] The quotation request list is processed by the automatic quotation system to obtain the quotation list corresponding to the quotation request list.

[0015] Thirdly, this application provides a computer device including a processor, which executes a computer program stored in a memory to implement the steps of the automatic commodity quotation method described above.

[0016] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described automatic commodity quotation method.

[0017] The automatic commodity pricing system and method based on a large model provided in this application involves a first intelligent agent module using a large language model to semantically extract the text content of the pricing demand list, obtaining structured commodity identification information and demand quantity information. Then, a second intelligent agent module combines the commodity data source and preset pricing strategy to determine the pricing result for each commodity item to be quoted. Thus, the correspondence, pricing, and strategy adjustment of each commodity item in the pricing demand list are completed collaboratively by the modules, reducing the amount of manual processing for each item. At the same time, since the pricing generation module generates a unified pricing sheet based on the pricing result, the field organization and expression of the pricing sheet are more consistent, improving the efficiency and consistency of pricing processing. Attached Figure Description

[0018] To more clearly illustrate the embodiments of this application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the structure of an automatic commodity pricing system based on a large model, provided in an embodiment of this application;

[0020] Figure 2 This is a schematic diagram of the overall architecture of an automatic commodity pricing system based on a large model, provided in an embodiment of this application.

[0021] Figure 3 This is a schematic diagram of an automatic product quotation process provided in an embodiment of this application;

[0022] Figure 4 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of this application. Detailed Implementation

[0023] The following specific embodiments illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] The applicant found that if the quotation request list was processed manually item by item, the business party would need to compare the name, model, and quantity of each item described in the request line by line with the company's product records, then find the corresponding product price, and calculate the price item by item according to pricing rules such as discount rules, customer level pricing, and bulk discounts. Finally, the results would be compiled and filled into the quotation form. Under this method, when there are many product items in the request list, or when the product descriptions in the request are inconsistent with the naming conventions of the company's product records, the amount of processing for comparison and calculation will increase accordingly, and it will be difficult to maintain the consistency of the processing results. In addition, when product prices or pricing rules are adjusted, it is necessary to manually track the latest data and update the existing quotations. The update process relies on manual recording and querying, which may result in untimely updates.

[0025] If a string matching method based on fixed rules is used to match the product description in the requirements with the product records, the matching results are often unsatisfactory when the names of the requirements description and the product records are inconsistent, there are synonyms, or the description is incomplete, requiring manual intervention again; moreover, fixed rules are difficult to adapt to situations where the requirements descriptions are diverse in form.

[0026] Based on the above findings, this application provides an automatic commodity pricing system and method based on a large model, thereby improving the problems existing in the above processing.

[0027] The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] Example 1:

[0029] Figure 1 This is a schematic diagram of the structure of an automatic commodity pricing system based on a large model, provided in an embodiment of this application. (Refer to...) Figure 1The system includes an information input module 101, a first intelligent agent module 102, a second intelligent agent module 103, and a quotation generation module 104.

[0030] The information input module 101 is used to obtain a list of quotation requirements;

[0031] The first intelligent agent module 102 is used to obtain the text content in the quotation demand list, and to extract the semantics of the text content through a large language model to obtain structured demand data including the product identification information and demand quantity information of each product item to be quoted.

[0032] The second intelligent agent module 103 is used to determine the commodity price information corresponding to each commodity item to be quoted from the commodity data source based on the structured demand data, and to adjust the commodity price information corresponding to each commodity item to be quoted based on the preset quoting strategy to obtain the quoting result corresponding to each commodity item to be quoted.

[0033] The quotation generation module 104 is used to generate a quotation sheet corresponding to the quotation requirement list based on the quotation results corresponding to each of the items to be quoted.

[0034] The quotation requirement list refers to the content carrier describing the requirements for the goods to be quoted, which can be a spreadsheet file, text file, document file, image file, or scanned copy. The information input module 101 can obtain the quotation requirement list in various ways, such as by uploading and receiving it through a client, receiving it as an email attachment, or receiving it through an interface connected to the client's system. This application embodiment does not limit the specific form of the quotation requirement list or the specific method by which the information input module 101 obtains the quotation requirement list.

[0035] In one example, when the quotation request list is in an unstructured form such as an image file or scanned document, the information input module 101 can first process it using optical character recognition (OCR) to convert the content into text before providing it to the first intelligent agent module 102. OCR refers to the process of recognizing text in an image and converting it into processable text. For example, if a customer submits a purchase request spreadsheet as an email attachment, the information input module 101 can read the cell contents of the spreadsheet and organize them into text. Alternatively, if a customer submits a photograph of a paper request form, the information input module 101 can first perform OCR on the photograph to obtain the corresponding text content.

[0036] An agent module refers to a functional unit that can invoke a large language model and perform processing tasks around a specific processing objective. A large language model refers to a processing model trained on a large-scale corpus that can understand, extract, rewrite, or generate text. The first agent module 102 and the second agent module 103 can invoke the same large language model, or they can invoke different large language models respectively.

[0037] After obtaining the quotation requirement list based on the above embodiments, the first intelligent agent module 102 can acquire the text content in the quotation requirement list. The text content refers to the text content obtained after reading or recognizing the quotation requirement list. Then, semantic extraction is performed on the text content using a large language model to obtain structured requirement data including product identification information and quantity information for each item to be quoted. Semantic extraction refers to the process of identifying and extracting target information from the text content based on an understanding of the text's meaning. Product identification information refers to information used to identify a product, which may include one or more of the following: product name, model, specifications, and brand, such as "product model A100 of a certain brand". Quantity information refers to the quantity required for each item to be quoted, such as "10 units". Structured requirement data refers to organizing the product items described in the text content into data organized by fields, such as recording the product identification information and quantity information of each product item in a list format.

[0038] For example, the text content in the quotation request list is "10 units of device A100 are needed, and 30 ink cartridges (compatible with A100) are also needed". After the first intelligent agent module 102 performs semantic extraction on the text content through the large language model, it can obtain two items to be quoted. The first item has the item identification information "device A100" and the required quantity information "10". The second item has the item identification information "ink cartridges (compatible with A100)" and the required quantity information "30". It organizes them into structured requirement data.

[0039] After acquiring the structured demand data, the second intelligent agent module 103 can determine the corresponding product price information for each product item to be quoted from the product data source. The product data source refers to a data source storing product records, which can be a database, a data file, or a service providing product records via an interface. A product record is a data entry containing product information, which may include product identification information and product price information. Product price information refers to price information associated with the product record, such as a base price or listed price. Then, the second intelligent agent module 103 adjusts the product price information corresponding to each product item to be quoted based on a preset pricing strategy, obtaining the corresponding pricing result for each product item. The preset pricing strategy refers to the rules used to adjust the product price information, which may include one or more of the following: discount rules, customer level pricing rules, and bulk discount rules. The pricing result refers to the pricing-related information determined for a specific product item to be quoted, which may include the individual price amount or the corresponding quantity subtotal amount.

[0040] After obtaining the quotation results, the quotation generation module 104 can generate a quotation sheet corresponding to the quotation requirement list based on the quotation results for each item to be quoted. The quotation sheet is a document generated based on the quotation results for each item to be quoted, used to present the quotation content. For example, the quotation generation module 104 can summarize the product identification information, required quantity information, unit price, subtotal, etc., of each item to be quoted into a single quotation sheet.

[0041] The first intelligent agent module uses a large language model to semantically extract the text content in the quotation request list, obtaining structured product identification information and required quantity information. Then, the second intelligent agent module combines the product data source and preset quotation strategy to determine the quotation result for each product item to be quoted. Thus, the correspondence, pricing and strategy adjustment of each product item in the quotation request list are completed by the modules collaboratively, reducing the amount of manual processing for each item. At the same time, since the quotation generation module generates a quotation sheet based on the quotation result, the field organization and expression of the quotation sheet are more consistent, improving the efficiency and consistency of quotation processing.

[0042] Example 2:

[0043] To accurately and quickly obtain product identification information and avoid inaccurate descriptions affecting price generation, based on the above embodiments, in this application, the first intelligent agent module 102 is specifically used for:

[0044] The text content is semantically extracted using the large language model to obtain preliminary descriptions of each item in the quotation request list.

[0045] Using the large language model, the preliminary description information is standardized based on the preset product description specifications to obtain the product identification information corresponding to each product item to be quoted.

[0046] The preliminary description information refers to the descriptive information of each item to be quoted, directly extracted from the text content. Since the quotation requests submitted by clients are often inconsistent in their wording, the preliminary description information may contain non-standard expressions, incomplete information, or inconsistent terminology. For example, for the text content "ten A100 devices," the first intelligent agent module 102, after semantic extraction using a large language model, can obtain the preliminary description information "A100 devices" and the corresponding quantity "ten units"; for the text content "ink cartridge (the A100 model)," the preliminary description information is "ink cartridge (the A100 model)." Then, the first intelligent agent module 102, using a large language model and based on preset product description specifications, standardizes the preliminary description information to obtain the product identification information corresponding to each item to be quoted. The preset product description specifications refer to pre-defined rules used to constrain the expression of product identification information, which may include conventions regarding field order, naming terms, and unit expressions. For example, it may stipulate that product identification information is expressed in the order of "brand + category + model," and uses a uniform model notation. Standardization processing refers to the process of organizing, supplementing, or rewriting the initial description information according to the product description specifications to make it conform to the specifications.

[0047] For example, for the initial description information "A100 device", the first intelligent agent module 102 can obtain the product identification information "device A100" after standardizing it according to the product description specification through the large language model; for the quantity "ten units" expressed in Chinese numerals in the initial description information, the required quantity information can be standardized to "10".

[0048] By first extracting preliminary description information and then standardizing it according to product description specifications, different forms of requirement descriptions can be organized into product identification information with a more uniform form. This facilitates subsequent matching with product records in the product data source, thereby improving the consistency of the matching process.

[0049] Example 3:

[0050] In order to accurately obtain the quotation results, based on the above embodiments, in this application, the second intelligent agent module 103 is specifically used for:

[0051] For any item to be quoted in the structured demand data, the product identification information of the item to be quoted is matched with the product records in the product data source to determine the benchmark price information corresponding to the product record that matches the item to be quoted.

[0052] Based on the benchmark price information and the preset pricing strategy, the pricing result corresponding to the item to be quoted is determined.

[0053] When determining the price information corresponding to each item to be quoted, the second intelligent agent module 103 can match the product identification information of any item to be quoted in the structured demand data with the product records in the product data source to determine the benchmark price information corresponding to the matching product record. Matching refers to the process of finding the product record corresponding to the item to be quoted in the product data source. The benchmark price information refers to the price information associated with the matched product record, which serves as the basis for price adjustment, such as the listed price or catalog price recorded in the product record. For example, for the item to be quoted with the product identification information "Equipment A100", the second intelligent agent module 103 finds the corresponding product record in the product data source and reads the benchmark price information in that product record as "2000 yuan / unit".

[0054] In one example, the second agent module 103 is specifically used for:

[0055] Using the large language model, the semantic similarity between the product identification information of the product item to be quoted and the product identification information of each product record in the product data source is determined;

[0056] Based on the semantic similarity, a set of candidate product records corresponding to the product item to be quoted is determined from the product data source;

[0057] The product record with the highest semantic similarity in the candidate product record set is identified as the product record that matches the product item to be quoted.

[0058] Semantic similarity is an indicator used to measure how close two pieces of text are in meaning. Its value can be a number within a certain range; a higher value indicates that the two pieces of text are closer in meaning. A candidate product record set refers to the collection of product records that may correspond to the product item to be quoted, selected from the product data source.

[0059] For example, the product identifier information for the item to be quoted is "A100 laser equipment," and the product data source contains product records such as "A100 laser equipment of a certain brand" and "B200 inkjet equipment of a certain brand." The second intelligent agent module 103 calculates the semantic similarity between the product identifier information and each product record using a large language model. Since "A100 laser equipment" and "A100 laser equipment of a certain brand" are closer in meaning, their semantic similarity is high. The second intelligent agent module 103 selects several product records with high semantic similarity as a candidate product record set, and determines "A100 laser equipment of a certain brand" with the highest semantic similarity as the matching product record.

[0060] There are several ways to obtain semantic similarity. In one example, a large language model can be used to generate corresponding semantic vectors for the product identifier information and the product identifier information of the product record, and the semantic similarity can be determined based on the distance or angle between the semantic vectors. In another example, the product identifier information of the product item to be quoted and the product identifier information of the product record can be input into the large language model, and the large language model can directly output the semantic similarity between the two.

[0061] By matching based on semantic similarity, even if the product identification information of the product item to be quoted differs from the naming method of the product record, there are synonyms, or the description is incomplete, the corresponding product record can still be identified, which helps to improve the adaptability of the matching process.

[0062] In one example, the second agent module 103 is also used for:

[0063] When the candidate product record set corresponding to the product item to be quoted is empty, or the highest semantic similarity in the candidate product record set is lower than a preset similarity threshold, the product item to be quoted is marked as a pending product item, and the quotation result corresponding to the pending product item is set to a preset pending state value.

[0064] The quotation generation module 104 is further configured to: identify the pending product items in the quotation and reserve fields corresponding to the pending product items in the quotation.

[0065] The preset similarity threshold refers to a pre-set value used to determine whether the semantic similarity is high enough. Pending product items refer to product items for which no matching product record can be found in the product data source. Pre-set pending status values ​​are values ​​used to indicate that the quotation result is not yet determined, such as null values, specific tags, or placeholders. Fields that can be supplemented are fields reserved in the quotation form for later supplementation of quotation-related information.

[0066] For example, the product identifier of the item to be quoted is "a certain special accessory," but there is no product record in the product data source with a semantic similarity that reaches the preset similarity threshold. In this case, the candidate product record set is empty, or the highest semantic similarity among them is lower than the preset similarity threshold. The second intelligent agent module 103 marks the item to be quoted as a pending item and sets its quotation result to a pending state value. The quotation generation module 104 marks the pending item in the generated quotation, for example, by marking it with a specific background color or remarks, and reserves a field for later addition in the price column corresponding to the item.

[0067] By using the above method, for product items for which no matching product record can be determined, the system does not directly provide a quote, but marks them as pending and reserves a field for supplementary entry. This avoids generating inaccurate quote results in the absence of matching basis, which helps to improve the reliability of the quote.

[0068] The second intelligent agent module 103 determines the corresponding quotation result for the item to be quoted based on the benchmark price information and the preset quotation strategy. For example, for an item to be quoted with a benchmark price of "2000 yuan / unit" and a required quantity of "10", the preset quotation strategy includes the rule of "a 5% discount for single items with a quantity of 10 or more". The second intelligent agent module 103 adjusts the benchmark price information accordingly to obtain a unit price of "1900 yuan / unit", and combines it with the required quantity to obtain the corresponding quotation result for the item to be quoted, such as a unit price of 1900 yuan and a subtotal of 19000 yuan.

[0069] In the above manner, the second intelligent agent module 103 completes the matching of product records, acquisition of benchmark prices and adjustment of strategies for each product item to be quoted, so as to obtain the quotation result corresponding to each product item to be quoted, making the quotation process have a clear processing basis and helping to improve the consistency of quotation results.

[0070] In one example, the preset pricing strategy includes multiple sub-strategies, each sub-strategy being associated with a triggering condition and a preset priority; the second intelligent agent module 103 is specifically used for:

[0071] For any of the items to be quoted, based on the item's identification information and demand quantity information, a target sub-strategy matching the triggering condition of the item to be quoted is determined from the preset quoting strategy.

[0072] When there are multiple target sub-strategies, the multiple target sub-strategies are sorted according to the preset priority associated with each target sub-strategy, and the product price information corresponding to the product item to be quoted is adjusted in sequence according to the sorted order to obtain the quotation result corresponding to the product item to be quoted.

[0073] Here, a sub-strategy refers to a relatively independent pricing rule within a pre-defined pricing strategy, such as bulk discount rules, customer tier pricing rules, or promotional discount rules. A trigger condition refers to the condition used to determine whether a sub-strategy applies to a specific product item to be quoted, such as "the required quantity reaches a certain amount" or "the product belongs to a certain category." Pre-defined priority refers to the pre-set order of application for each sub-strategy. A target sub-strategy is the sub-strategy whose trigger condition matches the product item to be quoted and needs to be applied.

[0074] For example, the preset pricing strategy includes a bulk discount sub-strategy (triggered when the demand quantity reaches 10 or more, the discount is 95%, priority is 1) and a customer level discount sub-strategy (triggered when the customer is of a specified level, the discount is an additional 95%, priority is 2). For a product item to be quoted with a demand quantity of "10" and a customer belonging to the specified level, the second intelligent agent module 103 determines that both of the above sub-strategies match the product item to be quoted, that is, there are multiple target sub-strategies. Therefore, according to the preset priority, the bulk discount sub-strategy is first applied to adjust the base price information, and then the customer level discount sub-strategy is applied to adjust it on this basis, thus obtaining the corresponding pricing result for the product item to be quoted.

[0075] By setting trigger conditions and preset priorities for each sub-strategy, and adjusting them sequentially according to the sorted order, the order of price adjustments when multiple sub-strategies are applied simultaneously has a clear basis, which is conducive to obtaining consistent pricing results.

[0076] Example 4:

[0077] To obtain standardized price lists for easy user review, based on the above embodiments, in this application, the price generation module 104 is specifically used for:

[0078] Based on the preset quotation template, the quotation results corresponding to each item to be quoted are filled into the field positions corresponding to the quotation template to obtain the quotation sheet; wherein, the output format of the quotation sheet includes one or more of the following: editable table format and layout document format.

[0079] A quotation template refers to a predefined format template used to organize the content of a quotation, which may include fields for filling in product identification information, quantity information, unit price, subtotal, total, etc. An editable table format refers to a table-like file whose content can be further edited, such as a spreadsheet file. A formatted document refers to a document with a fixed layout that is easy to view and send, such as a portable document format file.

[0080] For example, the quotation generation module 104 reads a preset quotation template, fills in the product identification information of each item to be quoted into the "Product Name" field, fills in the required quantity information into the "Quantity" field, fills in the unit price and subtotal of the quotation result into the "Unit Price" and "Amount" fields respectively, and fills in the sum of all subtotals into the total field of the template, thus obtaining a quotation. The quotation generation module 104 can output the quotation in an editable table format for further adjustments by the business party, or it can output it in a layout document format for sending to the customer for viewing.

[0081] By filling the quotation results into the corresponding field positions based on the quotation template, the generated quotation has a more uniform field organization and supports multiple output formats, which is beneficial for viewing, adjusting and sending the quotation.

[0082] In one example, the automatic product pricing system also includes an auditing interaction module and an update module.

[0083] The review interaction module is used to display the quotation on the user terminal, update the quotation in response to an adjustment operation for any quotation record in the quotation, determine the adjustment type corresponding to the adjustment operation based on the field affected by the adjustment operation, record the difference data before and after the adjustment corresponding to the adjustment type and the adjustment operation, and obtain adjustment feedback data; wherein, the adjustment type includes one or more of the following: adjustment of product identification information, adjustment of product price information, and adjustment of applied sub-strategy.

[0084] In this context, "user terminal" refers to the device used by users to view and operate the quotation, such as a personal computer or mobile terminal. "Quotation record" refers to a single record within the quotation corresponding to a specific item to be quoted. "Adjustment operation" refers to the modification operation performed by the user on the quotation record, such as changing the product name, price, or applied discount rules. "Adjustment type" refers to the category of the adjustment determined by the field affected by the adjustment operation. "Difference data" refers to the difference between the content before and after the adjustment operation, such as the value before and after the modification. "Adjustment feedback data" refers to data that records the adjustment type and difference data.

[0085] For example, the review and interaction module displays a generated quotation on the user's terminal. The user notices an error in the product name of one of the quotation records and corrects "a certain brand of inkjet equipment" to "a certain brand of laser equipment." The review and interaction module updates the quotation accordingly, and based on the adjustment applied to the product name field, determines the adjustment type as "adjustment of product identification information," records the difference data before and after the modification, and obtains the corresponding adjustment feedback data. As another example, if a user changes the unit price of a quotation record from "1900 yuan" to "1850 yuan," the review and interaction module determines the adjustment type as "adjustment of product price information" and records the corresponding difference data.

[0086] The update module is used to determine the corresponding update object based on the adjustment type in the adjustment feedback data, and to update the update object based on the difference data in the adjustment feedback data; wherein, when the adjustment type is an adjustment to product identification information, the update object is the large language model called by the first intelligent agent module 102; when the adjustment type is an adjustment to product price information, the update object is the price information of the corresponding product record in the product data source; when the adjustment type is an adjustment to the applied sub-strategy, the update object is the preset pricing strategy.

[0087] The update object refers to the object that needs to be updated based on the difference data, as determined by the adjustment type. For example, when the adjustment type is an adjustment to product identification information, the update module can use the corresponding adjustment feedback data as a sample to optimize the large language model called by the first intelligent agent module 102, such as as a basis for subsequent training or prompt adjustments, so that the results of subsequent semantic extraction and matching are closer to the user's actual needs; when the adjustment type is an adjustment to product price information, the update module updates the price information of the corresponding product record in the product data source based on the difference data, so that the price information in the product data source remains at a relatively recent value; when the adjustment type is an adjustment to the applied sub-strategy, the update module updates the preset pricing strategy based on the difference data, such as adjusting the discount value or triggering condition of the corresponding sub-strategy.

[0088] The review and interaction module records the user's adjustment operations and obtains adjustment feedback data. Then, the update module updates the corresponding update objects according to the adjustment type. This allows the user's adjustment results to be fed back to the large language model, product data source, or preset pricing strategy. This helps to make the subsequently generated quotation more closely meet the user's actual needs and keeps the price information in the product data source and the preset pricing strategy updated.

[0089] In one example, the quote generation module 104 is further configured to:

[0090] After the quotation is generated but before it is confirmed to be sent, monitor the product price information corresponding to each product item to be quoted in the product data source.

[0091] When the price information of any item to be quoted changes more than a preset change threshold relative to the price information recorded in the quotation, the quotation result corresponding to the item to be quoted is regenerated based on the current price information of the item in the item data source and the preset quotation strategy, and an update prompt is given for the item to be quoted in the quotation.

[0092] "Confirmation and Sending" refers to the user's confirmation of the quotation and its subsequent sending to the customer. "Monitoring" refers to the process of continuously or periodically viewing changes in product price information in the product data source. "Change Range" refers to the degree of change in the current product price information in the product data source relative to the price information recorded in the quotation; it can be expressed as a difference or a ratio. "Preset Change Threshold" refers to a pre-set value used to determine whether the change range needs to trigger an update. "Update Notification" refers to an indicator or explanation given in the quotation for a specific product item to be quoted, indicating that its quotation result has changed.

[0093] For example, if a quotation is generated but not yet sent, and during this period the price information for "Equipment A100" in the product data source changes from "2000 yuan / unit" to "2200 yuan / unit," exceeding a preset change threshold, the quotation generation module 104 will recalculate the quotation result for the product item to be quoted based on the current product price information "2200 yuan / unit" and the preset quotation strategy, and update the quotation with an update notification, such as marking "Price updated." If the change does not exceed the preset change threshold, regeneration will not be triggered.

[0094] By monitoring changes in product price information after the quotation is generated but before confirmation and sending, and regenerating the quotation result and providing an update prompt when the change exceeds a preset change threshold, it is beneficial to ensure that the quotation sent to the customer corresponds to the latest price information in the product data source, thereby reducing the situation where the quotation is inconsistent with the actual price due to price changes.

[0095] Example 5:

[0096] Figure 2 This is a schematic diagram of the overall architecture of an automatic commodity pricing system based on a large model, provided in an embodiment of this application. (Refer to...) Figure 2The system, in the order of data processing, includes an information input module 101, a first intelligent agent module 102, a second intelligent agent module 103, a quotation generation module 104, and an audit interaction module 105. Among them, the first intelligent agent module 102, the second intelligent agent module 103, and the quotation generation module 104 are in a self-learning mechanism.

[0097] The information input module 101 is used to obtain a quotation requirement list. As mentioned before, the quotation requirement list can be one or more forms, such as a paper document, a spreadsheet, or an image.

[0098] The first intelligent agent module 102 includes an optical character recognition engine 1021 and an information extraction intelligent agent 1022. The optical character recognition engine 1021 performs optical character recognition on the text in unstructured formats such as images, scanned copies, or photographs of paper documents when the quotation request list is in such a format, obtaining the text content. The information extraction intelligent agent 1022 uses a large language model to perform semantic extraction on the text content, obtaining structured demand data including product identification information and quantity information for each item to be quoted. Specifically, when the quotation request list itself is in a directly readable format such as a spreadsheet or text file, the text content can be directly read without processing by the optical character recognition engine 1021. When the quotation request list is in an unstructured format such as an image, the text content is first processed by the optical character recognition engine 1021 and then submitted to the information extraction intelligent agent 1022 for semantic extraction. For example, a customer submits a photo of a paper demand form. The first intelligent agent module 102 first uses the optical character recognition engine 1021 to recognize the text "equipment A100 ten units" in the photo. Then, the information extraction intelligent agent 1022 performs semantic extraction and normalization processing on the text content to obtain the product identification information "equipment A100" and the demand quantity information "10".

[0099] The second intelligent agent module 103 is used to determine the product price information corresponding to each product item to be quoted from the product data source based on structured demand data, and adjust it according to a preset quoting strategy to obtain the quotation result corresponding to each product item to be quoted. In one embodiment, the second intelligent agent module 103 includes a query data analysis unit 1031 for converting text to structured query statements (Text to SQL); in another embodiment, the second intelligent agent module 103 includes a query data analysis unit 1032 for converting text to API calls (Text to API); in yet another embodiment, the second intelligent agent module 103 includes both of the above.

[0100] The text-to-structured query analysis unit 1031 is a processing unit that converts the requirements of the product item to be quoted into a structured query statement based on a large language model, and then queries the database, which serves as the data source for the product, to obtain the price information of the matching product record. For example, for a product item with the product identifier "Equipment A100" to be quoted, the query data analysis unit 1031 can generate a structured query statement using the large language model to query the product record and its price, and then retrieve the corresponding benchmark price information from the database accordingly. The text-to-interface call query data analysis unit 1032 is a processing unit that converts the requirements of the product item to be quoted into an interface call request based on the large language model, and then initiates a query to the service interface, which serves as the data source for the product, to obtain the price information of the matching product record. For example, for the same product item to be quoted, the query data analysis unit 1032 can generate a corresponding interface call request using the large language model, and then call the query interface provided by the e-commerce platform to obtain the price information of the product.

[0101] It should be noted that the "product data source" in the aforementioned embodiments is a general term, which can be either a data source providing product records in the form of a database, corresponding to the processing method of the query data analysis unit 1031 for converting text to structured query statements; or a data source providing product records in the form of a service interface, corresponding to the processing method of the query data analysis unit 1032 for converting text to interface calls. Both of the above implementation methods can be used to determine the product price information corresponding to each product item to be quoted from the product data source. Those skilled in the art can choose one of them or combine the two according to the actual form in which the product data source is provided.

[0102] The quotation generation module 104 is used to generate and output a quotation sheet corresponding to the quotation requirement list based on the quotation results corresponding to each item to be quoted.

[0103] The review and interaction module 105 is used to display the generated quotation on the user terminal, and update the quotation in response to the adjustment operation of the quotation record in the quotation, while recording the adjustment feedback data corresponding to the adjustment operation.

[0104] The self-learning mechanism is used to update the corresponding update objects based on the adjustment feedback data recorded by the review interaction module 105. This self-learning mechanism can be implemented by the update module 106 described later. For example, when the adjustment feedback data indicates that the adjustment type is an adjustment to the product identification information, the large language model called by the first intelligent agent module 102 can be optimized; when the adjustment type is an adjustment to the product price information, the price information of the corresponding product record in the product data source can be updated; when the adjustment type is an adjustment to the applied sub-strategy, the preset pricing strategy can be updated.

[0105] Through the above overall architecture, the quotation request list forms a complete processing link from input, identification and extraction, price matching and strategy adjustment, quotation generation to manual review. The self-learning mechanism enables the processing results of each module to be continuously updated based on the feedback from manual review, which helps to improve the efficiency and consistency of quotation processing.

[0106] Example 6:

[0107] Based on the same inventive concept, this application provides an automatic product pricing method, which is applied to the automatic product pricing system described in the foregoing embodiments. Figure 3 This is a schematic diagram of an automatic product quotation process provided in an embodiment of this application. (Refer to...) Figure 3 The method includes:

[0108] S301: Obtain a list of quotation requirements.

[0109] S302: The quotation request list is processed by the automatic commodity quotation system to obtain the quotation sheet corresponding to the quotation request list.

[0110] The automatic product quotation method provided in this application is applied to computer equipment, which can be intelligent devices such as personal computers and smart terminals, or servers such as application servers and business servers.

[0111] In one implementation, the computer device first acquires a list of quotation requests, which can come from pending requests in the business system or from documents submitted by customers. After acquisition, the list of quotation requests is input into an automatic product quotation system. The system's information input module acquires the list of quotation requests, and then a first intelligent agent module performs semantic extraction on its text content using a large language model to obtain structured demand data including product identification information and quantity information for each product item to be quoted. Next, a second intelligent agent module, based on the structured demand data, determines the corresponding product price information for each product item to be quoted from the product data source and adjusts it based on a preset quotation strategy to obtain the quotation result for each product item to be quoted. Finally, the quotation generation module generates a quotation sheet corresponding to the quotation request list based on the quotation results and can output the quotation sheet.

[0112] It should be noted that the principle of the automatic quotation method for this product in solving the technical problem has been described in the above embodiments. For details, please refer to embodiments 1-4 above, and will not be elaborated here.

[0113] Using the above method, the matching, pricing, and strategy adjustment of each item in the quotation request list are completed collaboratively by the system, reducing the amount of manual processing for each item and improving the efficiency and consistency of quotation processing.

[0114] Example 7:

[0115] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of this application, such as... Figure 4 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 4 Take a processor 10 as an example.

[0116] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0117] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.

[0118] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device as shown by a landing page for an app. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, which can be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0119] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0120] The computer device also includes an input device 30 and an output device 40. The processor 10, memory 20, input device 30, and output device 40 can be connected via a bus or other means. Figure 4 Taking the example of a connection between China and Israel via a bus.

[0121] Input device 30 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the computer device, such as a touchscreen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 40 may include display devices, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibration motors). The aforementioned display devices include, but are not limited to, liquid crystal displays, light-emitting diodes, displays, and plasma displays. In some alternative embodiments, the display device may be a touchscreen.

[0122] Example 8:

[0123] Based on the above embodiments, this application also provides a computer-readable storage medium storing a computer program executable by a processor. When the program runs on the processor, it causes the processor to perform the following steps:

[0124] Obtain a list of quote requirements;

[0125] The quotation request list is processed by the automatic quotation system to obtain the quotation list corresponding to the quotation request list.

[0126] Since the principle of the computer-readable storage medium in solving the problem is similar to that of the automatic commodity pricing method, the implementation of the computer-readable storage medium can be found in the embodiments of the method, and repeated details will not be repeated.

[0127] The storage media described in the embodiments of this application are intended to include, but are not limited to, these and any other suitable types of memory.

[0128] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this application can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers.

[0129] Although this application has been described in detail above with general descriptions and specific embodiments, some modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of this application fall within the scope of protection claimed in this application.

Claims

1. An automatic commodity pricing system based on a large model, characterized in that, The system includes: The information input module is used to obtain a list of quotation requirements; The first intelligent agent module is used to obtain the text content in the quotation demand list, and to extract the semantics of the text content through a large language model to obtain structured demand data including the product identification information and demand quantity information of each product item to be quoted. The second intelligent agent module is used to determine the commodity price information corresponding to each commodity item to be quoted from the commodity data source based on the structured demand data, and adjust the commodity price information corresponding to each commodity item to be quoted based on the preset quoting strategy to obtain the quoting result corresponding to each commodity item to be quoted. The quotation generation module is used to generate a quotation sheet corresponding to the quotation requirement list based on the quotation results corresponding to each of the items to be quoted.

2. The system according to claim 1, characterized in that, The first intelligent agent module is specifically used for: The text content is semantically extracted using the large language model to obtain preliminary descriptions of each item in the quotation request list. Using the large language model, the preliminary description information is standardized based on the preset product description specifications to obtain the product identification information corresponding to each product item to be quoted.

3. The system according to claim 1, characterized in that, The second intelligent agent module is specifically used for: For any item to be quoted in the structured demand data, the product identification information of the item to be quoted is matched with the product records in the product data source to determine the benchmark price information corresponding to the product record that matches the item to be quoted. Based on the benchmark price information and the preset pricing strategy, the pricing result corresponding to the item to be quoted is determined.

4. The system according to claim 3, characterized in that, The second intelligent agent module is specifically used for: Using the large language model, the semantic similarity between the product identification information of the product item to be quoted and the product identification information of each product record in the product data source is determined; Based on the semantic similarity, a set of candidate product records corresponding to the product item to be quoted is determined from the product data source; The product record with the highest semantic similarity in the candidate product record set is identified as the product record that matches the product item to be quoted.

5. The system according to claim 4, characterized in that, The second intelligent agent module is also used for: When the candidate product record set corresponding to the product item to be quoted is empty, or the highest semantic similarity in the candidate product record set is lower than a preset similarity threshold, the product item to be quoted is marked as a pending product item, and the quotation result corresponding to the pending product item is set to a preset pending state value. The quotation generation module is also used for: The pending product items are marked in the quotation, and fields corresponding to the pending product items are reserved in the quotation.

6. The system according to claim 1, characterized in that, The preset pricing strategy includes multiple sub-strategies, each of which is associated with a triggering condition and a preset priority. The second intelligent agent module is specifically used for: For any of the items to be quoted, based on the item's identification information and demand quantity information, a target sub-strategy matching the triggering condition of the item to be quoted is determined from the preset quoting strategies. When there are multiple target sub-strategies, the multiple target sub-strategies are sorted according to the preset priority associated with each target sub-strategy, and the product price information corresponding to the product item to be quoted is adjusted in sequence according to the sorted order to obtain the quotation result corresponding to the product item to be quoted.

7. The system according to claim 1, characterized in that, The quotation generation module is specifically used for: Based on the preset quotation template, the quotation results corresponding to each item to be quoted are filled into the field positions corresponding to the quotation template to obtain the quotation sheet; The output format of the quotation includes one or more of the following: editable table format and layout document format.

8. The system according to claim 1, characterized in that, The system also includes: The review and interaction module is used to display the quotation on the user terminal, update the quotation in response to the adjustment operation for any quotation record in the quotation, determine the adjustment type corresponding to the adjustment operation based on the field affected by the adjustment operation, record the difference data before and after the adjustment corresponding to the adjustment type and the adjustment operation, and obtain adjustment feedback data. The adjustment types include one or more of the following: adjustments to product identification information, adjustments to product price information, and adjustments to the applied sub-strategies.

9. The system according to claim 8, characterized in that, The system also includes: The update module is used to determine the corresponding update object based on the adjustment type in the adjustment feedback data, and update the update object based on the difference data in the adjustment feedback data; Specifically, when the adjustment type is an adjustment to product identification information, the update object is the large language model called by the first intelligent agent module; when the adjustment type is an adjustment to product price information, the update object is the price information of the corresponding product record in the product data source; and when the adjustment type is an adjustment to the applied sub-strategy, the update object is the preset pricing strategy.

10. The system according to claim 1, characterized in that, The quotation generation module is also used for: After the quotation is generated but before it is confirmed to be sent, monitor the product price information corresponding to each product item to be quoted in the product data source. When the price information of any item to be quoted changes more than a preset change threshold relative to the price information recorded in the quotation, the quotation result corresponding to the item to be quoted is regenerated based on the current price information of the item in the item data source and the preset quotation strategy, and an update prompt is given for the item to be quoted in the quotation.

11. A method for automatically quoting prices of goods based on the system according to any one of claims 1-10, characterized in that, The method includes: Obtain a list of quote requirements; The quotation request list is processed by the automatic quotation system to obtain the quotation list corresponding to the quotation request list.

Images