ERP intelligent document management method and system based on natural language processing

Abstract

The embodiment of the invention provides an ERP (Enterprise Resource Planning) intelligent document management method and system based on natural language processing, which optimizes an initialized natural language processing network by obtaining the sample learning and verification sequence comprising the template after basic and semantic enhancement processing, not only realizes efficient natural language processing of an ERP intelligent document, but also improves the processing efficiency of the ERP intelligent document. And through an iterative verification and optimization mechanism, the accuracy and performance of the natural language processing network are ensured. Finally, under the support of the optimized natural language processing network, an accurate natural language processing result can be generated for the ERP intelligent document of the target online service, and the management efficiency and the intelligent level of the ERP intelligent document are remarkably improved.

Description

Technical Field

[0001] The present application relates to the field of natural language processing technology, and more specifically, to an ERP intelligent document management method and system based on natural language processing. Background Art

[0002] With the widespread adoption of enterprise resource planning (ERP) systems, the management of ERP intelligent documents has become a crucial component of enterprise information technology development. Traditional ERP document management methods rely heavily on manual operations, which are inefficient and error-prone. In recent years, the rapid development of natural language processing (NLP) technology has opened up new possibilities for automated management of ERP intelligent documents.

[0003] However, existing NLP-based ERP document management methods still have some shortcomings. On the one hand, these methods often lack optimization for the characteristics of ERP documents, resulting in poor processing results. On the other hand, when processing complex and diverse ERP documents, these methods often cannot accurately understand and extract key information from the documents, thus affecting the accuracy and efficiency of document management. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide an ERP intelligent document management method and system based on natural language processing.

[0005] In conjunction with the first aspect of the present application, a natural language processing-based ERP intelligent document management method is provided, which is applied to an ERP intelligent document management system based on natural language processing, and the method includes:

[0006] Obtaining a sample ERP smart document learning sequence and a sample ERP smart document verification sequence; the template ERP smart documents in the sample ERP smart document learning sequence and the sample ERP smart document verification sequence include a basic template ERP smart document and a template ERP smart document generated by semantically enhancing target text blocks of each of the basic template ERP smart documents;

[0007] Optimize and initialize a natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence;

[0008] Using the optimized and initialized natural language processing network to perform natural language processing on the verification template ERP smart document in the sample ERP smart document verification sequence, and generating an ERP smart document verification result corresponding to the verification template ERP smart document;

[0009] When it is determined based on the ERP smart document verification result that the optimization termination requirement is met, the optimized initialized natural language processing network is used as the natural language processing network; otherwise, the network parameter information of the initialized natural language processing network is updated, and the process returns to the step of optimizing the initialized natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence;

[0010] An ERP smart document of a target online service is obtained, an optimized natural language processing network is activated, and natural language processing is performed on the ERP smart document using the natural language processing network to generate a natural language processing result.

[0011] In a possible implementation of the first aspect, obtaining a sample ERP smart document learning sequence and a sample ERP smart document verification sequence includes:

[0012] Obtaining a basic template ERP smart document for target text blocks, enhancing the basic template ERP smart document, and generating a target number of template ERP smart documents for target text blocks;

[0013] The template ERP smart document is allocated into a sample ERP smart document learning sequence and a sample ERP smart document verification sequence based on a target ratio.

[0014] In a possible implementation of the first aspect, the template ERP smart document generated after semantic enhancement processing is performed on the target text blocks of each of the basic template ERP smart documents includes:

[0015] The text description paragraphs of the target text blocks of each basic template ERP smart document are respectively enhanced at multiple semantic levels to generate a template ERP smart document after the semantic enhancement processing.

[0016] In a possible implementation of the first aspect, optimizing and initializing a natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0017] Extracting multiple document feature vectors of each sample template ERP smart document in the sample ERP smart document learning sequence, wherein the document feature vectors include a structural feature vector, a semantic feature vector, and a format feature vector;

[0018] Based on the multiple document feature vectors of each sample template ERP smart document, the initialized natural language processing network is optimized.

[0019] In a possible implementation of the first aspect, extracting a structural feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0020] Obtaining a document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence;

[0021] Normalizing each document segmentation node in the document segmentation sequence, calculating the semantic weight and semantic direction of each document segmentation node, and generating a document position identifier, semantic weight, and semantic direction corresponding to each document segmentation node;

[0022] The semantic weight and semantic direction associated with each of the document word segmentation nodes are statistically calculated using a semantic distribution space, and the document word segmentation sequence is converted into a structural feature vector.

[0023] In a possible implementation of the first aspect, obtaining a document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0024] Obtain each document word segmentation node of each sample template ERP smart document in the sample ERP smart document learning sequence;

[0025] Calculating the weight value of each document segmentation node;

[0026] The document word segmentation nodes whose weight values ​​are less than the preset weight value are deleted to generate a document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence.

[0027] In a possible implementation of the first aspect, extracting the semantic feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0028] Using a preset semantic analysis framework as a feature extraction basis, a keyword association graph of each sample template ERP smart document in the sample ERP smart document learning sequence is constructed;

[0029] The keyword association map is used to extract the semantic core set of each sample template ERP smart document based on a preset association threshold, and the semantic feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence is described by the semantic core set.

[0030] In a possible implementation of the first aspect, extracting the format feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0031] Dividing each sample template ERP smart document in the sample ERP smart document learning sequence into a preset number of blocks;

[0032] Obtaining a baseline format identification value of each of the blocks, comparing the baseline format identification value with a preset format identification threshold corresponding to each of the blocks, and obtaining enhanced format feature data of each of the blocks;

[0033] Perform frequency graph statistics on each of the blocks after the enhanced format feature data to obtain a block document sub-segment frequency graph, and use the block document sub-segment frequency graph to describe the format feature vectors of each sample template ERP smart document in the sample ERP smart document learning sequence.

[0034] In combination with the second aspect of the present application, an ERP intelligent document management system based on natural language processing is provided. The ERP intelligent document management system based on natural language processing includes a machine-readable storage medium and a processor. The machine-readable storage medium stores machine-executable instructions. When the processor executes the machine-executable instructions, the ERP intelligent document management system based on natural language processing implements the aforementioned ERP intelligent document management method based on natural language processing.

[0035] In combination with the third aspect of the present application, a computer-readable storage medium is provided, in which computer-executable instructions are stored. When the computer-executable instructions are executed, the aforementioned ERP intelligent document management method based on natural language processing is implemented.

[0036] In combination with any of the above aspects, the embodiments of the present application optimize the initialization of the natural language processing network by acquiring a sample learning and verification sequence containing both basic and semantically enhanced templates. This not only enables efficient natural language processing of ERP smart documents, but also ensures the accuracy and performance of the natural language processing network through an iterative verification and optimization mechanism. Ultimately, with the support of the optimized natural language processing network, accurate natural language processing results can be generated for the ERP smart documents of the target online service, significantly improving the management efficiency and intelligence level of ERP smart documents. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained by combining these drawings without paying any creative work.

[0038] Figure 1A flowchart of an ERP intelligent document management method based on natural language processing provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0040] The terms "first," "second," and so on, in the specification and claims of the present invention and the accompanying drawings are used to distinguish between different items, not to describe a specific order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or end comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed therein, or may optionally include other steps or elements inherent to such process, method, product, or end.

[0041] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0042] Figure 1 The flowchart of the ERP intelligent document management method based on natural language processing provided by an embodiment of the present application is shown. It should be understood that in other embodiments, the order of some steps in the ERP intelligent document management method based on natural language processing of this embodiment can be shared based on actual needs, or some steps can be omitted or maintained. The details of the ERP intelligent document management method based on natural language processing include:

[0043] Step S110: Obtain a sample ERP smart document learning sequence and a sample ERP smart document verification sequence. The template ERP smart documents in the sample ERP smart document learning sequence and the sample ERP smart document verification sequence include a basic template ERP smart document and a template ERP smart document generated by semantically enhancing target text blocks of each basic template ERP smart document.

[0044] In this embodiment, the server is processing a smart document task related to the ERP system of a large manufacturing enterprise. In this enterprise's ERP system, there are various business processes, such as procurement, production, sales, inventory management, etc., and each business process has a corresponding basic template ERP smart document.

[0045] The ERP smart document, the foundational template for the procurement process, contains text blocks containing content such as purchase requisitions, purchase orders, and supplier information. The server first retrieves these foundational template ERP smart documents. It then performs semantic enhancement on the target text blocks within these foundational template ERP smart documents. For example, within the purchase requisition text block, there is a paragraph describing the purchased items, which might originally simply list the item name, quantity, and specifications. The server then performs semantic enhancement on this paragraph, operating at different semantic levels. At the vocabulary level, some common words were replaced with more specialized and precise ones, such as replacing "things" with the specific "electronic components." At the sentence level, the logical structure of descriptions was optimized, consolidating previously loose descriptions into more organized sentences, such as "We need to purchase 100 electronic components of specification model X. These components are used on specific equipment on the production line and are critical to their normal operation." At the paragraph level, more context and relevant information were added, such as "According to market forecasts and current production plans, demand for this electronic component will continue to increase over the next month. Currently, only 50 units remain in inventory. To avoid production interruptions, the purchase quantity is set at 100 units. Furthermore, specification model X of this component has been rigorously tested and is fully compatible with the equipment." Through this process, a new template ERP smart document was generated.

[0046] In this manner, the server performs semantic enhancement on the target text blocks of the basic template ERP smart documents corresponding to business processes such as production, sales, and inventory management, resulting in a series of template ERP smart documents. The server then allocates these template ERP smart documents into a sample ERP smart document learning sequence and a sample ERP smart document verification sequence based on a target ratio, such as an 80:20 ratio. The template ERP smart documents in the sample ERP smart document learning sequence are used to subsequently optimize the initial natural language processing network, while the template ERP smart documents in the sample ERP smart document verification sequence are used to verify the performance of the optimized network.

[0047] Step S120 , optimizing and initializing a natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence.

[0048] After receiving the sample ERP smart document learning sequence, the server begins optimizing the initialized natural language processing network. First, the server extracts multiple document feature vectors for each sample template ERP smart document in the sample ERP smart document learning sequence, including structural feature vectors, semantic feature vectors, and format feature vectors.

[0049] Taking a sample template ERP smart document in the procurement process as an example, the server obtains the document segmentation sequence of this document. Each word, phrase, etc. in this document is a document segmentation node. For example, there are segmentation nodes such as "purchase order number", "supplier name", "purchased items", "quantity", and "delivery date" in the document. The server calculates the weight value of each document segmentation node. The calculation of the weight value may be based on factors such as the frequency of occurrence of the node in the entire document and its importance in the business process. For those segmentation nodes that appear very rarely in the document and are not very important to the business process, such as some decorative words or segmentation nodes corresponding to rarely used backup information, if their weight value is less than the preset weight value, the server will delete them, thereby generating the document segmentation sequence of this sample template ERP smart document.

[0050] The server then normalizes each document segmentation node in the document segmentation sequence. For example, all uppercase letters may be converted to lowercase and some punctuation may be removed. The server then calculates the semantic weight and direction of each document segmentation node. For the segmentation node "Purchase Order Number," its semantic weight may be high because it uniquely identifies a purchase order, and its semantic direction points to a specific procurement business instance. Conjunctions such as "and" have lower semantic weights. The server generates a corresponding document location identifier, semantic weight, and semantic direction for each document segmentation node. For example, the document location identifier for "Purchase Order Number" might be the beginning of the document, with a semantic weight of 0.8 (hypothetically) and a positive semantic direction (indicating that it is important identifying information in normal business processes). Finally, the server uses a semantic distribution space to calculate the semantic weights and directions associated with each document segmentation node, converting the document segmentation sequence into a structural feature vector. This structural feature vector reflects the overall structural information of the document, such as which parts are concentrated with key information and which parts are auxiliary information.

[0051] The server uses a pre-set semantic analysis framework as the basis for feature extraction to construct a keyword association graph for this sample ERP smart document template for the procurement process. In this graph, "procurement" is the core keyword, and associations are established with keywords such as "supplier," "item," "quantity," and "delivery date." The strength of the association can be determined by factors such as the frequency of co-occurrence of the terms in the document and their semantic proximity. For example, "procurement" and "item" co-occur frequently and are directly semantically related, resulting in a relatively strong association.

[0052] The server then uses this keyword association graph to extract the semantic core set of the sample template ERP smart document based on a preset relevance threshold. Assuming a relevance threshold of 0.6, keywords with a relevance greater than 0.6 to "procurement," such as "supplier," "item," and "quantity," constitute the semantic core set. This semantic core set describes the semantic feature vector of the sample template ERP smart document, reflecting the document's core semantic content—the key elements surrounding procurement operations.

[0053] The server divides the sample template ERP smart document of this procurement process into a preset number of blocks. For example, the document is divided into three blocks according to the header, body content, and footer. For each block, the server obtains its baseline format identification value. For example, the font format, font size, alignment, etc. of the header constitute the baseline format identification value of the header. This baseline format identification value is then compared with the preset format identification threshold corresponding to each block to obtain the enhanced format feature data of each block. If the font size of the header should be between 12 and 14 (preset format identification threshold), and the actual font size is 16 (benchmark format identification value), then enhanced format feature data about the font size will be obtained.

[0054] The server then performs frequency graph statistics on each block after obtaining the enhanced format feature data, obtaining a block document sub-segment frequency graph. For example, within the main content block, different text sub-segments (such as different table rows or paragraphs) appear at different frequencies. By counting these frequencies and plotting them into a graph, a block document sub-segment frequency graph is obtained. This block document sub-segment frequency graph can describe the format feature vector of this sample template ERP smart document, reflecting the document's format characteristics, such as the degree of format standardization of different sections and the distribution of content.

[0055] Based on the structural, semantic, and format feature vectors extracted from each sample template ERP smart document, the server optimizes the initialized natural language processing network. For example, it adjusts the network's neuron connection weights and activation function parameters based on these feature vectors, enabling the network to better adapt to the processing requirements of ERP smart documents.

[0056] Step S130 , using the optimized initialized natural language processing network to perform natural language processing on the verification template ERP smart document in the sample ERP smart document verification sequence, and generating an ERP smart document verification result corresponding to the verification template ERP smart document.

[0057] In this embodiment, after optimizing and initializing the natural language processing network, the server begins processing the verification template ERP smart document in the sample ERP smart document verification sequence. Taking a verification template ERP smart document in the sales process as an example, this document contains sales order information, customer information, sales product information, etc.

[0058] The optimized natural language processing network first performs word segmentation on the document text and then analyzes it based on the structural, semantic, and format features learned during the previous optimization process. For sales order information, the network is able to identify whether the order number is formatted correctly because it has learned the structural characteristics of order numbers in similar documents. If the order number in the validation template ERP smart document is not in a standard format, such as containing disallowed characters, the network will detect this and mark it as an incorrectly formatted order number in the ERP smart document validation results.

[0059] For customer information, the network analyzes the plausibility of information such as the customer name and contact information based on semantic feature vectors. If the customer name is garbled or the contact information doesn't conform to standard phone number or email address formats, the network can accurately identify the issue in the ERP intelligent document verification results. For example, if the customer name contains special symbols that should not be included in a normal semantic understanding, the network will determine that the customer name has a semantic anomaly.

[0060] For sales product information, the network will check the accuracy and consistency of information such as product name, quantity, and price. If the product name does not match the company's internal product catalog, or the quantity and price values ​​do not conform to the business logic (for example, the quantity is negative or the price is too high or too low outside the reasonable range), the network will also give corresponding prompts in the ERP smart document verification results. By performing such natural language processing on each part of the entire verification template ERP smart document, the server generates the ERP smart document verification result corresponding to the verification template ERP smart document. This result records in detail the formatting, semantics and other issues or correct information in the document.

[0061] Step S140: When it is determined based on the ERP smart document verification result that the optimization termination requirements are met, the optimized and generated initialized natural language processing network is used as the natural language processing network; otherwise, the network parameter information of the initialized natural language processing network is updated, and the step of optimizing the initialized natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence is returned.

[0062] After obtaining the ERP smart document verification result corresponding to the verification template ERP smart document, the server determines whether it meets the optimization termination requirement. Assume that the optimization termination requirement is that the error rate in the verification template ERP smart document is lower than a certain threshold, such as 5%.

[0063] If the error rate in the validation results of the ERP smart document, a verification template for this sales process, is less than 5%, this indicates that the optimized natural language processing network has achieved relatively good performance when processing the ERP smart document. The server will then use the optimized initial natural language processing network as the final natural language processing network for subsequent processing of the ERP smart document for the target online service.

[0064] However, if the error rate is higher than 5%, it means that the network needs further optimization. The server will then update the network parameter information for the initialized natural language processing network. For example, based on the type and distribution of errors in the verification results, the network parameters related to structural feature analysis will be adjusted. If there are many format errors, the network parameter adjustment for the format feature vector processing part may be strengthened; if there are many semantic errors, the parameters related to semantic feature vector processing will be adjusted. The server will then return to the steps for optimizing the initialized natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence, and will again use the sample template ERP smart document in the sample ERP smart document learning sequence to optimize the network after the updated parameters. This process will be iterated continuously until the optimization termination requirements are met.

[0065] Step S150: Acquire the ERP smart document of the target online service, activate the optimized natural language processing network, use the natural language processing network to perform natural language processing on the ERP smart document, and generate a natural language processing result.

[0066] Assume that a large manufacturing enterprise has an online sales service platform that generates ERP smart documents containing a large amount of sales business data. After the server obtains the ERP smart document for this target online service, it activates the previously optimized natural language processing network to process it.

[0067] This ERP smart document contains a large amount of sales order information, including orders from different customers and sales figures for different products. The optimized natural language processing network first performs word segmentation on the document and then analyzes it based on previously learned structural, semantic, and format features.

[0068] For order information, the network can quickly and accurately determine order validity. If the customer information, product information, quantity, price, and other information in the order are reasonable and meet format requirements, the network will mark the order as valid in the natural language processing results. If there are any problems, such as incomplete customer information or product quantity and price calculations that do not comply with the company's pricing strategy, the network will clearly point out the problem in the natural language processing results.

[0069] For statistical information within the entire ERP smart document, such as sales figures and sales volume statistics, the network can verify this data based on semantic understanding. If sales figures are found to be inconsistent with the total amount of each order, the network will provide a prompt in the natural language processing results, indicating possible statistical errors. Through this process, the server uses the optimized natural language processing network to perform comprehensive natural language processing on the ERP smart document of the target online service, generating natural language processing results that include the accuracy and formatting of various information in the document. This result provides important reference for the company's ERP system management and decision-making.

[0070] Based on the above steps, this embodiment of the application optimizes the initialization of the natural language processing network by acquiring a sample learning and verification sequence containing both basic and semantically enhanced templates. This not only enables efficient natural language processing of ERP smart documents, but also ensures the accuracy and performance of the natural language processing network through an iterative verification and optimization mechanism. Ultimately, with the support of the optimized natural language processing network, accurate natural language processing results can be generated for the ERP smart documents of the target online service, significantly improving the management efficiency and intelligence level of ERP smart documents.

[0071] In a possible implementation, obtaining a sample ERP smart document learning sequence and a sample ERP smart document verification sequence includes:

[0072] A basic template ERP smart document of the target text block is obtained, the basic template ERP smart document is enhanced, and a template ERP smart document of the target text block of the target number is generated.

[0073] The template ERP smart document is allocated into a sample ERP smart document learning sequence and a sample ERP smart document verification sequence based on a target ratio.

[0074] In a possible implementation, the template ERP smart document generated after semantic enhancement processing is performed on the target text blocks of each basic template ERP smart document includes:

[0075] The text description paragraphs of the target text blocks of each basic template ERP smart document are respectively enhanced at multiple semantic levels to generate a template ERP smart document after the semantic enhancement processing.

[0076] In one possible implementation, optimizing and initializing a natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0077] Extract multiple document feature vectors of each sample template ERP smart document in the sample ERP smart document learning sequence, wherein the document feature vectors include a structural feature vector, a semantic feature vector, and a format feature vector.

[0078] Based on the multiple document feature vectors of each sample template ERP smart document, the initialized natural language processing network is optimized.

[0079] In a possible implementation, extracting the structural feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0080] Obtain a document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence.

[0081] Each document segmentation node in the document segmentation sequence is standardized, the semantic weight and semantic direction of each document segmentation node are calculated, and the document position identifier, semantic weight and semantic direction corresponding to each document segmentation node are generated.

[0082] The semantic weight and semantic direction associated with each of the document word segmentation nodes are statistically calculated using a semantic distribution space, and the document word segmentation sequence is converted into a structural feature vector.

[0083] In a possible implementation, obtaining the document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0084] Obtain each document word segmentation node of each sample template ERP smart document in the sample ERP smart document learning sequence.

[0085] Calculate the weight value of each document segmentation node.

[0086] The document word segmentation nodes whose weight values ​​are less than the preset weight value are deleted to generate a document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence.

[0087] In a possible implementation, extracting the semantic feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0088] Using the preset semantic analysis framework as the basis for feature extraction, a keyword association map of each sample template ERP smart document in the sample ERP smart document learning sequence is constructed.

[0089] The keyword association map is used to extract the semantic core set of each sample template ERP smart document based on a preset association threshold, and the semantic feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence is described by the semantic core set.

[0090] In a possible implementation, extracting the format feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes:

[0091] Each sample template ERP smart document in the sample ERP smart document learning sequence is divided into a preset number of blocks.

[0092] A baseline format identification value of each block is obtained, and the baseline format identification value is compared with a preset format identification threshold corresponding to each block to obtain enhanced format feature data of each block.

[0093] Perform frequency graph statistics on each of the blocks after the enhanced format feature data to obtain a block document sub-segment frequency graph, and use the block document sub-segment frequency graph to describe the format feature vectors of each sample template ERP smart document in the sample ERP smart document learning sequence.

[0094] In the above embodiment, the natural language processing-based ERP intelligent document management system for executing the above method embodiment has at least one processor, a control module (chip set) coupled to at least one of the (at least one) processors, a memory coupled to the control module, a non-volatile memory (NVM) / storage device coupled to the control module, at least one load / output device coupled to the control module, and a network interface coupled to the control module.

[0095] The processor may include at least one single-core or multi-core processor, and the processor may include any combination of general-purpose processors or special-purpose processors (such as graphics processors, application processors, baseband processors, etc.). For some alternative implementations, the ERP intelligent document management system based on natural language processing can be used as the gateway or other electronic device described in the embodiments of this application.

[0096] For some alternative embodiments, the ERP intelligent document management system based on natural language processing may include at least one computer-readable medium (e.g., a memory or NVM / storage device) having instructions and at least one processor integrated with the at least one computer-readable medium and configured to execute the instructions to implement a module to perform the actions described in the present disclosure.

[0097] For one embodiment, the control module may include any suitable interface controller to provide any suitable interface to at least one of the processor(s) and / or any suitable device or component in communication with the control module.

[0098] The control module may include a memory controller module to provide an interface to the memory. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0099] The memory may be used, for example, to load and store data and / or instructions for an ERP intelligent document management system based on natural language processing. For one embodiment, the memory may include any appropriate volatile memory, for example, an appropriate DRAM.

[0100] For one embodiment, the control module may include at least one load / output controller to provide an interface to the NVM / storage device and the (at least one) load / output device.

[0101] For example, NVM / storage devices may be used to store data and / or instructions. The NVM / storage devices may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (at least one) non-volatile storage device (e.g., at least one hard disk drive (HDD), at least one compact disc (CD) drive, and / or at least one digital versatile disc (DVD) drive).

[0102] The NVM / storage device may include a storage resource that is physically part of the device on which the natural language processing-based ERP intelligent document management system is installed, or it may be accessible to the device without being part of the device. For example, the NVM / storage device may be accessed via (at least one) load / output device over a network.

[0103] The (at least one) loading / output device may provide an interface for the ERP intelligent document management system based on natural language processing to communicate with any other appropriate device. The loading / output device may include a communication component, a pinyin component, a sensor component, etc. The network interface may provide an interface for the ERP intelligent document management system based on natural language processing to communicate based on at least one network. The ERP intelligent document management system based on natural language processing may wirelessly communicate with at least one component of a wireless network based on any of at least one wireless network priors and / or protocols, for example, access a wireless network based on communication priors.

[0104] For one embodiment, at least one of the (at least one) processors may be loaded together with the logic of at least one controller of a control module (e.g., a memory controller module). For one embodiment, at least one of the (at least one) processors may be loaded together with the logic of at least one controller of a control module to form a system-level load. For one embodiment, at least one of the (at least one) processors may be fused on the same die with the logic of at least one controller of a control module. For one embodiment, at least one of the (at least one) processors may be fused on the same die with the logic of at least one controller of a control module to form a system-on-chip (SoC).

[0105] The above is a detailed introduction to the embodiments of the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for those skilled in the art, based on the ideas of the present application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0106] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the ERP intelligent document management method based on natural language processing described in the aforementioned embodiment.

[0107] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the ERP intelligent document management method based on natural language processing described in the aforementioned embodiment.

[0108] The device embodiments described above are merely illustrative, wherein the modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, i.e., they may be located in one place or distributed across multiple modules. Some or all of the modules may be selected based on actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement the present invention without inventive effort.

[0109] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus the necessary general hardware platform, or of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, the storage medium including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electronically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of storing or storing data.

[0110] Finally, it should be noted that what is disclosed above is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An ERP intelligent document management method based on natural language processing, characterized in that: The method comprises: Obtaining a sample ERP smart document learning sequence and a sample ERP smart document verification sequence; the template ERP smart documents in the sample ERP smart document learning sequence and the sample ERP smart document verification sequence include a basic template ERP smart document, and a template ERP smart document generated by semantically enhancing target text blocks of each of the basic template ERP smart documents; Optimize and initialize the natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence; Using the optimized and initialized natural language processing network to perform natural language processing on the verification template ERP smart document in the sample ERP smart document verification sequence, and generating an ERP smart document verification result corresponding to the verification template ERP smart document; When it is determined based on the ERP smart document verification result that the optimization termination requirements are met, the optimized and generated initialized natural language processing network is used as the natural language processing network; otherwise, the network parameter information of the initialized natural language processing network is updated, and the step of optimizing the initialized natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence is returned; The ERP smart document of the target online service is obtained, the optimized natural language processing network is enabled, and the natural language processing network is used to perform natural language processing on the ERP smart document to generate a natural language processing result.

2. The ERP intelligent document management method based on natural language processing according to claim 1 is characterized in that: The obtaining of the sample ERP smart document learning sequence and the sample ERP smart document verification sequence includes: Obtain a basic template ERP smart document for target text blocks, enhance the basic template ERP smart document, and generate a template ERP smart document for a target number of target text blocks; The template ERP smart document is allocated into a sample ERP smart document learning sequence and a sample ERP smart document verification sequence based on a target ratio.

3. The ERP intelligent document management method based on natural language processing according to claim 1 is characterized in that: The template ERP smart document generated after semantic enhancement processing is performed on the target text blocks of each basic template ERP smart document includes: The text description paragraphs of the target text blocks of each of the basic template ERP smart documents are enhanced at multiple semantic levels to generate a template ERP smart document after the semantic enhancement processing.

4. The ERP intelligent document management method based on natural language processing according to claim 1 is characterized in that: The optimizing and initializing natural language processing network based on the sample template ERP smart document in the sample ERP smart document learning sequence includes: Extracting multiple document feature vectors of each sample template ERP smart document in the sample ERP smart document learning sequence, wherein the document feature vectors include a structural feature vector, a semantic feature vector, and a format feature vector; Based on the multiple document feature vectors of each of the sample template ERP smart documents, the initialized natural language processing network is optimized.

5. The ERP intelligent document management method based on natural language processing according to claim 4 is characterized in that: Extracting the structural feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes: Obtaining a document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence; Standardizing each document segmentation node in the document segmentation sequence, calculating the semantic weight and semantic direction of each document segmentation node, and generating a document position identifier, semantic weight and semantic direction corresponding to each document segmentation node; The semantic weight and semantic direction associated with each of the document word segmentation nodes are counted using a semantic distribution space, and the document word segmentation sequence is converted into a structural feature vector.

6. The ERP intelligent document management method based on natural language processing according to claim 5 is characterized in that: The step of obtaining the document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence includes: Obtain each document word segmentation node of each sample template ERP smart document in the sample ERP smart document learning sequence; Calculate the weight value of each document segmentation node; The document word segmentation nodes whose weight values ​​are less than the preset weight values ​​are deleted to generate the document word segmentation sequence of each sample template ERP smart document in the sample ERP smart document learning sequence.

7. The ERP intelligent document management method based on natural language processing according to claim 4 is characterized in that: Extracting the semantic feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes: Using the preset semantic analysis framework as the basis for feature extraction, a keyword association map of each sample template ERP smart document in the sample ERP smart document learning sequence is constructed; The keyword association map is used to extract the semantic core set of each sample template ERP smart document based on a preset association threshold, and the semantic feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence is described by the semantic core set.

8. The ERP intelligent document management method based on natural language processing according to claim 4 is characterized in that: Extracting the format feature vector of each sample template ERP smart document in the sample ERP smart document learning sequence includes: Divide each sample template ERP smart document in the sample ERP smart document learning sequence into a preset number of blocks; Acquire a reference format identification value of each of the blocks, compare the reference format identification value with a preset format identification threshold corresponding to each of the blocks, and obtain enhanced format feature data of each of the blocks; Perform frequency graph statistics on each of the blocks after the enhanced format feature data to obtain a block document sub-segment frequency graph, and use the block document sub-segment frequency graph to describe the format feature vectors of each sample template ERP smart document in the sample ERP smart document learning sequence.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are executed by a computer, the ERP intelligent document management method based on natural language processing described in any one of claims 1 to 8 is implemented.

10. An ERP intelligent document management system based on natural language processing, characterized in that: It includes a processor and a computer-readable storage medium, wherein the computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are executed by a computer, the ERP intelligent document management method based on natural language processing described in any one of claims 1 to 8 is implemented.

Images