Supply relation management method based on artificial intelligence

Abstract

The invention relates to a supply relation management method based on artificial intelligence. The method comprises the following steps: performing data analysis on public text data and communication records of suppliers by adopting an artificial intelligence technology based on deep learning, respectively extracting semantic feature representations of the public text data and the communication records of the suppliers, and performing semantic feature adaptive fusion processing on the public text data and the communication records of the suppliers; and the potential risk level of the supplier is judged based on cross validation analysis of public text data and communication records. In this way, an enterprise can be helped to identify and manage potential risks in a supply chain more accurately, and therefore the stability and safety of production and operation of the enterprise are improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular, to a supply relationship management method based on artificial intelligence. Background Art

[0002] In the chemical industry, supply relationship management between chemical companies and suppliers is an important part of corporate operations. It involves multiple levels such as raw material procurement, production collaboration, product distribution, etc., and is directly related to the production efficiency, cost control, product quality and market response speed of chemical companies.

[0003] Stable and high-quality suppliers can provide high-quality raw materials, improve the quality of corporate products, and ensure a stable supply of raw materials to avoid production delays or shutdowns caused by supply interruptions. With the increasingly fierce competition in the global market, chemical companies need to track and evaluate the performance of suppliers in real time and promptly identify potential supply risks, such as suppliers' financial stability, compliance, and delivery on time, in order to optimize supply chain management and enhance corporate competitiveness.

[0004] However, in terms of supplier qualification assessment and risk prediction, traditional management methods mainly rely on manual review and experience judgment, which is inefficient and easily affected by human factors, prone to omissions, and difficult to achieve comprehensive and accurate supplier management. Therefore, a supply relationship management method based on artificial intelligence is expected. Summary of the invention

[0005] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0006] In a first aspect, the present application provides a supply relationship management method based on artificial intelligence, the method comprising: Obtaining public text information of the tracked supplier and communication records with the tracked supplier; Performing segment-granular semantic coding on the communication record to obtain a sequence of communication record segment-granular semantic coding feature vectors; Taking the semantic features of the public text materials of the tracked supplier as background semantic features, and performing feature adaptive fusion with each communication record segment granularity semantic coding feature vector in the sequence of communication record segment granularity semantic coding feature vectors respectively to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics; Performing contextual semantic association coding on the sequence of the communication record segment granularity semantic coding feature vectors under the background semantics to obtain the communication record full text semantic coding feature vector under the background semantics; Based on the full-text semantic coding feature vector of the communication record under the background semantics, a level label of the potential risk of the tracked supplier is determined.

[0007] Optionally, the communication records are semantically encoded at segment granularity to obtain a sequence of communication record segment granularity semantic encoding feature vectors, including: segmenting the communication records to obtain a sequence of communication record segment descriptions; and semantically encoding the sequence of communication record segment descriptions to obtain a sequence of communication record segment granularity semantic encoding feature vectors.

[0008] Optionally, the semantic features of the public text materials of the tracked supplier are used as background semantic features, and are adaptively fused with each communication record segment granularity semantic coding feature vector in the sequence of communication record segment granularity semantic coding feature vectors to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics, including: semantically encoding the public text materials of the tracking supplier to obtain a tracking supplier public information semantic coding feature vector; using the tracking supplier public information semantic coding feature vector as the background semantic feature vector, and inputting each communication record segment granularity semantic coding feature vector in the sequence of communication record segment granularity semantic coding feature vectors and the background semantic feature vector into a threshold-based feature vector adaptive fusion module to obtain a sequence of communication record segment granularity semantic coding feature vectors under the background semantics.

[0009] Optionally, each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector is input into a threshold-based feature vector adaptive fusion module with the background semantic feature vector to obtain a sequence of communication record segment granularity semantic coding feature vectors under the background semantics, including: the communication record segment granularity semantic coding feature vector and the background semantic feature vector are fused through a multi-dimensional fusion module to obtain a first background semantics-communication record segment granularity semantic fusion feature vector, a second background semantics-communication record segment granularity semantic fusion feature vector and a third background semantics-communication record segment granularity semantic fusion feature vector. Combined feature vector; respectively calculating the equalization threshold values ​​of the first background semantics-communication record segment granularity semantic fusion feature vector, the second background semantics-communication record segment granularity semantic fusion feature vector and the third background semantics-communication record segment granularity semantic fusion feature vector to obtain the first equalization threshold value, the second equalization threshold value and the third equalization threshold value; based on the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are balanced and fused to obtain the communication record segment granularity semantic coding feature vector under the background semantics.

[0010] Optionally, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are passed through a multi-dimensional fusion module to obtain a first background semantic-communication record segment granularity semantic fusion feature vector, a second background semantic-communication record segment granularity semantic fusion feature vector and a third background semantic-communication record segment granularity semantic fusion feature vector, including: cascading the communication record segment granularity semantic coding feature vector and the background semantic feature vector to obtain the first background semantic-communication record segment granularity semantic fusion feature vector; adding the communication record segment granularity semantic coding feature vector and the background semantic feature vector by position to obtain the second background semantic-communication record segment granularity semantic fusion feature vector; performing positional point multiplication of the communication record segment granularity semantic coding feature vector and the background semantic feature vector to obtain the third background semantic-communication record segment granularity semantic fusion feature vector.

[0011] Optionally, the equalization threshold values ​​of the first background semantic-communication record segment granularity semantic fusion feature vector, the second background semantic-communication record segment granularity semantic fusion feature vector and the third background semantic-communication record segment granularity semantic fusion feature vector are calculated respectively to obtain a first equalization threshold value, a second equalization threshold value and a third equalization threshold value, including: using a first predetermined transformation vector to multiply the first background semantic-communication record segment granularity semantic fusion feature vector to obtain a first threshold scoring coefficient; adding the first threshold scoring coefficient and a first bias parameter and then passing through a sigmoid activation function to obtain the first equalization threshold value.

[0012] Optionally, based on the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are equalized and fused to obtain the communication record segment granularity semantic coding feature vector under the background semantics, including: determining a first weight parameter and a second weight parameter based on the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, wherein the first weight parameter is the average of the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, and the second weight parameter is the difference between one and the first weight parameter; weighting the communication record segment granularity semantic coding feature vector position by position with the first weight parameter to obtain a weighted communication record segment granularity semantic coding feature vector, and weighting the background semantic feature vector position by position with the second weight parameter to obtain a weighted background semantic feature vector; adding the corresponding elements of the weighted communication record segment granularity semantic coding feature vector and the weighted background semantic feature vector to obtain the communication record segment granularity semantic coding feature vector under the background semantics.

[0013] Optionally, context semantic association encoding is performed on the sequence of segment granularity semantic coding feature vectors of the communication record under the background semantics to obtain the full-text semantic coding feature vector of the communication record under the background semantics, including: inputting the sequence of segment granularity semantic coding feature vectors of the communication record under the background semantics into a context encoder based on a Bi-LSTM model to obtain the full-text semantic coding feature vector of the communication record under the background semantics.

[0014] Optionally, based on the full-text semantic encoding feature vector of the communication record under the background semantics, determining the level label of the potential risk of the tracked supplier includes: inputting the full-text semantic encoding feature vector of the communication record under the background semantics into a classifier-based supply relationship management generator to obtain a supply relationship management result, and the supply relationship management result is used to indicate the level label of the potential risk of the tracked supplier.

[0015] The above technical solution uses deep learning-based artificial intelligence technology to analyze the supplier's public text data and communication records, extracts the semantic feature representation of the supplier's public text data and communication records, and performs adaptive semantic feature fusion processing on the two, and then judges the supplier's potential risk level based on cross-validation analysis of public text data and communication records. This can help companies more accurately identify and manage potential risks in the supply chain, thereby improving the stability and security of their production operations.

[0016] Other features and advantages of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings: Figure 1 The present invention is a flowchart of a supply relationship management method based on artificial intelligence according to an exemplary embodiment.

[0018] Figure 2 It is a block diagram of a supply relationship management system based on artificial intelligence according to an exemplary embodiment.

[0019] Figure 3 It is a block diagram of an electronic device according to an exemplary embodiment.

[0020] Figure 4 It is an application scenario diagram of a supply relationship management method based on artificial intelligence according to an exemplary embodiment. DETAILED DESCRIPTION

[0021] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.

[0022] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0023] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0024] It should be noted that the concepts such as "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0025] It should be noted that the modifications of "one" and "plurality" mentioned in the present application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0026] The names of the messages or information exchanged between multiple devices in the embodiments of the present application are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0027] In the chemical industry, the application of artificial intelligence is ushering in a new era of supply chain management. The supply relationship management system based on artificial intelligence can optimize and automate complex supply networks through advanced data analysis and machine learning algorithms.

[0028] These systems can process large amounts of unstructured data, identify patterns and trends, thereby improving forecast accuracy, optimizing inventory levels, reducing the risk of supply chain disruptions, and improving overall operational efficiency. This technology is particularly important in the chemical industry because the products and raw materials it involves are often highly complex and variable. Artificial intelligence can help companies better understand and predict market demand and achieve more refined production plans and resource allocation. In addition, artificial intelligence can also support supplier selection and evaluation, helping companies make more informed cooperation decisions by analyzing suppliers' historical performance and risk status. Artificial intelligence can also help in sustainable development and environmental protection, such as reducing carbon footprints by optimizing transportation routes and reducing energy consumption. With the continuous advancement of technology, future chemical supply chain management will be more intelligent, automated and sustainable, bringing greater competitive advantages and social value to companies.

[0029] The specific implementation of the present application is described in detail below with reference to the accompanying drawings.

[0030] Figure 1 is a flow chart of a supply relationship management method based on artificial intelligence according to an exemplary embodiment. Figure 1 As shown, the method includes: Step 101: Obtaining public text information of the supplier being tracked, and communication records with the supplier being tracked; Step 102: performing segment granularity semantic coding on the communication record to obtain a sequence of communication record segment granularity semantic coding feature vectors; Step 103: using the semantic features of the public textual materials of the tracked supplier as background semantic features, and adaptively fusing the features with the communication record segment granularity semantic coding feature vectors in the sequence of communication record segment granularity semantic coding feature vectors to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics; Step 104, performing contextual semantic association coding on the sequence of the communication record segment granularity semantic coding feature vectors under the background semantics to obtain the communication record full text semantic coding feature vector under the background semantics; Step 105: Determine the level label of the potential risk of the tracked supplier based on the semantic encoding feature vector of the full text of the communication record under the background semantics.

[0031] In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to perform data analysis on the supplier's public text materials and communication records, extract the semantic feature representations of the supplier's public text materials and communication records respectively, and perform adaptive fusion processing of the semantic features of the two, and then judge the supplier's potential risk level based on cross-validation analysis of public text materials and communication records. In this way, it can help enterprises more accurately identify and manage potential risks in the supply chain, thereby improving the stability and security of the enterprise's production operations.

[0032] Based on this, in the technical solution of the present application, first, obtain the public text materials of the tracked supplier and the communication records with the tracked supplier. It should be understood that the public text materials include publicly available data such as the supplier's official website information, financial reports, industry rankings, compliance certificates, etc., which can fully reflect the supplier's basic situation, operating conditions and compliance. The communication records include email exchanges, purchase orders, delivery notices, etc. between suppliers and enterprises, which can reflect the actual interaction between suppliers and enterprises and business execution capabilities. In the technical solution of the present application, by comprehensively analyzing the supplier's communication records and public text materials, the supplier's risk situation can be cross-verified. For example, if it is found in the public text materials that the supplier's financial report shows signs of instability, and there are delivery delays or abnormal communication in the communication records, this will provide a more comprehensive perspective for the supplier's potential risk assessment. Through this multi-level analysis method, it is helpful to improve the accuracy of the supplier's potential risk assessment and reduce the impact of supplier problems on the operation of chemical companies.

[0033] Next, considering that the communication records with suppliers usually contain multiple information fragments, such as different email subjects, order details or discussion topics, these information fragments may involve different supply risk indicators. Therefore, in order to achieve more sophisticated semantic analysis of communication records, the communication records are further segmented to obtain a sequence of communication record segment descriptions, so as to perform fine-grained semantic encoding for each communication record segment description. In an embodiment of the present application, segmentation can be performed based on email subjects, time intervals or other logical separation points to ensure that each communication record segment description focuses on a specific discussion point or event.

[0034] In one embodiment of the present application, segment granularity semantic encoding is performed on the communication record to obtain a sequence of communication record segment granularity semantic encoding feature vectors, including: segmenting the communication record to obtain a sequence of communication record segment descriptions; semantically encoding the sequence of communication record segment descriptions to obtain a sequence of communication record segment granularity semantic encoding feature vectors.

[0035] Then, semantic encoding is performed on each communication record segment description in the sequence of communication record segment descriptions to capture the semantic relationship and contextual meaning between each word in the communication record segment description, understand its emotional tendency, and convert each communication record segment description into a vector representation of a fixed length to obtain a sequence of communication record segment granularity semantic encoding feature vectors. In an embodiment of the present application, a natural language processing model based on the Transformer architecture such as BERT or RoBERTa is used to extract semantic features of each communication record segment description in the sequence of communication record segment descriptions, and the model's word embedding encoding ability and self-attention mechanism are used to consider the order and importance of vocabulary to capture the complex semantic information in each communication record segment description.

[0036] Secondly, similarly, global context semantic encoding is performed on the public text materials of the tracking supplier. Through word segmentation, word embedding technology and self-attention mechanism, the key information and structured knowledge in the text materials are captured, and the global semantics of the text are understood, so as to convert the public text materials into a vector representation of a fixed length to obtain a semantic encoding feature vector of the public text materials.

[0037] In one embodiment of the present application, the semantic features of the public text materials of the tracked supplier are used as background semantic features, and feature adaptive fusion is performed with each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics, including: semantically encoding the public text materials of the tracking supplier to obtain a tracking supplier public information semantic coding feature vector; using the tracking supplier public information semantic coding feature vector as the background semantic feature vector, and inputting each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector and the background semantic feature vector into a threshold-based feature vector adaptive fusion module to obtain a sequence of communication record segment granularity semantic coding feature vectors under the background semantics.

[0038] Furthermore, considering that the public text material provides the supplier background information at the macro level, while the communication record provides the specific interaction details at the micro level, there is a relationship of mutual influence and complementation between the two. Therefore, when conducting a potential risk assessment of a supplier, the public text material can be used as background information to provide macro-level information support for the semantic analysis of the communication record. Based on this, in the technical solution of the present application, the semantic coding feature vector of the tracking supplier's public information is used as the background semantic feature vector, and each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector and the background semantic feature vector are input into a threshold-based feature vector adaptive fusion module to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics. Among them, the threshold-based feature vector adaptive fusion module can evaluate the feature importance of the background semantic feature vector and the communication record segment granularity semantic coding feature vector based on the multi-dimensional correlation feature between the background semantic feature vector and the communication record segment granularity semantic coding feature vector, so as to perform adaptive weight allocation to balance the information contribution of the two, thereby ensuring that the stability and credibility of the macro background information are considered in the potential risk assessment of the supplier, and the actual execution of the specific interaction details is emphasized, thereby improving the accuracy and comprehensiveness of the risk assessment.

[0039] In one embodiment of the present application, each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector is input into a threshold-based feature vector adaptive fusion module with the background semantic feature vector to obtain a sequence of communication record segment granularity semantic coding feature vectors under the background semantics, including: the communication record segment granularity semantic coding feature vector and the background semantic feature vector are fused through a multi-dimensional fusion module to obtain a first background semantics-communication record segment granularity semantic fusion feature vector, a second background semantics-communication record segment granularity semantic fusion feature vector and a third background semantics-communication record segment granularity semantic fusion feature vector. The method comprises the following steps: calculating the equalization threshold values ​​of the first background semantic-communication record segment granularity semantic fusion feature vector, the second background semantic-communication record segment granularity semantic fusion feature vector and the third background semantic-communication record segment granularity semantic fusion feature vector respectively to obtain the first equalization threshold value, the second equalization threshold value and the third equalization threshold value; based on the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, performing equalization fusion on the communication record segment granularity semantic coding feature vector and the background semantic feature vector to obtain the communication record segment granularity semantic coding feature vector under the background semantics.

[0040] Furthermore, in one embodiment of the present application, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are passed through a multi-dimensional fusion module to obtain a first background semantic-communication record segment granularity semantic fusion feature vector, a second background semantic-communication record segment granularity semantic fusion feature vector and a third background semantic-communication record segment granularity semantic fusion feature vector, including: cascading the communication record segment granularity semantic coding feature vector and the background semantic feature vector to obtain the first background semantic-communication record segment granularity semantic fusion feature vector; adding the communication record segment granularity semantic coding feature vector and the background semantic feature vector by position to obtain the second background semantic-communication record segment granularity semantic fusion feature vector; performing positional point multiplication of the communication record segment granularity semantic coding feature vector and the background semantic feature vector to obtain the third background semantic-communication record segment granularity semantic fusion feature vector.

[0041] Furthermore, in one embodiment of the present application, the equalization threshold values ​​of the first background semantic-communication record segment granularity semantic fusion feature vector, the second background semantic-communication record segment granularity semantic fusion feature vector and the third background semantic-communication record segment granularity semantic fusion feature vector are calculated respectively to obtain the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, including: using a first predetermined transformation vector to multiply the first background semantic-communication record segment granularity semantic fusion feature vector to obtain a first threshold scoring coefficient; adding the first threshold scoring coefficient and the first bias parameter and then passing through a sigmoid activation function to obtain the first equalization threshold value.

[0042] Furthermore, based on the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are equalized and fused to obtain the communication record segment granularity semantic coding feature vector under the background semantics, including: determining a first weight parameter and a second weight parameter based on the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, wherein the first weight parameter is the average of the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, and the second weight parameter is the difference between one and the first weight parameter; weighting the communication record segment granularity semantic coding feature vector position by position with the first weight parameter to obtain a weighted communication record segment granularity semantic coding feature vector, and weighting the background semantic feature vector position by position with the second weight parameter to obtain a weighted background semantic feature vector; adding the corresponding elements of the weighted communication record segment granularity semantic coding feature vector and the weighted background semantic feature vector to obtain the communication record segment granularity semantic coding feature vector under the background semantics.

[0043] Specifically, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are fused using the following adaptive fusion formula to obtain the communication record segment granularity semantic coding feature vector under the background semantics, wherein the adaptive fusion formula is: ; in, A semantic encoding feature vector for the communication record segment granularity, is the background semantic feature vector, , and are respectively a first predetermined transformation vector, a second predetermined transformation vector and a third predetermined transformation vector, , and are respectively a first bias parameter, a second bias parameter and a third bias parameter; represents the activation function, Indicates cascade processing, It means adding by position. It means point multiplication by position. , and are the first equalization threshold value, the second equalization threshold value and the third equalization threshold value respectively, and , A feature vector is encoded for the communication record segment granularity semantics under the background semantics.

[0044] Then, in order to obtain the full-text semantic representation of the communication record, the sequence of granular semantic encoding feature vectors of the communication record segments under the background semantics is further input into the context encoder based on the Bi-LSTM model for global context semantic association encoding. The bidirectional context association modeling capability of the Bi-LSTM model is used to capture the context semantic dependencies between the descriptions of each communication record segment, and integrate the global content semantic information of the communication record, so as to obtain the global semantic encoding feature vector of the communication record, providing a more comprehensive and in-depth feature basis for the subsequent potential risk assessment of suppliers.

[0045] In one embodiment of the present application, context semantic association encoding is performed on a sequence of segment granularity semantic coding feature vectors of communication records under the background semantics to obtain a full-text semantic coding feature vector of the communication records under the background semantics, including: inputting the sequence of segment granularity semantic coding feature vectors of communication records under the background semantics into a context encoder based on a Bi-LSTM model to obtain a full-text semantic coding feature vector of the communication records under the background semantics.

[0046] Next, the semantic encoding feature vector of the full text of the communication record under the background semantics is input into the supply relationship management generator based on the classifier to obtain the supply relationship management result, and the supply relationship management result is used to indicate the level label of the potential risk of the tracked supplier. That is, the classifier model is used to construct the supply relationship management generator, and the mapping relationship between the semantic encoding feature vector of the full text of the communication record under the background semantics and the preset potential risk level label is identified through training and learning, so as to realize the automatic evaluation and prediction of the potential risk level of the supplier. Then, after obtaining the supply relationship management result, the corresponding risk response mechanism is automatically triggered. For example, for high-risk suppliers, an early warning notice is automatically sent to the supply chain management department of the enterprise, and emergency measures are recommended, such as suspending procurement, increasing the frequency of quality inspections, etc. For medium-risk suppliers, the management department can be reminded to conduct further investigation and evaluation to determine whether targeted improvement measures are needed. For low-risk suppliers, the current cooperation status can be maintained, but regular risk assessments continue to be maintained to ensure the continuous stability of the supplier status.

[0047] Here, in the technical solution of the present application, the tracking supplier public information semantic coding feature vector represents the text semantic features of the public text materials of the tracked supplier, and each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector is used to represent the text semantic features of each communication record paragraph in the communication record of the tracked supplier. When the tracking supplier public information semantic coding feature vector is used as the background semantic feature vector, and each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector and the background semantic feature vector are input into the equalization threshold feature vector adaptive fusion module, the tracking supplier public information semantic coding feature vector is used as the background information interpolation unit to optimize the text information expression richness and accuracy of each communication record segment granularity semantic coding feature vector in the sequence of the communication record segment granularity semantic coding feature vector. However, when the sequence of the semantic encoding feature vectors of the communication record segment granularity under the background semantics is input into the context encoder based on the Bi-LSTM model, considering that the background semantic feature vectors are continuously semantically associated and encoded, the background information may grind the call record text information, resulting in the distribution discretization of the feature distribution of the full-text semantic encoding feature vectors of the communication record under the background semantics. That is, the feature manifold of the full-text semantic encoding feature vectors of the communication record under the background semantics has expression sparsity and uneven distribution in the high-dimensional feature space, which in turn affects the accuracy of the supply relationship management results obtained by the supply relationship management generator based on the classifier.

[0048] Preferably, the semantic encoding feature vector of the full text of the communication record under the background semantics is input into a classifier-based supply relationship management generator to obtain a supply relationship management result, including: Performing L random sampling on the semantic encoding feature vector of the full text of the communication record under the background semantics to obtain L communication record semantic random perturbation feature values; Arrange the L communication record semantic random perturbation eigenvalues ​​in order from large to small to obtain a communication record semantic random perturbation ordered vector; Calculating the inner product of the semantic coding feature vector of the full text of the communication record under the background semantics to obtain the communication record semantic energy modulation feature value; Based on the communication record semantic energy modulation feature value, the communication record semantic random perturbation ordered vector is semantically scaled to obtain a semantically enhanced communication record semantic random perturbation ordered vector; Calculating the product of the random perturbation ordered vector of the semantically enhanced communication record semantics and the transposed vector of the full-text semantic coding feature vector of the communication record under the background semantics to obtain the communication record semantics best sample prompt orthogonal modulation matrix; Calculating the matrix product between the best sample prompt orthogonal modulation matrix of the communication record semantics and the full text semantic coding feature vector of the communication record under the background semantics to obtain the optimized full text semantic coding feature vector of the communication record under the background semantics; The optimized semantic encoding feature vector of the full text of the communication record under the background semantics is input into the classifier-based supply relationship management generator to obtain the supply relationship management result.

[0049] The above optimization process is expressed as follows: in, represents a normal distribution, represents the Gaussian density probability function, and They represent the mean value and variance of the semantic encoding feature vector of the full text of the communication record under the background semantics, represents a random variable that follows a normal distribution, represents the number of random samplings, Indicates The semantic random perturbation feature value of the communication record, represents the semantic encoding feature vector of the full text of the communication record under the background semantics, The first character representing the semantic encoding feature vector of the full text of the communication record under the background semantics eigenvalues, It means to calculate the average value of the vector. represents the length of the semantic encoding feature vector of the full text of the communication record under the background semantics, It indicates that the semantic random perturbation feature values ​​of communication records are arranged in order from large to small. Represents the ordered vector of random disturbance of communication record semantics, Represents the semantic energy modulation feature value of the communication record, represents the square of the vector's two-norm, It means point multiplication by position. represents matrix multiplication, represents the transpose of a vector, The best example of communication record semantics prompts the orthogonal modulation matrix, Represents the semantic encoding feature vector of the full text of the communication record under optimized background semantics.

[0050] Accordingly, by performing L random sampling on the semantic encoding feature vector of the full text of the communication record under the background semantics to obtain a set of new feature values ​​based on the original features but with slight changes, randomness is introduced to explore different aspects of the feature space, and by simulating the noise or variation in the actual data, the model is better generalized to unseen data. Then, these communication record semantic random perturbation feature values ​​are arranged in order from large to small to form a communication record semantic random perturbation ordered vector to identify which perturbation directions may be the most important for the current problem. Such an ordering not only helps the algorithm focus on the feature differences that can best distinguish different categories, but also helps to reveal hidden data structures or patterns. Subsequently, the inner product of the semantic encoding feature vector of the full text of the communication record under the background semantics is calculated to obtain the communication record semantic energy modulation feature value, which reflects the overall strength or "energy" of the original feature vector and provides a benchmark for measuring the importance of the feature. Using this communication record semantic energy modulation feature value as a reference point, the previously formed communication record semantic random perturbation ordered vector is semantically scaled to construct a semantically enhanced communication record semantic random perturbation ordered vector. In this way, those disturbances that appear to be more important in the original feature space are amplified, thereby enhancing the model's sensitivity to changes in key features, while also retaining relatively small but potentially useful disturbance information.

[0051] Then, the product of the random perturbation ordered vector of semantically enhanced communication records and the transpose of the semantic encoding feature vector of the full text of communication records under background semantics is calculated to obtain the best example prompt orthogonal modulation matrix of communication record semantics. The purpose is to find the interaction mode between features under two different perspectives. The formed matrix can be regarded as a conversion tool for capturing and representing the complex linear relationship between features. Finally, the final feature structure modulation is completed by applying the best example prompt orthogonal modulation matrix of communication record semantics constructed above to the semantic encoding feature vector of the full text of communication records under the original background semantics, so as to make the features more in line with the requirements of the objective function, reduce redundant information, and maximize the retention of useful information, so as to improve learning efficiency and prediction performance.

[0052] In summary, the feature manifold optimization process for the semantic encoding feature vector of the full text of the communication record under the background semantics is explained, which extracts the best prompt examples based on the fine-grained target retrieval of random perturbations, and then guides adversarial robustness learning based on the best prompt examples to complete the structure of the high-dimensional manifold of the feature vector in the feature space to improve the detail and structure of the feature expression. In this way, the classification iteration effect of the supply relationship management generator based on the classifier of the feature set input of the semantic encoding feature vector of the full text of the communication record under the background semantics is improved, that is, the speed of classification training and the accuracy of classification results are improved.

[0053] In summary, the above scheme uses deep learning-based artificial intelligence technology to analyze the supplier's public text data and communication records, extracts the semantic feature representation of the supplier's public text data and communication records, and performs adaptive fusion processing of the semantic features of the two, and then judges the supplier's potential risk level based on cross-validation analysis of public text data and communication records. In this way, it can help enterprises more accurately identify and manage potential risks in the supply chain, thereby improving the stability and security of the enterprise's production operations.

[0054] Figure 2 FIG. 1 is a block diagram of a supply relationship management system based on artificial intelligence according to an exemplary embodiment. Figure 2 As shown, the supply relationship management system 200 includes: The communication data acquisition module 201 is used to acquire the public text data of the supplier being tracked, and the communication records with the supplier being tracked; A segment granularity semantic coding module 202, configured to perform segment granularity semantic coding on the communication record to obtain a sequence of communication record segment granularity semantic coding feature vectors; The feature adaptive fusion module 203 is used to use the semantic features of the public text materials of the tracked supplier as background semantic features, and perform feature adaptive fusion with each communication record segment granularity semantic coding feature vector in the sequence of communication record segment granularity semantic coding feature vectors to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics; A context semantic association coding module 204 is used to perform context semantic association coding on the sequence of the communication record segment granularity semantic coding feature vectors under the background semantics to obtain the communication record full text semantic coding feature vector under the background semantics; The potential risk level label determination module 205 is used to determine the level label of the potential risk of the tracked supplier based on the full-text semantic coding feature vector of the communication record under the background semantics.

[0055] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present application. The terminal device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0056] like Figure 3 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0057] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0058] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present application are executed.

[0059] It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0060] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0061] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0062] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0063] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0064] The modules involved in the embodiments described in this application can be implemented by software or hardware. The name of the module does not limit the module itself in some cases. For example, the test parameter acquisition module can also be described as a "module for acquiring device test parameters corresponding to the target device".

[0065] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0066] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0067] Figure 4 FIG. 1 is an application scenario diagram of a supply relationship management method based on artificial intelligence according to an exemplary embodiment. Figure 4 As shown, in this application scenario, first, obtain the public text information of the supplier being tracked (for example, Figure 4 C1 as indicated in the figure), and the communication records with the tracked supplier (e.g. Figure 4 Then, the obtained public text materials and communication records are input into a server that deploys an artificial intelligence-based supply relationship management algorithm (for example, Figure 4 In S) as shown in , the server is capable of processing the public text material and the communication records based on an artificial intelligence-based supply relationship management algorithm to determine a level label for the potential risk of the tracked supplier.

[0068] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

[0069] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or to be performed in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0070] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. A supply relationship management method based on artificial intelligence, characterized in that: include: Obtaining public text information of the tracked supplier and communication records with the tracked supplier; Performing segment-granular semantic coding on the communication record to obtain a sequence of communication record segment-granular semantic coding feature vectors; Taking the semantic features of the public text materials of the tracked supplier as background semantic features, and performing feature adaptive fusion with each communication record segment granularity semantic coding feature vector in the sequence of communication record segment granularity semantic coding feature vectors respectively to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics; Performing contextual semantic association coding on the sequence of the communication record segment granularity semantic coding feature vectors under the background semantics to obtain the communication record full text semantic coding feature vector under the background semantics; Based on the full-text semantic coding feature vector of the communication record under the background semantics, a level label of the potential risk of the tracked supplier is determined.

2. The supply relationship management method based on artificial intelligence according to claim 1, characterized in that: Performing segment granularity semantic coding on the communication record to obtain a sequence of communication record segment granularity semantic coding feature vectors, including: Segmenting the communication record to obtain a sequence of communication record segment descriptions; The sequence of the communication record segment descriptions is semantically encoded to obtain a sequence of the communication record segment granularity semantic encoding feature vectors.

3. The supply relationship management method based on artificial intelligence according to claim 2, characterized in that: The semantic features of the public text materials of the tracked supplier are used as background semantic features, and feature adaptive fusion is performed on each communication record segment granularity semantic coding feature vector in the sequence of communication record segment granularity semantic coding feature vectors to obtain a sequence of communication record segment granularity semantic coding feature vectors under background semantics, including: Performing semantic coding on the public textual data of the tracking supplier to obtain a semantic coding feature vector of the public information of the tracking supplier; Taking the tracking supplier public information semantic coding feature vector as the background semantic feature vector, each communication record segment granularity semantic coding feature vector in the sequence of communication record segment granularity semantic coding feature vectors is respectively input into a threshold-based feature vector adaptive fusion module with the background semantic feature vector to obtain a sequence of communication record segment granularity semantic coding feature vectors under the background semantics.

4. The supply relationship management method based on artificial intelligence according to claim 3 is characterized in that: Inputting each communication record segment granular semantic coding feature vector in the sequence of the communication record segment granular semantic coding feature vector and the background semantic feature vector into a threshold-based feature vector adaptive fusion module to obtain a sequence of communication record segment granular semantic coding feature vectors under the background semantics, including: The communication record segment granularity semantic coding feature vector and the background semantic feature vector are passed through a multi-dimensional fusion module to obtain a first background semantic-communication record segment granularity semantic fusion feature vector, a second background semantic-communication record segment granularity semantic fusion feature vector and a third background semantic-communication record segment granularity semantic fusion feature vector; Respectively calculating the equalization threshold values ​​of the first background semantics-communication record segment granularity semantics fusion feature vector, the second background semantics-communication record segment granularity semantics fusion feature vector, and the third background semantics-communication record segment granularity semantics fusion feature vector to obtain a first equalization threshold value, a second equalization threshold value, and a third equalization threshold value; Based on the first equalization threshold value, the second equalization threshold value and the third equalization threshold value, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are equalized and fused to obtain the communication record segment granularity semantic coding feature vector under the background semantics.

5. The supply relationship management method based on artificial intelligence according to claim 4 is characterized in that: The communication record segment granularity semantic coding feature vector and the background semantic feature vector are subjected to a multi-dimensional fusion module to obtain a first background semantic-communication record segment granularity semantic fusion feature vector, a second background semantic-communication record segment granularity semantic fusion feature vector and a third background semantic-communication record segment granularity semantic fusion feature vector, including: Cascading the communication record segment granularity semantic encoding feature vector and the background semantic feature vector to obtain the first background semantic-communication record segment granularity semantic fusion feature vector; Adding the communication record segment granularity semantic coding feature vector and the background semantic feature vector by position to obtain the second background semantic-communication record segment granularity semantic fusion feature vector; The communication record segment granularity semantic coding feature vector and the background semantic feature vector are multiplied point by point to obtain the third background semantic-communication record segment granularity semantic fusion feature vector.

6. The supply relationship management method based on artificial intelligence according to claim 5 is characterized in that: Respectively calculating the equalization threshold values ​​of the first background semantics-communication record segment granularity semantics fusion feature vector, the second background semantics-communication record segment granularity semantics fusion feature vector, and the third background semantics-communication record segment granularity semantics fusion feature vector to obtain a first equalization threshold value, a second equalization threshold value, and a third equalization threshold value, including: Multiplying the first background semantics-communication record segment granularity semantics fusion feature vector by a first predetermined transformation vector to obtain a first threshold scoring coefficient; The first threshold scoring coefficient and the first bias parameter are added together and then passed through a sigmoid activation function to obtain the first equalization threshold value.

7. The supply relationship management method based on artificial intelligence according to claim 6 is characterized in that: Based on the first equalization threshold value, the second equalization threshold value, and the third equalization threshold value, the communication record segment granularity semantic coding feature vector and the background semantic feature vector are equalized and fused to obtain the communication record segment granularity semantic coding feature vector under the background semantics, including: Determine a first weight parameter and a second weight parameter based on the first equalization threshold value, the second equalization threshold value, and the third equalization threshold value, wherein the first weight parameter is an average of the first equalization threshold value, the second equalization threshold value, and the third equalization threshold value, and the second weight parameter is a difference between one and the first weight parameter; The communication record segment granularity semantic coding feature vector is weighted position by position with the first weight parameter to obtain a weighted communication record segment granularity semantic coding feature vector, and the background semantic feature vector is weighted position by position with the second weight parameter to obtain a weighted background semantic feature vector; The weighted communication record segment granularity semantic coding feature vector and the weighted background semantic feature vector are added together to obtain the communication record segment granularity semantic coding feature vector under the background semantics.

8. The supply relationship management method based on artificial intelligence according to claim 7 is characterized in that: Performing context semantic association coding on the sequence of the communication record segment granularity semantic coding feature vectors under the background semantics to obtain the communication record full text semantic coding feature vector under the background semantics, including: The sequence of the segment-granular semantic encoding feature vectors of the communication record under the background semantics is input into a context encoder based on the Bi-LSTM model to obtain the full-text semantic encoding feature vector of the communication record under the background semantics.

9. The supply relationship management method based on artificial intelligence according to claim 8, characterized in that: Based on the semantic encoding feature vector of the full text of the communication record under the background semantics, determining the level label of the potential risk of the tracked supplier includes: The semantic encoding feature vector of the full text of the communication record under the background semantics is input into a supply relationship management generator based on a classifier to obtain a supply relationship management result, and the supply relationship management result is used to indicate a grade label of the potential risk of the tracked supplier.

Images