Raw material property prediction method, device, equipment, medium and program product

Abstract

The present invention provides a raw material attribute prediction method, apparatus, device, medium, and program product. The method relates to the fields of computer application technologies such as artificial intelligence and deep learning. The method comprises: encoding and decoding a six-tuple to be completed, consisting of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information, to obtain a true complex embedding vector of the existing attribute and a predicted complex embedding vector for predicting the attribute to be completed, wherein the existing attribute represents other attributes in the six-tuple to be completed, excluding the attribute to be completed; calculating a candidate matching evaluation value representing the degree of match between the predicted result of the attribute to be completed and the true information of the existing attribute based on the true complex embedding vector and the predicted complex embedding vector; and determining a target result for predicting the attribute to be completed based on the candidate matching evaluation value.

Description

Technical Field

[0001] The present invention relates to the field of computer application technology, in particular to the fields of artificial intelligence and deep learning, and more specifically to a method, device, equipment, medium and program product for predicting raw material properties. Background Art

[0002] A crucial aspect of mine economic evaluation is estimating the costs and properties of raw materials, fuel, and power required for mining. This process typically relies on manual price inquiries or extrapolations based on historical value attributes, which consumes significant effort. Furthermore, due to rapid changes in market supply and demand, relevant attribute information fluctuates dramatically and frequently, making traditional methods inaccurately estimating these attributes. This directly impacts evaluation results and investment decisions. Furthermore, data collection accuracy and methods often result in gaps and omissions, making it difficult to accurately estimate these attributes. Summary of the Invention

[0003] In view of this, the present invention provides a raw material property prediction method, apparatus, device, medium and program product.

[0004] One aspect of the present invention provides a method for predicting raw material attributes, including: encoding and decoding a six-tuple to be completed consisting of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information, to obtain a real complex embedding vector of the existing attributes in the six-tuple to be completed and a predicted complex embedding vector for predicting the attributes to be completed in the six-tuple to be completed, wherein the attributes to be completed include any of the following: raw material type information, raw material name information, raw material unit information, raw material origin information, and raw material value attribute information, and the existing attributes represent other attributes in the six-tuple to be completed except the attributes to be completed; based on the real complex embedding vector and the predicted complex embedding vector, calculating a candidate matching evaluation value representing the matching degree between the prediction result of the attribute to be completed and the real information of the existing attribute; and determining a target result for predicting the attribute to be completed based on the candidate matching evaluation value.

[0005] Another aspect of the present invention provides a raw material attribute prediction device, including: a prediction module, used to encode and decode a six-tuple to be completed consisting of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information, to obtain a real complex embedding vector of the existing attributes in the six-tuple to be completed and a predicted complex embedding vector for the predicted attributes to be completed in the six-tuple to be completed, wherein the attributes to be completed include any of the following: raw material type information, raw material name information, raw material unit information, raw material origin information, and raw material value attribute information, and the existing attributes represent other attributes in the six-tuple to be completed except the attributes to be completed; a candidate matching evaluation value calculation module, used to calculate a candidate matching evaluation value representing the matching degree between the predicted result of the attribute to be completed and the real information of the existing attribute based on the real complex embedding vector and the predicted complex embedding vector; a target result determination module, used to determine the target result predicted for the attribute to be completed based on the candidate matching evaluation value.

[0006] Another aspect of the present invention provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the raw material property prediction method of the present invention.

[0007] Another aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed, are used to implement the raw material property prediction method of the present invention.

[0008] Another aspect of the present invention provides a computer program product, which includes computer executable instructions. When the instructions are executed, they are used to implement the raw material property prediction method of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0010] Figure 1 An exemplary system architecture to which a raw material property prediction method according to an embodiment of the present invention can be applied is shown;

[0011] Figure 2 A flowchart of a method for predicting raw material properties according to an embodiment of the present invention is shown;

[0012] Figure 3 A schematic diagram showing the technical principle of the MFIA scoring function according to an embodiment of the present invention is shown;

[0013] Figure 4Ashows a functional module diagram of an MFIA model according to an embodiment of the present invention;

[0014] Figure 4B shows an overall flow chart of the process of designing and using the MFIA model according to an embodiment of the present invention;

[0015] Figure 5 A block diagram of a raw material property prediction device according to an embodiment of the present invention is shown;

[0016] Figure 6 A block diagram of an electronic device suitable for implementing a raw material property prediction method according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0017] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concept of the present invention.

[0018] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the presence of the features, steps, operations and / or components, but do not exclude the presence or addition of one or more other features, steps, operations or components.

[0019] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0020] When expressions such as "at least one of A, B, and C, etc." are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0021] In the embodiments of the present invention, the collection, updating, analysis, processing, use, transmission, provision, disclosure, and storage of all data involved (including, but not limited to, user personal information) comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures are taken to prevent unauthorized access to user personal information and maintain information security.

[0022] Figure 1 FIG. 4 shows an exemplary system architecture to which the raw material property prediction method according to an embodiment of the present invention can be applied. It should be noted that Figure 1 The examples shown are merely examples of system architectures to which the embodiments of the present invention may be applied, to help those skilled in the art understand the technical content of the present invention, but do not mean that the embodiments of the present invention cannot be used in other devices, systems, environments or scenarios.

[0023] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0024] A user may use a first terminal device 101, a second terminal device 102, or a third terminal device 103 to interact with a server 105 via a network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, or the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software (for example only).

[0025] The first terminal device 101 , the second terminal device 102 , and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0026] The server 105 may be a server that provides various services, such as a background management server (for example only) that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process received data such as user requests, and feed back processing results (e.g., web pages, information, or data obtained or generated based on user requests) to the terminal devices.

[0027] It should be noted that the raw material property prediction method provided in the embodiment of the present invention can generally be executed by the server 105. Accordingly, the raw material property prediction device provided in the embodiment of the present invention can generally be set in the server 105. The raw material property prediction method provided in the embodiment of the present invention can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the raw material property prediction device provided in the embodiment of the present invention can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Alternatively, the raw material property prediction method provided in the embodiment of the present invention can also be executed by the first terminal device 101, the second terminal device 102, or the third terminal device 103, or can also be executed by other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103. Correspondingly, the raw material property prediction device provided in the embodiment of the present invention can also be set in the first terminal device 101, the second terminal device 102, or the third terminal device 103, or in other terminal devices different from the first terminal device 101, the second terminal device 102, or the third terminal device 103.

[0028] For example, the six-tuple to be completed may be originally stored in any one of the first terminal device 101, the second terminal device 102, or the third terminal device 103 (for example, the first terminal device 101, but not limited thereto), or stored on an external storage device and imported into the first terminal device 101. Then, the first terminal device 101 may locally execute the raw material attribute prediction method provided by the embodiment of the present invention, or send the six-tuple to be completed to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the six-tuple to be completed may execute the raw material attribute prediction method provided by the embodiment of the present invention.

[0029] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0030] Figure 2 A flow chart of a raw material property prediction method according to an embodiment of the present invention is shown.

[0031] like Figure 2 As shown, the method includes operations S201 to S203.

[0032] In operation S201, the six-tuple to be completed, which is composed of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information, is encoded and decoded to obtain real complex embedding vectors of existing attributes in the six-tuple to be completed and predicted complex embedding vectors for the attributes to be completed in the six-tuple to be completed. The attributes to be completed include any one of the following: raw material type information, raw material name information, raw material unit information, raw material origin information, and raw material value attribute information. The existing attributes represent other attributes in the six-tuple to be completed except for the attributes to be completed.

[0033] According to an embodiment of the present invention, in a raw material attribute prediction scenario, the six-tuple used can be represented as (raw material type, raw material name, raw material unit, raw material origin, raw material value attribute, timestamp). Using a null value (NULL) to represent the attributes to be completed in the six-tuple, the six-tuple to be completed can be represented as (NULL, raw material name, raw material unit, raw material origin, raw material value attribute, timestamp), (raw material type, NULL, raw material unit, raw material origin, raw material value attribute, timestamp), (raw material type, raw material name, NULL, raw material origin, raw material value attribute, timestamp), (raw material type, raw material name, raw material unit, NULL, raw material value attribute, timestamp), (raw material type, raw material name, raw material unit, raw material origin, NULL, timestamp), etc.

[0034] According to an embodiment of the present invention, the codec is mainly used to encode the relevant attribute information into a complex embedding vector for representation. The codec process can use various codec modules capable of encoding and decoding information into a complex embedding vector, which is not limited here.

[0035] For example, in the raw material value attribute prediction scenario, the six-tuple to be completed may be (raw material type, raw material name, raw material unit, raw material origin, NULL, timestamp). In this case, by encoding the raw material type, raw material name, raw material unit, raw material origin, and timestamp, a true complex embedding vector can be obtained, which is determined by the complex embedding vectors of the raw material type, raw material name, raw material unit, raw material origin, and timestamp. Then, by encoding and decoding the complex embedding vectors of the raw material type, raw material name, raw material unit, raw material origin, and timestamp, the complex embedding vector of the raw material value attribute can be predicted as the prediction result for the NULL in (raw material type, raw material name, raw material unit, raw material origin, NULL, timestamp).

[0036] In operation S202 , a candidate matching evaluation value representing the matching degree between the predicted result of the attribute to be completed and the real information of the existing attribute is calculated based on the real complex embedding vector and the predicted complex embedding vector.

[0037] According to an embodiment of the present invention, a prediction result can be determined based on existing results in a six-tuple dataset. The six-tuple dataset can represent a collection of six-tuple data obtained by collecting and structuring historical raw material data. There can be one or more prediction results. Each prediction result can have a corresponding candidate matching evaluation value.

[0038] In operation S203 , a target result for the attribute to be completed prediction is determined according to the candidate matching evaluation value.

[0039] According to an embodiment of the present invention, a preset threshold may be set in advance, and a target result may be determined based on a prediction result corresponding to a candidate matching evaluation value greater than the preset threshold. If no candidate matching evaluation value greater than the preset threshold exists, a no result or error message may be output.

[0040] Through the above-mentioned embodiments of the present invention, by encoding and decoding relevant attributes into complex embedding vectors, it is possible to deeply analyze and study the problem of raw material attribute inference and data missing problems based on the powerful representation ability of complex space, thereby improving the accuracy of raw material attribute inference, further improving the scientific nature of economic evaluation, and making progress in improving the level of mine economic evaluation work.

[0041] In conjunction with specific embodiments, Figure 2 The method shown is further explained.

[0042] According to an embodiment of the present invention, operation S201 may include: performing one-hot encoding on the existing attribute to obtain the existing attribute binary vector; performing complex conversion on the existing attribute binary vector to obtain a true complex embedding vector; and encoding and decoding the true complex embedding vector to obtain a predicted complex embedding vector.

[0043] According to an embodiment of the present invention, when obtaining six-tuple data, the relevant attribute information in the six-tuple data can also be firstly one-hot encoded. Expressed as a one-hot encoded binary vector of dimension f, let The one-hot encoding process ends when the i-th element of is equal to 1 and the other elements are set to 0. After that, the one-hot encoding result is subjected to complex embedding encoding to obtain the complex embedding vector of the relevant attribute.

[0044] For example, in the raw material value attribute prediction scenario, the raw material type information, raw material name information, raw material unit information, raw material origin information, and timestamp information can first be one-hot encoded to obtain a binary vector of raw material type, a binary vector of raw material name, a binary vector of raw material unit, a binary vector of raw material origin, and a binary vector of timestamp. Then, the binary vectors of raw material type, raw material name, raw material unit, raw material origin, and timestamp can be complex-converted to obtain a complex embedding vector of raw material type, a complex embedding vector of raw material name, a complex embedding vector of raw material unit, a complex embedding vector of raw material origin, and a complex embedding vector of timestamp. After that, the complex embedding vectors of raw material type, raw material name, raw material unit, raw material origin, and timestamp can be encoded and decoded to obtain a complex embedding vector of raw material value attributes.

[0045] According to an embodiment of the present invention, corresponding to the above-mentioned raw material value attribute prediction scenario, the above-mentioned operation S202 may include: calculating a first matching evaluation value between the raw material value attribute prediction result and the raw material type information, raw material name information, raw material unit information, raw material origin information, and timestamp information based on the raw material type plural embedding vector, raw material name plural embedding vector, raw material unit plural embedding vector, raw material origin information, and timestamp information; calculating a second matching evaluation value between the raw material category information and the raw material unit-value attribute information based on the raw material type plural embedding vector, raw material name plural embedding vector, raw material unit plural embedding vector, raw material value attribute plural embedding vector, and timestamp plural embedding vector; and calculating a candidate matching evaluation value based on at least one of the first matching evaluation value and the second matching evaluation value.

[0046] According to an embodiment of the present invention, the raw material category information may represent information obtained by combining the raw material type information and the raw material name information. The raw material unit-value attribute information may represent information obtained by combining the raw material unit information and the raw material value attribute information.

[0047] According to embodiments of the present invention, based on the existing complex embedding vector, a suitable matching calculation method can be set according to business needs to calculate the first matching evaluation value and the second matching evaluation value, which are not limited here. Based on obtaining the first matching evaluation value and the second matching evaluation value, a suitable calculation method can also be set according to business needs to calculate the candidate matching evaluation value, which are not limited here.

[0048] According to an embodiment of the present invention, the process of calculating the first matching evaluation value may include: calculating, using a Hermitian operator, an origin-timestamp interaction complex embedding vector between the raw material origin complex embedding vector and the timestamp unit complex embedding vector obtained by normalizing the timestamp complex embedding vector; calculating, using a Hermitian operator, a category-origin-timestamp interaction complex embedding vector between the raw material category complex embedding vector obtained by concatenating the raw material type complex embedding vector and the raw material name complex embedding vector and the origin-timestamp interaction unit complex embedding vector obtained by normalizing the origin-timestamp interaction complex embedding vector; calculating a first difference between a first argument of the category-origin-timestamp interaction complex embedding vector and a second argument of the raw material unit-value attribute complex embedding vector obtained by concatenating the raw material unit complex embedding vector and the raw material value attribute complex embedding vector; calculating a second difference between a first absolute value of the category-origin-timestamp interaction complex embedding vector and a second absolute value of the raw material unit-value attribute complex embedding vector; and performing exponential calculation on the first and second differences to obtain the first matching evaluation value.

[0049] In the raw material value attribute prediction scenario, for example, the raw material type complex number can be first embedded in the vector and the raw material name plural embedding vector The spliced ​​raw material category complex embedding vector is named Embed the raw unit complex number into a vector and the complex embedding vector of raw material value attributes The spliced ​​raw material unit-value attribute complex embedding vector is named .

[0050] Furthermore, we can combine formula (1) to embed the timestamp complex number into the vector Divide by its norm to get the timestamp unit complex embedding vector .

[0051] (1)

[0052] in, Represents the norm.

[0053] Next, we can combine formula (2) and use the Hermitian operator to realize the origin and Interaction between:

[0054] (2)

[0055] in, represents the Hermitian operator, A complex embedding vector representing the origin-timestamp interaction.

[0056] Then, the origin-timestamp interaction complex can be embedded into the vector Convert to origin-timestamp interaction unit complex embedding vector Then, we can combine formula (3) and use the Hermitian operator to realize and Interaction between:

[0057] (3)

[0058] in, Represents the complex embedding vector of category-origin-timestamp interactions.

[0059] Finally, the approximate feature interaction module score function for calculating the first matching evaluation value can be defined as shown in formula (4) through a complex composite error metric.

[0060] (4)

[0061] in, represents the argument of a complex number, Indicates the absolute value, Indicates the first matching evaluation value.

[0062] According to an embodiment of the present invention, the process of calculating the second matching evaluation value may include: calculating a category-timestamp interaction complex embedding vector between the raw material category complex embedding vector and the timestamp unit complex embedding vector using a Hermitian operator. Calculating a unit-value attribute-timestamp interaction complex embedding vector between the raw material unit-value attribute complex embedding vector and the timestamp unit complex embedding vector using a Hermitian operator. Calculating a third argument of the Hermitian operator calculation result between a preset dependency factor and the category-timestamp interaction complex embedding vector using a Hermitian operator. Performing an exponential calculation on the difference between a third absolute value of the third argument and a fourth absolute value of a fourth argument of the unit-value attribute-timestamp interaction complex embedding vector to obtain the second matching evaluation value.

[0063] In the raw material value attribute prediction scenario, for example, we can first combine formula (5) to and Do the Hermitian operator to realize the interaction and obtain the category-timestamp interaction complex embedding vector .

[0064] (5)

[0065] Then we can combine formula (6) to get and The Hermitian operator is used to interact with each other to obtain the complex embedding vector of the unit-value attribute-timestamp interaction. .

[0066] (6)

[0067] Furthermore, the preset dependency factor can be used collect and The time series feature dependency module score function used to calculate the second matching evaluation value is obtained by using the different granularities at different times at the same time. The time series feature dependency module score function can be defined as shown in formula (7).

[0068] (7)

[0069] in, Represents the second matching evaluation value.

[0070] According to an embodiment of the present invention, a mine multi-feature interaction and aggregation (MFIA) score function for calculating a candidate matching evaluation value can be determined based on the approximate feature interaction module score function and the temporal feature dependency module score function. For example, the score function can be expressed as shown in formula (8).

[0071] (8)

[0072] Figure 3 A schematic diagram showing the technical principle of the MFIA scoring function according to an embodiment of the present invention is shown.

[0073] like Figure 3 As shown, the technical principles of the above formulas (2) to (3) can be described in virtual box 310. The technical principles of the above formula (4) can be described in virtual box 320. The technical principles of the above formulas (5) to (7) can be described in virtual box 330. The technical principles of the above formula (8) can be described in virtual box 340.

[0074] It should be noted that the above embodiment only describes the calculation method of the MFIA score function for the raw material value attribute prediction scenario. In other scenarios, such as the raw material type prediction scenario, the raw material name prediction scenario, the raw material unit prediction scenario, and the raw material origin prediction scenario, a MFIA score function applicable to the corresponding scenario can be constructed based on a method similar to the above raw material value attribute prediction scenario. It is sufficient to ensure that the complex embedding vector on the right side of the equal sign in formula (4) includes the complex embedding vector of the attribute to be completed for the prediction of the attribute to be completed. For example, it can be the complex embedding vector of the attribute to be completed itself, or it can be a concatenated complex embedding vector obtained by concatenating the complex embedding vector of the attribute to be completed and other real complex embedding vectors. There is no limitation here.

[0075] The design methods for other scenarios will not be elaborated here.

[0076] Through the above-mentioned embodiments of the present invention, the powerful representation ability of complex space can be utilized to construct multi-feature cross-learning technology, thereby improving the accuracy of raw material property estimation, further improving the scientific nature of economic evaluation, and making progress in improving the level of mine economic evaluation work.

[0077] According to an embodiment of the present invention, when there are multiple prediction results, operation S202 may include: calculating, based on each predicted complex embedding vector and the true complex embedding vector, a candidate matching evaluation value between each prediction result and the true information, thereby obtaining multiple candidate matching evaluation values. Based on this, operation S203 may include: determining the prediction result that is used to calculate the candidate matching evaluation value with the largest value among the multiple candidate matching evaluation values ​​as the target result.

[0078] For example, for each given sextuple to be completed, the MFIA score function can be calculated using the trained multi-feature cross-learning perception technology, and the prediction result with the highest score is automatically determined as the target result for completing the attribute to be completed.

[0079] According to an embodiment of the present invention, to implement the above raw material property prediction method, an MFIA model suitable for implementing the complex-based mine technical and economic evaluation raw material property prediction method can be constructed and trained based on the above MFIA score function.

[0080] Figure 4A FIG. 4 shows a functional module diagram of an MFIA model according to an embodiment of the present invention.

[0081] like Figure 4AAs shown, the MFIA model 400 includes a raw material attribute data import module 420, a data preprocessing module 430, an attribute prediction module 440, and a system security maintenance module 450. The data preprocessing module 430 may include a missing value processing submodule 431, an outlier detection submodule 432, a data normalization submodule 433, and a data structuring submodule 434.

[0082] Figure 4B The figure shows an overall flow chart of the process of designing and using the MFIA model according to an embodiment of the present invention.

[0083] like Figure 4B As shown, the process includes operations S401 to S410.

[0084] In operation S401 , a raw material attribute data file to be estimated is imported into and stored in a database.

[0085] For example, the raw material attribute data import module 420 can be combined to complete the import operation by clicking "Raw Material Attribute Data Import" on the page, selecting the data file to be imported in the folder, and clicking "Upload".

[0086] In operation S402 , the data file is preprocessed to obtain a structured data set.

[0087] This operation can be performed in conjunction with the data preprocessing module 430, including: missing value processing, outlier detection, data standardization, and data normalization. For missing value processing, the missing value processing submodule 431 can be combined to read the selected data set in the database, and the imported data can be filled with the missing raw material attribute data using the mean filling method. For outlier detection, the outlier detection submodule 432 can be combined to use statistical methods such as the interquartile range (IQR) to detect abnormal data and perform reasonable processing. For data normalization, the data normalization submodule 433 can be combined to normalize or standardize the data to improve the model convergence speed. For data structuring, the data structuring submodule 434 can be combined to use the Language Technology Platform (LTP) to perform entity extraction, relationship extraction, and entity unification on the data set in sequence. Entity extraction, that is, the recognition of entities, includes entity detection and classification. Feature extraction can generally be understood as multi-tuple extraction. This means that a dataset can be represented as a six-tuple consisting of (raw material type, raw material name, raw material unit, raw material origin, raw material value attribute, and timestamp). Entity unification, such as "coal" and "coal," requires entity unification.

[0088] In operation S403 , the data set is split into a training set, a validation set, and a test set.

[0089] For example, the dataset can be divided into a training set T, a validation set V, and a test set S in a 6:1:3 ratio. The training set is used for model training to ensure data diversity and improve generalization. The validation set is used to adjust hyperparameters to prevent overfitting. The test set is used to ultimately evaluate the model's completion performance.

[0090] In operation S404 , the data set is encoded.

[0091] This operation can include two stages: categorical variable processing and numerical feature conversion, or it can include only the numerical feature conversion stage. During the categorical variable processing stage, one-hot encoding can be used to obtain a binary vector. During the numerical feature conversion stage, a logarithmic transformation can be performed on the numerical variables to convert the binary vector into a four-element embedding vector.

[0092] In operation S405 , a complex number-based MFIA model is constructed, where the model includes a score function and a loss function.

[0093] In operation S406 , model hyperparameters are set.

[0094] Hyperparameters may include model initial learning rate, batch size, ent_vec_dim, rel_vec_di, etc., as shown in Table 1, but are not limited thereto.

[0095] Table 1:

[0096]

[0097] In operation S407 , training is performed using the training set until the model loss function converges.

[0098] For example, the MFIA model can be trained using training set data and parameters optimized through backpropagation, including: using the Adam optimization algorithm to adjust model weights; monitoring the downward trend of the loss function to prevent gradient vanishing or exploding; evaluating the model on the validation set and adjusting hyperparameters based on the results to ensure model generalization capabilities.

[0099] Specifically, for a given training set T, the loss function is used to calculate the loss of the training set data until the loss function of the model converges, and the model training is completed. The loss function can be shown as formula (9).

[0100] (9)

[0101] in, is the probability vector predicted by the model, is the label vector, which is set to 1 for the true sextuple and 0 for the predicted sextuple, is the loss function. Indicates the Sextuples. is the total number of sextuples. express dimensional real embedding vector, Represents the total number of entities of the six categories represented by the six-tuple determined by (raw material type, raw material name, raw material unit, raw material origin, raw material value attribute, timestamp).

[0102] In operation S408 , the trained MFIA model is used to complete the test data set.

[0103] This operation can be performed in conjunction with the attribute prediction module 440, including: entering the data into the model, calculating the score using the scoring function, where the higher the score, the closer it is to the inferred attribute. All data in the test set are traversed to complete the test set inference.

[0104] In operation S409 , the attribute inference result is visualized.

[0105] In operation S410 , the effectiveness of the MFIA model is evaluated and verified through experiments.

[0106] For example, the error between predicted and real data can be calculated to assess the accuracy of the completion. The completed data can then be recorded and compared with existing methods. Specifically, the mean absolute error (MAE), root mean square error (RMSE), and Pearson correlation coefficient (CC) can be calculated to evaluate the model's effectiveness. To better balance these three metrics, comparative experiments can be used to evaluate and validate the performance of the technology.

[0107] Table 2 summarizes the experimental evaluation indicators, showing the comparison results between MFIA and the decision tree (DT) model, autoregressive model (AR), support vector regression with polynomial kernel (SVR-POLY) model, support vector regression with radial basis function kernel (SVR-RBF) model, support vector regression with linear kernel (SVR-LINEAR) model, long short-term memory (LSTM) network, gated recurrent unit (GRU), convolutional neural network + long short-term memory (CNN+LSTM) combination, convolutional neural network + Gated recurrent unit (CNN+GRU), and empirical mode decomposition (EMD) model.

[0108] Table 2:

[0109]

[0110] In the entire process of the above operations S401 to S410, real-time maintenance can be performed in conjunction with the system security maintenance module 450.

[0111] Through the above-described embodiments of the present invention, an intelligent method for estimating raw material properties for technical and economic evaluation of mines is provided. This method, based on complex space, utilizes the interaction of different features using Hermitian operators. Furthermore, by leveraging latent dependency factors, the temporal granularity of different complex embedding vectors at the same time is collected. Ultimately, reasonable estimation of raw material properties is achieved, and the results are visually presented via a web page. Compared to other methods, this method leverages the powerful capacity of complex space to estimate raw material properties with greater efficiency and interpretability, resulting in more accurate cost estimates for mine economic evaluations. This method incorporates mathematical principles into system applications, presenting them in a black-box format. The system interface is simple and aesthetically pleasing, and requires no technical expertise from the operator.

[0112] Figure 5 A block diagram of a raw material property prediction device according to an embodiment of the present invention is shown.

[0113] like Figure 5 As shown, the raw material attribute prediction 500 includes a prediction module 510 , a candidate matching evaluation value calculation module 520 and a target result determination module 530 .

[0114] The prediction module 510 is used to encode and decode the to-be-completed six-tuple consisting of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information, to obtain the real complex embedding vector of the existing attributes in the to-be-completed six-tuple and the predicted complex embedding vector for the predicted attributes to be completed in the to-be-completed six-tuple, where the attributes to be completed include any one of the following: raw material type information, raw material name information, raw material unit information, raw material origin information, and raw material value attribute information. The existing attributes represent other attributes in the to-be-completed six-tuple except the attributes to be completed.

[0115] The candidate matching evaluation value calculation module 520 is used to calculate a candidate matching evaluation value representing the matching degree between the predicted result of the attribute to be completed and the real information of the existing attribute based on the real complex embedding vector and the predicted complex embedding vector.

[0116] The target result determination module 530 is used to determine the target result predicted for the attribute to be completed according to the candidate matching evaluation value.

[0117] According to an embodiment of the present invention, the attributes to be completed are attributes representing raw material value attribute information, the prediction result is the raw material value attribute prediction result, the actual complex embedding vectors are the complex embedding vectors of the raw material type, the complex embedding vectors of the raw material name, the complex embedding vectors of the raw material unit, the complex embedding vectors of the raw material origin, and the complex embedding vectors of the timestamp. The predicted complex embedding vector is the complex embedding vector of the raw material value attribute.

[0118] According to an embodiment of the present invention, the candidate matching evaluation value calculation module includes a first matching evaluation value calculation unit, a second matching evaluation value calculation unit, and a first candidate matching evaluation value calculation unit.

[0119] The first matching evaluation value calculation unit is used to calculate the first matching evaluation value between the raw material value attribute prediction result and the raw material type information, raw material name information, raw material unit information, raw material origin information and timestamp information based on the raw material type complex embedding vector, raw material name complex embedding vector, raw material unit complex embedding vector, raw material origin complex embedding vector, timestamp complex embedding vector and raw material value attribute complex embedding vector.

[0120] The second matching evaluation value calculation unit is used to calculate the second matching evaluation value between the raw material category information and the raw material unit-value attribute information based on the raw material type complex embedding vector, the raw material name complex embedding vector, the raw material unit complex embedding vector, the raw material value attribute complex embedding vector and the timestamp complex embedding vector.

[0121] The first candidate matching evaluation value calculation unit is configured to calculate a candidate matching evaluation value according to at least one of the first matching evaluation value and the second matching evaluation value.

[0122] According to an embodiment of the present invention, the first matching evaluation value calculation unit includes a first Hermitian operator subunit, a second Hermitian operator subunit, a first difference calculation subunit, a second difference calculation subunit and a first exponent calculation subunit.

[0123] The first Hermitian operator subunit is used to calculate, by using the Hermitian operator, an origin-timestamp interaction complex embedding vector between the raw material origin complex embedding vector and the timestamp unit complex embedding vector obtained by normalizing the timestamp complex embedding vector.

[0124] The second Hermitian operator subunit is used to calculate, using the Hermitian operator, a category-origin-timestamp interaction complex embedding vector between a raw material category complex embedding vector obtained by concatenating the raw material type complex embedding vector and the raw material name complex embedding vector and an origin-timestamp interaction unit complex embedding vector obtained by normalizing the origin-timestamp interaction complex embedding vector.

[0125] The first difference calculation subunit is used to calculate the first difference between the first argument of the category-origin-timestamp interaction complex embedding vector and the second argument of the raw material unit-value attribute complex embedding vector obtained by splicing the raw material unit complex embedding vector and the raw material value attribute complex embedding vector.

[0126] The second difference calculation subunit is used to calculate a second difference between the first absolute value of the category-origin-timestamp interaction complex embedding vector and the second absolute value of the raw material unit-value attribute complex embedding vector.

[0127] The first index calculation subunit is configured to perform index calculation on the first difference and the second difference to obtain a first matching evaluation value.

[0128] According to an embodiment of the present invention, the second matching degree evaluation value calculation unit includes a third Hermitian operator subunit, a fourth Hermitian operator subunit, a fifth Hermitian operator subunit, and a second exponent calculation subunit.

[0129] The third Hermitian operator subunit is used to calculate a category-timestamp interaction complex embedding vector between the raw material category complex embedding vector and the timestamp unit complex embedding vector using the Hermitian operator.

[0130] The fourth Hermitian operator subunit is configured to use a Hermitian operator to calculate a unit-value attribute-timestamp interaction complex embedding vector between the unit-value attribute complex embedding vector of the raw material and the unit complex embedding vector of the timestamp. The fifth Hermitian operator subunit is configured to use a Hermitian operator to calculate a third argument of the Hermitian operator calculation result between a preset dependency factor and the category-timestamp interaction complex embedding vector.

[0131] The second exponential calculation subunit is used to perform exponential calculation on the difference between the third absolute value of the third argument and the fourth absolute value of the fourth argument of the unit-value attribute-timestamp interactive complex embedding vector to obtain a second matching evaluation value.

[0132] According to an embodiment of the present invention, the prediction module includes a one-hot encoding unit, a complex conversion unit and a coding and decoding unit.

[0133] The one-hot encoding unit is used to perform one-hot encoding on existing attributes to obtain a binary vector of the existing attributes.

[0134] The complex conversion unit is used to perform complex conversion on the existing attribute binary vector to obtain a real complex embedding vector.

[0135] The encoding and decoding unit is used to encode and decode the real complex embedding vector to obtain the predicted complex embedding vector.

[0136] According to an embodiment of the present invention, there are multiple prediction results. The candidate matching evaluation value calculation module includes a second candidate matching evaluation value calculation unit.

[0137] The second candidate matching evaluation value calculation unit is used to calculate the candidate matching evaluation value between each prediction result and the real information according to each predicted complex embedding vector and the real complex embedding vector, so as to obtain multiple candidate matching evaluation values.

[0138] The target result determination module includes a target result determination unit.

[0139] The target result determination unit is configured to determine the prediction result of the candidate matching evaluation value having the largest value among the plurality of candidate matching evaluation values ​​as the target result.

[0140] Any number of the modules, units, and sub-units according to the embodiments of the present invention, or at least part of the functions of any number of them, can be implemented in a single module. Any one or more of the modules, units, and sub-units according to the embodiments of the present invention can be split into multiple modules for implementation. Any one or more of the modules, units, and sub-units according to the embodiments of the present invention can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented in hardware or firmware through any other reasonable method of circuit integration or packaging, or can be implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, one or more of the modules, units, and sub-units according to the embodiments of the present invention can be at least partially implemented as a computer program module, which can perform the corresponding functions when executed.

[0141] For example, any number of the prediction module 510, the candidate match evaluation value calculation module 520, and the target result determination module 530 can be combined into a single module / unit / sub-unit, or any one of these modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functionality of one or more of these modules / units / sub-units can be combined with at least part of the functionality of other modules / units / sub-units and implemented in a single module / unit / sub-unit. According to an embodiment of the present invention, at least one of the prediction module 510, the candidate match evaluation value calculation module 520, and the target result determination module 530 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented in hardware or firmware by any other reasonable means of circuit integration or packaging, or can be implemented in any one of the three implementation methods of software, hardware, and firmware, or any appropriate combination of any of these. Alternatively, at least one of the prediction module 510 , the candidate matching evaluation value calculation module 520 , and the target result determination module 530 may be at least partially implemented as a computer program module, which may perform corresponding functions when executed.

[0142] Figure 6 A block diagram of an electronic device suitable for implementing a raw material property prediction method according to an embodiment of the present invention is shown. Figure 6 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0143] like Figure 6As shown, an electronic device 600 according to an embodiment of the present invention includes a processor 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 unit 608 into a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0144] Various programs and data required for the operation of the electronic device 600 are stored in the RAM 603. The processor 601, ROM 602, and RAM 603 are connected to each other via a bus 604. The processor 601 executes the programs in the ROM 602 and / or RAM 603 to perform various operations according to the method flow of the embodiment of the present invention. It should be noted that the programs may also be stored in one or more memories other than the ROM 602 and RAM 603. The processor 601 may also execute the programs stored in the one or more memories to perform various operations according to the method flow of the embodiment of the present invention.

[0145] According to an embodiment of the present invention, electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to bus 604. System 600 may also include one or more of the following components connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including devices such as a cathode ray tube (CRT), liquid crystal display (LCD), and speakers; a storage section 608 including a hard disk; and a communication section 609 including a network interface card such as a LAN card or modem. Communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. Removable media 611, such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed in drive 610 as needed, so that computer programs read from the removable media can be installed into storage section 608 as needed.

[0146] According to an embodiment of the present invention, the method flow according to an embodiment of the present invention can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable storage 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 part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, the above-mentioned functions defined in the system of the embodiment of the present invention are executed. According to an embodiment of the present invention, the system, device, apparatus, module, unit, etc. described above can be implemented by a computer program module.

[0147] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently and not incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.

[0148] According to embodiments of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium. Examples include, but are not limited to, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0149] For example, according to an embodiment of the present invention, the computer-readable storage medium may include the ROM 602 and / or the RAM 603 described above and / or one or more memories other than the ROM 602 and the RAM 603 .

[0150] An embodiment of the present invention also includes a computer program product, which includes a computer program, which contains program code for executing the method provided by the embodiment of the present invention. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the raw material property prediction method provided by the embodiment of the present invention.

[0151] When the computer program is executed by the processor 601, the above functions defined in the system / device of the embodiment of the present invention are performed. According to the embodiment of the present invention, the above-described systems, devices, modules, units, etc. can be implemented by computer program modules.

[0152] In one embodiment, the computer program may be stored on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may be transmitted and distributed in the form of a signal on a network medium, downloaded and installed via the communication portion 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0153] According to an embodiment of the present invention, the program code for executing the computer program provided by the embodiment of the present invention can be written in any combination of one or more programming languages. Specifically, these computer programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect via the Internet).

[0154] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of the systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the boxes may occur in an order different from that marked in the accompanying drawings. For example, two boxes shown in succession may actually be executed substantially in parallel, or they may sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, as well as the combination of boxes in the block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or may be implemented using a combination of dedicated hardware and computer instructions. It will be understood by those skilled in the art that the features described in the various embodiments and / or claims of the present invention may be combined and / or coupled in various ways, even if such combinations or couplings are not explicitly described in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features described in the various embodiments and / or claims of the present invention may be combined and / or coupled in various ways, and all such combinations and / or couplings fall within the scope of the present invention.

[0155] The above describes embodiments of the present invention. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. Although each embodiment has been described separately above, this does not mean that the measures in each embodiment cannot be advantageously used in combination. The scope of the present invention is defined by the appended claims and their equivalents. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which are intended to fall within the scope of the present invention.

Claims

1. A method for predicting raw material properties, characterized in that: The method comprises: Encoding and decoding a six-tuple to be completed consisting of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information to obtain real complex embedding vectors of existing attributes in the six-tuple to be completed and predicted complex embedding vectors for predicted attributes to be completed in the six-tuple to be completed, where the existing attributes represent other attributes in the six-tuple to be completed except the attributes to be completed; Calculating a candidate matching evaluation value representing a matching degree between a prediction result of the attribute to be completed and the real information of the existing attribute according to the real complex embedding vector and the predicted complex embedding vector; Determining a target result for predicting the attribute to be completed based on the candidate matching evaluation value; The attributes to be completed are attributes representing the value attribute information of the raw materials, the prediction results are the prediction results of the raw material value attributes, the real complex embedding vectors are the complex embedding vectors of the raw material types, the complex embedding vectors of the raw material names, the complex embedding vectors of the raw material units, the complex embedding vectors of the raw material origins, and the complex embedding vectors of the timestamps; and the predicted complex embedding vectors are the complex embedding vectors of the raw material value attributes. The calculating, based on the real complex embedding vector and the predicted complex embedding vector, a candidate matching evaluation value representing a matching degree between the predicted result of the attribute to be completed and the real information of the existing attribute includes: According to the raw material type complex embedding vector, the raw material name complex embedding vector, the raw material unit complex embedding vector, the raw material origin complex embedding vector, the timestamp complex embedding vector and the raw material value attribute complex embedding vector, calculate the first matching evaluation value between the raw material value attribute prediction result and the raw material type information, the raw material name information, the raw material unit information, the raw material origin information and the timestamp information; wherein, using the Hermitian operator, calculate the origin-timestamp interaction complex embedding vector between the raw material origin complex embedding vector and the timestamp unit complex embedding vector obtained by normalizing the timestamp complex embedding vector; using the Hermitian operator, calculate the matching evaluation value between the raw material type complex embedding vector and the raw material unit complex embedding vector. a category-origin-timestamp interactive complex embedding vector between the raw material category complex embedding vector obtained by concatenating the material name complex embedding vectors and the origin-timestamp interactive unit complex embedding vector obtained by normalizing the origin-timestamp interactive complex embedding vector; calculating a first difference between a first argument of the category-origin-timestamp interactive complex embedding vector and a second argument of the raw material unit-value attribute complex embedding vector obtained by concatenating the raw material unit complex embedding vector and the raw material value attribute complex embedding vector; calculating a second difference between a first absolute value of the category-origin-timestamp interactive complex embedding vector and a second absolute value of the raw material unit-value attribute complex embedding vector; performing exponential calculation on the first difference and the second difference to obtain a first matching evaluation value; Calculating a second matching evaluation value between the raw material category information and the raw material unit-value attribute information based on the raw material type plural embedding vector, the raw material name plural embedding vector, the raw material unit plural embedding vector, the raw material value attribute plural embedding vector, and the timestamp plural embedding vector; The candidate matching evaluation value is calculated according to at least one of the first matching evaluation value and the second matching evaluation value.

2. The method according to claim 1, characterized in that Calculating a second matching evaluation value between the raw material category information and the raw material unit-value attribute information based on the raw material type plural embedding vector, the raw material name plural embedding vector, the raw material unit plural embedding vector, the raw material value attribute plural embedding vector, and the timestamp plural embedding vector includes: Calculating a category-timestamp interaction complex embedding vector between the raw material category complex embedding vector and the timestamp unit complex embedding vector using the Hermitian operator; Calculating a unit-value attribute-timestamp interaction complex embedding vector between the raw material unit-value attribute complex embedding vector and the timestamp unit complex embedding vector using the Hermitian operator; Utilizing the Hermitian operator, calculating a third argument of a Hermitian operator calculation result between a preset dependency factor and the category-timestamp interaction complex embedding vector; An exponential calculation is performed on the difference between the third absolute value of the third argument and the fourth absolute value of the fourth argument of the unit-value attribute-timestamp interactive complex embedding vector to obtain the second matching evaluation value.

3. The method according to claim 1, characterized in that The encoding and decoding of the to-be-completed six-tuple consisting of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information to obtain the real complex embedding vector of the existing attribute in the to-be-completed six-tuple and the predicted complex embedding vector for the predicted attribute in the to-be-completed six-tuple includes: Performing one-hot encoding on the existing attributes to obtain a binary vector of the existing attributes; Performing complex conversion on the existing attribute binary vector to obtain the real complex embedding vector; The real complex embedding vector is encoded and decoded to obtain the predicted complex embedding vector.

4. The method according to claim 1, wherein There are multiple prediction results; The calculating, based on the real complex embedding vector and the predicted complex embedding vector, a candidate matching evaluation value representing a matching degree between the predicted result of the attribute to be completed and the real information of the existing attribute includes: Calculating a candidate matching evaluation value between each of the predicted results and the real information based on each of the predicted complex embedding vectors and the real complex embedding vector to obtain a plurality of candidate matching evaluation values; Determining a target result for the attribute to be completed prediction based on the candidate matching evaluation value includes: The prediction result used to calculate the candidate matching evaluation value with the largest value among the plurality of candidate matching evaluation values ​​is determined as the target result.

5. A raw material property prediction device, characterized in that: The device comprises: a prediction module, configured to encode and decode a six-tuple to be completed consisting of raw material type information, raw material name information, raw material unit information, raw material origin information, raw material value attribute information, and timestamp information, to obtain a true complex embedding vector of an existing attribute in the six-tuple to be completed and a predicted complex embedding vector for the attribute to be completed in the six-tuple to be completed, wherein the existing attribute represents other attributes in the six-tuple to be completed except the attribute to be completed; a candidate matching evaluation value calculation module, configured to calculate a candidate matching evaluation value representing a matching degree between a predicted result of the attribute to be completed and the real information of the existing attribute based on the real complex embedding vector and the predicted complex embedding vector; A target result determination module is used to determine a target result for the attribute to be completed prediction based on the candidate matching evaluation value; The attributes to be completed are attributes representing the value attribute information of the raw materials, the prediction results are the prediction results of the raw material value attributes, the real complex embedding vectors are the complex embedding vectors of the raw material types, the complex embedding vectors of the raw material names, the complex embedding vectors of the raw material units, the complex embedding vectors of the raw material origins, and the complex embedding vectors of the timestamps; and the predicted complex embedding vectors are the complex embedding vectors of the raw material value attributes. The candidate matching evaluation value calculation module includes: The first matching evaluation value calculation unit is used to calculate the first matching evaluation value between the raw material value attribute prediction result and the raw material type information, the raw material name information, the raw material unit information, the raw material origin information and the timestamp information based on the raw material type complex embedding vector, the raw material name complex embedding vector, the raw material unit complex embedding vector, the raw material origin complex embedding vector, the timestamp complex embedding vector and the raw material value attribute complex embedding vector; wherein, the first matching evaluation value calculation unit includes: a first Hermitian operator subunit, used to calculate the origin-timestamp interaction complex embedding vector between the raw material origin complex embedding vector and the timestamp unit complex embedding vector obtained by normalizing the timestamp complex embedding vector using the Hermitian operator; a second Hermitian operator subunit, used to calculate the raw material type complex embedding vector using the Hermitian operator a category-origin-timestamp interactive complex embedding vector between the raw material category complex embedding vector obtained by concatenating the complex embedding vector and the raw material name complex embedding vector and the origin-timestamp interactive unit complex embedding vector obtained by normalizing the origin-timestamp interactive complex embedding vector; a first difference calculation subunit, configured to calculate a first difference between a first argument of the category-origin-timestamp interactive complex embedding vector and a second argument of the raw material unit-value attribute complex embedding vector obtained by concatenating the raw material unit complex embedding vector and the raw material value attribute complex embedding vector; a second difference calculation subunit, configured to calculate a second difference between a first absolute value of the category-origin-timestamp interactive complex embedding vector and a second absolute value of the raw material unit-value attribute complex embedding vector; a first index calculation subunit, configured to perform index calculation on the first difference and the second difference to obtain a first matching evaluation value; a second matching evaluation value calculation unit, configured to calculate a second matching evaluation value between the raw material category information and the raw material unit-value attribute information based on the raw material type plural embedding vector, the raw material name plural embedding vector, the raw material unit plural embedding vector, the raw material value attribute plural embedding vector, and the timestamp plural embedding vector; The first candidate matching evaluation value calculation unit is configured to calculate the candidate matching evaluation value according to at least one of the first matching evaluation value and the second matching evaluation value.

6. An electronic device comprising: one or more processors; a memory for storing one or more programs, The method is characterized in that when the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having executable instructions stored thereon, characterized in that: When the executable instructions are executed by a processor, the processor is caused to implement the method according to any one of claims 1 to 4.

8. A computer program product comprising a computer program, characterized in that The computer program implements the method according to any one of claims 1 to 4 when executed by a processor.

Images