CN116738067A

CN116738067A - Vendor recommendation method and system based on big data

Info

Publication number: CN116738067A
Application number: CN202311032878.9A
Authority: CN
Inventors: 彭剑兵; 吴景文; 湛亮; 熊大为
Original assignee: Hunan Valin E Commerce Co ltd
Current assignee: Hunan Valin E Commerce Co ltd
Priority date: 2023-08-16
Filing date: 2023-08-16
Publication date: 2023-09-12
Anticipated expiration: 2043-08-16
Also published as: CN116738067B

Abstract

The invention relates to the technical field of data processing, in particular to a provider recommendation method and system based on big data. The method comprises the following steps: acquiring information data of suppliers and buyers and analyzing the information data to acquire market data; carrying out data characteristic extraction processing on the supply market data so as to generate supply market characteristic data; performing digital signal code conversion processing on the market feature data to generate a market digital signal; carrying out validity screening on the market digital signals according to a preset valid time period, so as to obtain valid digital signals; performing gray image format conversion processing on the effective digital signal, thereby generating supply market image data; and carrying out image gray value similarity calculation on the supplied market image data to generate image similarity data. The method for recommending the suppliers is realized by calculating the matching degree of the supplier information and the purchasing party information in the supply market.

Description

Vendor recommendation method and system based on big data

Technical Field

The invention relates to the technical field of data processing, in particular to a provider recommendation method and system based on big data.

Background

The current bidding and purchasing items are generally performed after the purchasing party issues the bidding and purchasing items, or after the purchasing party actively invites the suppliers to bid or the suppliers find the bidding and purchasing items by themselves, and there may be a case that the proper suppliers do not find the item. The enterprise is increased in scale, the required products are increased, when the enterprise needs to purchase the products, the enterprise mainly searches for the proper suppliers by manually screening enterprise information, along with the huge number of suppliers in a library in a bidding system, each supplier information data can not be comprehensively inquired during manual inquiry, and the manual judgment mode is easy to generate errors, so that the suppliers meeting the actual requirements of the enterprise can not be accurately selected, and if the suppliers are recommended and matched with a purchasing party, the benefits of the two parties can be well facilitated. However, the conventional supplier recommendation method can only recommend the proper suppliers, but when the recommended suppliers are more, the proper suppliers need to be searched, so that the labor time is wasted, and the matching recommendation of the suppliers cannot be accurately performed.

Disclosure of Invention

Based on the above, the present invention provides a provider recommendation method and system based on big data, so as to solve at least one of the above technical problems.

To achieve the above object, a vendor recommendation method based on big data, the method comprising the steps of:

step S1: acquiring information data of suppliers and purchasing parties and analyzing the information data to obtain market data; carrying out data characteristic extraction processing on the supply market data so as to generate supply market characteristic data;

step S2: performing digital signal code conversion processing on the market feature data to generate a market digital signal;

step S3: carrying out validity screening on the market digital signals according to a preset valid time period, so as to obtain valid digital signals;

step S4: performing gray image format conversion processing on the effective digital signal, thereby generating supply market image data;

step S5: performing image gray value similarity calculation on the supplied market image data to generate image similarity data; comparing and calculating the image similarity data with a preset supply market recommendation threshold, removing the image similarity data when the image similarity data is smaller than the supply market recommendation threshold, and generating supplier recommendation information when the image similarity data is larger than or equal to the supply market recommendation threshold;

step S6: and carrying out optimal provider recommendation information extraction processing on the provider recommendation information according to the image similarity data, thereby generating provider priority recommendation information.

According to the embodiment, the information data of the suppliers and the purchasing parties are obtained and analyzed to obtain the supply market data, a data basis is provided for the subsequent steps, the obtained supply market data is subjected to data characteristic extraction processing, key characteristic information is extracted from the data characteristic extraction processing, and a foundation is laid for the subsequent digital signal code conversion and image similarity calculation; the market feature data is subjected to digital signal coding conversion treatment, so that the market feature data is converted into a digital signal which can be processed by a computer, the data after the digital signal coding is easier to process by the computer, and various complex mathematical operations and analyses can be performed, thereby providing convenience for subsequent effectiveness screening and gray image format conversion; unnecessary data can be removed through a preset effective time period, the effectiveness of the digital signals is improved, the data quantity required to be processed in the subsequent calculation process can be reduced after the ineffective digital signals are removed, and therefore the calculation efficiency and response speed of the whole system are improved, the effective digital signals are screened out, unnecessary errors and noise interference can be avoided, and the accuracy and reliability of the data are ensured; the effective digital signals are converted into a gray image format, abstract digital signals can be converted into specific image data, and further analysis and processing are carried out by applying an image processing technology, such as image similarity calculation, market data are displayed more intuitively, and the market data are more visual and intelligible; the gray value similarity between the image data of the supply market is calculated, the similarity between the image data of the supply market and the gray value similarity is quantitatively evaluated, the accuracy and precision of the supplier recommendation can be improved through image similarity calculation and comparison, better supplier selection recommendation is provided for a purchasing party, the image comparison data and a preset supply market recommendation threshold value are compared and calculated, and the information data of suppliers meeting the conditions are screened out, so that unnecessary information interference and misjudgment are avoided; and re-sequencing and extracting optimal supplier recommendation information according to the image similarity data, further optimizing a supplier recommendation result, improving the selection reference value of a purchasing party to a supplier, and more accurately matching the requirement of the purchasing party and the capacity of the supplier, thereby improving the matching degree of the supplier, reducing the purchasing risk, providing more targeted and credible decision support for the purchasing party and helping the purchasing party to make better business decisions. Therefore, the provider recommendation method based on big data can carry out recommendation matching on the provider and the purchasing party, find out optimal matching recommendation to the provider and the purchasing party in the matching result, does not need to find out the optimal provider through searching, saves labor time and can carry out accurate provider matching recommendation.

In one embodiment of the present specification, step S1 includes the steps of:

step S11: acquiring information data of suppliers and buyers and analyzing the information data to acquire market data;

step S12: carrying out data feasibility filtering processing on the supplied market data to generate market filtering data;

step S13: performing data cleaning treatment on the market filtering data to generate market cleaning data;

step S14: carrying out data noise reduction processing on the market cleaning data by using a supplied market data noise reduction formula to generate market noise reduction data;

step S15: carrying out time sequence data integration processing on the market noise reduction data to generate market sorting data;

step S16: and carrying out data characteristic extraction processing on the market ranking data so as to generate supply market characteristic data.

In the embodiment, the information data of the suppliers and the purchasing parties are obtained and analyzed to obtain the market data, so that a data basis is provided for the subsequent steps; through the data feasibility filtering process, the consistency of the data can be ensured, errors and deviations caused by the difference of data sources or formats are avoided, and the accuracy of subsequent processing is ensured; the market filtering data is subjected to data cleaning treatment, so that useless information and redundant information in the data can be removed, and the accuracy and the credibility of the data are improved; noise and abnormal values in the data can be removed by applying a supplied market data noise reduction formula to carry out noise reduction treatment on the market cleaning data, so that the data is smoother and more continuous, and errors and deviations caused by data jitter or discontinuity are reduced; the method has the advantages that the time sequence data integration processing is carried out on the market noise reduction data, the data in different time periods can be integrated together to form a continuous data sequence taking time as a standard, the data structure is simplified, the complexity of data storage and processing is reduced, the efficiency and performance of subsequent processing are improved, the data characteristic extraction processing is carried out on the market sorting data, the data rules and specific data characteristics contained in the data are found, the processing of redundant data is reduced, and the efficiency and performance of data processing and calculation are improved.

In one embodiment of the present description, the supply market data noise reduction formula in step S14 is as follows:

；

in the method, in the process of the invention,expressed as supplied market data noise reduction index, +.>Expressed as constant +.>Abnormal value coefficient expressed as supply market data, < +.>Expressed as generating noise reduction index from supply market data, < > and->Denoted as +.>Weight information of abnormal noise data of individual supply market data,/->Denoted as +.>Abnormal noise data of individual supply market data, < +.>Expressed as noise reduction data initial adjustment value, +.>Expressed as market evaluation index data, +.>Expressed as noise reduction information correction item->An outlier adjustment represented as a market data noise reduction index.

The present embodiment provides a supply market data noise reduction formula that fully considers outlier coefficients of supply market dataGenerating a noise reduction index from supply market data>First->Weight information of abnormal noise data of individual supply market data +.>First->Abnormal noise data of individual supply market data +.>Noise reduction data initial adjustment value->Market evaluation index data ∈>Noise reduction information correction item->And the interaction relationship with each other to form a functional relationship Repairing abnormal values in abnormal noise data by using noise reduction indexes, adjusting by supplying market evaluation index data and noise reduction data initial adjustment values, adjusting noise reduction data initial adjustment value parameters so as to adapt to different changes of supplying market data, repairing the abnormal noise to obtain a repairing noise function filling the abnormal noise>The noise reduction information is obtained by summing the repairing noise point functions, the noise reduction information is supplemented by utilizing the noise reduction information repairing items, the situation of over fitting or under fitting in the noise reduction process is effectively avoided, interference signals of abnormal data in the data are removed, so that data noise reduction information for repairing each noise point is obtained, the data of the noise point is repaired better while the integrity of the data is ensured, errors and deviations caused by data jitter or discontinuity are reduced, and the abnormal noise point data is processed by the methodThe derivation is carried out, the data dimension is reduced, the calculation is carried out to achieve rapid convergence, the data quantity and the calculation force are saved, and the abnormal adjustment value of the noise reduction index of the market data is supplied>Correction is performed to reduce the error influence on the supply market data caused by the external error data, and the supply market data noise reduction index is generated more accurately >The accuracy and the reliability of the noise reduction processing of the supplied market data are improved. Meanwhile, parameters such as constant items, weight information and adjustment items in the formula can be adjusted according to actual conditions and are applied to different supply market data, so that the flexibility and applicability of the algorithm are improved.

In one embodiment of the present specification, wherein the market digital signal includes a provider digital signal and a purchasing side digital signal, the step S2 includes the steps of:

step S21: performing feature code conversion on the market feature data to generate market discrete feature data;

step S22: carrying out data standardization processing on the market discrete feature data to generate market standard feature data;

step S23: carrying out normalization processing on the market standard characteristic data to generate market normalized characteristic data;

step S24: carrying out feature dimension reduction processing on the market normalized feature data by using a supplied market feature data dimension reduction formula to generate market dimension reduction feature data;

step S25: performing feature vector conversion on the market dimension reduction feature data to generate a supplier feature vector and a purchasing party feature vector; wherein the vendor feature vector comprises: supply product service vector, supply specification vector, supply price vector and release supply product time vector, purchasing party feature vector includes: a demand product service vector, a demand specification vector, a demand price vector, and a release demand product time vector;

Step S26: carrying out undirected graph conversion on the supply product service vector, the supply specification vector, the supply price vector and the release supply product time vector to generate a provider undirected graph;

step S27: carrying out undirected graph conversion on the demand product service vector, the demand specification vector, the demand price vector and the time vector for issuing the demand product by the purchasing party to generate a purchasing party undirected graph;

step S28: carrying out provider digital signal code conversion processing on the provider undirected graph to generate a provider digital signal;

step S29: and carrying out coding conversion processing on the purchasing side digital signal on the purchasing side undirected graph to generate a purchasing side digital signal.

The embodiment performs feature code conversion on the market feature data, so that continuous data is discretized and becomes a classification variable, and the method has readability and interpretability; the market discrete feature data is subjected to data standardization processing, feature data of different magnitudes and measurement modes are converted into the same standard unit and range, the comparability of the data is improved, the standardized market feature data has the same numerical range and distribution condition, the subsequent processing and calculation are easier to perform, and the difficulty and complexity of data processing are reduced; the market characteristic data after normalization processing has the same numerical range and distribution condition, and subsequent processing and calculation are easier to carry out. Meanwhile, the average value and the variance of the normalized characteristic data are changed into 0 and 1, so that optimization of some calculation modes is facilitated, influence caused by different numerical ranges is avoided, errors and deviations caused by the data range are reduced, and the accuracy of data processing and analysis is improved; the feature dimension reduction processing is carried out on the market normalized feature data by utilizing the supplied market feature data dimension reduction formula, irrelevant or redundant feature data can be removed, the complexity and redundancy of a data set are reduced, the dimension of the data after dimension reduction is lower, the calculated amount is correspondingly reduced, the calculation efficiency and the algorithm speed are improved, the main features of the data are reserved, the structure and the rule of the data are clearer, the subsequent analysis and processing are convenient, and meanwhile, the problem of excessive fitting caused by excessive features is avoided; the market dimension reduction feature data is converted into feature vectors of suppliers and purchasing parties, and key information and specific features in the data can be extracted, so that the market current situation and the demands of all parties are better described, and further use and analysis are facilitated; the provider feature vector and the purchasing side feature vector are subjected to undirected graph conversion, so that the relationships among the different feature vectors can be described, potential cooperation opportunities can be analyzed and mined conveniently, the abstract of the data can be reduced through presenting the data in the undirected graph form, and the understanding and the use are convenient; the undirected graph is subjected to digital signal code conversion processing, a group of digital signals can be generated to describe the relation between different undirected graphs, potential cooperation opportunities are conveniently analyzed and mined, a foundation is provided for subsequent mathematical operation and recommendation system construction, analysis and calculation of markets are conveniently carried out, meanwhile, the digital signals are easy to transmit and process, and calculation resources and time cost are saved.

In one embodiment of the present specification, the supply market feature data dimension reduction formula in step S24 is as follows:

；

in the method, in the process of the invention,expressed as a dimension-reduction index of the supplied market feature data, < >>Denoted as +.>Personal potential supply market characteristics data,/->Denoted as +.>Weight information of individual potential supply market feature data, < +.>Denoted as +.>Personal historical supply market characteristic data, < >>Denoted as +.>Weight information of individual historical supply market feature data, < ->Dimension-reducing initial adjustment value expressed as supply market feature data +.>Dimension reduction rationalization index expressed as supplied market feature data,/->Represented as according to->Dimension reduction adjustment values generated by the potential supply market feature data and the historical supply market feature data, +.>An outlier represented as a dimension reduction index of the supply market feature data.

The present embodiment provides a supply market feature data dimension reduction formula that fully considers the firstPersonal potential supply market characteristics data->First->Weight information of individual potential supply market feature data +.>First, the/>Personal historical supply market characteristics data->First->Weight information of individual historical supply market feature data +.>Dimension reduction initial adjustment value +. >Dimension reduction rationalization index for market character data>According to->Dimension reduction adjustment value generated by potential supply market feature data and historical supply market feature data +.>And the interaction relationship with each other to form a functional relationshipThe method has the advantages that the potential supply market characteristic data and the historical supply market characteristic data are compared, the core data of the supply market characteristic data are accurately extracted, the data dimension reduction is carried out by taking the core data as a basis, the supply market characteristic data are reduced from high latitude to low latitude, the calculation is enabled to reach the convergence stage rapidly, the calculation error caused by the dimension of the redundant part is reduced, the obtained data are more accurate, the data are mapped in smaller space data for calculation, the data processing speed is improved under the condition of ensuring the accuracy of the data, the calculation force is saved, the pressure of hardware processing data is reduced, the interrelationship among the supply market data is clearly reflected, and the method is convenient to use laterContinuous data processing and by providing an abnormality adjustment value of the dimension reduction index of the market characteristic data +.>Correcting, reducing error influence of abnormal data on the supply market characteristic data, and generating the dimension reduction index of the supply market characteristic data more accurately >The accuracy and the reliability of the dimension reduction processing of the market feature data are improved. Meanwhile, parameters such as rationalization index, weight information and adjustment items in the formula can be adjusted according to actual conditions and are applied to different market feature data, so that the flexibility and applicability of the algorithm are improved.

In one embodiment of the present specification, wherein the valid digital signals include a provider valid digital signal and a buyer valid digital signal, the step S3 includes the steps of:

step S31: according to a preset effective time period, carrying out release time effectiveness screening on release and supply product time vectors to obtain effective time data of suppliers;

step S32: extracting and processing the effective data of the provider digital signal according to the provider effective time data to generate a provider effective digital signal;

step S33: according to a preset effective time period, carrying out release time validity screening on the release demand product time vector to obtain purchasing party effective time data;

step S34: and extracting and processing the effective data of the purchasing side digital signal according to the purchasing side effective time data to generate the purchasing side effective digital signal.

Through time range screening, the embodiment can eliminate data which does not meet the preset time requirement, extract effective data of suppliers and buyers, remove invalid, redundant or erroneous data, improve the quality and accuracy of the data, facilitate subsequent analysis and use, greatly reduce the data processing amount, optimize the processing efficiency and the computing speed, and simultaneously avoid the influence caused by unnecessary errors and deviations, thereby providing a foundation for subsequent image analysis and other operations.

In one embodiment of the present specification, wherein the supply market image data includes supplier image data and purchasing party image data, step S4 includes the steps of:

step S41: carrying out gray level image format conversion processing on the provider effective digital signal to generate provider image data;

step S42: and carrying out gray level image format conversion processing on the purchasing side effective digital signal to generate purchasing side image data.

According to the embodiment, the provider effective digital signal and the purchasing side effective digital signal are converted into the gray image data, the characteristic difference and the similarity between different providers and purchasing sides can be more intuitively described, the data are presented in the form of images, the abstract of the data can be reduced, the readability and the interpretability of the data are improved, and the understanding and the use are convenient.

In one embodiment of the present specification, step S5 includes the steps of:

step S51: performing image gray value similarity calculation on the provider image data and the purchasing side image data to generate image similarity data;

step S52: and comparing the image similarity data with a preset supply market recommendation threshold value, removing the image similarity data when the image similarity data is smaller than the supply market recommendation threshold value, and generating supplier recommendation information when the image similarity data is larger than or equal to the supply market recommendation threshold value.

According to the method and the device for recommending the image gray value, through image gray value similarity calculation and selection of the preset supply market recommending threshold, recommending results can be adjusted according to actual conditions, the recommending results are more in accordance with requirements and requirements, the similarity between different suppliers and purchasing parties is quantized, recommending accuracy is improved, data with the similarity lower than the preset threshold are removed, computing resources are saved, and processing efficiency is improved.

In one embodiment of the present specification, step S6 includes the steps of:

step S61: carrying out matching degree calculation processing on the image similarity data between the suppliers and the purchasing parties to generate supply market matching data;

step S62: matching degree sorting processing is carried out on the supplier recommendation information according to the supply market matching data, and the supplier sorting recommendation information is generated;

step S63: and carrying out optimal provider recommendation information extraction processing on the provider ranking recommendation information, thereby generating provider priority recommendation information.

According to the embodiment, the matching degree between different suppliers and purchasing parties can be quantified through matching degree calculation processing, and the recommendation accuracy is improved; sequencing the provider recommendation information according to the matching degree from high to low through sequencing processing, and generating a recommendation result which meets the requirements; redundant information and unnecessary data can be removed in a targeted manner through the optimal provider recommended information extraction processing, so that the data processing is simplified, the processing efficiency is improved, and necessary data support is provided for subsequent operation. And simultaneously, better quality recommendation service is provided for the user.

In one embodiment of the present specification, there is provided a big data based vendor recommendation system, comprising:

at least one processor; the method comprises the steps of,

a memory communicatively coupled to the at least one processor; wherein, the liquid crystal display device comprises a liquid crystal display device,

the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the big data based vendor recommendation method as described above.

The invention collects the supplier data and the purchasing side data of the supply market through the web crawler technology, and generates the supply market feature data by processing the data and extracting the feature data, thereby reducing redundant data, invalid data and abnormal data, and further ensuring that the data is more accurate; the method comprises the steps of carrying out code conversion on supply market feature data, converting each piece of data into a specific digital signal according to the supply market feature data, changing the difference between comparison data, converting the data into vectors and undirected graphs in the conversion process, saving the data amount, accelerating the data processing speed, converting each undirected graph into the digital signal according to the feature vectors, screening the digital signal according to the effective time, screening the possibly selected suppliers and purchasing parties, obtaining the latest digital signals of the supplier data and purchasing party data, converting the digital signal into gray image data, conveniently calculating the matching degree between the suppliers and purchasing parties by comparing gray images of the suppliers and purchasing parties, clearly showing the matching degree to a user, extracting the information of the optimal suppliers to be recommended to the purchasing party, saving the time and energy of the selected suppliers or purchasing parties, recommending the suppliers and purchasing parties to be unable to find, and realizing the most suitable connection between the suppliers and purchasing parties based on the large data. Therefore, the provider recommendation method based on big data can carry out recommendation matching on the provider and the purchasing party, find out optimal matching recommendation to the provider and the purchasing party in the matching result, does not need to find out the optimal provider through searching, saves labor time and can carry out accurate provider matching recommendation.

Drawings

FIG. 1 is a flowchart illustrating a vendor recommendation method based on big data according to the present invention;

FIG. 2 is a detailed flowchart illustrating the implementation of step S1 in FIG. 1;

FIG. 3 is a detailed flowchart illustrating the implementation of step S2 in FIG. 1;

FIG. 4 is a flowchart illustrating the detailed implementation of step S3 in FIG. 1;

FIG. 5 is a flowchart illustrating the detailed implementation of step S6 in FIG. 1;

the achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.

Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.

It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

The embodiment of the application provides a provider recommendation method and a provider recommendation system based on big data, wherein the cloud server comprises the following components: at least one of real estate engineering provider cloud server and material engineering provider cloud service. The market digital signals include, but are not limited to: at least one of digital signals composed of 0 or 1, and the like. The supply market image data includes, but is not limited to: at least one of image data composed of black-and-white image lattices.

To achieve the above object, referring to fig. 1 to 5, a vendor recommendation method based on big data, the method comprises the following steps:

Step S1: acquiring information data of suppliers and buyers and analyzing the information data to acquire market data; carrying out data characteristic extraction processing on the supply market data so as to generate supply market characteristic data;

In the embodiment of the present invention, as described with reference to fig. 1, a flowchart of steps of a big data based vendor recommendation method of the present invention is provided, where in this example, the steps of the big data based vendor recommendation method include:

in the embodiment of the invention, the supply market data is obtained by acquiring information data of the suppliers and the purchasing parties which are acquired from outside and analyzing the information relationship between the suppliers and the purchasing parties and the data between the suppliers and the purchasing parties, and a proper data characteristic extraction method is designed according to the requirements, for example, the acquired data is subjected to cleaning, de-duplication, normalization and the like, and the acquired data is processed according to the designed method to generate the supply market characteristic data.

in the embodiment of the invention, a suitable digital signal coding scheme is defined according to the requirements and the nature of the characteristic data, for example, binary coding or other coding modes can be selected. According to the defined digital signal coding scheme, a corresponding digital signal coding conversion method is designed to carry out digital signal coding conversion processing on the market feature data, so that a market digital signal is generated.

in the embodiment of the invention, a proper effective time period is determined according to the requirements and market characteristics. For example, a recent week, a recent month or other time period may be selected, so as to screen the market digital signal according to a preset effective time period, and reject the data outside the effective time period, so as to obtain an effective digital signal.

in the embodiment of the invention, the proper gray image format conversion method is determined according to the requirements and the properties of the effective digital signals, for example, gray conversion, binarization or other format conversion modes can be selected. A corresponding gray image format conversion algorithm is designed for converting the effective digital signal into a gray image format, for example determining whether the image cell is white or black based on 0 or 1 of the digital signal. And carrying out gray image format conversion processing on the effective digital signals according to a designed method, thereby generating market image data.

In the embodiment of the invention, gray value similarity calculation is carried out on the provider image and the purchasing party in the image data of the supply market, black and white grids of the image are matched, if the black and white grids of the image are similar, whether the black and white grids of the matched image are overlapped or not is judged, and if the overlapping degree is higher, the generated image similarity data is higher. And designing a preset supply market recommendation threshold, if the designed supply market threshold is 80%, comparing and calculating the image comparison data with the preset supply market recommendation threshold, removing when the image similarity data is smaller than 80%, and generating supplier recommendation information when the image similarity data is larger than or equal to 80%.

In the embodiment of the invention, the optimal provider recommendation information extraction processing is performed on the provider recommendation information according to the image similarity data, and the provider recommendation information corresponding to the image similarity data with the largest value is subjected to data extraction, so that the provider priority recommendation information is generated.

In one embodiment of the present specification, step S1 includes the steps of:

Step S11: acquiring information data of suppliers and purchasing parties from a cloud server by utilizing a web crawler technology to obtain market data;

As an example of the present invention, referring to fig. 2, a detailed implementation step flow diagram of step S1 in fig. 1 is shown, where step S1 includes:

in the embodiment of the invention, the information data of the suppliers and the purchasing parties are acquired through the outside, and the information relationship between the suppliers and the purchasing parties and the data between the suppliers and the purchasing parties are analyzed to obtain the market supply data.

in the embodiment of the invention, invalid data such as image data, audio data, useless data and the like which are mixed in the crawled market data are filtered by using programming languages and related data processing libraries to generate market filtering data.

in the embodiment of the invention, the abnormal value, redundant data and null value data in the data are subjected to data cleaning by using a programming language and a related data processing library thereof, so that market cleaning data are generated.

in the embodiment of the invention, a supplied market data noise reduction formula is designed, the supplied market data noise reduction formula is utilized to carry out data noise reduction processing on market cleaning data, and abnormal noise data in the data are repaired, so that market noise reduction data with good noise reduction effect is obtained.

in the embodiment of the invention, according to the release time in the market noise reduction data as a time sequence standard, the market noise reduction data is subjected to time sequence data integration processing, such as the market noise reduction data with the latest release time is arranged at the forefront, so that market sequencing data is generated.

In the embodiment of the invention, the feature information related to the market data is defined, the feature information is used as a standard, the machine learning algorithm is utilized to perform feature extraction processing on the market sorting data, the feature variables of the data are generated, and all the feature variables are assembled into a data set to form the market data.

；

in the method, in the process of the invention,expressed as supplied market data noise reduction index, +.>Expressed as constant +.>Abnormal value coefficient expressed as supply market data, < +.>Expressed as generating noise reduction index from supply market data, < > and->Denoted as +.>Weight information of abnormal noise data of individual supply market data,/->Denoted as +.>Abnormal noise data of individual supply market data, < +.>Expressed as noise reduction data initial adjustment value, +.>Representation ofFor supplying market evaluation index data, < > for>Expressed as noise reduction information correction item->An outlier adjustment represented as a market data noise reduction index.

The present embodiment provides a supply market data noise reduction formula that fully considers outlier coefficients of supply market dataGenerating a noise reduction index from supply market data>First->Weight information of abnormal noise data of individual supply market data +.>First->Abnormal noise data of individual supply market data +.>Noise reduction data initial adjustment value->Market evaluation index data ∈>Noise reduction information correction item->And the interaction relationship with each other to form a functional relationship Abnormal noise point by using noise reduction indexRepairing abnormal values in the data, adjusting through supply market evaluation index data and noise reduction data initial adjustment values, adjusting noise reduction data initial adjustment value parameters to adapt to different change conditions of supply market data, repairing abnormal noise points to obtain a repairing noise point function filling the abnormal noise points>The method comprises the steps of summing up the repairing noise point functions to obtain initial noise reduction information, supplementing the initial noise reduction information by utilizing noise reduction information repairing items, effectively avoiding the situation of over fitting or under fitting in the noise reduction process, removing interference signals of abnormal data in data, obtaining data noise reduction information for repairing each noise point, repairing the noise point data better while guaranteeing the integrity of the data, reducing errors and deviations caused by jitter or discontinuity of the data, and reducing the dimension of the data by seeking the abnormal noise point data so as to achieve rapid convergence, saving the data quantity and calculation force, and supplying an abnormal adjustment value of market data noise reduction index>Correction is performed to reduce the error influence on the supply market data caused by the external error data, and the supply market data noise reduction index is generated more accurately >The accuracy and the reliability of the noise reduction processing of the supplied market data are improved. Meanwhile, parameters such as constant items, weight information and adjustment items in the formula can be adjusted according to actual conditions and are applied to different supply market data, so that the flexibility and applicability of the algorithm are improved.

As an example of the present invention, referring to fig. 3, a detailed implementation step flow diagram of step S2 in fig. 1 is shown, where step S2 includes:

in the embodiment of the invention, the coding mode, such as single-hot coding or transition coding, is selected according to the market feature data. And performing code conversion on the market feature data by using the selected coding mode, and combining the converted result into a new discrete feature data set so as to generate the market discrete feature data.

in the embodiment of the invention, the Z-score standardization is selected to carry out standardization processing on the market discrete feature data, the same standard is selected for each data to be processed according to the standardization method, so that the average value of the features of the data is 0, the standard deviation is 1, and all the standardized feature variables are combined into one data set, namely the market standard feature data.

In the embodiment of the invention, a preset data proportion interval is utilized, the maximum value of data is set as the maximum value of the data proportion interval, the minimum value is set as the minimum value of the data proportion interval, and Min-Max normalization processing is carried out on market standard characteristic data, so that market normalization characteristic data are generated.

in the embodiment of the invention, a supplied market feature dimension reduction formula is selected as a data dimension reduction method, the core dimension of the market normalized feature data is extracted, the redundant dimension is removed, and feature dimension reduction processing is carried out on the market normalized feature data to generate the market dimension reduction feature data.

In the embodiment of the invention, market dimension reduction feature data are divided according to service requirements and suppliers and purchasing parties to obtain different data sets, then feature vector conversion is carried out on the data, and feature vectors obtained in each data set are combined according to corresponding suppliers or purchasing parties, so that supplier feature vectors and purchasing party feature vectors are generated; wherein the vendor feature vector comprises: supply product service vector, supply specification vector, supply price vector and release supply product time vector, purchasing party feature vector includes: a demand product service vector, a demand specification vector, a demand price vector, and a release demand product time vector.

in the embodiment of the invention, a service vector of a supplied product, a supply specification vector, a supply price vector and a time vector for releasing the supplied product for constructing the undirected graph are selected, each vector is converted as a node of the undirected graph, the similarity or distance between the nodes on the selected vector is calculated, whether edge connection exists or not is determined according to a set threshold value, each provider feature vector is enabled to generate a specific undirected graph, and the undirected graph is integrated to generate the provider undirected graph.

in the embodiment of the invention, a service vector of a demand product, a demand specification vector, a demand price vector and a time vector for issuing the demand product for constructing the undirected graph are selected, each vector is converted as a node of the undirected graph, the similarity or distance between the nodes on the selected vector is calculated, whether edge connection exists or not is determined according to a set threshold value, each purchasing party feature vector is enabled to generate a specific undirected graph, and the undirected graph is integrated to generate the purchasing party undirected graph.

in the embodiment of the invention, for each node of the provider undirected graph, the degree of the undirected graph is calculated, the degree is added to the node as a mark, the mark on each node is converted into a binary code according to a set bit number (for example, 8 bits), the binary codes of all nodes are connected according to a certain sequence to form a long binary sequence, and encryption processing is carried out on binary training by using encryption modes such as AES, DES and the like, so that a provider digital signal is generated.

In the embodiment of the invention, for each node of the undirected graph of the purchasing party, the degree of the undirected graph is calculated, the degree is used as a mark to be added to the node, the mark on each node is converted into a binary code according to a set bit number (for example, 8 bits), the binary codes of all the nodes are connected according to a certain sequence to form a long binary sequence, and encryption processing is carried out on binary training by using encryption modes such as AES, DES and the like, so that the purchasing party digital signal is generated.

；/>

in the method, in the process of the invention,expressed as a dimension-reduction index of the supplied market feature data, < >>Denoted as +.>Each diveIn the supply of market feature data, < >>Denoted as +.>Weight information of individual potential supply market feature data, < +.>Denoted as +.>Personal historical supply market characteristic data, < >>Denoted as +.>Weight information of individual historical supply market feature data, < ->Dimension-reducing initial adjustment value expressed as supply market feature data +. >Dimension reduction rationalization index expressed as supplied market feature data,/->Represented as according to->Dimension reduction adjustment values generated by the potential supply market feature data and the historical supply market feature data, +.>An outlier represented as a dimension reduction index of the supply market feature data.

The present embodiment provides a supply market feature data dimension reduction formula that fully considers the firstPersonal potential supply market characteristics data->First->Weight information of individual potential supply market feature data +.>First->Personal historical supply market characteristics data->First->Weight information of individual historical supply market feature data +.>Dimension reduction initial adjustment value +.>Dimension reduction rationalization index for market character data>According to->Dimension reduction adjustment value generated by potential supply market feature data and historical supply market feature data +.>And the interaction relationship with each other to form a functional relationshipBy comparing the potential supply market feature data with the historical supply market feature data, the core data of the supply market feature data is accurately extractedThe core data is used as a basis for data dimension reduction, supplied market characteristic data is reduced from high latitude to low latitude, so that calculation reaches a convergence stage rapidly, calculation errors caused by redundant part dimension are reduced, obtained data is more accurate, the data is mapped in smaller space data for calculation, data processing speed is improved under the condition that the data is accurate, calculation force is saved, pressure of hardware processing data is reduced, interrelationships among the supplied market data are clearly reflected, subsequent data processing is facilitated, and abnormal adjustment values of dimension reduction indexes of the supplied market characteristic data are realized >Correcting, reducing error influence of abnormal data on the supply market characteristic data, and generating the dimension reduction index of the supply market characteristic data more accurately>The accuracy and the reliability of the dimension reduction processing of the market feature data are improved. Meanwhile, parameters such as rationalization index, weight information and adjustment items in the formula can be adjusted according to actual conditions and are applied to different market feature data, so that the flexibility and applicability of the algorithm are improved.

As an example of the present invention, referring to fig. 4, a detailed implementation step flow diagram of step S3 in fig. 1 is shown, where step S3 includes:

in the embodiment of the present invention, the preset valid period is determined according to the service requirement and the analysis purpose, for example, a time window of 30 days before and after each other may be set as the valid period, and the release time exceeding the time window is regarded as the invalid time. For each supplier, each element in the time vector of the release and supply product is compared with a preset effective time period, whether the time is within the effective time is judged, if the time is within the effective time, the time is considered to be the effective time, and otherwise, the time is not effective time. All of its time of validity is recorded and stored as a separate data set, i.e., vendor time of validity data.

in the embodiment of the invention, the provider digital signals reaching the effective time period and the provider digital signals not reaching the effective time period are divided according to the provider effective time data, the provider digital signals reaching the effective time period are extracted, and the provider digital signals not reaching the effective time period are removed, so that the provider effective digital signals are generated.

in the embodiment of the present invention, the preset valid period is determined according to the service requirement and the analysis purpose, for example, a time window of 30 days before and after each other may be set as the valid period, and the release time exceeding the time window is regarded as the invalid time. And comparing each element in the time vector of the release demand product of each purchasing party with a preset effective time period, judging whether the time is within the effective time, if so, considering the time as the effective time, and otherwise, judging the time as the ineffective time. All of its time of validity is recorded and stored as an independent data set, namely purchasing party time of validity data.

In the embodiment of the invention, the purchasing side digital signals reaching the effective time period and the purchasing side digital signals not reaching the effective time period are divided according to the purchasing side effective time data, the purchasing side digital signals reaching the effective time period are extracted, and the purchasing side digital signals not reaching the effective time period are removed, so that the provider effective digital signals are generated.

In the embodiment of the invention, the provider effective digital signal and the purchasing side effective digital signal are subjected to gray image conversion, and black grids and white grids in an image block are generated according to 0 or 1 of the digital signals. Converting the provider effective digital signal into an identification image of a black-and-white grid according to a preset provider image data standard, so as to generate provider image data; and converting the purchasing side effective digital signal into an identification image of a black-white grid according to a preset purchasing side image data standard, so as to generate purchasing side image data.

In one embodiment of the present specification, step S5 includes the steps of:

In the embodiment of the invention, the image gray value calculation is carried out on the image data of the supplier and the image data of the purchasing party, the two photographs are compared, the black-white grids are the matching factors when appearing in the same place, the black-white grids are the non-matching factors when not appearing in the same place, and the ratio occupied by the matching factors is calculated, so that the image similarity data is generated. And designing a preset supply market recommendation threshold, if the designed supply market threshold is 80%, comparing and calculating the image comparison data with the preset supply market recommendation threshold, removing when the image similarity data is smaller than 80%, and generating supplier recommendation information when the image similarity data is larger than or equal to 80%.

In one embodiment of the present specification, step S6 includes the steps of:

As an example of the present invention, referring to fig. 5, a detailed implementation step flow diagram of step S6 in fig. 1 is shown, where step S6 includes:

in the embodiment of the invention, the matching degree of the provider and the purchasing party corresponding to the image can be calculated through the image similarity data, and the generated matching degree data is integrated, so that the supply market matching data is generated, and the supply market matching data and the image similarity data are in a direct proportion relation.

in the embodiment of the invention, matching degree ranking processing is performed on the corresponding supplier recommendation information according to the size of the supply market matching data, if the maximum value of the supply market matching data is 98%, the A supplier recommendation information corresponding to the supply market matching data is ranked at the forefront, and the like, so that the supplier ranking recommendation information is generated.

In the embodiment of the invention, the optimal supplier recommendation information in the supplier sequencing recommendation information is extracted, for example, the supplier sequencing recommendation information is arranged at the forefront A supplier recommendation information for data extraction, so that the supplier priority recommendation information is generated.

at least one processor; the method comprises the steps of,

Claims

1. A vendor recommendation method based on big data, the method comprising the steps of:

2. The big data based vendor recommendation method according to claim 1, wherein step S1 comprises the steps of:

3. The big data based vendor recommendation method according to claim 2, wherein the supply market data noise reduction formula in step S14 is as follows:

；

in the method, in the process of the invention,expressed as supplied market data noise reduction index, +.>Expressed as constant +.>Abnormal value coefficient expressed as supply market data, < +.>Expressed as generating noise reduction index from supply market data, < > and- >Denoted as +.>Weight information of abnormal noise data of individual supply market data,/->Denoted as +.>Abnormal noise data of individual supply market data, < +.>Expressed as noise reduction data initial adjustment value, +.>Expressed as market evaluation index data, +.>Expressed as noise reduction information correction item->Representation ofTo supply an abnormal adjustment value of the market data noise reduction index.

4. The big data based supplier recommendation method of claim 2, wherein the market digital signal includes a supplier digital signal and a purchasing side digital signal, step S2 includes the steps of:

5. The big data based vendor recommendation method according to claim 4, wherein the supply market feature data dimension reduction formula in step S24 is as follows:

；

in the method, in the process of the invention,expressed as a dimension-reduction index of the supplied market feature data, < >>Denoted as +.>Personal potential supply market characteristics data,/->Denoted as +.>Weight information of individual potential supply market feature data, < +.>Denoted as +.>Personal historical supply market characteristic data, < >>Denoted as +.>Weight information of individual historical supply market feature data, < - >Dimension-reducing initial adjustment value expressed as supply market feature data +.>Dimension reduction rationalization index expressed as supplied market feature data,/->Represented as according to->Dimension reduction adjustment values generated by the potential supply market feature data and the historical supply market feature data, +.>An outlier represented as a dimension reduction index of the supply market feature data.

6. The big data based vendor recommendation method of claim 4, wherein the valid digital signals comprise a vendor valid digital signal and a buyer valid digital signal, and step S3 comprises the steps of:

7. The big data based supplier recommendation method according to claim 6, wherein the supply market image data includes supplier image data and purchasing party image data, and step S4 includes the steps of:

8. The big data based vendor recommendation method according to claim 7, wherein step S5 comprises the steps of:

9. The big data based vendor recommendation method according to claim 8, wherein step S6 comprises the steps of:

10. A big data based vendor recommendation system comprising:

at least one processor; the method comprises the steps of,

the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the big data based vendor recommendation method according to any of claims 1 to 9.