CN106600356B - Multi-platform e-commerce information aggregation method and system - Google Patents
Multi-platform e-commerce information aggregation method and system Download PDFInfo
- Publication number
- CN106600356B CN106600356B CN201610957162.3A CN201610957162A CN106600356B CN 106600356 B CN106600356 B CN 106600356B CN 201610957162 A CN201610957162 A CN 201610957162A CN 106600356 B CN106600356 B CN 106600356B
- Authority
- CN
- China
- Prior art keywords
- information
- neural network
- commerce
- platform
- commodity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0641—Shopping interfaces
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a multi-platform e-commerce information aggregation method and a multi-platform e-commerce information aggregation system, which comprise the following steps: acquiring page category information of each E-commerce platform about sample commodities; carrying out similarity analysis, analyzing the minimum information elements of the sample commodities based on the analysis result, combining the minimum information elements to obtain the minimum information element set of the sample commodities on each E-commerce platform, and forming a comprehensive sample set by the minimum information element sets of the sample commodities; aiming at each e-commerce platform, taking the minimum information element set and category information acquired from one e-commerce platform as a training sample pair, and carrying out neural network training with the weight coefficient value of 0 or 1; inputting the minimum information element set of the commodities to be aggregated according to the neural network aggregation models of different e-commerce platforms, and obtaining aggregated category information meeting the requirement of the corresponding e-commerce platform commodity page through model calculation. The defects of large maintenance workload, low efficiency, easy error, incapability of finding change in time and the like caused by non-uniform interfaces of each platform in the prior art are overcome.
Description
Technical Field
The invention relates to an electronic commerce information technology, in particular to an information aggregation technology for commodity publishing and maintenance of multi-platform cross-border electronic commerce activities.
Background
Nowadays, with the development of the world economy integration and globalization, the influence of electronic commerce on the economic development and social transition of each country is gradually enhanced. China occupies the most population in the world and has huge market foundation and economic vitality. As early as 2010, China has become the country with the highest global manufacturing yield. Under the background of rapid development of electronic commerce in the world, small and medium-sized enterprises in China combine electronic commerce technology, particularly cross-border electronic commerce, and the enterprises in China are pushed to enter the international market by virtue of a new trade mode, so that the electronic commerce is more and more feasible. The medium and small enterprises in China enter the international market by using cross-border electronic commerce, and have important and profound significance for expanding the international market share of products in China, expanding a foreign trade marketing network, breeding new export competitive advantages, converting the transaction mode of the traditional international trade, promoting the industry transformation and the industry upgrade of the foreign trade in China and improving the competitiveness of the foreign trade in the international market in China.
The cross-border electronic commerce refers to an international business trading activity in which transaction subjects belonging to different countries or regions respectively complete transactions, pay settlement and cross-border logistics distribution of commodities through an electronic commerce platform. It features globality, instantaneity and convenience. In 2013, the cross-border e-commerce transaction scale in China is 3.1 trillion yuan, the growth rate is 31.3%, and the cross-border e-commerce transaction scale accounts for 11.9% of the total volume of import and export trades, although the percentage of the cross-border e-commerce transaction scale is low, along with globalization of the e-commerce in China, the position of the cross-border e-commerce in the foreign trade in China is higher and higher, and the proportion of the cross-border e-commerce transaction scale in 2017 in the total volume of import and export trades is expected to reach about.
The government of China actively promotes the development of cross-border electronic commerce, and 7 cities or regions such as Shanghai, Chongqing, Hangzhou, Ningbo, Zhengzhou, Shenzhen, the former sea and Qingdao and the like have successively obtained cross-border electronic commerce pilot cities at present. Meanwhile, the intensive emergence of a series of cross-border e-commerce policies also greatly promotes the policy environment, laws, regulations, standard systems, support and guarantee levels and the like of cross-border e-commerce in China, and the development momentum of the cross-border e-commerce is strong.
One of the important obstacles for domestic small and medium-sized enterprises to enter the international market is the high cost of information acquisition, and the cost of information acquisition is reduced by optimizing the supply chain flow in electronic commerce, the integrated operation level of enterprise supply chain management is improved, and further, the management cost and the information acquisition cost are also greatly reduced. Enterprises involved in cross-border electronic commerce firstly need to build and maintain the electronic informatization level of the enterprises, and only about 7000 enterprises create the electronic information platforms in about 25 ten thousand of small and medium-sized enterprises engaged in the cross-border electronic commerce, while other numerous small and medium-sized enterprises still need to rely on third party platforms to conduct external commerce. However, overseas markets are huge, third-party cross-border electronic platforms are numerous, release specifications and requirements of commodity information are different, workload and specialization degree of information maintenance are high, at present, maintenance of most of different platforms needs to be completed manually by professionals, and maintenance of a multi-platform electronic commerce system becomes a heavy burden for small and micro enterprises with limited resources. The construction of a cross-border e-commerce management platform requires the integration of a large number of distributed heterogeneous information resources, which may be distributed in different third-party platform systems, which is a complex and arduous task. How to effectively reuse information resources of different portals, automatically construct an information management and maintenance system of commodities, and automatically publish and maintain commodities of different platforms, thereby greatly improving the cross-border electronic commerce commodity information management efficiency and reducing the publishing and maintenance cost of commodities on a plurality of third-party cross-border electronic commerce platforms becomes a key technical problem to be solved urgently.
At present, cross-border electronic commerce platforms are numerous, mainstream platforms such as Sumaitong, Amazon, eBay, Wish, Lanting trend, Dunhuang, Walmart, Newegg and the like are adopted, technical standards and background databases adopted by each platform are different, and for enterprises engaged in cross-border, information and description of commodities need to be issued on different platforms respectively according to requirements of the platforms, and dynamic management is carried out on the commodities, including timely and synchronous updating of information of different platforms of the same commodity; remapping the commodity description items changed by a certain platform, and the like; and editing the attributes of the new product to adapt to different requirements of each platform and pushing the attributes to the platform to finish commodity loading. In the past, the release, maintenance and update of commodities need manual processing, the workload is large, the efficiency is low, and errors are easy to occur, therefore, some platforms release data interfaces, cross-border enterprises can release and update batch products through the data interfaces, however, the interfaces and data formats of different platforms are different, for cross-border enterprises which operate on a plurality of platforms at the same time, a large amount of manual operations still need to be performed, and to realize the computer automatic processing of the process, the following problems need to be specifically solved:
1) unstructured of the data. Most of data information of the modern cross-border e-commerce platform is full of a large amount of unstructured data, and the proportion of the structured data is still low, which is a big problem for accessing the database, and only the unstructured data is converted into the structured data in a uniform data format.
2) Mapping and combination among commodity minimum information elements, data class item difference among platforms, description categories of each platform for commodities are different, some attributes (such as commodity names) are common attributes of each platform, some attributes are unique to individual platforms, and the others are different in description requirements and information aggregation degrees although the platforms are common, so that some differences exist, and some complex attributes are formed by combining simple attributes. Cross-border enterprises need to perform uniform maintenance of commodities, so a large commodity data category containing all platform attributes needs to be established, commodity data is uniformly organized and planned, and an original attribute comprehensive set needs to be screened and merged.
3) And respectively generating platform commodity descriptions in batches according to the platform where the commodity is located, wherein the commodity description categories and formats of each platform are different, and dynamic mapping relations between different platform information and description information sets need to be established, so that a commodity attribute list and contents of a target platform are dynamically generated according to the attribute comprehensive set and the target platform, and the commodity information of different platforms is uniformly updated and maintained.
In cross-border electronic commerce activities, a large number of small and medium-sized enterprises which rely on a third-party platform to conduct external trade exist, the small and medium-sized enterprises are weak in technical strength and deficient in human resources, and the problem becomes a technical problem which is urgently needed to be solved and influences the small and medium-sized cross-border enterprises to rapidly expand product sales markets.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a multi-platform e-commerce information aggregation method and system, and overcome the defects that in the prior art, when a plurality of cross-border e-commerce platforms are released and information is updated, the maintenance workload is large, the efficiency is low, errors are easy to occur, the change of a platform system cannot be found in time, and the information updating cannot be carried out due to the fact that the platform interfaces are not uniform.
In order to solve the above problems, the present invention provides a multi-platform e-commerce information aggregation method, which includes the following steps:
acquiring page category information of each E-commerce platform about sample commodities;
carrying out similarity analysis on all the obtained category information of the same sample commodity on different E-commerce platform pages, analyzing the minimum information element of the sample commodity based on the similarity analysis result, combining the minimum information elements to obtain the minimum information element set of the sample commodity, and forming a comprehensive sample set by the minimum information element sets of a plurality of sample commodities;
aiming at each e-commerce platform, taking a minimum information element set of a sample commodity and category information acquired from one e-commerce platform as a training sample pair, and carrying out neural network training with the weight coefficient value of 0 or 1 to obtain a neural network aggregation model of each e-commerce platform;
inputting the minimum information element set of the commodity to be aggregated according to the neural network aggregation models of different e-commerce platforms, and obtaining category information corresponding to the commodity page requirements of the e-commerce platforms through model calculation.
According to an embodiment of the present invention, further comprising:
when the commodity information is changed, the commodity information is input into the neural network aggregation model again according to the updated minimum information element set, updated category information of each e-commerce platform is obtained, and the category information is uploaded to the corresponding e-commerce platform.
According to an embodiment of the present invention, further comprising:
when the category information of the monitored E-commerce platform sample commodity page changes, carrying out similarity analysis and merging to obtain a new minimum information element set, constructing a new sample pair of a target E-commerce platform, retraining a new neural network aggregation model, inputting the original minimum information element set of the commodity to be aggregated into the updated neural network aggregation model to obtain the changed category information, and uploading the changed category information to the target E-commerce platform.
According to one embodiment of the invention, related commodity webpage resources of each e-commerce platform are directionally captured by adopting a focusing crawler based on a target data mode so as to obtain category information of each e-commerce platform about commodities.
According to an embodiment of the invention, each neural network aggregation model is a mapping model of the minimum information element set to the category information of an e-commerce platform, and comprises an input layer, a plurality of layers or a single layer of hidden layers and an output layer, wherein the input node number is determined according to the element number of the minimum unit information set, the number of the hidden layers and the node number are determined according to the granularity of information aggregation, and the output node number is determined according to the category number of the corresponding e-commerce platform.
According to one embodiment of the invention, a neural network feedback correction algorithm is used to train and optimize parameters of the neural network aggregation model.
According to one embodiment of the present invention, the training and optimizing comprises:
adopting a binary GA algorithm, wherein the weight parameter of the neural network aggregation model is taken as 0 or 1, and the hidden node and the output node perform weighted summation on input information to obtain the synthesis of different information elements;
and comparing the output value of the neural network aggregation model with the output value of the sample pair, and changing the weight parameter according to the comparison result to obtain the neural network aggregation model suitable for each E-commerce platform.
According to one embodiment of the invention, the neural network training includes an offline training mode, comprising the steps of:
a1: randomly generating a chromosome set of a binary GA algorithm, selecting an initial chromosome as an initial weight coefficient value of a neural network, calculating a fitness value, and assigning the initial chromosome to a historical optimal value;
a2: copying, mutating and crossing chromosome groups to obtain a chromosome population collection after copying, mutating and crossing, respectively taking chromosomes in the chromosome population collection as weight coefficient values of a neural network, and calculating each fitness value;
a3: selecting the chromosome with the minimum fitness in the fitness values, if the minimum fitness is smaller than the historical optimum, assigning the corresponding chromosome to the historical optimum, otherwise keeping the historical optimum unchanged, screening a new chromosome set generated by variation based on a fitness function by adopting a roulette mode to obtain a next generation chromosome set, and waiting for iteration;
a4: and (3) judging whether the model outputs of all the sample pairs are consistent with the sample outputs when the chromosomes with the historical optimal values are applied to the neural network aggregation model, if so, finishing iterative learning and outputting the optimal chromosomes, and otherwise, selecting a next generation chromosome set and returning to the step A2.
According to one embodiment of the invention, the neural network training further comprises an online operation mode comprising the steps of:
b1: selecting a weight coefficient and a structure of a corresponding neural network aggregation model according to the target e-commerce platform to generate a corresponding neural network aggregation model;
b2: and inputting a minimum information element set, and calling a neural network aggregation model to calculate the category information of the commodity page of the target e-commerce platform.
The invention also provides a multi-platform e-commerce information aggregation system, which comprises:
the platform information acquisition module is used for acquiring page category information of each E-commerce platform about the sample commodity;
the information similarity analysis integration module is used for carrying out similarity analysis on all the obtained category information of the same sample commodity on different cross-border e-commerce platform pages, analyzing the minimum information elements of the sample commodity based on the similarity analysis result, combining the minimum information elements to obtain the minimum information element set of the sample commodity, and forming a comprehensive sample set by the minimum information element sets of the sample commodities;
the aggregation model training module is used for carrying out neural network training with the weight coefficient value of 0 or 1 by taking the minimum information element set of the sample commodity and the category information acquired from one e-commerce platform as a training sample pair aiming at each e-commerce platform to obtain a neural network aggregation model of each e-commerce platform;
and the aggregation information calculation module is used for inputting the minimum information element set of the commodity to be aggregated according to the neural network aggregation models of different e-commerce platforms, and obtaining category information corresponding to the commodity page requirements of the e-commerce platforms through model calculation.
According to an embodiment of the present invention, further comprising:
the aggregation model dynamic updating module is used for inputting the updated minimum information element set into the neural network aggregation model again when the commodity information is changed, obtaining the updated category information of each e-commerce platform and uploading the category information to the corresponding e-commerce platform; and/or when the category information of the monitored E-commerce platform sample commodity page changes, carrying out similarity analysis and merging to obtain a new minimum information element set, constructing a new sample pair of the target E-commerce platform, retraining a new neural network aggregation model, inputting the original minimum information element set of the commodity to be aggregated into the updated neural network aggregation model to obtain the changed category information, and uploading the changed category information to the target E-commerce platform.
After the technical scheme is adopted, compared with the prior art, the invention has the following beneficial effects:
a neural network aggregation model is established through the 0-1 connection weight coefficient, so that the aggregation of basic elements of commodity information can be effectively carried out, category information conforming to each target e-commerce platform page can be quickly obtained, and the method can be used for information aggregation for multi-platform cross-border e-commerce;
the invention provides a redesigned neural network model for better achieving the purpose of information aggregation, wherein the parameters adopt 0-1 integer, the conversion function adopts addition function, and the training adopts 2-system GA algorithm, which are not described in the traditional or current technology;
the problem of model parameter discontinuity is caused by the introduction of an integer 0-1 neural network aggregation model, and the problems of learning and training of the neural network aggregation model on a sample machine are effectively solved by a mapping method and a training method of the chromosome of the binary GA and the weight of the neural network aggregation model;
the dynamic detection and update mechanism of the system detects the change condition of the commodity information and the cross-border E-commerce platform page category dynamically, and immediately performs page category content recalculation or retraining of the neural network aggregation model once the change is detected, so that the problems of time delay and asynchronous page content update brought by the traditional manual update mode are solved.
Drawings
FIG. 1 is a flowchart illustrating a multi-platform e-commerce information aggregation method according to an embodiment of the present invention;
FIG. 2 is a block diagram of a neural network aggregation model according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating an operation principle of a multi-platform e-commerce information aggregation method according to an embodiment of the present invention;
FIG. 4 is a schematic flow chart of neural network training in an offline training mode according to an embodiment of the present invention;
FIG. 5 is a flow chart illustrating the operation of the neural network in the online operation mode according to an embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating an online dynamic update workflow of a neural network aggregation model according to an embodiment of the present invention;
FIG. 7 is a diagram illustrating the correspondence between weight coefficients of a neural network aggregation model and binary GA chromosomes according to an embodiment of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein, but rather construed as limited to the embodiments set forth herein.
Referring to fig. 1, the multi-platform e-commerce information aggregation method of the embodiment of the present invention includes the following steps:
s1: acquiring page category information of each E-commerce platform about sample commodities;
s2: analyzing the similarity of all the acquired category information, analyzing the minimum information elements of the sample commodities based on the similarity analysis result, combining the minimum information elements to obtain the minimum information element set of the sample commodities, and forming a comprehensive sample set by the minimum information element sets of the sample commodities;
s3: aiming at each e-commerce platform, taking a minimum information element set of a sample commodity and category information acquired from one e-commerce platform as a training sample pair, and carrying out neural network training with the weight coefficient value of 0 or 1 to obtain a neural network aggregation model of each e-commerce platform;
s4: inputting the minimum information element set of the commodity to be aggregated according to the neural network aggregation models of different e-commerce platforms, and obtaining category information corresponding to the commodity page requirements of the e-commerce platforms through model calculation.
Each e-commerce platform may have more or less differences in the classification and description data of the commodity attributes of its own commodity page, for example, the volume of the commodity defined by one e-commerce platform may be defined as the length, width and height in another e-commerce platform. In step S1, the product pages of the e-commerce platforms may be captured to obtain different attribute classifications and description data of the products, which may be obtained in other manners. In other words, category information about the goods is first acquired for each e-commerce platform. The category information is information obtained after the e-commerce platform classifies the attributes of the commodities and can include categories and category contents of corresponding classifications.
Optionally, in step S1, the focused crawler is used to directionally capture the related commodity webpage resources of each e-commerce platform based on the target data mode, so as to obtain the category information of each e-commerce platform about the commodity. Specifically, the method is completed by three parts, namely a controller, a resolver and a resource library. The main task of the controller is to be responsible for assigning work tasks to each crawler thread in multiple threads. The controller is a central controller of the web crawler and is mainly responsible for distributing a thread according to a URL link transmitted by the system and then starting the thread to call a process of crawling the web pages. The main work of the resolver is to download a webpage, process the webpage, mainly process some JS script tags, CSS code content, space characters, HTML tags and other content, the basic work of the crawler is completed by the resolver, and the work content is as follows: the function of downloading the webpage, the function of processing the text of the webpage, such as a filtering function, a function of extracting a special HTML label, a data analysis function and the like. The resource library is used for storing downloaded web page resources, generally adopts a large database storage, such as an Oracle database, establishes an index for the large database storage, is used for storing a container of data records downloaded from a web page, and provides a target source for generating the index. Medium and large database products are: oracle, Sql Server, etc.
In step S2, similarity analysis is performed on all the category information acquired in step S1, and each minimum information element is analyzed based on the result of the similarity analysis, specifically, similar elements are retained one, dissimilar elements are retained all, and elements that are combined together are decomposed, for example, a long-wide high element is decomposed into three elements, such that all the minimum information elements of a commodity are obtained, and the minimum information elements are combined to obtain a minimum information element set, which is used as a comprehensive sample set of the commodity information for learning, training and operation.
In step S3, for each e-commerce platform, each e-commerce platform trains its own neural network aggregation model, performs neural network training with a weighting coefficient value of 0 or 1 each time taking the minimum information element set obtained in step S2 and the category information obtained from the e-commerce platform as a sample pair, obtains a neural network aggregation model of the e-commerce platform, and trains each e-commerce platform to obtain its own neural network model.
Each neural network aggregation model is a mapping model of the minimum information element set to the category information of an e-commerce platform, and comprises an input layer, a plurality of layers or a single-layer hidden layer and an output layer, wherein the number of input nodes is determined according to the number of elements of the minimum unit information set, the number of hidden layers and the number of hidden nodes are determined according to the granularity of information aggregation to realize the integration of different information, and the number of output nodes is determined according to the category number (which can be understood as the data number in the category information) of the corresponding e-commerce platform. The number of input nodes and the number of output nodes of the model can be dynamically adjusted according to actual conditions, namely, the structure and parameters of the model are dynamically changed so as to adapt to the maintenance and upgrading requirements of the information of the commodity and the page of the platform.
In one embodiment, a neural network feedback correction algorithm is used to train and optimize parameters of the neural network aggregation model. The 2-system GA algorithm is preferably adopted in the training algorithm, and the feedback learning problem of the adopted discontinuous parameter neural network model is solved.
Specifically, performing training and optimizing may include: adopting a binary GA algorithm, wherein the weight parameter of the neural network aggregation model is taken as 0 or 1, and the hidden node and the output node perform weighted summation on input information to obtain the synthesis of different information elements; and comparing the output value of the neural network aggregation model with the output value of the sample pair, and changing the weight parameter according to the comparison result to obtain the neural network aggregation model suitable for each E-commerce platform.
In step S4, after the training is completed, model operation may be performed to obtain category information of each e-commerce platform, the minimum information element set of the commodity to be aggregated is input according to the neural network aggregation models of different e-commerce platforms, and the category information corresponding to the commodity page requirement of the e-commerce platform is obtained through model calculation. The manual operation in the prior art is omitted for establishing and maintaining the table of category information, the running speed is greatly improved compared with the manual operation, and the accuracy is greatly improved. The minimum information element set of the commodities to be aggregated can be obtained by dividing and summarizing the commodity attributes provided by the commodity manufacturer according to the format of the minimum information element set of the sample commodities, and can be continuously used after being obtained without dividing and summarizing.
Since the information of the goods and the category information of the goods page of the e-commerce platform change, the update of the category information of the goods page and the update of the aggregation model need to be dynamically performed, and the category and the content required by the goods page of the e-commerce platform after the update are uploaded to the e-commerce platform.
Preferably, the multi-platform e-commerce information aggregation method further includes: when the commodity information is changed, namely the elements in the minimum information element set are changed, the updated minimum information element set is input into the neural network aggregation model again according to the updated minimum information element set, updated category information of each e-commerce platform is obtained, and the category information is uploaded to the corresponding e-commerce platform.
Preferably, the multi-platform e-commerce information aggregation method further includes: when the category information of the monitored E-commerce platform sample commodity page changes (namely the aggregation relation between the minimum information element set of the commodity and the page category information changes), carrying out similarity analysis and merging to obtain a new minimum information element set, constructing a new sample pair of the target E-commerce platform, retraining a new neural network aggregation model, inputting the original minimum information element set of the commodity to be aggregated into the updated neural network aggregation model to obtain changed category information, and uploading the changed category information to the target E-commerce platform.
The methods of the embodiments of the present invention are further described below.
As shown in fig. 2, the input of the neural network aggregation model is the minimum information element set of the commodity information, the output is the category information required by the commodity page of the cross-border e-commerce platform, the number of input nodes of the neural network is the number j of the minimum information elements of the commodity information, the number of output nodes is the number i of the category information, the number of hidden nodes and the number of layers are determined by the degree of information aggregation, in the embodiment, the number of the hidden nodes is selected to be k, one layer is selected from the number of layers of the hidden nodes, and the input node p (p belongs to [1,2 … j ] is selected from the input node p (p belongs to [1]) To the hidden node q (q e [1,2 … k ]]) Has a weight coefficient of ωpqFrom hidden node q to output node m (m e [1,2 … n)]) Has a weight coefficient vqmWherein ω ispqV and vqmAll values of (A) are 0 or 1, i.e. omegapq∈[0,1]，νqm∈[0,1]The weighting coefficients of the cross-border e-commerce information neural network aggregation model are both valued between 0 and 1, and therefore the cross-border e-commerce information neural network aggregation model can be called an integer 0-1 neural network. The transfer function of hidden node and output of the neural network is linear addition function, namely the hidden node output yq＝Σxp*ωpq,yqFor the q hidden node output, xpThe input of the p-th node and the output node are similar to the hidden node, the transfer function also adopts a linear addition function, and the purpose of adopting the integer 0-1 neural network and the linear addition function is to describe the category information required by the cross-border e-commerce platform commodity page through different commodity information minimum element combinations. During the neural network training, based on a binary GA algorithm, according to a sample pair consisting of the minimum elements of commodity information and information categories required by the commodity page of the cross-border E-commerce platform, the connection weight parameter omega of the neural network is usedpqV and vqmTraining is carried out to obtain a model capable of reflecting the combination and mapping relation between the minimum elements of the different e-commerce page information and the commodity information.
The neural network training may include an offline training mode. The input of the off-line training mode is a sample pair set formed by a cross-border E-commerce platform commodity page information minimum element information set after similarity analysis and integration, the sample pair comprises a commodity minimum information element set U and page category information Y, if a knapsack and a man are two minimum elements, the corresponding content of the platform name category of the commodity, namely the man knapsack, is a combined mapping result of the two minimum elements; in the category of 'size', three minimum information elements of the backpack are 480mm long, 350mm high and 150mm deep to jointly form specification category information '350 mm 480mm 150 mm' in a platform page, different platforms have different mapping relations, matching and combination of the minimum information to page category contents are manually completed by professionals at present, and the combination relation can be automatically generated through a computer program so as to facilitate enterprises to update related commodity contents and aggregation models.
The set formed by the sample pairs of the category contents and the minimum element set acquired by the same cross-border e-commerce platform is input into the neural network aggregation model in the figure 2, wherein the minimum element set of the commodity information is an input value, the commodity page category information is an output value, and through comparison and judgment, if the model output values of all samples are completely matched with the commodity page category information, the error is zero, which indicates that the model can truly match the commodity description mapping and combination relation of the E-commerce platform, if any item does not conform to the commodity description mapping and combination relation, if the connection weight of the model is not trained yet, GA is still needed for optimization, the iterative optimization training process is continued until the output values of each model on the training sample set are all in accordance with the contents of the commodity page information category, and obtaining the correct mapping and combination relation from the minimum element set of the commodity information to the content of the commodity page information category on the target cross-border e-commerce platform.
Specifically, referring to fig. 3 and 4, the neural network training in the offline training mode includes the following steps a1-a 4.
A1: and randomly generating a chromosome set of a binary GA algorithm, selecting an initial chromosome as an initial weight coefficient value of the neural network, calculating a fitness value, and assigning the initial chromosome to a historical optimal value.
More specifically, let N be ωpqV and vqmIf N is p q + q m, then the GA algorithm generates a set of randomly generated N-length binary codes (e.g., 10 … 01 … 101 … 10 … 1) for the chromosome, and the set of chromosomes is S ═ S [, for example1,S2,…,Sr-1,Sr]R is the number of chromosomes, in this example, r is 20, the weights are assigned to the neural networks of integers 0 to 1 in a one-to-one correspondence manner according to the mode of fig. 7, then the neural network corresponding to the first chromosome is selected, the fitness value is calculated, and the fitness value is assigned to the historical optimal value Sbest。
A2: and copying, mutating and crossing the chromosome group to obtain a chromosome population collection after copying, mutating and crossing, respectively taking the chromosomes in the chromosome population collection as weight coefficient values of the neural network, and calculating each fitness value.
More specifically, standard 2-ary GA selection, crossover and mutation operations are performed on all 20 chromosomes, and the parameters are mapped to an integer 0-1 neural network with reference to fig. 6.
i. Randomly selecting 1 chromosome from S and copying the chromosome according to the selection opportunity determined by the selection probability p (si), performing N times in total, and forming a population S1 by the copied N chromosomes;
ii. Randomly determining c chromosomes from S1 according to the determined number c of chromosomes participating in crossing, pairing to perform crossing operation, and replacing the original chromosomes with the generated chromosomes to form a population S2;
iii, randomly determining m chromosomes from S2 according to the determined mutation times m, performing mutation operations respectively, and replacing the original chromosomes with the generated new chromosomes to form a population S3;
the fitness value is calculated on the sample set by a newly generated integer 0-1 neural network resulting from mapping the collection consisting of the S1, S2, and S3 subsets.
A3: and selecting the chromosome with the minimum fitness in the fitness values, if the minimum fitness is smaller than the historical optimum, assigning the corresponding chromosome to the historical optimum, otherwise, keeping the historical optimum unchanged, screening a new chromosome set generated by the variation based on a fitness function by adopting a roulette mode to obtain a next generation chromosome set, and waiting for iteration.
More specifically, the chromosomes with the minimum fitness are selected according to the fitness function from small to large, and if the fitness is higher than the historical optimal value SbestSmaller, it is better than the historical optimum SbestAssign it to the historical optimum value SbestOtherwise, keeping the historical optimum value SbestAnd (4) unchanged, then, screening a new chromosome set generated by the mutation by adopting a roulette mode based on a fitness function to obtain a chromosome set of the next generation of chromosomes.
A4: and (3) judging whether the model outputs of all the sample pairs are consistent with the sample outputs when the chromosomes with the historical optimal values are applied to the neural network aggregation model, if so, finishing iterative learning and outputting the optimal chromosomes, and otherwise, selecting a next generation chromosome set and returning to the step A2.
In one embodiment, the neural network training further comprises an online operation mode, referring to fig. 3 and 5, the neural network training in the online operation mode comprises the following steps: b1: selecting a weight coefficient and a structure of a corresponding neural network aggregation model according to the target e-commerce platform to generate a corresponding neural network aggregation model; b2: and inputting a minimum information element set, and calling a neural network aggregation model to calculate the category information of the commodity page of the target e-commerce platform.
And in the online operation mode, the minimum information element set of the commodity and the category information of the target platform are input into the aggregation model on the basis of the neural network aggregation model obtained in the offline training mode, and the target cross-border E-commerce platform commodity page category information is obtained through calculation. Through the mode, the system can dynamically generate the commodity page category information of the target cross-border E-commerce platform at any time, and compared with the traditional mode of manually arranging and integrating information, the system can automatically generate the commodity page information category content through a computer, so that the working efficiency is greatly improved, and the error rate is reduced.
In practical application, with the difference of suppliers and raw materials, sometimes the commodity information needs to be changed, and in addition, the page description category of the cross-border e-commerce platform changes with the platform update of the cross-border e-commerce service provider, and the manual work cannot track the changes and update the model and the information in real time, so that on the basis of obtaining the neural network aggregation model, the page category content and category are dynamically optimized and adjusted by dynamically detecting and monitoring the change of the commodity information and the cross-border e-commerce platform sample commodity page category, as shown in fig. 6, the specific flow is as follows:
(1) starting a monitoring period, and dynamically acquiring a system commodity minimum information set stored locally and category information of a cross-border e-commerce platform sample commodity page;
(2) judging the change condition of the minimum information set of the commodity stored locally, and realizing the change condition in a comparison mode, if the change condition does not occur, turning to the step (3), and if the change condition occurs, performing the following processing:
i. inputting the updated commodity minimum element set into the neural network aggregation model again for calculation to obtain updated page category information;
ii. Outputting the updated page category information to a target e-commerce platform for updating;
iii, judging whether all the platforms are updated, if so, entering the step (3), and if not, returning to the step i for circular execution until the pages of the commodity on all the E-commerce platforms are updated;
(3) judging the category information change condition of the monitored E-commerce platform sample commodity page, if the category information change condition does not change, returning to the step (1) to enter the next monitoring period, and if the category information change condition changes, performing the following processing:
i. acquiring commodity page information of a change platform to construct a new learning sample;
ii. Retraining the target platform neural network aggregation model;
and iii, inputting the original minimum information element set of the to-be-aggregated commodity into the updated neural network aggregation model for calculation, obtaining commodity page category information after the aggregation relation is changed, uploading the commodity page category information to the target cross-border e-commerce platform, and returning to the step (1) to enter the next monitoring period.
The invention also provides a multi-platform e-commerce information aggregation system, which comprises:
the platform information acquisition module is used for acquiring page category information of each E-commerce platform about the sample commodity;
the information similarity analysis integration module is used for carrying out similarity analysis on all the obtained category information of the same sample commodity on different cross-border e-commerce platform pages, analyzing the minimum information elements of the sample commodity based on the similarity analysis result, combining the minimum information elements to obtain the minimum information element set of the sample commodity, and forming a comprehensive sample set by the minimum information element sets of the sample commodities;
the aggregation model training module is used for carrying out neural network training with the weight coefficient value of 0 or 1 by taking the minimum information element set of the sample commodity and the category information acquired from one e-commerce platform as a training sample pair aiming at each e-commerce platform to obtain a neural network aggregation model of each e-commerce platform;
and the aggregation information calculation module is used for inputting the minimum information element set of the commodity to be aggregated according to the neural network aggregation models of different e-commerce platforms, and obtaining category information corresponding to the commodity page requirements of the e-commerce platforms through model calculation.
According to an embodiment of the present invention, further comprising:
the aggregation model dynamic updating module is used for inputting the updated minimum information element set into the neural network aggregation model again when the commodity information is changed, obtaining the updated category information of each e-commerce platform and uploading the category information to the corresponding e-commerce platform; and/or when the category information of the monitored E-commerce platform sample commodity page changes (namely, when the aggregation relation between the minimum information element set of the commodity and the page category information changes), performing similarity analysis and merging to obtain a new minimum information element set, constructing a new sample pair of the target E-commerce platform, retraining a new neural network aggregation model, inputting the original minimum information element set of the commodity to be aggregated into the updated neural network aggregation model to obtain the changed category information, and uploading the changed category information to the target E-commerce platform.
For the specific content of the multi-platform e-commerce information aggregation system of the present invention, reference may be made to the description of the multi-platform e-commerce information aggregation method, which is not repeated herein.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to limit the scope of the claims, and those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention.
Claims (9)
1. A multi-platform e-commerce information aggregation method is characterized by comprising the following steps:
acquiring page category information of each E-commerce platform about sample commodities;
carrying out similarity analysis on all the obtained category information of the same sample commodity on different E-commerce platform pages, analyzing the minimum information element of the sample commodity based on the similarity analysis result, combining the minimum information elements to obtain the minimum information element set of the sample commodity, and forming a comprehensive sample set by the minimum information element sets of a plurality of sample commodities;
aiming at each e-commerce platform, taking a minimum information element set of a sample commodity and category information acquired from one e-commerce platform as a training sample pair, and carrying out neural network training with the weight coefficient value of 0 or 1 to obtain a neural network aggregation model of each e-commerce platform;
inputting a minimum information element set of the commodity to be aggregated according to the neural network aggregation models of different e-commerce platforms, and obtaining category information corresponding to the commodity page requirements of the e-commerce platforms through model calculation;
when the commodity information is changed, the commodity information is input into the neural network aggregation model again according to the updated minimum information element set, updated category information of each e-commerce platform is obtained, and the category information is uploaded to the corresponding e-commerce platform.
2. The multi-platform e-commerce information aggregation method of claim 1, further comprising:
when the category information of the monitored E-commerce platform sample commodity page changes, carrying out similarity analysis and merging to obtain a new minimum information element set, constructing a new sample pair of a target E-commerce platform, retraining a new neural network aggregation model, inputting the original minimum information element set of the commodity to be aggregated into the updated neural network aggregation model to obtain the changed category information, and uploading the changed category information to the target E-commerce platform.
3. The multi-platform e-commerce information aggregation method of any one of claims 1 to 2, wherein the category information of each e-commerce platform about the goods is obtained by directionally crawling related goods web page resources of each e-commerce platform based on the target data mode by using a focused crawler.
4. The method as claimed in claim 1, wherein each neural network aggregation model is a mapping model of a minimum information element set to category information of an e-commerce platform, and comprises an input layer, a plurality of or single hidden layers, and an output layer, the number of input nodes is determined according to the number of elements of the minimum unit information set, the number of hidden layers and the number of nodes is determined according to the granularity of information aggregation, and the number of output nodes is determined according to the number of categories of the corresponding e-commerce platform.
5. The multi-platform e-commerce information aggregation method of claim 1 or 4, wherein parameters of the neural network aggregation model are trained and optimized using a neural network feedback correction algorithm.
6. The method of claim 5, wherein the training and optimizing comprises:
adopting a binary GA algorithm, wherein the weight parameter of the neural network aggregation model is taken as 0 or 1, and the hidden node and the output node perform weighted summation on input information to obtain the synthesis of different information elements;
and comparing the output value of the neural network aggregation model with the output value of the sample pair, and changing the weight parameter according to the comparison result to obtain the neural network aggregation model suitable for each E-commerce platform.
7. The method for multi-platform e-commerce information aggregation according to any one of claims 1-2 and 4, wherein the neural network training comprises an offline training mode, comprising the steps of:
a1: randomly generating a chromosome set of a binary GA algorithm, selecting an initial chromosome as an initial weight coefficient value of a neural network, calculating a fitness value, and assigning the initial chromosome to a historical optimal value;
a2: copying, mutating and crossing chromosome groups to obtain a chromosome population collection after copying, mutating and crossing, respectively taking chromosomes in the chromosome population collection as weight coefficient values of a neural network, and calculating each fitness value;
a3: selecting the chromosome with the minimum fitness in the fitness values, if the minimum fitness is smaller than the historical optimum, assigning the corresponding chromosome to the historical optimum, otherwise keeping the historical optimum unchanged, screening a new chromosome set generated by variation based on a fitness function by adopting a roulette mode to obtain a next generation chromosome set, and waiting for iteration;
a4: and (3) judging whether the model outputs of all the sample pairs are consistent with the sample outputs when the chromosomes with the historical optimal values are applied to the neural network aggregation model, if so, finishing iterative learning and outputting the optimal chromosomes, and otherwise, selecting a next generation chromosome set and returning to the step A2.
8. The method of multi-platform e-commerce information aggregation of claim 7, wherein neural network training further comprises an online mode of operation comprising the steps of:
b1: selecting a weight coefficient and a structure of a corresponding neural network aggregation model according to the target e-commerce platform to generate a corresponding neural network aggregation model;
b2: and inputting a minimum information element set, and calling a neural network aggregation model to calculate the category information of the commodity page of the target e-commerce platform.
9. A multi-platform e-commerce information aggregation system, comprising:
the platform information acquisition module is used for acquiring page category information of each E-commerce platform about the sample commodity;
the information similarity analysis integration module is used for carrying out similarity analysis on all the obtained category information of the same sample commodity on different cross-border e-commerce platform pages, analyzing the minimum information elements of the sample commodity based on the similarity analysis result, combining the minimum information elements to obtain the minimum information element set of the sample commodity, and forming a comprehensive sample set by the minimum information element sets of the sample commodities;
the aggregation model training module is used for carrying out neural network training with the weight coefficient value of 0 or 1 by taking the minimum information element set of the sample commodity and the category information acquired from one e-commerce platform as a training sample pair aiming at each e-commerce platform to obtain a neural network aggregation model of each e-commerce platform;
the aggregation information calculation module is used for inputting the minimum information element set of the commodities to be aggregated according to the neural network aggregation models of different e-commerce platforms, and obtaining category information corresponding to the commodity page requirements of the e-commerce platforms through model calculation;
the aggregation model dynamic updating module is used for inputting the updated minimum information element set into the neural network aggregation model again when the commodity information is changed, obtaining the updated category information of each e-commerce platform and uploading the category information to the corresponding e-commerce platform; and/or when the category information of the monitored E-commerce platform sample commodity page changes, carrying out similarity analysis and merging to obtain a new minimum information element set, constructing a new sample pair of the target E-commerce platform, retraining a new neural network aggregation model, inputting the original minimum information element set of the commodity to be aggregated into the updated neural network aggregation model to obtain the changed category information, and uploading the changed category information to the target E-commerce platform.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610957162.3A CN106600356B (en) | 2016-10-27 | 2016-10-27 | Multi-platform e-commerce information aggregation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610957162.3A CN106600356B (en) | 2016-10-27 | 2016-10-27 | Multi-platform e-commerce information aggregation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106600356A CN106600356A (en) | 2017-04-26 |
| CN106600356B true CN106600356B (en) | 2019-12-24 |
Family
ID=58590357
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610957162.3A Active CN106600356B (en) | 2016-10-27 | 2016-10-27 | Multi-platform e-commerce information aggregation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106600356B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108335164A (en) * | 2017-01-20 | 2018-07-27 | 阿里巴巴集团控股有限公司 | A kind of method, apparatus and electronic equipment for realizing shopping at network |
| CN109389450B (en) * | 2017-08-08 | 2022-04-19 | 阿里巴巴集团控股有限公司 | Method and system for associating attribute information and generating data set |
| CN107464170A (en) * | 2017-08-15 | 2017-12-12 | 合肥蓝胖子科技有限公司 | Mobile terminal electric business service system based on internet |
| CN107483582A (en) * | 2017-08-16 | 2017-12-15 | 广州环球梦电子商务股份有限公司 | It is a kind of to borrow the method for pushing and system for selling platform commodity |
| CN109947782A (en) * | 2017-11-03 | 2019-06-28 | 中国移动通信有限公司研究院 | A kind of update method of big data real-time application system, apparatus and system |
| CN108257031B (en) * | 2017-11-09 | 2022-04-12 | 中国平安人寿保险股份有限公司 | Medical insurance product issuing method and device and storage medium |
| CN109224453B (en) * | 2018-06-28 | 2021-07-09 | 平安科技（深圳）有限公司 | Game supervision method and system, computer equipment and computer-readable storage medium |
| CN108665358A (en) * | 2018-08-03 | 2018-10-16 | 北京引领海逛科技有限公司 | The method and apparatus of the multiple platforms of product information Rapid matching |
| CN112085568B (en) * | 2020-09-09 | 2023-10-27 | 百度在线网络技术（北京）有限公司 | Commodity and rich media aggregation display method and equipment, electronic equipment and medium |
| CN112231311A (en) * | 2020-10-23 | 2021-01-15 | 食亨(上海)科技服务有限公司 | Method, system and computer readable medium for integrating cross-platform commodity data |
| CN112685320B (en) * | 2021-01-09 | 2023-04-25 | 北京信息科技大学 | Software defect repairing method and device based on multiple candidate programs |
| WO2022269336A1 (en) * | 2021-06-23 | 2022-12-29 | Ddb Mudra Max Private Limited | System and method for intelligent cross platform marketing resource optimization |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101118610A (en) * | 2007-09-10 | 2008-02-06 | 东北大学 | Sparseness data process modeling approach |
| CN103914478A (en) * | 2013-01-06 | 2014-07-09 | 阿里巴巴集团控股有限公司 | Webpage training method and system and webpage prediction method and system |
| CN104504086A (en) * | 2014-12-25 | 2015-04-08 | 北京国双科技有限公司 | Clustering method and device for webpage |
| CN104765728A (en) * | 2014-01-08 | 2015-07-08 | 富士通株式会社 | Method and device for training neural network and method for determining sparse feature vector |
| CN105005917A (en) * | 2015-07-07 | 2015-10-28 | 上海晶赞科技发展有限公司 | Universal method for correlating single items of different e-commerce websites |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080306830A1 (en) * | 2007-06-07 | 2008-12-11 | Cliquality, Llc | System for rating quality of online visitors |
-
2016
- 2016-10-27 CN CN201610957162.3A patent/CN106600356B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101118610A (en) * | 2007-09-10 | 2008-02-06 | 东北大学 | Sparseness data process modeling approach |
| CN103914478A (en) * | 2013-01-06 | 2014-07-09 | 阿里巴巴集团控股有限公司 | Webpage training method and system and webpage prediction method and system |
| CN104765728A (en) * | 2014-01-08 | 2015-07-08 | 富士通株式会社 | Method and device for training neural network and method for determining sparse feature vector |
| CN104504086A (en) * | 2014-12-25 | 2015-04-08 | 北京国双科技有限公司 | Clustering method and device for webpage |
| CN105005917A (en) * | 2015-07-07 | 2015-10-28 | 上海晶赞科技发展有限公司 | Universal method for correlating single items of different e-commerce websites |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106600356A (en) | 2017-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106600356B (en) | Multi-platform e-commerce information aggregation method and system | |
| US11327935B2 (en) | Intelligent data quality | |
| US7574379B2 (en) | Method and system of using artifacts to identify elements of a component business model | |
| Shi et al. | Analytics for cross‐border e‐commerce: inventory risk management of an online fashion retailer | |
| US20180330282A1 (en) | Optimizing Data-to-Learning-to-Action | |
| Zhen et al. | Capacitated closed-loop supply chain network design under uncertainty | |
| CN113674013B (en) | Advertisement bidding adjustment method and system based on merchant custom rules | |
| CN103336791A (en) | Hadoop-based fast rough set attribute reduction method | |
| Chen et al. | Development and application of big data platform for garlic industry chain | |
| Zhou et al. | Capacitated disassembly scheduling with random demand and operation time | |
| CN108628882A (en) | Method and system for prejudging problem | |
| Ming-Te et al. | Using data mining technique to perform the performance assessment of lean service | |
| Shang et al. | An intelligent risk assessment system for distributed construction teams | |
| CN116340726A (en) | Energy economy big data cleaning method, system, equipment and storage medium | |
| Shamim et al. | Cloud Computing and AI in Analysis of Worksite | |
| Kolková et al. | Demand forecasting: AI-based, statistical and hybrid models vs practicebased models-the case of SMEs and large enterprises | |
| Zhou et al. | [Retracted] Research on Intelligent Solution of Service Industry Supply Chain Network Optimization Based on Genetic Algorithm | |
| Qu et al. | Parallel genetic algorithm model based on AHP and neural networks for enterprise comprehensive business | |
| CN113722564A (en) | Visualization method and device for energy and material supply chain based on space map convolution | |
| Zeisel | A framework and use cases for big data in procurement | |
| CN110597796B (en) | Big data real-time modeling method and system based on full life cycle | |
| Xiong et al. | WebATP: a Web-based flexible available-to-promise computation system | |
| Ocampo et al. | A sustainable manufacturing strategy from different strategic responses under uncertainty | |
| Sulova | Big data processing in the logistics industry | |
| Ocampo-Martinez et al. | Multi-Objective-Based Tuning of Economic Model Predictive Control of Drinking Water Transport Networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20201010 Address after: Room 301-303, 305-308, building 2, No. 567, Jiangling Road, Xixing street, Binjiang District, Hangzhou City, Zhejiang Province Patentee after: HANGZHOU PINGPONG INTELLIGENT TECHNICAL Co.,Ltd. Address before: Hangzhou City, Zhejiang province 310000 Binjiang District West Street No. 567 Jiangling Road, building 2, floor 5 Patentee before: HANGZHOU WANGDAO TECHNOLOGY Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| CP01 | Change in the name or title of a patent holder |
Address after: Room 301-303, 305-308, building 2, No. 567, Jiangling Road, Xixing street, Binjiang District, Hangzhou City, Zhejiang Province Patentee after: Hangzhou ping pong Intelligent Technology Co., Ltd Address before: Room 301-303, 305-308, building 2, No. 567, Jiangling Road, Xixing street, Binjiang District, Hangzhou City, Zhejiang Province Patentee before: HANGZHOU PINGPONG INTELLIGENT TECHNICAL Co.,Ltd. |
|
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 310000 rooms 301-303, 305-308, building 2, No.567 Jiangling Road, Xixing street, Binjiang District, Hangzhou City, Zhejiang Province Patentee after: Hangzhou ping pong Intelligent Technology Co.,Ltd. Address before: 310000 rooms 301-303, 305-308, building 2, No.567 Jiangling Road, Xixing street, Binjiang District, Hangzhou City, Zhejiang Province Patentee before: Hangzhou ping pong Intelligent Technology Co., Ltd |
|
| CP01 | Change in the name or title of a patent holder |