Information generation method and device

Abstract

The invention provides an information generation method and device, and the method comprises the steps: generating the market loss information of a first product based on a first product feature and first product value information obtained by a first generator in a generative adversarial network; generating second product value information based on market loss information and second product features acquired by a second generator in the generative adversarial network; wherein the market increment information, represented by the second product value information and the second product feature, of the second product is matched with the market loss information of the first product, and the second product is a new product having market correlation with the first product.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to an information generation method and device. Background Art

[0002] When new products are introduced to the market, they often affect the market utility of old products related to their market. For example, a new generation of products in the same series has a substitution relationship with the previous generation of products. At the end of the life cycle of the previous generation of products, the previous generation of products will gradually withdraw from the market and the new generation of products will replace the previous generation of products.

[0003] Generally, when a new product is launched, the merchant expects that the "market utility" gained by the new product is equal to the "market utility" lost by the old product, so that the new and old products can reach a balance in their target markets and smoothly replace the old with the new. However, product price is a key factor affecting the market utility of the product, so the price of the new product needs to be reasonably customized. Summary of the invention

[0004] One aspect of the present disclosure provides an information generation method, including: generating market loss information of a first product based on first product features and first product value information obtained by a first generator in a generative adversarial network; generating second product value information based on market loss information and second product features obtained by a second generator in the generative adversarial network; wherein market incremental information of the second product represented by the second product value information and the second product features matches the market loss information of the first product, and the second product is a new product having market relevance to the first product.

[0005] According to an embodiment of the present disclosure, the method also includes: training a generative adversarial network so that the market loss information of the first product output by the first generator in the generative adversarial network matches the market increment information of the second product.

[0006] According to an embodiment of the present disclosure, training is performed in a generative adversarial network, including: obtaining multiple groups of first sample product data and second sample product data, wherein the first sample product data includes first sample product characteristics and first sample product value information, and the second sample product data includes second sample product characteristics and second sample product value information; in response to the first sample product data being input to the first generator, outputting market loss information of the first sample product, and in response to the second sample product data being input to the first generator, outputting market increment information of the second sample product.

[0007] According to an embodiment of the present disclosure, training the generative adversarial network also includes: adjusting parameters of the discriminator based on market loss information of a first sample product and market increment information of a second sample product obtained by a discriminator in the generative adversarial network to increase the difference between the market loss information and the market increment information discriminated by the discriminator; wherein the difference between the market loss information and the market increment information represents the difference between the market loss information and the market increment information; and adjusting parameters of the first generator to reduce the difference between the market loss information of the first sample product generated by the generator and the market increment information of the second sample product.

[0008] According to an embodiment of the present disclosure, multiple sets of first product sample data and second product sample data are used to adjust the parameters of the first generator and the discriminator multiple times; wherein, when adjusting the parameters of the first generator, the parameters of the discriminator are kept unchanged; and when adjusting the parameters of the discriminator, the parameters of the first generator are kept unchanged.

[0009] According to an embodiment of the present disclosure, training is performed on a generative adversarial network, including at least one of the following: taking the market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator as a first input, the first sample product feature of the first sample product as a second input, and the first sample product value information as an output, to train the second generator; taking the market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator as a first input, the second sample product feature of the second sample product as a second input, and the second sample product value information as an output, to train the second generator.

[0010] According to an embodiment of the present disclosure, market loss information of a first product is generated, including: based on two or more first product features and first product value information of the first products, generating two or more market loss information corresponding to the first product; generating second product value information includes: based on two or more market loss information corresponding to the first product, obtaining two or more second product value information corresponding to the second product.

[0011] According to an embodiment of the present disclosure, the degree of difference is obtained based on a first expression and a second expression, the first expression is calculated based on market loss information, and the second expression is calculated based on market incremental information; wherein, the first expression is positively correlated with the market loss information, and the second expression is negatively correlated with the market incremental information.

[0012] According to an embodiment of the present disclosure, the product features include at least one of the following: product hardware data, product function data, product appearance data, product time data, and product region data.

[0013] Another aspect of the present disclosure also provides an information generating device, including: a first generating module, which generates market loss information of a first product based on first product features and first product value information obtained by a first generator in a generative adversarial network; a second generating module, which generates second product value information based on market loss information and second product features obtained by a second generator in the generative adversarial network; wherein the market incremental information of the second product represented by the second product value information and the second product features matches the market loss information of the first product, and the second product is a new product having market relevance to the first product. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] For a more complete understanding of the present disclosure and its advantages, reference will now be made to the following description taken in conjunction with the accompanying drawings, in which:

[0015] Figure 1 A flowchart schematically illustrates an information generation method according to an embodiment of the present disclosure;

[0016] Figure 2 The schematic diagram of the application side of the generative adversarial network according to the embodiment of the present disclosure is schematically shown;

[0017] Figure 3 Schematically illustrates a flow chart of generating adversarial network training according to an embodiment of the present disclosure;

[0018] Figure 4 A schematic diagram of the training side of a generative adversarial network according to an embodiment of the present disclosure is shown;

[0019] Figure 5 A block diagram schematically illustrates an information generating device according to an embodiment of the present disclosure;

[0020] Figure 6 A schematic block diagram shows an example electronic device that can be used to implement the method of the embodiments of the present disclosure. DETAILED DESCRIPTION

[0021] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0022] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0023] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0024] Some block diagrams and / or flow charts are shown in the accompanying drawings. It should be understood that some blocks or combinations thereof in the block diagrams and / or flow charts may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that these instructions, when executed by the processor, may create a device for implementing the functions / operations described in these block diagrams and / or flow charts.

[0025] Therefore, the technology of the present disclosure can be implemented in the form of hardware and / or software (including firmware, microcode, etc.). In addition, the technology of the present disclosure can take the form of a computer program product on a computer-readable medium storing instructions, which can be used by an instruction execution system or in combination with an instruction execution system. In the context of the present disclosure, a computer-readable medium can be any medium that can contain, store, transmit, propagate, or transmit instructions. For example, a computer-readable medium can include, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, device, or propagation medium. Specific examples of computer-readable media include: magnetic storage devices, such as magnetic tape or hard disk (HDD); optical storage devices, such as compact disk (CD-ROM); memory, such as random access memory (RAM) or flash memory; and / or wired / wireless communication links.

[0026] This disclosure first takes mobile phone products as an example to introduce the application scenarios of this disclosure. Table 1 shows the product features of new and old mobile phone products in the embodiments of this disclosure, specifically including product hardware data, product function data, product appearance data, product time data and product region data:

[0027]

[0028] Table 1

[0029] It can be seen from Table 1 of the embodiments of the present disclosure that the new mobile phone products are different from the old mobile phone products in various product features. Specifically, the new mobile phone products are improved in hardware, functions and appearance compared with the old mobile phone products. Then when the new mobile phone products are launched into the market, they will inevitably impact the market share of the old mobile phone products, that is, the market utility of the old mobile phone products will be lost. It should be emphasized that the product market utility mentioned in the present disclosure refers to data information that can reflect the economic value of the product. The data information can be a numerical value, such as a simple sales volume multiplied by the unit price (sales), or for example, with the addition of product features, that is, the sales volume, unit price, and product features are calculated to obtain the numerical value; the data information can also be a set of vector expressions, which covers the above-mentioned sales volume, unit price and product features, and then expresses the market utility of the product. At the same time, users can formulate corresponding market utility expressions for products based on the specificity of the products. For example, for mobile phone products, users can increase the weight of hardware, functions, appearance and other data in the market utility expression, because these features can have a greater impact on the sales of mobile phone products. For example, for daily products such as toilet paper, data such as its appearance is not a key factor affecting its sales. Users can reduce the data weight of the appearance features and increase the data weight of the functional features of toilet paper (water absorbency).

[0030] Let's take the example shown in Table 1 of the present disclosure as an example. Before the new mobile phone product is launched, the monthly sales volume of the old mobile phone product is 2,000 units, and the unit price of the old mobile phone product is 5,000 yuan / unit. If the market utility is expressed by simple sales volume, the current market utility of the old mobile phone product is 10 million yuan. When the new mobile phone product is launched, it is estimated (the "estimate" here can be inferred based on historical data or calculated based on a data model) that the monthly sales volume of the old mobile phone product will drop to 800 units. Then the market utility lost by the old mobile phone product is (2,000-800) × 5,000 = 6 million yuan (i.e., the market loss information proposed later). In order for the new and old products to have a smooth transition and achieve market equilibrium, merchants expect that the market utility obtained when the new mobile phone product is launched can make up for the market utility lost by the old mobile phone product (6 million yuan). If the new mobile phone product is priced too high or too low, the market utility lost by the old mobile phone product (6 million yuan) will not be achieved. Therefore, it is necessary to formulate a reasonable price for the new product to achieve the above purpose.

[0031] However, as mentioned above, there are many factors that affect the market utility of products. Product physical characteristics, product time characteristics, product geographical characteristics, and price characteristics are all key factors that affect the market utility of products. Users need to objectively formulate product market utility expressions based on these characteristics. Therefore, when the product physical characteristics, product time characteristics, and product geographical characteristics are relatively fixed, how to formulate product prices to meet the market utility balance when the old and new products are replaced is a problem that needs to be solved in the present disclosure.

[0032] The present invention is based on a generative adversarial network (GAN), and uses historical data of old and new products as data samples for training, so that the "old product loss utility" generated by the generative adversarial network based on these sample data is close to the "new product utility", and then the "old product loss utility", old product feature data, and old product price data are used to train a price generation model, so that when the trained price generation model is applied, it can obtain the price of a new generation of products according to the product utility of the current product and the feature data of the current product. At the same time, the price obtained by the price generation model combined with the "new product utility" obtained by the feature data of the new generation of products is close to the "old product loss utility" of the current product.

[0033] Figure 1 The flowchart of the information generation method according to the embodiment of the present disclosure is schematically shown. Figure 2 The schematic diagram of the application side of the generative adversarial network according to the embodiment of the present disclosure is shown below. Figure 1 and Figure 2 The information generating method of the embodiment of the present disclosure is introduced.

[0034] Combination Figure 1 and Figure 2 The information generation method shown includes the following operations:

[0035] S101, generating market loss information of the first product based on the first product features and the first product value information obtained by the first generator 101 in the generative adversarial network.

[0036] Under the premise that the training of the generative adversarial network is completed, the present disclosure inputs the product features and product value information of the old product (first product) into the first generator 101 of the generative adversarial network, and the first generator 101 generates the market loss information of the old product based on the input information. As mentioned above, users can use different product features to express the market loss information of the product according to the specificity of the product (of course, the same product features need to be used for training on the training side). Taking the mobile phone product in the previous article as an example, the first generator 101 can convert the utility values ​​of each product feature of the old product, as shown in Table 2:

[0037]

[0038] Table 2

[0039] Similarly, the first generator 101 can convert the value information of the old product into a utility value to obtain a value utility value. The first generator 101 calculates the market loss information of the old product ("old product loss utility") based on the product's characteristic total utility value and value utility value. For example, the first generator 101 calculates based on the price of the old mobile phone product of 5,000 yuan and obtains a value utility value of 1,000. Combined with the characteristic total utility of 3,590, the "old product loss utility" is 4,590.

[0040] S102, based on the market loss information and second product characteristics obtained by the second generator 103 in the generative adversarial network, generate second product value information; wherein the market increment information of the second product represented by the second product value information and the second product characteristics matches the market loss information of the first product, and the second product is a new product having market relevance to the first product.

[0041] Figure 2 The second generator 103 shown in is the price generation model mentioned above, and the second generator 103 is also trained in the generative adversarial network. The market loss information ("old product loss utility") and the product features of the new product (second product) are input into the second generator 103, and the trained second generator 103 obtains the product value information of the new product based on the input information. Taking the mobile phone product mentioned above as an example, the second generator 103 performs utility value conversion based on the product features of the new mobile phone product, as shown in Table 3:

[0042]

[0043] Table 3

[0044] It can be seen that the second generator 103 obtains the utility value that the new mobile phone product needs to satisfy based on the "lost utility of the old product" (4590) and the total utility value of the product characteristics of the new mobile phone product (2510), which is 4590-2510=2080. That is, the value utility value of the new mobile phone product needs to reach 2080. That is, the second generator 103 calculates the price of the new mobile phone product as 5700 yuan based on the value utility value of the new mobile phone product (2080). The calculated price combined with its product characteristics can make the total utility value (market incremental information) of the new product consistent with the "lost utility of the old product", thereby achieving the purpose of utility balance when the old and new products are replaced.

[0045] In addition, the disclosed embodiment can generate multiple corresponding market loss information based on two or more old product data, and then obtain multiple new product prices corresponding to the new product based on the market loss information of the multiple old products. For example, for the upcoming fourth-generation mobile phone, the relevant data of the first-generation mobile phone, the second-generation mobile phone, and the third-generation mobile phone can be used to generate three prices corresponding to the fourth-generation mobile phone for comprehensive reference by users.

[0046] It should be pointed out that the data provided in the above embodiment is only used to facilitate the explanation of this embodiment, and the corresponding product utility is also simplified for the convenience of understanding. In fact, as mentioned above, the utility expression of the product can be a numerical value or a set of vector expressions, which covers the above sales volume, unit price and product characteristics, and then expresses the market utility of the product. The actual calculation process in the network is also calculated based on the network parameters of the trained neural network. The specific calculation method and network parameter setting are obtained by training based on sample data, which is well known in the art and will not be introduced in depth in this disclosure.

[0047] The above has been combined Figure 1 and Figure 2 This article introduces how the Generative Adversarial Network obtains the value information of new products on the application side. The following will introduce how to train the Generative Adversarial Network based on training samples on the training side.

[0048] Figure 3 The flowchart of generating adversarial network training according to an embodiment of the present disclosure is schematically shown. Figure 4 The schematic diagram of the training side of the generative adversarial network according to the embodiment of the present disclosure is shown below. Figure 3 and Figure 4 The information generating method of the embodiment of the present disclosure is introduced.

[0049] The training objectives of the generative adversarial network training in the disclosed embodiment are two: one is that after the relevant data of the new and old products are input into the trained first generator 101, the output market loss information (market loss utility) of the first product (old product) and the market increment information (new product market utility) of the second product (new product) are close; the other is that after the market loss information (market loss utility) and the relevant data of the new product are input into the trained second generator 103, the output price of the new product satisfies the utility balance of the new and old products mentioned above.

[0050] Combination Figure 3 and Figure 4 The training of the generative adversarial network shown here consists of the following repeated operations:

[0051] S301, obtaining multiple sets of first sample product data and second sample product data, wherein the first sample product data includes first sample product characteristics and first sample product value information, and the second sample product data includes second sample product characteristics and second sample product value information.

[0052] It should be mentioned that the first sample product data and the second sample product data in the embodiment of the present disclosure are historical data, and their corresponding product features and the value information of new and old products are real data. The embodiment of the present disclosure uses these historical real data to train the generative adversarial network, thereby achieving the two purposes mentioned above.

[0053] S302, in response to the first sample product data being input to the first generator 101, outputting the market loss information of the first sample product, and in response to the second sample product data being input to the first generator 101, outputting the market increment information of the second sample product. Specifically, it can be expressed by the following expression:

[0054]

[0055]

[0056] Among them, is the market loss information of the first sample product, is the first sample product feature, is the product value information of the second sample, is the market increment information of the second sample product, is the first sample product feature, is the second sample product value information, and G is the first generator 101.

[0057] During the initialization stage of the generative adversarial network, the first generator 101 will calculate the first sample product data or the second sample product data according to its initialization parameters, and output the initial market loss information and market increment information. The difference between the two information outputs is likely to be very large, so a differentiated discrimination feedback is required to adjust the parameters of the first generator 101 and recalculate based on the first sample product data or the second sample product data to obtain new market loss information and market increment information.

[0058] S303, based on the market loss information of the first sample product and the market increment information of the second sample product obtained by the discriminator 102 in the generative adversarial network, adjust the parameters of the discriminator 102 to increase the difference between the market loss information and the market increment information discriminated by the discriminator 102; wherein the difference between the market loss information and the market increment information represents the difference between the market loss information and the market increment information.

[0059] The degree of difference proposed in the embodiment of the present disclosure is obtained based on a first expression and a second expression, wherein the first expression is calculated based on the market loss information, and the second expression is calculated based on the market increment information; wherein the first expression is positively correlated with the market loss information, and the second expression is negatively correlated with the market increment information.

[0060] In this operation, the function of the discriminator 102 in the generative adversarial network is to maximize the difference between the market loss information and the market increment information. Specifically, the parameters of the discriminator 102 are adjusted. , so that the difference is maximized:

[0061]

[0062] The goal of the discriminator is to maximize the function , which is used to distinguish the generated market loss information and market growth information , when the first expression Close to 1 and the second expression When it is close to 0, the discriminator 102 performs best. In the embodiment of the present disclosure, when updating the parameters of the discriminator 102 When , we use the gradient ascent method to maximize ,Right now:

[0063]

[0064] To facilitate the role of the discriminator 102, let's take the old and new mobile phones mentioned above as an example. If the value represented by the market loss information is 4590 and the value represented by the market increment information is 4000, then when input into the discriminator 102, its initial difference is 4590-4000=590. By adjusting the parameters of the discriminator 102, the difference between the two is 1.1×4590-1.1×4000=649. Obviously, after updating the parameters, the difference between the two discriminators 102 has changed from 590 to 649. The coefficient 1.1 here can be understood as an adjustment from 1.0 to 1.1, that is, the parameter has changed. Of course, in actual application scenarios, market loss information and market increment information may be multidimensional data, so the difference between the two, that is, the parameter adjustment, is also correspondingly complex.

[0065] S304, adjusting the parameters of the first generator 101 to reduce the difference between the market loss information of the first sample product generated by the generator 101 and the market increment information of the second sample product.

[0066] In this operation, it is necessary to adjust the parameters of the first generator 101 according to the discrimination result of the discriminator 102, so that the difference between the market loss information and the market increment information generated by the first generator 101 is as small as possible. , so that the difference is minimized:

[0067]

[0068] The goal of the first generator 101 is to minimize the function , which makes it impossible for the discriminator 102 to distinguish the generated market loss information and market growth information , the disclosed embodiment updates the parameters of the first generator 101 When , we use gradient descent to minimize :

[0069]

[0070] Taking the previous text as an example, if the discriminator 102 determines that the difference between the market loss information and the market incremental information is 1.1×4590-1.1×4000=649, the first generator 101 is adjusted so that the market loss information generated by the first generator 101 is 4600 and the market incremental information is 4300, then the discriminator 102 again determines that the difference between the two is 1.1×4600-1.1×4300=330, and then further amplifies the difference between 4600 and 4300.

[0071] Therefore, the first generator 101 and the discriminator 102 in the generative adversarial network of the present invention form an "adversary" with each other. The first generator 101 generates market loss information and market incremental information with smaller differences as much as possible, while the discriminator 102 amplifies the differences between the generated market loss information and market incremental information as much as possible, thereby prompting the first generator 101 to generate market loss information and market incremental information with smaller differences, thereby achieving the purpose of training the first generator 101.

[0072] The embodiment of the present disclosure uses multiple sets of first product sample data and second product sample data to perform multiple iterative adjustments when adjusting the parameters of the first generator 101 and the discriminator 102. When adjusting the parameters of the first generator 101, the parameters of the discriminator 102 are kept unchanged; when adjusting the parameters of the discriminator 102, the parameters of the first generator 101 are kept unchanged.

[0073] Recombination Figure 4As shown, in order to achieve that after the market loss information (market loss utility) and relevant data of the new product are input into the trained second generator 103, the output new product price satisfies the utility balance of the new and old products mentioned above, the second generator 103 needs to be trained. The specific training process can be any of the following methods:

[0074] The market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator 101 is used as the first input, the first sample product feature of the first sample product is used as the second input, and the first sample product value information is used as the output, to train the second generator 103;

[0075] The market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator 101 is used as the first input, the second sample product feature of the second sample product is used as the second input, and the second sample product value information is used as the output to train the second generator 103. It should be noted that, compared with the previous training method for the second generator 103, the training sample data used in the input and output of this training method are different, but since the market loss information and market increment information output by the first generator 101 during the training process are close to the same, whether for the first sample product or the second sample product, it can be understood that, under the premise that the market loss information and the market increment information are gradually close to each other, in the later stage, whether the market loss information or the market increment information is used as the first input, and the first sample product feature or the second sample product feature is used as the second input, the training effect on the second generator 103 is consistent, but in the early stage, it may be necessary to use the first sample product feature corresponding to the market loss information as the first input as the first input, and the second sample product feature corresponding to the sampling time increment information as the first input as the second input, so as to realize the value information corresponding to the market loss information or the market increment information.

[0076] It is worth mentioning that since the training data used are consistent, after each training data is input into the generative adversarial network, the first generator 101, the discriminator 102 and the second generator 103 complete a parameter adjustment, that is, the training of the first generator 101, the discriminator 102 and the second generator 103 is synchronized.

[0077] The above introduces the process of training the generative adversarial network according to the embodiment of the present disclosure. After training the first generator 101 and the second generator 103 according to the content of the embodiment of the present disclosure, the following purposes can be achieved: first, after the relevant data of the new and old products are input into the trained first generator 101, the output market loss information (market loss utility) of the first product (old product) and the market increment information (new product market utility) of the second product (new product) are close; second, after the market loss information (market loss utility) and the relevant data of the new product are input into the trained second generator 103, the output price of the new product satisfies the utility balance of the new and old products mentioned above.

[0078] Figure 5 The block diagram schematically shows an information generating device according to an embodiment of the present disclosure.

[0079] like Figure 5 As shown, the information generating device 500 includes a first generating module 510 and a second generating module 520. The information generating device 500 can execute the above reference Figure 1 A method is described to generate value information for a second product.

[0080] Specifically, the first generation module 510 generates market loss information of the first product based on the first product characteristics and first product value information obtained by the first generator 101 in the generative adversarial network; the second generation module 520 generates second product value information based on the market loss information and second product characteristics obtained by the second generator 103 in the generative adversarial network.

[0081] Specifically, the information generating device 500 also includes: a training module, which is used to train the generative adversarial network so that the market loss information of the first product output by the first generator in the generative adversarial network matches the market increment information of the second product.

[0082] Specifically, the training module in the embodiment of the present disclosure trains the generative adversarial network, including: obtaining multiple groups of first sample product data and second sample product data, wherein the first sample product data includes first sample product characteristics and first sample product value information, and the second sample product data includes second sample product characteristics and second sample product value information; in response to the first sample product data being input to the first generator, outputting market loss information of the first sample product, and in response to the second sample product data being input to the first generator, outputting market increment information of the second sample product.

[0083] Specifically, the training module in the embodiment of the present disclosure trains the generative adversarial network, and also includes: adjusting the parameters of the discriminator based on the market loss information of the first sample product and the market increment information of the second sample product obtained by the discriminator in the generative adversarial network to increase the difference between the market loss information and the market increment information discriminated by the discriminator; wherein the difference between the market loss information and the market increment information represents the difference between the market loss information and the market increment information; adjusting the parameters of the first generator to reduce the difference between the market loss information of the first sample product generated by the generator and the market increment information of the second sample product.

[0084] Specifically, when the training module in the embodiment of the present disclosure trains the generative adversarial network, the parameters of the first generator and the discriminator are adjusted multiple times using multiple groups of first product sample data and second product sample data; wherein, when adjusting the parameters of the first generator, the parameters of the discriminator are kept unchanged; when adjusting the parameters of the discriminator, the parameters of the first generator are kept unchanged.

[0085] Specifically, the training module in the embodiment of the present disclosure trains the generative adversarial network, including at least one of the following: taking the market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator as the first input, the first sample product feature of the first sample product as the second input, and the first sample product value information as the output, to train the second generator; taking the market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator as the first input, the second sample product feature of the second sample product as the second input, and the second sample product value information as the output, to train the second generator.

[0086] Specifically, when the training module in the embodiment of the present disclosure trains the generative adversarial network, market loss information of the first product is generated, including: generating two or more market loss information corresponding to the first product based on two or more first product features and first product value information of the first product; generating second product value information, including: obtaining two or more second product value information corresponding to the second product based on two or more market loss information corresponding to the first product.

[0087] Specifically, when the training module in the embodiment of the present disclosure trains the generative adversarial network, the difference is obtained based on the first expression and the second expression, the first expression is calculated based on the market loss information, and the second expression is calculated based on the market incremental information; wherein the first expression is positively correlated with the market loss information, and the second expression is negatively correlated with the market incremental information.

[0088] According to an embodiment of the present disclosure, the product features include at least one of the following: product hardware data, product function data, product appearance data, product time data, and product region data.

[0089] It is understandable that the first generation module 510 and the second generation module 520 included in the information generation device 500 can be combined in one module, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present application, at least one of the first generation module 510 and the second generation module 520 included in the information generation device 500 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented in hardware or firmware in any other reasonable way of integrating or packaging the circuit, or in a proper combination of software, hardware and firmware. Alternatively, at least one of the first generation module 510 and the second generation module 520 included in the information generation device 500 can be at least partially implemented as a computer program module, and when the program is run by a computer, the function of the corresponding module can be executed.

[0090] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement the method of an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0091] like Figure 6 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device can also be stored. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0092] Multiple components in the electronic device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a disk, an optical disk, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0093] The computing unit 601 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as the application running method. For example, in some embodiments, the application running method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the application running method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to execute the application running method in any other appropriate manner (e.g., by means of firmware).

[0094] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0095] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0096] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0097] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0098] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for generating information, comprising: Generate market loss information of the first product based on the first product feature and the first product value information obtained by the first generator in the generative adversarial network; Generate second product value information based on the market loss information and second product features obtained by the second generator in the generative adversarial network; The market increment information of the second product represented by the second product value information and the second product characteristics matches the market loss information of the first product, and the second product is a new product having market relevance to the first product.

2. The method according to claim 1, further comprising: The generative adversarial network is trained so that the market loss information of the first product output by the first generator in the generative adversarial network matches the market increment information of the second product.

3. The method according to claim 2, wherein: The training of the generative adversarial network comprises: Acquire multiple sets of first sample product data and second sample product data, wherein the first sample product data includes first sample product characteristics and first sample product value information, and the second sample product data includes second sample product characteristics and second sample product value information; In response to the first sample product data being input to the first generator, market loss information of the first sample product is output; in response to the second sample product data being input to the first generator, market increment information of the second sample product is output.

4. The method according to claim 3, wherein the training of the generative adversarial network further comprises: Based on the market loss information of the first sample product and the market increment information of the second sample product obtained by the discriminator in the generative adversarial network, the parameters of the discriminator are adjusted to increase the difference between the market loss information and the market increment information discriminated by the discriminator; wherein the difference between the market loss information and the market increment information represents the difference between the market loss information and the market increment information; The parameters of the first generator are adjusted to reduce the difference between the market loss information of the first sample product and the market increment information of the second sample product generated by the generator.

5. The method according to claim 4, using multiple sets of first product sample data and second product sample data to adjust the parameters of the first generator and the discriminator multiple times; wherein, When adjusting the parameters of the first generator, the parameters of the discriminator are kept unchanged; when adjusting the parameters of the discriminator, the parameters of the first generator are kept unchanged.

6. The method according to claim 3, wherein the training of the generative adversarial network comprises at least one of the following: The second generator is trained by using the market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator as a first input, the first sample product feature of the first sample product as a second input, and the first sample product value information as an output; The second generator is trained by taking the market loss information of the first sample product and / or the market increment information of the second sample product generated by the first generator as the first input, the second sample product feature of the second sample product as the second input, and the second sample product value information as the output.

7. The method according to claim 1, wherein generating the market loss information of the first product comprises: generating two or more pieces of market loss information corresponding to the first products according to the first product characteristics and first product value information of the two or more first products; The generating the second product value information includes: obtaining two or more second product value information corresponding to the second product according to two or more market loss information corresponding to the first product.

8. The method according to claim 4, wherein the difference is obtained according to a first expression and a second expression, wherein the first expression is calculated according to the market loss information, and the second expression is calculated according to the market increment information; in, The first expression is positively correlated with the market loss information, and the second expression is negatively correlated with the market increment information.

9. The method according to claim 1, wherein: The product characteristics include at least one of the following: product hardware data, product function data, product appearance data, product time data and product region data.

10. An information generating device, comprising: A first generating module generates market loss information of the first product based on the first product feature and the first product value information obtained by the first generator in the generative adversarial network; A second generating module generates second product value information based on the market loss information and second product features obtained by the second generator in the generative adversarial network; The market increment information of the second product represented by the second product value information and the second product characteristics matches the market loss information of the first product, and the second product is a new product having market relevance to the first product.

Images