E-commerce platform commodity introduction method, electronic equipment and e-commerce system

By constructing a feature vector of consumer interaction traces and attention mechanisms, it solves the problem that e-commerce platforms are difficult to capture the dynamic evolution of consumer interests, and improves the accuracy and transparency of product recommendations.

CN120509946APending Publication Date: 2025-08-19NANJING XIAOCHU INFORMATION TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510608045.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-05-13
Publication Date
2025-08-19

AI Technical Summary

Technical Problem

In the field of online shopping, e-commerce platforms are difficult to capture the dynamic evolution process of consumer product interests, resulting in a low order rate.

Method used

By constructing consumer interaction traces, using neural network models to generate consumer consumption trajectories in the perimeter and full growth cycles, using attention mechanisms to fusion feature vectors, and predict the probability of interaction between consumers and commodities.

Benefits of technology

It improves the accuracy and transparency of product recommendations on e-commerce platforms, captures the changes in consumer interests over time, and provides more meaningful reasons for transparency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120509946A_ABST
    Figure CN120509946A_ABST
Patent Text Reader

Abstract

The invention relates to an e-commerce platform commodity introduction method, an electronic device and an e-commerce system.The e-commerce platform commodity introduction method comprises the steps that on the basis of consumer interaction traces, a weekly consumer consumption track and a full-growth-cycle consumer consumption track are generated through a neural network model; the weekly consumer consumption trajectory and the full-growth-cycle consumer consumption trajectory are coded into multi-dimensional feature vectors through a language model, weekly consumer feature vectors and full-growth-cycle consumer feature vectors are obtained, the weekly consumer feature vectors and the full-growth-cycle consumer feature vectors are dynamically fused by using an attention mechanism, and the multi-dimensional feature vectors are obtained. Obtaining consumer feature vectors; and connecting the consumer feature vector with the corresponding commodity feature vector, and predicting the probability of interaction between the consumer and the commodity through a multi-layer neural network, so that the e-commerce platform introduces the determined commodity to the consumer. According to the invention, the effectiveness of introducing consumer commodities by the e-commerce platform can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce platforms, and in particular to a method for introducing goods into an e-commerce platform, electronic equipment, and an e-commerce system. Background Art

[0002] In the field of online shopping, it has become a daily behavior for consumers to interact with products online based on e-commerce platforms. Although there are many products on e-commerce platforms, it is difficult to capture the dynamic evolution of consumers' interest in products, thereby reducing the success rate of online shopping. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention proposes a method for introducing products into an e-commerce platform, an electronic device and an e-commerce system.

[0004] The technical solutions for achieving the purpose of the present invention are:

[0005] In a first aspect, the present invention provides a method for introducing products to an e-commerce platform, the method comprising:

[0006] Based on the e-commerce platform data, a consumer set and a product set are constructed. Consumers in the consumer set interact with some of the products in the favorite product set, and the time nodes of the interactions are retained to construct consumer interaction traces; based on the consumer interaction traces, a neural network model is used to generate weekly consumer consumption trajectories that are liked weekly and expressed in natural language, as well as full-growth cycle consumer consumption trajectories that are liked throughout the growth cycle and expressed in natural language; the weekly consumer consumption trajectories and the full-growth cycle consumer consumption trajectories are encoded into multi-dimensional feature vectors through a language model to obtain weekly consumer feature vectors and full-growth cycle consumer feature vectors; the attention mechanism is used to dynamically fuse the weekly consumer feature vectors and obtain a consumer feature vector; the consumer feature vector is connected with the corresponding product feature vector, and a multi-layer neural network is used to predict the probability of interaction between consumers and products, so that the e-commerce platform introduces certain products to consumers.

[0007] In some embodiments, the consumer interaction trace is represented by:

[0008] S p ={(q,t p,q )|q∈Q p}

[0009] S p It is represented as the consumer interaction trace, q is the product, t p,q Q is the time point when consumers in the consumer set interact with some of the products in the product set. p Represents a subset of items in a set.

[0010] In some embodiments, the attention mechanism calculates personalized importance weights for weekly and full growth cycle favorites, and the attention weights are calculated as:

[0011]

[0012] represents the attention weight assigned to the weekly consumer feature vector, represents the attention weight assigned to the feature vector of consumers in the entire growth cycle, M is a learnable parameter vector that calculates the consumer interest weight, and exp(·) represents the exponential function. represents the weekly consumer feature vector, Represents the consumer feature vector of the entire growth cycle.

[0013] In some embodiments, the consumer feature vector is obtained by the weighted sum of the weekly consumer feature vector and the full growth cycle consumer feature vector:

[0014]

[0015] z p represents the consumer feature vector, represents the attention weight assigned to the weekly consumer feature vector, represents the attention weight assigned to the feature vector of consumers throughout the growth cycle, represents the weekly consumer feature vector, Represents the consumer feature vector of the entire growth cycle.

[0016] In a second aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the method for introducing goods into an e-commerce platform as described in any one of the first aspects is implemented.

[0017] In a third aspect, the present invention provides an e-commerce system, which includes: a support unit having an internal space, a plurality of electronic devices as described in the second aspect, and a temperature control device for controlling the temperature of the electronic devices, all of the electronic devices being arranged in the internal space of the support unit.

[0018] Compared with the prior art, the present invention has the following significant advantages:

[0019] (1) Time awareness: By explicitly modeling weekly and full-cycle preferences and integrating them through the attention mechanism, the model better captures the changes in consumer preferences over time, thereby improving its performance.

[0020] (2) Intrinsic interpretability: Each consumer’s weekly and full-cycle interaction traces can be directly exposed, and the attention weight indicates whether recent or ongoing behavior drives product introduction. This provides a more meaningful and transparent reasoning than a single average vector of past items.

[0021] (3) The synergy of LLM-based text generation and attention-based fusion in the present invention produces highly semantic consumer feature vectors, which not only improves recommendation indicators but also enables end consumers and e-commerce platform staff to accurately view which aspects of consumer temporal interaction behavior are influencing each product introduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 Flowchart of a method for introducing products to an e-commerce platform according to one embodiment of the present invention.

[0023] Figure 2 FIG. 1 is a schematic diagram of the structure of an e-commerce system proposed in one embodiment of the present invention.

[0024] Figure 3 FIG. 1 is a schematic structural diagram of the internal part of an e-commerce system proposed in one embodiment of the present invention.

[0025] Figure 4 FIG. 1 is a schematic structural diagram of the internal part of an e-commerce system proposed in another embodiment according to the present invention.

[0026] Figure 5 FIG. 1 is a schematic structural diagram of an e-commerce system proposed in another embodiment according to the present invention.

[0027] Figure 6 FIG. 1 is a schematic structural diagram of an e-commerce system proposed in another embodiment according to the present invention.

[0028] Figure 7 FIG. 1 is a schematic structural diagram of an e-commerce system proposed in another embodiment according to the present invention.

[0029] Figure 8 FIG. 1 is a schematic structural diagram of an e-commerce system proposed in another embodiment according to the present invention.

[0030] Figure 9 FIG. 1 is a schematic structural diagram of an e-commerce system proposed in another embodiment according to the present invention. DETAILED DESCRIPTION

[0031] like Figure 1 As shown, the present invention proposes a method for introducing products to an e-commerce platform, and the method for introducing products to an e-commerce platform includes:

[0032] Step 1: Construct the consumer set P and product set Q based on the e-commerce platform data, where P = {p1, p2, ..., p |p|}, Q={q1,q2,...,q |q| For example, q1 = smart air conditioner, q2 = energy-saving refrigerator, q3 = flagship mobile phone; each consumer p in the consumer set P and a part of the product Q in the product set Q p Interact, and the interaction is retained at time node t p,q , and through S p ={(q,t p,q )|q∈Q p} represents the consumer interaction trace. For example, Q p ={q1=smart air conditioner, q2=energy-saving refrigerator, q3=flagship mobile phone}; and S p ={q2=Energy-saving Refrigerator, 2023-07-15), / / Purchase; (q3=Flagship Mobile Phone, 2024-03-20), / / Browse; (q1=Smart Air Conditioner, 2025-05-10) / / Recently Browsed}. The above only shows part of the data. Sorted by time, the longer the time, the richer the behavior data, including multiple repurchase records.

[0033] The goal of this e-commerce platform product introduction method is to learn the consumer feature vector for each consumer p and product q. and product feature vector In order to accurately predict the interaction between consumer p and product q (described in detail below). Specifically, it aims to estimate a function This function predicts the probability of interaction as follows:

[0034]

[0035] in represents the predicted possibility of interaction between consumer p and product q, and f is implemented as a multi-layer neural network.

[0036] Step 2, such as Figure 1 As shown in Figure 2, the proposed LLM-driven consumer consumption trajectory architecture utilizes temporal information and semantic representation to generate personalized and interpretable consumer feature vectors. The model consists of three components: (1) consumer consumption trajectory establishment, (2) feature vector representation, and (3) product introduction generation.

[0037] (1) Establishment of consumer consumption trajectory

[0038] Based on consumer interaction traces, a neural network model is used to generate weekly consumer consumption trajectories expressed in natural language, as well as full-cycle consumer consumption trajectories expressed in natural language.

[0039] Specifically, in order to construct a consumer consumption trajectory, two different natural language representations of consumer preferences are generated: weekly preferences and full growth cycle preferences. Let S p ={(q,t p,q )|q∈Q p} represents the interaction trace of consumer p, where t p,q Represents the time points of consumer-product interaction. Interaction traces are sorted in ascending order of time points to maintain chronological order. The entire consumer interaction trace is processed twice using a large language model (LLM), using different prompts to generate separate representations of weekly likes and likes over the entire growth cycle (from the establishment of the interaction behavior to the present):

[0040] Category 1: Weekly consumer consumption trajectory generation. Complete interaction trace S p This input is provided to the LLM along with a task-specific prompt. For example, the prompt might be: "Based on consumers' browsing, add-to-cart, and purchase history over the past week, summarize their consumption characteristics, focusing on immediate needs such as high-frequency products, price sensitivity, and promotional response."

[0041] The instructional model extracts the consumer’s weekly preferences, giving more weight to recent interactions while still considering the temporal context. The LLM generates a natural language representation of the weekly preferences, denoted as:

[0042]

[0043] Predefined prompt word templates guide the LLM to extract weekly features. For example, the LLM generates a natural language representation of weekly preferences - output: "The consumer paid attention to a certain brand of smart air conditioner (model 001) within a week, browsing for more than 10 minutes, but did not participate in promotional activities; interactive behavior data shows that the consumer is not price sensitive and prefers to directly purchase flagship models."

[0044] Category 2: Generation of consumer consumption trajectories throughout the entire growth cycle. The same complete interaction traces S p The input is again passed to the LLM, but this time with a different prompt. For example, the prompt might be: "Analyze consumer behavior throughout their growth cycle to extract brand preferences, category trends, and consumption levels."

[0045] The indicator model generates a full-cycle consumer preference trajectory by considering the consumer's entire trace, focusing on long-term consumer interaction. The LLM generates a natural language representation of the full-cycle preference, denoted as:

[0046]

[0047] Predefined prompt word templates guide the LLM to extract features throughout the entire growth cycle. For example, the LLM generates a natural language representation of preferences throughout the entire growth cycle—output: "Long-term preference for home appliances, particularly trust in brand XX, and average order value concentrated in the 2,000-5,000 yuan range."

[0048] (2) Feature vector representation

[0049] The natural language representations generated by the weekly consumer consumption trajectory and the full growth cycle consumer consumption trajectory are encoded into high-dimensional feature vectors through the pre-trained BERT model, generating short-weekly consumer feature vectors and full growth cycle consumer feature vectors, as shown below:

[0050] 1. Weekly consumer feature vector: For example, the weekly consumer feature vector: [-0.1, 1.8, ..., 0.6], captures the recent attention to smart air conditioners;

[0051] 2. Consumer characteristic vectors throughout the entire growth cycle: For example, the full growth cycle consumer feature vector: [0.9, -0.3, ..., 1.2], reflects long-term brand loyalty and energy-saving preferences;

[0052] and are d-dimensional feature vectors, representing the consumer's weekly consumer feature vector and full growth cycle consumer feature vector, respectively. Formulas (4) and (5) describe the encoding process of BERT to convert each natural language profile into a feature vector representation. These feature vectors capture the semantic richness of consumer preferences and the contextual nuances in the interaction traces, serving as the basis for constructing the final consumer trace. Similarly, the prompt word description is also processed by BERT to generate the consumer feature vector, ensuring that the consumer and feature vector are in the same feature vector space.

[0053] To construct the overall consumer feature vector, an attention mechanism is applied to dynamically fuse weekly and lifecycle consumer feature vectors. This attention mechanism calculates personalized importance weights for weekly and lifecycle preferences, enabling the model to adaptively emphasize weekly interactions for consumers with rapidly changing interests or highlight stable, overall trends for consumers with persistent behavior. This dynamic fusion mechanism is a key difference from centralized methods, which create consumer traces by simply averaging item feature vectors, thus ignoring subtle differences over time.

[0054] set up and They represent the attention weights assigned to the weekly and full growth cycle favorite feature vectors respectively. The attention weights are calculated as:

[0055]

[0056] in is a learnable parameter vector, M = [0.4, -0.2, ..., 1.0], and exp(·) represents the exponential function. Formula (6) calculates the attention weights of the weekly favorites, while Formula (7) ensures that the sum of these attention weights is equal to 1.

[0057] In some embodiments, consumer A There is a strong willingness to buy during promotions on e-commerce platforms.

[0058] Then, the overall consumer feature vector z p Obtained as the weighted sum of the weekly and full growth cycle eigenvectors:

[0059]

[0060] in represents the final consumer feature vector. As shown in formula (8), the attention mechanism enables the model to dynamically balance the influence of weekly and full growth cycle preferences based on the consumer’s specific interaction pattern.

[0061] Learned attention weights and can be displayed in a real system to convey the relative importance of a consumer’s recent activity versus past activity. For example, if Very high for a particular consumer, indicating that the model believes that this consumer's most recent interaction dominates their current item introduction.

[0062] (3) Product introduction and generation

[0063] Final consumer feature vector z p , obtained through the attention mechanism, and the corresponding product feature vector z q (For example, smart air conditioner q1 =[1.2,...,0.8](Energy-saving technology upgrade)) and pass through a multi-layer neural network to predict the probability of interaction between consumer p and product:

[0064]

[0065] in Represents the consumer feature vector z p and product feature vector z q connection, Represents the predicted probability of interaction between consumer p and product q. For example, predict the interaction probability between consumer p and product (smart air conditioner)

[0066] As shown in formula (9), MLP acts as a scoring function to learn the nonlinear interaction between the consumer and product feature vectors. The final output is a probability representing the possibility of interaction. MLP consists of multiple fully connected layers and ReLU activation functions, followed by a sigmoid output layer to ensure that the output is in the range [0,1]. The model is trained using binary cross entropy loss, where the target label Indicates whether consumer p interacts with product q:

[0067]

[0068] in Represents the set of all consumer-merchant interaction pairs in the training dataset. Formula (10) predicts the probability and the true label y p,q The difference between , prompting the model to learn accurate predictions.

[0069] The e-commerce platform then introduces specific products to consumers. For example, if the probability of consumer p interacting with a smart air conditioner is 0.88, the e-commerce platform introduces the smart air conditioner to the consumer based on recent browsing behavior and long-term brand loyalty, recommending an upgraded model with energy-saving technology.

[0070] This approach to introducing products to e-commerce platforms offers two key advantages:

[0071] (1) Time awareness: By explicitly modeling weekly and full-cycle preferences and integrating them through the attention mechanism, the model better captures the changes in consumer preferences over time, thereby improving its performance.

[0072] (2) Intrinsic interpretability: Each consumer’s weekly and full-cycle interaction traces can be directly exposed, and the attention weight indicates whether recent or ongoing behavior drives product introduction. This provides a more meaningful and transparent reasoning than a single average vector of past items.

[0073] Essentially, the synergy of LLM (Large Language Model)-based text generation and attention-based fusion produces highly semantic consumer feature vectors, which not only improves product introduction indicators, but also enables end consumers to accurately see which aspects of consumers' weekly and full-life cycle behavior affect each product introduction.

[0074] In embodiment 1, the present invention proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the e-commerce platform product introduction method when executing the computer program.

[0075] like Figure 3As shown, in some embodiments, the present invention provides an e-commerce system. The e-commerce system 1 includes a support unit 10 , a plurality of electronic devices 20 , and a temperature control device 2 for controlling the temperature of the electronic devices 20 .

[0076] In some embodiments, the support unit 10 has a shape that extends along a first direction D1. The support unit 10 can accommodate multiple electronic devices 20 arranged along the first direction D1. The support unit 10 includes a support body 11, a first protective wall 12, an intermediate outer wall 13, and a second protective wall 14. The support body 11 has a rectangular parallelepiped shape. The first protective wall 12 is disposed at the bottom of the support unit 11, forming the bottom of the support unit 10. The intermediate outer wall 13 is disposed on the side of the support unit 11. The intermediate outer wall 13 is arranged to face each other in a direction perpendicular to the vertical direction D1. The lower ends of the pair of intermediate outer wall portions 13 are connected to each other by the first protective wall 12. The second protective wall 14 is disposed at the top of the support unit 11. The second protective wall 14 connects the upper ends of the pair of intermediate outer wall portions 13. Hereinafter, the opposite direction between the pair of intermediate outer wall portions 13 is referred to as the "second direction D2," and the direction perpendicular to the vertical direction D1 and the second direction D2 is referred to as the "third direction D3."

[0077] A plurality of electronic devices 20 are accommodated so that they are arranged along the first direction D1 in the support unit 10. In this embodiment, the electronic devices 20 are accommodated in the support unit 10. The number of electronic devices 20 accommodated in the support unit 10 can be changed as needed. Figure 2 As shown, the electronic device 20 includes a protective cover 21 , a component board portion 22 and a heat dissipation source portion 23 .

[0078] The protective cover 21 is formed into a horizontally extending rectangular parallelepiped shape. The protective cover 21 is fixed to the middle outer wall portion 13. The front and rear surfaces of the protective cover 21 are provided with a plurality of ventilation holes 24 (see FIG. Figure 2 Protective cover 21 houses component board 22 and heat sink 23. Component board 22 is a printed circuit board (PCB) on which multiple electronic components are mounted. Component board 22 extends horizontally. Heat sink 23 is an electronic component mounted on component board 22. Component board 22 is provided with multiple heat sinks 23.

[0079] The plurality of heat dissipation source portions 23 include a first heat dissipation source portion 23 a having a relatively low calorific value and a second heat dissipation source portion 23 b having a relatively large calorific value.

[0080] The first heat sink 23a is, for example, a low-load heat sink 23 of 80W or less. Examples of the first heat sink 23a include memory devices. For example, the first heat sink 23a is arranged behind the front and rear portions D3 in the front-to-back direction. Multiple first heat sinks 23a are arranged in a row, extending to D2 in the second direction. Two rows of first heat sinks 23a are formed longitudinally, aligned with D3.

[0081] For example, the second heat sink 23b is a heat sink 23 with a high load exceeding 80W. Examples of the second heat sink 23b include chips such as CPUs and GPUs. The second heat sink 23b is, for example, located on the front side of the front-to-back direction D3. Multiple second heat sinks 23b are arranged in a row, extending along the second direction D2. The row of second heat sinks 23b forms two rows along the third direction D3.

[0082] The temperature control device 2 can cool down each heat source 23 in the electronic device 20. The temperature control device 2 includes a heat conducting portion 30, a cooling fluid output channel 40, a cooling fluid delivery channel 50, a cooling fluid outlet channel 60 and a cooler 70.

[0083] In some embodiments, a plurality of heat conducting portions 30 are provided to correspond to the heat dissipation source portion 23 of each electronic device 20 , and the heat conducting portion 30 is in contact with the corresponding heat dissipation source portion 23 .

[0084] Each heat conducting portion 30 is arranged to contact multiple heat dissipation sources 23. The heat conducting portion 30 is formed as a horizontally extending rectangular plate. The heat conducting portion 30 extends along the left and right sides by a distance D2. A groove portion (not shown) is formed on the surface of the heat conducting portion 30 facing the heat dissipation sources 23, and the heat dissipation sources 23 are mounted in the groove. A cooling fluid R1 for cooling the heat dissipation sources 23 is encapsulated within the heat conducting portion 30.

[0085] Examples of the cooling fluid R1 include a fluid that reduces temperature, such as water, etc. The cooling fluid R1 is supplied to the heat transfer portion 30 from the cooling fluid output channel 30 , and is discharged through the cooling fluid outlet channel 30 .

[0086] The plurality of heat conducting parts 30 include a first heat conducting part 30 a and a second heat conducting part 30 b .

[0087] In the first heat transfer section 30a, the cooling fluid R1 flows in a single-phase state. In this embodiment, the first heat transfer section 30a is arranged on the cooling fluid output channel 40 side of the second heat transfer section 30b, along the flow direction of the cooling fluid R1. The first heat transfer section 30a is arranged in each row of the first heat dissipation source sections 23a. The first heat transfer section 30a is arranged along a corresponding row of first heat dissipation source sections 23a. In the first heat transfer section 30a, the cooling fluid R1 receives heat from the first heat dissipation source sections 23a and flows in a fluid phase.

[0088] The second heat conducting portions 30b are connected in series with the first heat conducting portions 30a and the cooling fluid R1 in the direction of flow. The second heat conducting portions 30b are disposed in each row of the second heat dissipation source portions 23b. The second heat conducting portions 30b are arranged along a corresponding row of second heat dissipation source portions 23b. In the second heat conducting portions 30b, the cooling fluid R1 receives heat from the second heat dissipation source portions 23b and partially vaporizes. Therefore, in the second heat conducting portions 30b, the cooling fluid R1 flows in a two-phase state of fluid and vapor.

[0089] The cooling fluid output channel 40 supplies cooling fluid R1 to each heat conducting portion 30. Multiple cooling fluid output channels 40 are arranged along the first direction D1. A cooling fluid output channel 20 is provided for each electronic device 40. The cooling fluid output channel 40 includes a cooling fluid output main pipe 41 and a cooling fluid output branch pipe 42.

[0090] The cooling fluid output pipe 41 is connected to the cooler 70. The cooling fluid R1 is supplied from the cooler 70 to the cooling fluid output pipe 41. The cooling fluid output pipe 41 penetrates the protective cover 21 of the electronic device 20.

[0091] A plurality of cooling fluid output branch pipes 42 are provided in each cooling fluid output main pipe 41. The cooling fluid output branch pipes 42 are connected to the plurality of heat conducting parts 30 in the protective cover 21. The cooling fluid output branch pipes 42 supply the cooling fluid R1 from the cooling fluid output main pipe 41 to each connected heat conducting part 30. In the present embodiment, the cooling fluid output branch pipe 42 is connected to the first heat conducting part 30a among the plurality of heat conducting parts 30. The cooling fluid output branch pipe 42 is connected to the longitudinal center portion of the corresponding first heat conducting part 30a. The cooling fluid delivery channel 50 connects the longitudinal center portion of the first heat conducting part 30a and the longitudinal center portion of the second heat conducting part 30b. In the present embodiment, the cooling fluid delivery channel 50 guides the cooling fluid R1 from the first heat conducting part 30a to the second heat conducting part 30b.

[0092] In some embodiments, a cooling fluid outlet channel 60 discharges cooling fluid R1 through each heat conducting portion 30. Multiple cooling fluid outlet channels 60 are arranged along a first direction D1. A cooling fluid outlet channel 60 is provided for each electronic device 20. The cooling fluid outlet channels 60 include a cooling fluid outlet main pipe 61 and a cooling fluid outlet branch pipe 62.

[0093] The cooling fluid outlet main pipe 61 is connected to the cooler 70 . The cooling fluid outlet main pipe 61 penetrates the protective cover 21 of the electronic device 20 .

[0094] Each cooling fluid outlet main pipe 61 is provided with a plurality of cooling fluid outlet branch pipes 62. The cooling fluid outlet branch pipes 62 are connected to the plurality of heat conducting parts 30 in the sleeve 21. The cooling fluid outlet branch pipes 62 discharge the cooling fluid R1 from each connected heat conducting part 30 to the cooling fluid outlet main pipe 61. The cooling fluid R1 discharged into the cooling fluid outlet main pipe 61 is guided to the cooler 70. In this embodiment, the cooling fluid outlet branch pipe 62 is connected to the second heat conducting part 30b among the plurality of heat conducting parts 30. The cooling fluid outlet branch pipe 62 is provided with a terminal for each second heat conducting part 30b. The cooling fluid outlet branch pipe 62 is connected to the longitudinal center portion of the corresponding second heat conducting part 30b. For an electronic device 20, the connection port between the cooling fluid outlet branch pipe 62 and the heat conducting part 30 is higher than the connection port between the cooling fluid output branch pipe 42 and the heat conducting part 30.

[0095] In some embodiments, a cooler 70 cools the cooling fluid R1 through each cooling fluid outlet channel 60 and introduces the cooled cooling fluid R1 into each cooling fluid output channel 40. For example, the cooler 70 is a vertically mounted cooling water circulation device. From the perspective of reducing the pressure drop of the cooling fluid R1 during the cooling cycle of the temperature control device 2, the cooler 70 is preferably positioned near the support unit 10. The cooler 70 includes a cooler protective cover 71, a heat exchanger 72, a first fluid body pipe 73, a first fluid body flow pipe 74, a second fluid body pipe 75, a second fluid body flow pipe 76, and a pump 77.

[0096] The desuperheater protection cover 71 accommodates a heat exchanger 72, a first fluid pipe 73, a first fluid flow pipe 74, a second fluid pipe 75, a second fluid flow pipe 76, and a pump 77. In this embodiment, the desuperheater protection cover 71 is formed in a rectangular shape extending along the first direction to D1.

[0097] Heat exchanger 72 is a condenser used to cool and condense cooling fluid R1 from each cooling fluid outlet channel 60. In this embodiment, heat exchanger 72 provides cooling water w. Heat exchanger 72 cools cooling fluid R1 by exchanging heat with cooling water w and cooling fluid R1. Heat exchanger 72 is located on top of cooler protective cover 71.

[0098] A first fluid pipe 73 is connected to the heat exchanger 72. Multiple cooling fluid output channels 40 are connected to the first fluid pipe 73. The first fluid pipe 73 directs cooling fluid R1, cooled by the heat exchanger 72, to each cooling fluid output channel 40. The first main connector 73 extends upward and downward by a distance D1. In this embodiment, the lower end of the first fluid pipe 73 is connected to the heat exchanger 72 and the first fluid flow pipe 74.

[0099] A second fluid pipe 75 is connected to the heat exchanger 72. Multiple cooling fluid outlet channels 60 are connected to the second fluid pipe 75. The second fluid pipe 75 is heated by each heat conducting portion 30 and guides cooling fluid R1 through each cooling fluid outlet channel 60 and discharges it into the heat exchanger 72. The second main connector 75 extends in the first direction to D1. In this embodiment, the upper end of the second fluid pipe 75 is connected to the second fluid flow pipe 76 via the heat exchanger 72.

[0100] The pump 77 pumps the cooling fluid R1 cooled by the heat exchanger 72 to each cooling fluid output channel 40. The pump 77 is a lower portion of the cooler protection cover 71 and is provided in the first fluid flow pipe 74.

[0101] Subsequently, the circulation of the temperature-reducing fluid R1 in the electronic commerce system 1 will be described.

[0102] First, cooling fluid R1 in the fluid phase of the cooler 70 is pumped by pump 77 and distributed to each cooling fluid output channel 40 via the first fluid body pipe 73. Cooling fluid R1 is further distributed to each cooling fluid output branch pipe 42 via the cooling fluid output main pipe 41 and then to the connected heat transfer unit 30. In the heat transfer unit 30, cooling fluid R1 exchanges heat with the heat source 23. As a result, the heat source 23 is cooled, while cooling fluid R1 is heated.

[0103] In this embodiment, cooling fluid R1 is first supplied from cooling fluid outlet branch pipe 42 to first heat transfer section 30a. Within first heat transfer section 30a, cooling fluid R1 exchanges heat with first heat sink 23a while maintaining a fluid phase. Consequently, first heat sink 23a is cooled, while cooling fluid R1 is heated.

[0104] The cooling fluid R1 is then supplied to the downstream second heat transfer section 30b via the cooling fluid delivery channel 41. In the second heat transfer section 30b, the cooling fluid R1 exchanges heat with the second heat sink 23b. As a result, the second heat sink 23b is cooled, while the cooling fluid R1 is heated. At this point, a portion of the cooling fluid R1 in the second heat transfer section 30b boils and evaporates due to the heat from the second heat sink 23b. Consequently, the cooling fluid R1 exists in both a fluid phase and a vapor phase in the second heat transfer section 30b.

[0105] The cooling fluid R1 passing through the heat transfer portion 30 returns to the cooler 70 through the cooling fluid outlet passage 60. The cooling fluid R1 discharged through each cooling fluid outlet passage 60 is collected in the second fluid pipe 75. The cooling fluid R1 in the second fluid pipe 75 is guided to the heat exchanger 72 through the second fluid flow pipe 76.

[0106] In heat exchanger 72, heat exchange occurs between cooling fluid R1 and cooling water W. Consequently, cooling fluid R1, heated by heat transfer unit 30, is cooled. Consequently, cooling fluid R1 in the vapor phase condenses and becomes a fluid phase. Cooling fluid R1 in the fluid phase in heat exchanger 72 is again directed through first fluid flow pipe 74 to first fluid pipe 73 and distributed to each cooling fluid output channel 40. In this manner, cooling fluid R1 circulates within e-commerce system 1.

[0107] According to the electronic commerce system 1 of this embodiment, the following operational effects are exhibited.

[0108] In this embodiment, the e-commerce system 1 includes a temperature control device 2 capable of cooling each heat source 23. The temperature control device 2 includes a heat conducting portion 30, a cooling fluid output channel 40, a cooling fluid outlet channel 60, and a cooler 70. Multiple heat conducting portions 30 are provided to correspond to and contact the heat source 23 of each electronic device 20. The cooling fluid outlet channel 40 supplies cooling fluid R1 to each heat conducting portion 30. The cooling fluid outlet channel 60 discharges cooling fluid R1 through each heat conducting portion 30. The cooler 70 cools the cooling fluid R1 through each cooling fluid outlet channel 60 and introduces it into the cooling fluid output channel 40.

[0109] The cooling fluid R1 exchanges heat with the heat sink 23 in each heat transfer section 30, absorbing heat from the heat sink 23. Consequently, the heat sink 23 is cooled, and the cooling fluid R1 is heated. In this embodiment, the heated cooling fluid R1 is directed to the cooler 70 through each cooling fluid outlet channel 60. The cooling fluid R1 is cooled by the cooler 70 and then supplied to each heat transfer section 30 again through the cooling fluid outlet channel 40. Thus, the cooling fluid R1 heated by each heat transfer section 30 is also cooled by the cooler 70. Therefore, according to this embodiment, the heat sink 23 can be effectively cooled while achieving compactness.

[0110] In this embodiment, each heat transfer portion 30 is arranged to contact multiple heat dissipation source portions 23. The multiple heat transfer portions 30 include a first heat transfer portion 30a, in which the cooling fluid R1 flows in a single-phase state, and a second heat transfer portion 30b, in which the cooling fluid R1 boils and flows in a two-phase state of fluid and gas. The second heat transfer portion 30b is connected in series with the first heat transfer portion 30a in the direction of flow of the cooling fluid R1.

[0111] According to the above configuration, the heat conducting portion 30 contacts multiple heat dissipation sources 23. Therefore, the number of heat conducting portions 30 can be reduced compared to providing one heat conducting portion 30 for each heat dissipation source 23. Consequently, the number of components in the e-commerce system 1 can be reduced. Furthermore, the multiple heat conducting portions 30 include a first heat conducting portion 30a and a second heat conducting portion 30b connected in series. Therefore, the e-commerce system 1 can perform heat exchange between the cooling fluid R1 and the heat dissipation sources 23 in stages. Consequently, the e-commerce system 1 can cascade the cooling fluid R1 according to the arrangement of the heat dissipation sources 23 to be dissipated. This further improves the cooling efficiency of the e-commerce system 1.

[0112] In this embodiment, the first heat conducting portion 30 a is disposed on one side of the cooling fluid output channel 40 more than the second heat conducting portion 30 b , and the cooling fluid R1 flows in a fluid phase.

[0113] In the above configuration, cooling fluid R1 exchanges heat with the fluid heat sink 23 in the first heat transfer section 30a. Cooling fluid R1 is then supplied to the second heat transfer section 30b via the first heat transfer section 30a. Cooling fluid R1 absorbs heat from the heat sink 23 in the heat transfer section 30a, boiling and evaporating. Consequently, in the second heat transfer section 30b, the heat of vaporization of cooling fluid R1 is removed from the heat sink 23, significantly cooling the heat sink 23.

[0114] In this embodiment, a first heat dissipation source 23a, such as a memory device, is provided with a first heat conduction portion 30a, while a second heat dissipation source 23b, such as a CPU or GPU chip, is provided with a second heat conduction portion 30b. Therefore, the first heat dissipation source 23a can be cooled in the fluid phase using a cooling fluid R1, while the second heat dissipation source 23b can be cooled by evaporating the cooling fluid R1. Consequently, the e-commerce system 1 can effectively cool heat dissipation sources 23 with different calorific values.

[0115] Subsequently, the first embodiment is modified. In some embodiments, as Figure 4 As shown, in the temperature control device 2A of this modification, for example, multiple first heat dissipation source portions 23a are provided in the front-to-back direction of D3. For example, the second heat dissipation source portion 23b is provided in the rear side of D3 in the third direction. To this end, among the multiple heat transfer portions 30, first heat transfer portions 30a are provided in front of and behind D3, and second heat transfer portions 30b are provided in front of and behind D3.

[0116] The cooling fluid output channel 40 is connected to the second heat transfer portion 30b. In the second heat transfer portion 30b, the cooling fluid R1 supplied from the cooling fluid output channel 40 receives heat from the second heat dissipation source 23b and boils. Consequently, the cooling fluid R1 flows in the second heat transfer portion 30b in a two-phase state: a fluid phase and a gas phase. The cooling fluid R1, while in the fluid phase, changes to a gas phase in the second heat transfer portion 30b.

[0117] In this modification, the first heat transfer portion 30a is positioned on the cooling fluid outlet channel 60 side of the second heat transfer portion 30b, along the flow direction of the cooling fluid R1. The cooling fluid outlet channel 60 is connected to the first heat transfer portion 30a. The first heat transfer portion 30a supplies the cooling fluid R1, which has passed through the second heat transfer portion 30b and completely evaporated. In the first heat transfer portion 30a, the cooling fluid R1 flows in a vapor phase.

[0118] According to the modified e-commerce system 1A, the following action effects are demonstrated.

[0119] In this modification, the first heat conducting portion 30 a is provided more on the temperature-lowering fluid outlet passage 60 side than the second heat conducting portion 30 b , and the temperature-lowering fluid R1 flows in a gas phase.

[0120] In the above configuration, the second heat transfer section 30b receives heat from the heat sink 23, boils, and evaporates. Subsequently, the vapor-phase cooling fluid R1 is supplied to the first heat transfer section 30a. Consequently, the vapor-phase cooling fluid R1 flows through the first heat transfer section 30a. Consequently, the cooling fluid R1 flows at a high velocity within the first heat transfer section 30a. Consequently, the cooling efficiency of the first heat transfer section 30a, which is connected to the single-phase heat transfer section 23a, is higher.

[0121] In some embodiments, the electronic commerce system 201 according to the second embodiment of the present disclosure will be described below with reference to the drawings. For the configuration as in the above-mentioned first embodiment, the description is appropriately omitted by attaching the same names and the same numbers.

[0122] like Figure 5 and Figure 6 As shown, the e-commerce system 201 of this embodiment includes a support unit 10, a plurality of electronic devices 20 and a temperature control device 202. In this embodiment, a plurality of electronic devices 20 (for example, four) are provided, and there is space on the top of the support unit 10. Each electronic device 20 has a plurality of heat dissipation source parts 23. The plurality of heat dissipation source parts 23 include a first heat dissipation source part 23a and a second heat dissipation source part 23b. Figure 4 , a part of the configuration of the temperature control device 202 is omitted.

[0123] The temperature control device 202 includes a heat conducting portion 230 , a cooling fluid output channel 240 , a cooling fluid outlet channel 260 , a cooler 270 , a wind power component protective cover 203 , a wind power component 204 , a second cooler 280 , a first connecting portion 205 and a second connecting portion 206 .

[0124] In some embodiments, a plurality of heat conducting portions 230 are provided to correspond to the heat dissipation source portion 23 of each electronic device 20. The heat conducting portion 230 contacts the corresponding heat dissipation source portion 23. In this embodiment, the heat conducting portion 230 is provided in the second heat dissipation source portion 23b of the heat dissipation source portion 23.

[0125] A cooling fluid output channel 240 is provided for each electronic device 20. The cooling fluid output channel 240 is connected to each corresponding heat conducting portion 230 and supplies the cooling fluid R1 from the cooler 270 to each heat conducting portion 230. In this embodiment, the cooling fluid output channel 240 extends horizontally.

[0126] In some embodiments, a cooling fluid outlet channel 260 is provided for each electronic device 20 , and the cooling fluid outlet channel 260 is connected to each corresponding heat conducting portion 230 and discharges the cooling fluid R1 to the cooler 270 through each heat conducting portion 230 .

[0127] In addition, for an electronic device 20 , it is desirable that the connection port between the cooling fluid outlet channel 260 and the heat conducting portion 230 is higher than the connection port between the cooling fluid output channel 240 and the heat conducting portion 230 .

[0128] In some embodiments, the cooler 270 cools down the cooling fluid R1 through each cooling fluid outlet channel 260 and introduces the cooling fluid R1 into the cooling fluid output channel 240. The detailed configuration of the cooler 270 will be described later.

[0129] The wind turbine protection cover 203 is provided at the rear of the support unit 10. The wind turbine protection cover 203 is formed in a rectangular shape extending to D1 in the first direction and is open at both sides in the front-rear direction of D3.

[0130] In some embodiments, multiple wind assemblies 204 are positioned within the wind assembly protective cover 203 along the first direction D1. The wind assemblies 204 are connected to the support unit 10. Each wind assembly 204 is provided for at least one electronic device 20 and is positioned to face the corresponding electronic device 20 and the third direction D3. In this embodiment, the wind assemblies 204 are also positioned above the support unit 10 in a space where the electronic devices 20 are not displaced, at a position perpendicular to the third direction D3, overlapping the space D1. The wind assemblies 204 draw air A through the heat dissipation source 23 in the electronic device 20.

[0131] The second cooling unit 280 is provided between the support unit 10 and the wind power assembly 204. The second cooling unit 280 cools down the air A that has passed through the heat dissipation source 23. The second cooling unit 280 includes a cooling pipe group 281.

[0132] In some embodiments, a cooling tube assembly 281 is installed at an opening on the front of the wind power module protective cover 203. The cooling tube assembly 281 extends in a first direction D1 and a second direction D2. For example, the cooling tube assembly 281 is a finned tube type cooling tube assembly. A second cooling fluid R2, which exchanges heat with the ambient air A of the second cooler 280, flows through the cooling tube assembly 281. Examples of the second cooling fluid R2 include water.

[0133] The cooler 270 is disposed in the cooling tube group 281 , and the cooling fluid R1 exchanges heat with the second cooling fluid R2 flowing through the cooling tube group 281 , thereby cooling the cooling fluid R1 . The cooler 270 has a protective cover 271 .

[0134] In some embodiments, the protective cover 271 is set at a vertical position D1, which overlaps with the space where the electronic device 20 is not located on the upper part of the support unit 10 and the third direction D3. A portion of the cooling tube group 281 is set inside the protective cover 271. Each cooling fluid output channel 240 and each cooling fluid outlet channel 260 are connected to the protective cover 271 through the first connecting portion 205 and the second connecting portion 206 described later. The cooling fluid R1 discharged through the cooling fluid outlet channel 260 is supplied in the protective cover 271. In the protective cover 271, heat exchange is performed between the cooling fluid R1 and the second cooling fluid R2 in the cooling tube group 281, and the cooling fluid R1 is cooled. The cooling fluid R1 cooled in the protective cover 271 is supplied from each cooling fluid output channel 240 to each electronic device 20.

[0135] The first connecting portion 205 connects the protective cover 271 and the plurality of cooling fluid output channels 240. The first connecting portion 205 guides the cooling fluid R1 cooled in the protective cover 271 to each cooling fluid output channel 240. The first connecting portion 205 extends to D1 in the first direction.

[0136] The second connecting portion 206 connects the protective cover 271 and the plurality of cooling fluid outlet channels 260. The second connecting portion 206 is heated by each heat conducting portion 30 and guides the cooling fluid R1 discharged through each cooling fluid outlet channel 260 into the protective cover 271. The second connecting portion 206 extends in the first direction to D1.

[0137] Subsequently, the circulation of the cooling fluid R1 in the electronic commerce system 201 will be described.

[0138] First, cooling fluid R1 in protective sleeve 271 is distributed to each cooling fluid output channel 240 via first connection portion 205. Cooling fluid R1 is supplied to heat transfer portion 230, to which each cooling fluid output channel 240 is connected. Within heat transfer portion 230, cooling fluid R1 exchanges heat with heat source portion 23. As a result, heat source portion 23 is cooled, while cooling fluid R1 is heated.

[0139] The cooling fluid R1 passing through each heat conducting portion 230 is collected in the second connecting header 206 by the cooling fluid outlet channel 260 . Then, the cooling fluid R1 returns to the protective jacket 271 .

[0140] In protective jacket 271, heat exchange occurs between cooling fluid R1 and a second cooling fluid R2. Consequently, cooling fluid R1, heated by heat transfer unit 230, is cooled. Cooling fluid R1 in the fluid phase within protective jacket 271 is redistributed to each cooling fluid output channel 240. In this manner, cooling fluid R1 circulates within e-commerce system 201.

[0141] The cooling fluid R1 can boil and vaporize in the heat transfer portion 230, condense in the protective jacket 271, and circulate in both a fluid phase and a gas phase within the circulation of the e-commerce system 201. In this case, the cooling fluid R1 naturally circulates within the e-commerce system 201 due to the upward flow of the cooling fluid R1 in the gas phase generated within the heat transfer portion 230. Alternatively, the cooling fluid R1 can circulate in a single phase within the circulation of the e-commerce system 201 without boiling within the heat transfer portion 230. In this case, for example, a pump (not shown) for pumping the cooling fluid R1 can be installed within the e-commerce system 201 to forcefully circulate the cooling fluid R1.

[0142] According to the electronic commerce system 201 of this embodiment, the following operational effects are exhibited.

[0143] In this embodiment, the temperature control device 202 includes a wind assembly 204 and a second cooling unit 280. The wind assembly 204 is connected to the support unit 10 and draws air A through the heat sink 23. The second cooling unit 280 is disposed between the support unit 10 and the wind assembly 204 and is used to cool the air A that has passed through the heat sink 23. The second cooling unit 280 includes a cooling tube assembly 281, through which flows a second cooling fluid R2, which exchanges heat with the surrounding air A.

[0144] According to the above configuration, the e-commerce system 201 can draw air A through the wind power component 204 and pass the air A through the heat dissipation source 23. Therefore, the e-commerce system 201 can use the cooling fluid R1 in the heat conduction part 230 and the air A drawn in by the wind power component 204 to cool the heat dissipation source 23.

[0145] For example, in this embodiment, by attaching the heat conducting portion 230 to the second heat sink 23b of the heat sink 23, the e-commerce system 201 can simultaneously cool the cooling fluid R1 and the air A drawn in by the wind power assembly 204 to cool the second heat sink 23b. Therefore, the e-commerce system 201 can cool the first heat sink 23a and the second heat sink 23b at a set air volume, which only allows the first heat sink 23a to be cooled between the heat sinks 23. Consequently, the e-commerce system 201 can reduce the power consumption required to drive the wind power assembly 204. Furthermore, since the noise generated by the wind power assembly 204 is also reduced, the working environment is improved.

[0146] In this embodiment, the cooler 270 is disposed in the cooling tube group 281 and performs heat exchange between the cooling fluid R1 and the second cooling fluid R2.

[0147] According to the above configuration, the cooling fluid R1, heated by heat exchange with the heat sink 23, is cooled by the second cooling fluid R2 in the cooling tube assembly 281. Therefore, there is no need for a separate device for cooling or heating the cooling fluid R1. This allows for miniaturization of the e-commerce system 201 and saves space.

[0148] In this embodiment, the cooling fluid output channel 240 has been described as extending horizontally, but the present invention is not limited thereto. However, when the cooling fluid R1 is naturally circulated by convection within the e-commerce system 201, as in this embodiment, the cooling fluid output channel 240 extends horizontally in the front-to-back direction, or toward the front of D3. However, when the cooling fluid R1 is forcibly circulated by, for example, a pump (not shown) or the like, this is not the case, and the cooling fluid output channel 240 extends in an upward gradient shape, allowing the cooling fluid R1 to circulate within the e-commerce system 201, for example, in the front-to-back direction toward the front of D3.

[0149] Then, the first modification is performed according to the second embodiment. In some embodiments, Figure 7 As shown, in this modification, as in the first embodiment, a plurality of electronic devices 20 are arranged substantially uniformly in the support unit 10 along the first direction D1. For example, the electronic devices 20 are arranged in 5. In each modification described later, the arrangement of the plurality of electronic devices 20 is the same as that of the first embodiment. In the temperature control device 202A of this modification, the cooler 270 is located above the wind power component 204. The cooler 270 includes a protective cover 271 and a cooling tube group not shown provided in the protective cover 271. In the protective cover 271, the cooling fluid R1 is supplied from each cooling fluid outlet channel 260 through the second connecting portion 206. In the protective cover 271, the provided cooling fluid R1 is cooled by the internal cooling tube group (not shown) and the air A outside the protective cover 271.

[0150] The electronic commerce system 201A according to this modification exhibits the following operational effects: In this modification, the cooler 270 is located above the wind power component 204 .

[0151] According to the above configuration, the cooling unit 270 does not block the flow of air A by the wind power assembly 204. Therefore, the e-commerce system 201A can more effectively cool the heat dissipation source 23. In addition, since the noise generated by the driving of the wind power assembly 204 is reduced, the working environment is further improved.

[0152] Then, a second modification is made according to the second embodiment. Figure 8 As shown, in this modification, the temperature control device 202B further has a second wind power component 207 .

[0153] The second wind force component 207 is disposed on the second protective wall portion 14 of the support unit 10. In some embodiments, the second wind force component 207 discharges the air A in the support unit 10 upward.

[0154] The cooler 270 is disposed in the support unit 10 , and firstly, the electronic device 20 is disposed in the support unit 10 . In addition, the cooler 270 is disposed below the second wind power component 207 .

[0155] The modified e-commerce system 201B demonstrates the following operational effects. In this modification, the temperature control device 202B includes a second wind force assembly 207. This second wind force assembly 207 is mounted on the second protective wall 14 of the support unit 10 and exhausts air A from the support unit 10 upward. A cooling unit 270 is positioned above all electronic devices 20 in the support unit 10 and below the second wind force assembly 207.

[0156] With this configuration, e-commerce system 201B can deliver air A to protective cover 271 of cooling unit 270 via second wind power assembly 207. Air A blown out by second wind power assembly 207 cools cooling fluid R1 within protective cover 271. Furthermore, since cooling unit 270 and second wind power assembly 207 are located within support unit 10, temperature control device 202B can be compacted. This allows for miniaturization of e-commerce system 201B and reduces space requirements.

[0157] Then, a third modification is made according to the second embodiment. In some embodiments, as Figure 9As shown, in this modification, a cooling unit 270 is disposed within the support unit 10, and the electronic device 20 is initially disposed within the support unit 10. The cooling unit 270 is arranged so as to extend longitudinally and longitudinally to D3. The cooling unit 270 is disposed at an angle. Therefore, the front end of the cooling unit 270 is positioned above the rear end of the cooling unit 270. A first connecting portion 205 is connected to the rear end of the cooling unit 270, and each second connecting portion 206 is connected to the front end of the cooling unit 270. For example, the cooling unit 270 is a plate heat exchanger.

[0158] Temperature control device 202C further includes a fluid inlet channel 208 and a fluid outlet channel 209. Fluid inlet channel 208 connects cooling unit 270 and cooling tube assembly 281 of second cooling unit 280. Fluid inlet channel 208 communicates with cooling tube assembly 281 of second cooling unit 280 to direct the second cooling fluid R2 to cooling unit 270. Fluid outlet channel 209 is disposed within cooling unit 270. Fluid outlet channel 209 communicates with cooler 270 and discharges the second cooling fluid R2 from cooler 270 to the exterior of cooler 270.

[0159] The modified e-commerce system 201C exhibits the following operational effects. In this modification, a cooler 270 is positioned above all electronic devices 20 in the support unit 10. The temperature control device 202C includes a fluid inlet channel 208 and a fluid outlet channel 209. The fluid inlet channel 208 communicates with the second cooler 280 to direct the second cooling fluid R2 to the cooler 270. The fluid outlet channel 209 communicates with the cooler 270 to discharge the second cooling fluid R2 from the cooler 270.

[0160] According to the above configuration, the temperature reducer 270 is disposed above all electronic devices 20. Therefore, the dead space above the electronic devices 20 can be used to compress the temperature control device 202C.

[0161] Furthermore, the e-commerce system 201C is capable of guiding the second cooling fluid R2 of the cooling tube group 281 to the cooling unit 270. Therefore, the cooling fluid R1 is cooled by heat exchange with the second cooling fluid R2. Therefore, the configuration of the cooler 270 can be simplified. Therefore, the temperature control device 202C can be designed to be more compact. Thus, according to the current modification, the miniaturization of the e-commerce system 201C can be achieved, and space saving of the e-commerce system 201C can be achieved. In this modification, it is assumed that the cooler 270 is set in the support unit 10, but it is not limited to this. The cooling unit 270 can be set outside the support unit 10.

[0162] The foregoing is to be considered merely as an illustration of the principles of the present invention. Furthermore, since many modifications and variations will readily occur to those skilled in the art, it is not desired to limit the present invention to the exact structure and operation shown and described. Although preferred embodiments have been described, the details may be changed without departing from the essential concept of the present invention. Therefore, the technical solutions involved in the present invention include not only the technical solutions disclosed above, but also technical solutions consisting of equivalent replacements of the technical features in the technical solutions involved in the present invention. Matters not covered by the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A method for introducing products to an e-commerce platform, characterized in that: include: Based on e-commerce platform data, a consumer set and a product set are constructed. Consumers in the consumer set interact with some of the products in the favorite product set, and the time nodes of the interactions are retained to construct consumer interaction traces. Based on the consumer interaction traces, a neural network model is used to generate weekly consumer consumption trajectories expressed in natural language, as well as full-growth cycle consumer consumption trajectories expressed in natural language based on weekly preferences. Encoding the weekly consumer consumption trajectory and the full growth cycle consumer consumption trajectory into a multidimensional feature vector through a language model to obtain a weekly consumer feature vector and a full growth cycle consumer feature vector; Use the attention mechanism to dynamically fuse the weekly consumer feature vector and the full growth cycle consumer feature vector to obtain the consumer feature vector; The consumer feature vector is connected to the corresponding product feature vector and a multi-layer neural network is used to predict the probability of interaction between the consumer and the product, so that the e-commerce platform can introduce certain products to the consumer.

2. The method for introducing products to an e-commerce platform according to claim 1, characterized in that: The consumer interaction trace is represented as: S p ={(q,t p,q )|q∈Q p } S p It is represented as the consumer interaction trace, q is the product, t p,q Q is the time point when consumers in the consumer set interact with some of the products in the product set. p Represents a subset of products in a product set.

3. The method for introducing products to an e-commerce platform according to claim 1, characterized in that: The attention mechanism calculates the personalized importance weights of weekly and full growth cycle favorites. The calculation of attention weights is: represents the attention weight assigned to the weekly consumer feature vector, represents the attention weight assigned to the feature vector of consumers in the entire growth cycle, M is a learnable parameter vector that calculates the consumer interest weight, and exp(·) represents the exponential function. represents the weekly consumer feature vector, Represents the consumer feature vector of the entire growth cycle.

4. The method for introducing products to an e-commerce platform according to claim 2, characterized in that: The consumer characteristic vector is obtained by the weighted sum of the weekly consumer characteristic vector and the full growth cycle consumer characteristic vector: z p represents the consumer feature vector, represents the attention weight assigned to the weekly consumer feature vector, represents the attention weight assigned to the feature vector of consumers throughout the growth cycle, represents the weekly consumer feature vector, Represents the consumer feature vector of the entire growth cycle.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the method for introducing goods into an e-commerce platform as described in any one of claims 1 to 4 is implemented.

6. E-commerce system, characterized in that, include: A support unit having an internal space, a plurality of electronic devices as claimed in claim 5 and a temperature control device for controlling the temperature of the electronic devices, wherein all the electronic devices are arranged in the internal space of the support unit.

7. The electronic commerce system according to claim 6, characterized in that: Each of the electronic devices includes a heat dissipation source; the temperature control device includes: The heat conducting parts, the number of which is the same as the heat dissipation source parts of all the electronic devices, are in contact with the corresponding heat dissipation source parts of the electronic devices; a cooling fluid output channel for outputting cooling fluid to each heat conducting part; a cooling fluid outlet channel for outlet cooling fluid passing through each heat conducting portion; and A cooler is used to cool the cooling fluid led out through each cooling fluid lead-out channel and introduce the cooled cooling fluid into the cooling fluid output channel.

8. The electronic commerce system according to claim 7, wherein: The heat dissipation source portion includes: a first heat dissipation source portion and a second heat dissipation source portion, wherein the first heat dissipation source portion has a lower calorific value than the second heat dissipation source portion; The heat conducting part includes a first heat conducting part and a second heat conducting part, the first heat conducting part is arranged in contact with a row of corresponding first heat dissipation source parts, the cooling fluid in the first heat conducting part receives heat from the first heat dissipation source parts and flows in a liquid state, the second heat conducting part is arranged in contact with a row of corresponding second heat dissipation source parts, the second heat conducting part is connected in series with the first heat conducting part along the flow direction of the cooling fluid, the cooling fluid in the second heat conducting part receives heat from the second heat dissipation source parts and flows in a gas-liquid mixed state.