A transport information pushing method, device and electronic equipment

Abstract

Embodiments of the present application provide a method and device for pushing transport capacity information and electronic equipment. The method comprises: for any to-be-processed freight information, determining a target feature vector of the to-be-processed freight information based on a route code and a price code of the to-be-processed freight information; comparing the target feature vector of the to-be-processed freight information with historical behavior feature vectors of candidate transporters; wherein the historical behavior feature vector of any candidate transporter is determined based on a target feature vector of historical freight information corresponding to the candidate transporter; selecting a to-be-pushed transporter from the candidate transporters based on the comparison result, and pushing transport capacity information of the to-be-pushed transporter to a shipping party corresponding to the to-be-processed freight information. The above method pushes the transport capacity information of the transporter suitable for the route and price of the to-be-processed freight information, and improves the order forming efficiency.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of logistics, and in particular, to a transport capacity information pushing method and device and electronic equipment. BACKGROUND

[0002] In the freight process of a logistics trunk line, when a consignor (a shipper) needs to send goods to a designated destination, the consignor sends the goods to the destination through a carrier after reaching an agreement with a driver (or a carrier).

[0003] In related technologies, the carrier publishes transport capacity information (the transport capacity information contains relevant information of the carrier), and the shipper finds suitable transport capacity information from a large amount of transport capacity information to reach an agreement.

[0004] However, the above method requires the shipper to spend a lot of time selecting transport capacity information, which is time-consuming and laborious. SUMMARY

[0005] Embodiments of the present application provide a transport capacity information pushing method and device and electronic equipment to push suitable transport capacity information to a shipper and improve order formation efficiency.

[0006] In a first aspect, embodiments of the present application provide a transport capacity information pushing method, which includes:

[0007] For any to-be-processed freight information, a target feature vector of the to-be-processed freight information is determined based on a route code and a price code of the to-be-processed freight information;

[0008] The target feature vector of the to-be-processed freight information and historical behavior feature vectors of candidate carriers are compared, wherein the historical behavior feature vector of any candidate carrier is determined based on a target feature vector of historical freight information corresponding to the candidate carrier;

[0009] Based on the comparison result, a to-be-pushed carrier is selected from the candidate carriers, and transport capacity information of the to-be-pushed carrier is pushed to a shipper corresponding to the to-be-processed freight information.

[0010] The above scheme determines a target feature vector representing route features and price features of any to-be-processed freight information based on a route code and a price code of the to-be-processed freight information; by comparing the target feature vector of the to-be-processed freight information with historical behavior feature vectors of candidate carriers, a to-be-pushed carrier suitable for the route and price of the to-be-processed freight information can be selected, and the transport capacity information of the to-be-pushed carrier is pushed to the corresponding shipper, without the shipper selecting from all transport capacity information or the carrier selecting from all freight information, thereby improving order formation efficiency.

[0011] In some optional embodiments, the target feature vector of the to-be-processed freight source information is determined based on the route code and the price code of the to-be-processed freight source information, comprising:

[0012] The route code of the to-be-processed freight source information is input into a first conversion matrix to obtain a first vector representing route features, and the price code of the to-be-processed freight source information is input into a second conversion matrix to obtain a second vector representing price features;

[0013] The target feature vector of the to-be-processed freight source information is determined according to the first vector and the second vector of the to-be-processed freight source information.

[0014] The above scheme obtains the first vector representing route features and the second vector representing price features through the first conversion matrix and the second conversion matrix, and then accurately determines the target feature vector representing the route features and the price features of the to-be-processed freight source information based on the two vectors.

[0015] In some optional embodiments, the first conversion matrix is obtained in the following manner:

[0016] The route codes of the first (M-1) freight source information in each sample sequence are input as input, and the similarity between the output result and the route code of the Mth freight source information in the corresponding sample sequence is taken as an optimization condition to train a first initial model to obtain a trained route model; wherein the first initial model comprises a conversion matrix, a summation layer, and a restoration layer;

[0017] The conversion matrix in the route model is determined as the first conversion matrix;

[0018] The second conversion matrix is obtained in the following manner:

[0019] The price codes of the first (M-1) freight source information in each sample sequence are input as input, and the similarity between the output result and the price code of the Mth freight source information in the corresponding sample sequence is taken as an optimization condition to train a second initial model to obtain a trained price model; wherein the second initial model comprises a conversion matrix, a summation layer, and a restoration layer;

[0020] The conversion matrix in the price model is determined as the second conversion matrix.

[0021] The above scheme trains the first initial model based on route coding of the source information in the sample sequence, so that the trained first initial model can learn the association between the routes of each source information, and then the first initial model after training can be used as a route model, and the conversion matrix in the route model can be used as a first conversion matrix. Thereafter, the first vector obtained through the first conversion matrix can accurately represent the route characteristics of the corresponding source information. The second initial model is trained based on the price coding of the source information in the sample sequence, so that the trained second initial model can learn the association between the prices of each source information, and then the second initial model after training can be used as a price model, and the conversion matrix in the price model can be used as a second conversion matrix. Thereafter, the second vector obtained through the second conversion matrix can accurately represent the price characteristics of the corresponding source information.

[0022] In some optional embodiments, any sample sequence source information includes M source information continuously viewed by the same carrier, and the interval between the time when the first source information in the sample sequence is viewed and the time when the Mth source information is viewed does not exceed a preset time interval.

[0023] In some optional embodiments, the dimensions of the first vector and the second vector are the same; and the target feature vector of the to-be-processed source information is determined according to the first vector and the second vector of the to-be-processed source information, including:

[0024] The average of the first vector and the second vector of the to-be-processed source information is determined as the target feature vector of the to-be-processed source information; or

[0025] The weighted average of the first vector and the second vector of the to-be-processed source information is determined as the target feature vector of the to-be-processed source information based on the first weight coefficient corresponding to the route and the second weight coefficient corresponding to the price.

[0026] The above scheme can conveniently and efficiently determine the target feature vector of the to-be-processed source information through the average of the first vector and the second vector of the to-be-processed source information; and can more accurately determine the target feature vector of the to-be-processed source information through the weighted average of the first vector and the second vector of the to-be-processed source information; thereby meeting the needs of different application scenarios.

[0027] In some optional embodiments, the historical behavior feature vector of each candidate carrier is determined in the following manner:

[0028] For any candidate carrier, the average of the target feature vectors of the historical source information corresponding to the candidate carrier is determined as the historical behavior feature vector of the candidate carrier; or

[0029] For any candidate carrier, determine the weight coefficient of the behavior type of each historical cargo information corresponding to the candidate carrier; and determine the weighted average of the target feature vectors of the historical cargo information corresponding to the candidate carrier as the historical behavior feature vector of the candidate carrier; wherein, the behavior type includes some or all of viewing, order grabbing, and order acceptance.

[0030] The above solution can conveniently and efficiently determine the historical behavior feature vector of the candidate carrier by averaging the target feature vectors of the historical cargo information corresponding to the candidate carrier; and can more accurately determine the historical behavior feature vector of the candidate carrier by averaging the weighted average of the target feature vectors of the historical cargo information corresponding to the candidate carrier; thus meeting the needs of different application scenarios.

[0031] In some optional implementations, the target feature vector of the cargo information to be processed is compared with the historical behavioral feature vectors of each candidate carrier, including:

[0032] Determine the target feature vector of the cargo information to be processed and its similarity to the historical behavior feature vectors of each candidate carrier;

[0033] Based on the comparison results, carriers to be pushed to are selected from the candidate carriers, including:

[0034] The N candidate carriers with the highest similarity are determined as the carriers to be pushed, where N is an integer greater than or equal to 1; or, the candidate carriers with a similarity greater than a preset similarity are determined as the carriers to be pushed.

[0035] The above scheme determines the similarity between the target feature vector of the cargo information to be processed and the historical behavior feature vector of each candidate carrier; then, the N candidate carriers with the highest similarity, or the candidate carriers with similarity greater than the preset similarity, are determined as the carriers to be pushed. This makes the historical behavior feature vector of the carrier to be pushed more similar to the target feature vector of the cargo information to be processed, thereby improving the matching degree between the capacity information of the carrier to be pushed and the needs of the shipper.

[0036] In some optional implementations, before comparing the target feature vector of the cargo information to be processed with the historical behavioral feature vectors of each candidate carrier, the following steps are also included:

[0037] Carriers that meet the target demand indicators contained in the cargo information to be processed are identified as candidate carriers.

[0038] The scheme avoids pushing the capacity information of the carrier that does not meet the target demand index contained in the to-be-processed cargo source information to the shipper, thereby saving the viewing time of the shipper.

[0039] In a second aspect, the embodiments of the present application further provide a carrier capacity information pushing device, comprising:

[0040] A vector determination module is configured to determine a target feature vector of any to-be-processed cargo source information based on a route code and a price code of the to-be-processed cargo source information.

[0041] A vector comparison module is configured to compare the target feature vector of the to-be-processed cargo source information with historical behavior feature vectors of each candidate carrier, wherein the historical behavior feature vector of any candidate carrier is determined based on a target feature vector of historical cargo source information corresponding to the candidate carrier.

[0042] An information pushing module is configured to select a to-be-pushed carrier from each candidate carrier based on a comparison result, and push the capacity information of the to-be-pushed carrier to a shipper corresponding to the to-be-processed cargo source information.

[0043] In a third aspect, the embodiments of the present application provide an electronic device, comprising at least one processor and at least one memory, wherein the memory stores a computer program, and when the program is executed by the processor, the processor executes the carrier capacity information pushing method of any of the first aspect.

[0044] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, which stores a computer program executable by an electronic device, and when the program runs on the electronic device, the electronic device executes the carrier capacity information pushing method of any of the first aspect.

[0045] In addition, the technical effects brought by any of the implementation manners of the second to fourth aspects can refer to the technical effects brought by different implementation manners of the first aspect, which will not be described here. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0047] Figure 1 The application scenario provided by the embodiments of the present application is shown in the figure.

[0048] Figure 2 A flowchart of a first transport capacity information pushing method provided by an embodiment of the present application is shown in FIG. 1.

[0049] Figure 3 A flowchart of a first conversion matrix determination method provided by an embodiment of the present application is shown in FIG. 2.

[0050] Figure 4 A first initial model provided by an embodiment of the present application is shown in FIG. 3.

[0051] Figure 5 A flowchart of a second conversion matrix determination method provided by an embodiment of the present application is shown in FIG. 4.

[0052] Figure 6 A structure diagram of a transport capacity information pushing device provided by an embodiment of the present application is shown in FIG. 5.

[0053] Figure 7 A structure diagram of an electronic device provided by an embodiment of the present application is shown in FIG. 6. DETAILED DESCRIPTION

[0054] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0055] The terms "first", "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0056] The term "transport capacity information" includes relevant information of a carrier.

[0057] The term "cargo source information" includes relevant information required to complete a cargo transport task.

[0058] In the description of the present application, it should be noted that, unless otherwise specified and limited, the term "connection" should be understood broadly, for example, it can be directly connected, or indirectly connected through an intermediate medium, or the connection inside two devices. For those skilled in the art, the specific meaning of the above term in the present application can be understood according to the specific circumstances.

[0059] In the freight process of logistics trunk lines, when the shipper needs to send the goods to a designated destination, after reaching an agreement with the carrier, the carrier will send the goods to the destination.

[0060] In some embodiments, the carrier publishes capacity information, and the shipper finds suitable capacity information from a large amount of capacity information, thereby reaching an agreement.

[0061] However, the above methods require shippers to spend a lot of time selecting shipping capacity information, which is time-consuming and laborious.

[0062] In other embodiments, the shipper publishes cargo information, and the carrier finds suitable cargo information from a large amount of cargo information, thereby reaching an agreement.

[0063] However, the above methods require carriers to spend a lot of time selecting capacity information, which is time-consuming and labor-intensive.

[0064] In view of this, embodiments of this application propose a method, apparatus, and electronic device for pushing transportation capacity information. The method includes: for any cargo information to be processed, determining a target feature vector of the cargo information to be processed based on the route code and price code of the cargo information to be processed; comparing the target feature vector of the cargo information to be processed with the historical behavior feature vectors of each candidate carrier; wherein, the historical behavior feature vector of any candidate carrier is determined based on the target feature vector of the historical cargo information corresponding to the candidate carrier; based on the comparison result, selecting a carrier to be pushed from each candidate carrier, and pushing the transportation capacity information of the carrier to be pushed to the shipper corresponding to the cargo information to be processed.

[0065] The above scheme, for any pending cargo information, determines a target feature vector that can characterize the route and price features of the pending cargo information based on the route and price codes of the pending cargo information; by comparing the target feature vector of the pending cargo information with the historical behavior feature vectors of each candidate carrier, a suitable carrier for the route and price of the pending cargo information can be selected for push, and then the capacity information of the carrier to be pushed is pushed to the corresponding shipper, eliminating the need for the shipper to select from all capacity information or the carrier to select from all cargo information, thus improving the efficiency of order completion.

[0066] See Figure 1 The image shows an application scenario provided in this embodiment. This scenario includes an electronic device 100, a carrier connected to the electronic device 100, and multiple shippers connected to the electronic device 100. Figure 1 The example uses 5 carriers (carriers 201 to 205). In practice, there may be more or fewer carriers.Figure 1 The example uses six shippers (shippers 301 to 306). In practice, there may be more or fewer shippers.

[0067] Electronic devices can transmit data with various carriers and shippers through a variety of communication methods, such as Local Area Network (LAN), Wireless Local Area Network (WLAN), or other networks.

[0068] The aforementioned electronic device is used to determine the target feature vector of any pending cargo information from any shipper, based on the route code and price code of the pending cargo information; compare the target feature vector of the pending cargo information with the historical behavior feature vectors of each candidate carrier; wherein the historical behavior feature vector of any candidate carrier is determined based on the target feature vector of the historical cargo information corresponding to the candidate carrier; based on the comparison result, select a carrier to be pushed from each candidate carrier, and push the capacity information of the carrier to be pushed to the shipper corresponding to the pending cargo information.

[0069] In the embodiments of this application, the specific implementation methods of the above-mentioned electronic devices, carriers, and shippers are not limited.

[0070] The above application scenarios are merely examples of application scenarios for implementing the embodiments of this application, and the embodiments of this application are not limited to the above application scenarios.

[0071] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with reference to the accompanying drawings and specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0072] This application provides a first method for pushing transportation capacity information, applied to the aforementioned electronic device, such as... Figure 2 As shown, it includes the following steps:

[0073] Step S201: For any source information to be processed, determine the target feature vector of the source information to be processed based on the route code and price code of the source information to be processed.

[0074] The above pending cargo information refers to cargo information that has not yet been accepted by the carrier.

[0075] This embodiment does not specifically limit the content of the source information, and may include some or all of the following information:

[0076] Route information (i.e., loading and unloading locations), pricing information, loading time information, unloading time information, and required vehicle type information, etc.

[0077] The route coding described above is obtained by encoding the route information in the cargo source information. In some optional implementation schemes, one-hot coding is used to encode the route information to obtain an x-dimensional vector as the route code.

[0078] The price code described above is obtained by encoding the price information in the source information. In some optional implementations, one-hot encoding is used to determine the price range in which the price information is located, and the code (y-dimensional vector) corresponding to the price range is used as the price code. This embodiment does not specifically limit the implementation method of the price range. For example, within 20,000 yuan, every 100 yuan is a price range, and there are 200 price ranges within 20,000 yuan, namely price range 1 (0-100 yuan), price range 2 (100-200 yuan), price range 3 (200-300 yuan), ..., price range 199 (19,800-19,900 yuan) and price range 200 (19,900-20,000 yuan); although 5,210 yuan and 5,272 yuan are different in specific price, they are in the same price range, so the price code is the same. This example is only to illustrate how to set price ranges more clearly, but this application is not limited to this.

[0079] In this embodiment, the dimensions of the price code and the route code can be the same or different, that is, the values ​​of x and y can be the same or different.

[0080] Step S202: Compare the target feature vector of the cargo information to be processed with the historical behavior feature vectors of each candidate carrier. The historical behavior feature vector of any candidate carrier is determined based on the target feature vector of the historical cargo information corresponding to that candidate carrier.

[0081] For example, a candidate carrier can operate on multiple historical cargo information, such as viewing cargo information, bidding for cargo information, and accepting cargo information. This cargo information is the historical cargo information corresponding to that carrier.

[0082] In this embodiment, the historical behavioral feature vector of each candidate carrier can be determined through, but is not limited to, the following methods:

[0083] 1) For any candidate carrier, the average value of the target feature vector of the historical cargo information corresponding to the candidate carrier is determined as the historical behavior feature vector of the candidate carrier.

[0084] As mentioned above, each candidate carrier can correspond to multiple historical cargo information. Since the dimension of each target feature vector is the same, the average value of the feature values ​​on the same dimension of multiple target feature vectors corresponding to any candidate carrier can be determined as the behavioral feature value on this dimension, thereby obtaining the historical behavioral feature vector of the candidate carrier.

[0085] 2) For any candidate carrier, determine the weight coefficient of the behavior type of each historical cargo information corresponding to the candidate carrier; and determine the weighted average of the target feature vectors of the historical cargo information corresponding to the candidate carrier as the historical behavior feature vector of the candidate carrier; wherein, the behavior type includes some or all of viewing, order grabbing and order acceptance.

[0086] As mentioned above, candidate carriers may perform different operations on multiple historical cargo information. Obviously, the candidate carriers' level of interest and competitiveness in cargo information with different behavior types are different. For example, if a carrier only views a certain cargo information but does not bid for it, it indicates that the carrier's level of interest in that cargo information is average; if a carrier bids for a certain cargo information but fails to do so, that is, does not accept the order for that cargo information, it indicates that the carrier's competitiveness in that cargo information is average.

[0087] Based on this, this embodiment determines the weight coefficient of the behavior type of each historical cargo information corresponding to the candidate carrier and the weighted average of the target feature vector of the historical cargo information corresponding to the candidate carrier, which can more accurately determine the historical behavior feature vector of the candidate carrier.

[0088] For example, since the dimension of each target feature vector is the same, the weighted average of the feature values ​​of multiple target feature vectors corresponding to any candidate carrier in the same dimension can be determined as the behavioral feature value in this dimension, thereby obtaining the historical behavioral feature vector of the candidate carrier.

[0089] The above-described method for determining the historical behavioral feature vectors of each candidate carrier is merely an illustrative example, and this application does not impose any specific limitations on it.

[0090] Step S203: Based on the comparison results, select the carrier to be pushed from each candidate carrier, and push the capacity information of the carrier to be pushed to the shipper corresponding to the cargo information to be processed.

[0091] The above scheme, for any pending cargo information, determines a target feature vector that can characterize the route and price features of the pending cargo information based on the route and price codes of the pending cargo information; by comparing the target feature vector of the pending cargo information with the historical behavior feature vectors of each candidate carrier, a suitable carrier for the route and price of the pending cargo information can be selected for push, and then the capacity information of the carrier to be pushed is pushed to the corresponding shipper, eliminating the need for the shipper to select from all capacity information or the carrier to select from all cargo information, thus improving the efficiency of order completion.

[0092] In some optional implementations, step S201 above can be implemented in, but is not limited to, the following ways:

[0093] The route code of the cargo information to be processed is input into the first transformation matrix to obtain a first vector representing the route characteristics; and the price code of the cargo information to be processed is input into the second transformation matrix to obtain a second vector representing the price characteristics.

[0094] Based on the first vector and the second vector of the source information to be processed, the target feature vector of the source information to be processed is determined.

[0095] The above scheme obtains a first vector representing route characteristics and a second vector representing price characteristics through a first transformation matrix and a second transformation matrix, respectively; then, based on these two vectors, it accurately determines the target feature vectors that can represent the route characteristics and price characteristics of the source information to be processed.

[0096] In some optional implementations, the first vector and the second vector have the same dimension; the determination of the target feature vector of the source information to be processed based on the first vector and the second vector of the source information to be processed can be achieved in the following ways, but are not limited to:

[0097] 1) The average value of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed.

[0098] In this embodiment, since the first vector and the second vector have the same dimension, the average value of the feature values ​​of the first vector and the second vector in the same dimension can be determined as the target feature value in this dimension, thereby obtaining the target feature vector of the source information to be processed.

[0099] 2) Based on the first weight coefficient corresponding to the route and the second weight coefficient corresponding to the price, the weighted average of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed.

[0100] In some scenarios, the impact of route and price on order completion varies. Based on this, this embodiment sets a first weight coefficient corresponding to the route and a second weight coefficient corresponding to the price; and determines the weighted average of the first vector and the second vector of the source information to be processed, which can more accurately determine the target feature vector of the source information to be processed.

[0101] For example, since the first vector and the second vector have the same dimension, the weighted average of the feature values ​​of the first vector and the second vector in the same dimension can be determined as the target feature value in this dimension, thereby obtaining the target feature vector of the source information to be processed.

[0102] The above method for determining the target feature vector of the source information to be processed is merely an illustrative example, and this application does not impose any specific limitations on it.

[0103] See Figure 3 As shown, the first transformation matrix can be obtained in, but is not limited to, the following ways:

[0104] Step S301: Take the route codes of the first (M-1) cargo information in each sample sequence as input, and use the similarity between the output result and the route code of the Mth cargo information in the corresponding sample sequence as the optimization condition to train the first initial model and obtain the trained route model; wherein the first initial model includes a transformation matrix, a summation layer and a restoration layer.

[0105] See Figure 4 As shown, exemplarily, the first initial model includes a transformation matrix, a summation layer, and a reduction layer.

[0106] The cargo information in the sample sequence is sorted according to the order in which it was viewed by the carrier. A sample sequence can contain M cargo information entries (M is a positive integer, and the specific value can be preset). If M = 5, according to the order in which the carrier viewed the cargo information, a sample sequence can include cargo information A, cargo information B, cargo information C, cargo information D, and cargo information E, where cargo information E is the Mth cargo information entry. Cargo information A corresponds to route code 1 and price code 1; cargo information B corresponds to route code 2 and price code 2; cargo information C corresponds to route code 3 and price code 3; cargo information D corresponds to route code 4 and price code 4; and cargo information E corresponds to route code 5 and price code 5.

[0107] After inputting route codes 1 to 4 into the first initial model, the transformation matrix W in the first initial model is multiplied by the route codes respectively. That is, the transformation matrix W is multiplied by route code 1 to obtain the first first vector; the transformation matrix W is multiplied by route code 2 to obtain the second first vector; the transformation matrix W is multiplied by route code 3 to obtain the third first vector; and the transformation matrix W is multiplied by route code 4 to obtain the fourth first vector.

[0108] The summation layer of the first initial model performs a summation operation on the above four first vectors to obtain a single line summation vector. For example, the encoded values ​​of the four first vectors in the same dimension are added together to obtain the target encoded value in that dimension. Taking a 10-dimensional first vector as an example, the encoded values ​​of the four first vectors in the first dimension are added together to obtain the target encoded value in the first dimension; the encoded values ​​of the four first vectors in the second dimension are added together to obtain the target encoded value in the second dimension; ...; the encoded values ​​of the four first vectors in the tenth dimension are added together to obtain the target encoded value in the tenth dimension, thus obtaining a 10-dimensional line summation vector.

[0109] The reduction layer of the first initial model restores the route summation vector into an output with the same dimension as the route encoding. This embodiment does not limit the specific implementation of the dimension reduction of the vector by the reduction layer, such as multiplying with a specific matrix.

[0110] Since there are multiple sample sequences, each sample sequence has a corresponding output result. Furthermore, the route code (i.e., route code 5) for the Mth source information in the sample sequence can be obtained. This allows us to correlate the output result with the route code of the Mth source information based on different sample sequences, thus obtaining multiple sets of output results and corresponding route codes. The similarity between the output result and the corresponding route code is used as an optimization condition. That is, by optimizing the above transformation matrix during training, the similarity between the output result and the corresponding route code can be improved. Therefore, the first initial model after training is used as the route model, and the transformation matrix in the route model can be used as the first transformation matrix. This embodiment does not limit the specific algorithm used in the above optimization process, such as using the least squares method.

[0111] It is understandable that the summation layer only sums multiple first vectors, and the reduction layer only reduces the dimension. Therefore, the summation layer and the reduction layer remain unchanged during training, meaning that the summation layer and the reduction layer of the first initial model are the same as those of the route model.

[0112] In some optional implementations, any sample sequence of cargo information includes M cargo information items viewed consecutively by the same carrier, and the interval between the time when the first cargo information item in the sample sequence is viewed and the time when the Mth cargo information item is viewed does not exceed a preset time interval.

[0113] For example, there are M cargo information in the above sample sequence. The cargo information in a sample sequence is viewed continuously by the same carrier, and the time interval between the viewing of cargo information in a sample sequence by the carrier cannot be too long.

[0114] For example, selecting cargo information from all the cargo information viewed by all carriers to form a sample sequence is a large workload. Since some carriers rarely accept orders and basically do not view cargo information, these carriers can be filtered out first.

[0115] Based on this, the sample sequence can be determined in, but is not limited to, the following ways:

[0116] Step 1: Select carriers that have viewed more than M cargo listings within the preset time period as sample carriers.

[0117] The preset duration can be set according to the actual application scenario, such as within six months before the current time, or within one month before the current time.

[0118] The specific value of M can also be set according to the actual application scenario, see above. Figure 4 In an example, it can be set to 5.

[0119] For example, the aforementioned electronic device can record the actions of each carrier in viewing cargo information, and thus, after setting the preset duration and the specific value of M, the aforementioned sample carriers can be identified.

[0120] Step 2: From the cargo information continuously viewed by the sample carrier, select cargo information that has been continuously viewed by the same sample carrier and whose latest viewing time is no more than M times the preset time interval, and use this as a sample sequence, thereby selecting multiple sample sequences of cargo information.

[0121] The aforementioned preset time interval can be set in advance, such as 7 days, 10 days, etc.

[0122] The following is a specific example to illustrate this:

[0123] In this example, M is 5, the preset time interval is 10 days, and the preset duration is within six months prior to the current moment.

[0124] For example, if carrier A and carrier B have viewed more than 5 cargo listings in the six months prior to the current moment, then the sample carriers are carrier A and carrier B.

[0125] Obtain all viewing records of each sample carrier, or viewing records within a certain period of time (which can be the preset duration mentioned above, or other durations);

[0126] Based on the order of viewing time in the viewing record of carrier A, carrier A viewed cargo information 1, cargo information 2, ..., cargo information 6, cargo information 7, these 7 cargo information;

[0127] Carrier A viewed cargo information 1 on August 5, 2021; Carrier A viewed cargo information 2-5 on August 6, 2021; Carrier A viewed cargo information 6 on August 11, 2021; and Carrier A viewed cargo information 7 on August 12, 2021.

[0128] The viewing time interval between source information 1 and source information 5 is 1 day, which does not exceed the preset time interval (7 days). Source information 1 to source information 5 can be considered as a sample sequence. The viewing time interval between source information 2 and source information 6 is 5 days, which does not exceed the preset time interval (7 days). Source information 2 to source information 6 can be considered as a sample sequence. The viewing time interval between source information 3 and source information 7 is 6 days, which does not exceed the preset time interval (7 days). Source information 3 to source information 7 can be considered as a sample sequence. Therefore, the above 7 source information items can form 3 sample sequences.

[0129] Based on the order of viewing time in the viewing record of Carrier B, Carrier B viewed cargo information 8, cargo information 9, ..., cargo information 12, cargo information 13, these 6 cargo information;

[0130] Carrier B viewed cargo information 8 and 9 on August 6, 2021; Carrier B viewed cargo information 10 and 11 on August 7, 2021; Carrier B viewed cargo information 12 on August 8, 2021; and Carrier B viewed cargo information 13 on August 20, 2021.

[0131] The viewing time interval between source information 8 and source information 12 is 2 days, which does not exceed the preset time interval (7 days). Source information 8 to source information 12 can be used as a sample sequence. The viewing time interval between source information 9 and source information 13 is 14 days, which exceeds the preset time interval (7 days). Therefore, source information 9 to source information 13 cannot be used as a sample sequence. Thus, the above 6 source information items can form 1 sample sequence.

[0132] The same source information may be found in the above source information 1 to 13. Furthermore, the above examples are only for clearer illustration of how to determine the sample sequence, but this application is not limited thereto.

[0133] Step S302: Determine the transformation matrix in the route model as the first transformation matrix.

[0134] The above scheme trains a first initial model by encoding the route information of the cargo sources in the sample sequence. This allows the trained first initial model to learn the correlation between the routes of each cargo source, thus serving as a route model. Simultaneously, the transformation matrix in the route model can be used as the first transformation matrix. Subsequently, the first vector obtained through this first transformation matrix can accurately represent the route features of the corresponding cargo source information.

[0135] See Figure 5 As shown, the second transformation matrix can be obtained in, but is not limited to, the following ways:

[0136] Step S501: Take the price codes of the first (M-1) source information in each sample sequence as input, and use the similarity between the output result and the price code of the Mth source information in the corresponding sample sequence as the optimization condition to train the second initial model and obtain the trained price model; wherein the second initial model includes a transformation matrix, a summation layer and a restoration layer.

[0137] The specific implementation of the sample sequence can be referred to the above embodiments, and will not be repeated here.

[0138] The second initial model also includes a transformation matrix, a summation layer, and a reduction layer; however, the transformation matrix in the second initial model may be different from the transformation matrix in the first initial model, and the number of rows and columns of the two transformation matrices may not be the same.

[0139] Taking the sample sequence as an example, which includes the above source information A to source information E, where the Mth source information is source information E:

[0140] The corresponding price codes 1 to 5 can be obtained based on the source information A to source information E;

[0141] After inputting price codes 1 to 4 into the second initial model, the transformation matrix in the second initial model is multiplied by the price codes respectively. That is, the transformation matrix is ​​multiplied by price code 1 to obtain the first second vector; the transformation matrix is ​​multiplied by price code 2 to obtain the second second vector; the transformation matrix is ​​multiplied by price code 3 to obtain the third second vector; and the transformation matrix is ​​multiplied by price code 4 to obtain the fourth second vector.

[0142] The summation layer of the second initial model performs a summation operation on the above four second vectors to obtain a single price summation vector. For example, the encoded values ​​of the four second vectors in the same dimension are added to obtain the target encoded value in that dimension. Taking a 10-dimensional second vector as an example, the encoded values ​​of the four second vectors in the first dimension are added to obtain the target encoded value in the first dimension; the encoded values ​​of the four second vectors in the second dimension are added to obtain the target encoded value in the second dimension; ...; the encoded values ​​of the four second vectors in the tenth dimension are added to obtain the target encoded value in the tenth dimension, thus obtaining a 10-dimensional price summation vector.

[0143] The price summation vector is restored to an output with the same dimension as the price encoding through the reduction layer of the second initial model.

[0144] Since there are multiple sample sequences, each sample sequence has a corresponding output result. Furthermore, the price code (i.e., price code 5) of the Mth source information in the sample sequence can be obtained. This allows us to correlate the output result with the price code of the Mth source information based on different sample sequences, thus obtaining multiple sets of output results and corresponding price codes. The similarity between the output result and the corresponding price code is used as an optimization condition. That is, by optimizing the above transformation matrix during training, the similarity between the output result and the corresponding price code can be improved. Therefore, the second initial model after training is used as the price model, and the transformation matrix in the price model can be used as the second transformation matrix. This embodiment does not limit the specific algorithm used in the above optimization process, such as using the least squares method.

[0145] It is understandable that the summation layer only sums multiple second vectors, and the reduction layer only reduces the dimension. Therefore, the summation layer and the reduction layer remain unchanged during training, meaning that the summation layer and the reduction layer of the second initial model are the same as those of the price model.

[0146] Furthermore, the dimensions of the route encoding and price encoding are not necessarily the same, but the dimensions of the first vector and the second vector are the same. Therefore, in some optional implementations, the number of rows and columns of the transformation matrix in the first initial model can be determined based on the target dimension (the dimensions of the first vector and the second vector) and the dimension of the route encoding; and the number of rows and columns of the transformation matrix in the second initial model can be determined based on the target dimension and the dimension of the price encoding. For example:

[0147] The route code is a 100-dimensional vector, the price code is a 200-dimensional vector, and the target dimension is 10-dimensional. Therefore, the first transformation matrix has 100 rows and 10 columns, so after inputting the route code into the first transformation matrix, a 10-dimensional first vector is obtained. The second transformation matrix has 200 rows and 10 columns, so after inputting the price code into the second transformation matrix, a 10-dimensional second vector is obtained.

[0148] The dimensions of the target, route encoding, and price encoding mentioned above are illustrative examples. In practical applications, various vectors can have more or fewer dimensions.

[0149] Step S502: Determine the transformation matrix in the price model as the second transformation matrix.

[0150] The above scheme trains a second initial model based on price encoding of supply information in the sample sequence. This allows the trained second initial model to learn the correlation between the prices of various supply information items. The trained second initial model can then be used as a price model, and the transformation matrix in the price model can be used as a second transformation matrix. The second vector obtained from this second transformation matrix can then accurately represent the price characteristics of the corresponding supply information.

[0151] In some optional implementations, step S202 above can be implemented in, but is not limited to, the following ways:

[0152] Determine the target feature vector of the cargo information to be processed and its similarity to the historical behavior feature vectors of each candidate carrier;

[0153] Correspondingly, step S203 above can be implemented in, but is not limited to, the following ways:

[0154] The N candidate carriers with the highest similarity are determined as the carriers to be pushed, where N is an integer greater than or equal to 1; or, the candidate carriers with a similarity greater than a preset similarity are determined as the carriers to be pushed.

[0155] For example, the similarity between the target feature vector of the cargo information to be processed and the historical behavior feature vector of each candidate carrier can be determined by calculating the cosine distance between the target feature vector of the cargo information to be processed and the historical behavior feature vector of each candidate carrier; wherein, the larger the cosine distance, the smaller the similarity.

[0156] In some embodiments, candidate carriers are sorted in ascending order of cosine distance (in descending order of similarity), and the top N candidate carriers are determined as the carriers to be pushed. N can be set according to the actual application scenario.

[0157] In other embodiments, each cosine distance can be compared with a preset cosine distance. Candidate carriers with a cosine distance smaller than the preset cosine distance are considered as candidate carriers with a similarity greater than the preset cosine distance, and these candidate carriers are identified as carriers to be pushed.

[0158] The above example uses the cosine distance between vectors to determine similarity. In practical applications, other methods can also be used to determine the similarity between vectors, which will not be illustrated here.

[0159] The above scheme determines the similarity between the target feature vector of the cargo information to be processed and the historical behavior feature vector of each candidate carrier; then, the N candidate carriers with the highest similarity, or the candidate carriers with similarity greater than the preset similarity, are determined as the carriers to be pushed, so that the historical behavior feature vector of the carrier to be pushed is more similar to the target feature vector of the cargo information to be processed, thereby improving the matching degree between the capacity information of the carrier to be pushed and the needs of the shipper.

[0160] In this embodiment, the transportation capacity information includes the carrier's registered vehicle model, contact information, and other relevant information of the carrier. After the transportation capacity information of the carrier to be pushed is pushed to the shipper corresponding to the above-mentioned cargo information to be processed, the shipper can complete communication and negotiation and even reach a transportation agreement based on the transportation capacity information.

[0161] In some optional implementations, the following steps are included before step S202:

[0162] Carriers that meet the target demand indicators contained in the cargo information to be processed are identified as candidate carriers.

[0163] As mentioned above, cargo information includes some or all of the following: route information (i.e., loading and unloading locations), price information, loading time information, unloading time information, and required vehicle type information.

[0164] The target demand indicators may include the loading location, loading time information, and required vehicle type information mentioned above. By using the loading location, loading time information, and required vehicle type information in the pending cargo information as filtering conditions, candidate carriers that can undertake the transportation tasks in the pending cargo information can be selected.

[0165] The following is a specific example to illustrate this:

[0166] First, determine the distance between each carrier and the loading location in the cargo information to be processed, and filter out carriers whose distance is greater than the preset distance.

[0167] Based on the distance between the current location of each carrier and the loading location in the pending cargo information, the travel time is determined; based on the travel time, it is determined whether the carrier can arrive at the loading location before the loading time.

[0168] If so, the required vehicle type information in the pending cargo information will be matched with the carrier's registered vehicle type. If a match is successful, the carrier will be selected as a candidate carrier.

[0169] The above example is only one possible implementation for determining candidate carriers. Other methods can also be used to determine candidate carriers in this application, which will not be illustrated here.

[0170] The above solution identifies carriers that meet the target demand indicators contained in the cargo information to be processed as candidate carriers, thus avoiding the waste of the shipper's viewing time by pushing the capacity information of carriers that do not meet the target demand indicators contained in the cargo information to the shipper.

[0171] Based on the same inventive concept, embodiments of this application provide a capacity information push device, see below. Figure 6 As shown, the capacity information push device 600 includes:

[0172] The vector determination module 601 is used to determine the target feature vector of any source information to be processed based on the route code and price code of the source information to be processed.

[0173] The vector comparison module 602 is used to compare the target feature vector of the cargo information to be processed with the historical behavior feature vector of each candidate carrier; wherein, the historical behavior feature vector of any candidate carrier is determined based on the target feature vector of the historical cargo information corresponding to the candidate carrier.

[0174] The information push module 603 is used to select a carrier to be pushed from each candidate carrier based on the comparison results, and push the capacity information of the carrier to be pushed to the shipper corresponding to the cargo information to be processed.

[0175] In some optional implementations, the vector determination module 601 is specifically used for:

[0176] The route code of the cargo information to be processed is input into the first transformation matrix to obtain a first vector representing the route characteristics; and the price code of the cargo information to be processed is input into the second transformation matrix to obtain a second vector representing the price characteristics.

[0177] Based on the first vector and the second vector of the source information to be processed, the target feature vector of the source information to be processed is determined.

[0178] In some optional implementations, the first transformation matrix is ​​obtained in the following manner:

[0179] The route codes of the first (M-1) cargo information in each sample sequence are used as input, and the similarity between the output result and the route code of the Mth cargo information in the corresponding sample sequence is used as the optimization condition to train the first initial model, thereby obtaining the trained route model; wherein the first initial model includes a transformation matrix, a summation layer and a restoration layer;

[0180] The transformation matrix in the route model is determined as the first transformation matrix;

[0181] The second transformation matrix is ​​obtained in the following manner:

[0182] The price codes of the first (M-1) source information in each sample sequence are used as input, and the similarity between the output result and the price code of the Mth source information in the corresponding sample sequence is used as the optimization condition to train the second initial model, thereby obtaining the trained price model; wherein the second initial model includes a transformation matrix, a summation layer and a restoration layer;

[0183] The transformation matrix in the price model is determined as the second transformation matrix.

[0184] In some optional implementations, any sample sequence of cargo information includes M cargo information items viewed consecutively by the same carrier, and the interval between the time when the first cargo information item in the sample sequence is viewed and the time when the Mth cargo information item is viewed does not exceed a preset time interval.

[0185] In some optional implementations, the first vector and the second vector have the same dimension; the vector determination module 601 is specifically used for:

[0186] The average of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed; or

[0187] Based on the first weight coefficient corresponding to the route and the second weight coefficient corresponding to the price, the weighted average of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed.

[0188] In some alternative implementations, the historical behavioral feature vector of each candidate carrier is determined in the following manner:

[0189] For any candidate carrier, the average value of the target feature vectors of the historical cargo information corresponding to the candidate carrier is determined as the historical behavior feature vector of the candidate carrier; or

[0190] For any candidate carrier, determine the weight coefficient of the behavior type of each historical cargo information corresponding to the candidate carrier; and determine the weighted average of the target feature vectors of the historical cargo information corresponding to the candidate carrier as the historical behavior feature vector of the candidate carrier; wherein, the behavior type includes some or all of viewing, order grabbing, and order acceptance.

[0191] In some optional implementations, the vector alignment module 602 is specifically used for:

[0192] Determine the target feature vector of the cargo information to be processed and its similarity to the historical behavior feature vectors of each candidate carrier;

[0193] Information push module 603 is specifically used for:

[0194] The N candidate carriers with the highest similarity are determined as the carriers to be pushed, where N is an integer greater than or equal to 1; or, the candidate carriers with a similarity greater than a preset similarity are determined as the carriers to be pushed.

[0195] In some optional implementations, before comparing the target feature vector of the cargo information to be processed with the historical behavior feature vectors of each candidate carrier, the vector comparison module 602 is further configured to:

[0196] Carriers that meet the target demand indicators contained in the cargo information to be processed are identified as candidate carriers.

[0197] Since this device is the same as the device in the method of this application embodiment, and the principle of the device in solving the problem is similar to that of the method, the implementation of the device can be referred to the implementation of the method, and the repeated parts will not be described again.

[0198] Based on the same technical concept, this application also provides an electronic device 700, such as... Figure 7 As shown, it includes at least one processor 701 and a memory 702 connected to at least one processor. In this embodiment, the specific connection medium between the processor 701 and the memory 702 is not limited. Figure 7 Taking the connection between the processor 701 and the memory 702 via a bus 703 as an example, the bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, Figure 7 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0199] The processor 701 is the control center of the electronic device, capable of connecting various parts of the device via various interfaces and lines. It performs data processing by running or executing instructions stored in the memory 702 and retrieving data stored in the memory 702. Optionally, the processor 701 may include one or more processing units. The processor 701 may integrate an application processor and a modem processor. The application processor primarily handles the operating system, user interface, and applications, while the modem processor primarily handles issuing instructions. It is understood that the modem processor may not be integrated into the processor 701. In some embodiments, the processor 701 and the memory 702 may be implemented on the same chip; in other embodiments, they may be implemented on separate chips.

[0200] The processor 701 can be a general-purpose processor, such as a central processing unit (CPU), digital signal processor, application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the transportation capacity information push method can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0201] Memory 702, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. Memory 702 may include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage, magnetic disk, optical disk, etc. Memory 702 can be any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. In the embodiments of this application, memory 702 can also be a circuit or any other device capable of implementing storage functions for storing program instructions and / or data.

[0202] In this embodiment, the memory 702 stores a computer program, which, when executed by the processor 701, causes the processor 701 to perform the following:

[0203] For any pending source of goods information, the target feature vector of the pending source of goods information is determined based on the route code and price code of the pending source of goods information;

[0204] The target feature vector of the cargo information to be processed is compared with the historical behavior feature vector of each candidate carrier; wherein, the historical behavior feature vector of any candidate carrier is determined based on the target feature vector of the historical cargo information corresponding to the candidate carrier.

[0205] Based on the comparison results, a carrier to be pushed is selected from each candidate carrier, and the capacity information of the carrier to be pushed is pushed to the shipper corresponding to the cargo information to be processed.

[0206] In some alternative implementations, processor 701 specifically performs:

[0207] The route code of the cargo information to be processed is input into the first transformation matrix to obtain a first vector representing the route characteristics; and the price code of the cargo information to be processed is input into the second transformation matrix to obtain a second vector representing the price characteristics.

[0208] Based on the first vector and the second vector of the source information to be processed, the target feature vector of the source information to be processed is determined.

[0209] In some optional implementations, the first transformation matrix is ​​obtained in the following manner:

[0210] The route codes of the first (M-1) cargo information in each sample sequence are used as input, and the similarity between the output result and the route code of the Mth cargo information in the corresponding sample sequence is used as the optimization condition to train the first initial model, thereby obtaining the trained route model; wherein the first initial model includes a transformation matrix, a summation layer and a restoration layer;

[0211] The transformation matrix in the route model is determined as the first transformation matrix;

[0212] The second transformation matrix is ​​obtained in the following manner:

[0213] The price codes of the first (M-1) source information in each sample sequence are used as input, and the similarity between the output result and the price code of the Mth source information in the corresponding sample sequence is used as the optimization condition to train the second initial model, thereby obtaining the trained price model; wherein the second initial model includes a transformation matrix, a summation layer and a restoration layer;

[0214] The transformation matrix in the price model is determined as the second transformation matrix.

[0215] In some optional implementations, any sample sequence of cargo information includes M cargo information items viewed consecutively by the same carrier, and the interval between the time when the first cargo information item in the sample sequence is viewed and the time when the Mth cargo information item is viewed does not exceed a preset time interval.

[0216] In some optional implementations, the first vector and the second vector have the same dimension; processor 701 specifically executes:

[0217] The average of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed; or

[0218] Based on the first weight coefficient corresponding to the route and the second weight coefficient corresponding to the price, the weighted average of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed.

[0219] In some alternative implementations, the historical behavioral feature vector of each candidate carrier is determined in the following manner:

[0220] For any candidate carrier, the average value of the target feature vectors of the historical cargo information corresponding to the candidate carrier is determined as the historical behavior feature vector of the candidate carrier; or

[0221] For any candidate carrier, determine the weight coefficient of the behavior type of each historical cargo information corresponding to the candidate carrier; and determine the weighted average of the target feature vectors of the historical cargo information corresponding to the candidate carrier as the historical behavior feature vector of the candidate carrier; wherein, the behavior type includes some or all of viewing, order grabbing, and order acceptance.

[0222] In some alternative implementations, processor 701 specifically performs:

[0223] Determine the target feature vector of the cargo information to be processed and its similarity to the historical behavior feature vectors of each candidate carrier;

[0224] The N candidate carriers with the highest similarity are determined as the carriers to be pushed, where N is an integer greater than or equal to 1; or, the candidate carriers with a similarity greater than a preset similarity are determined as the carriers to be pushed.

[0225] In some optional implementations, before comparing the target feature vector of the cargo information to be processed with the historical behavioral feature vectors of each candidate carrier, the processor 701 further performs the following:

[0226] Carriers that meet the target demand indicators contained in the cargo information to be processed are identified as candidate carriers.

[0227] Since the electronic device is the same as the electronic device in the method of this application embodiment, and the principle of the electronic device in solving the problem is similar to that of the method, the implementation of the electronic device can refer to the implementation of the method, and the repeated parts will not be described again.

[0228] Based on the same technical concept, embodiments of this application also provide a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of the above-described capacity information push method.

[0229] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0230] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0231] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0232] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0233] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0234] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for pushing transportation capacity information, characterized in that, The method includes: For any pending source of goods information, the target feature vector of the pending source of goods information is determined based on the route code and price code of the pending source of goods information; The target feature vector of the cargo information to be processed is compared with the historical behavior feature vector of each candidate carrier; wherein, the historical behavior feature vector of any candidate carrier is determined based on the target feature vector of the historical cargo information corresponding to the candidate carrier. Based on the comparison results, a carrier to be pushed is selected from each candidate carrier, and the capacity information of the carrier to be pushed is pushed to the shipper corresponding to the cargo information to be processed. Based on the route coding and price coding of the cargo information to be processed, the target feature vector of the cargo information to be processed is determined, including: The route code of the cargo information to be processed is input into the first transformation matrix to obtain a first vector representing the route characteristics; and the price code of the cargo information to be processed is input into the second transformation matrix to obtain a second vector representing the price characteristics. Based on the first vector and the second vector of the source information to be processed, determine the target feature vector of the source information to be processed; The first transformation matrix is ​​obtained in the following manner: The route codes of the first M-1 cargo information in each sample sequence are used as input, and the similarity between the output result and the route code of the Mth cargo information in the corresponding sample sequence is used as the optimization condition to train the first initial model, thereby obtaining the trained route model; wherein the first initial model includes a transformation matrix, a summation layer and a restoration layer; The transformation matrix in the route model is determined as the first transformation matrix; The second transformation matrix is ​​obtained in the following manner: The price codes of the first M-1 source information in each sample sequence are used as input, and the similarity between the output result and the price code of the Mth source information in the corresponding sample sequence is used as the optimization condition to train the second initial model, thereby obtaining the trained price model; wherein the second initial model includes a transformation matrix, a summation layer and a restoration layer; The transformation matrix in the price model is determined as the second transformation matrix; Each sample sequence contains M pieces of information about the source of goods.

2. The method as described in claim 1, characterized in that, Any sample sequence of cargo information includes M cargo information items viewed consecutively by the same carrier, and the time interval between the viewing time of the first cargo information item and the viewing time of the Mth cargo information item in the sample sequence does not exceed a preset time interval.

3. The method as described in claim 1, characterized in that, The first vector and the second vector have the same dimension; based on the first vector and the second vector of the source information to be processed, the target feature vector of the source information to be processed is determined, including: The average of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed; or Based on the first weight coefficient corresponding to the route and the second weight coefficient corresponding to the price, the weighted average of the first vector and the second vector of the source information to be processed is determined as the target feature vector of the source information to be processed.

4. The method as described in claim 1, characterized in that, The historical behavioral feature vectors of each candidate carrier are determined using the following method: For any candidate carrier, the average value of the target feature vectors of the historical cargo information corresponding to the candidate carrier is determined as the historical behavior feature vector of the candidate carrier. or For any candidate carrier, determine the weight coefficient of the behavior type of each historical cargo information corresponding to the candidate carrier; and determine the weighted average of the target feature vectors of the historical cargo information corresponding to the candidate carrier as the historical behavior feature vector of the candidate carrier; wherein, the behavior type includes some or all of viewing, order grabbing, and order acceptance.

5. The method as described in claim 1, characterized in that, The target feature vector of the cargo information to be processed is compared with the historical behavior feature vectors of each candidate carrier, including: Determine the target feature vector of the cargo information to be processed and its similarity to the historical behavior feature vectors of each candidate carrier; Based on the comparison results, carriers to be pushed to are selected from the candidate carriers, including: The N candidate carriers with the highest similarity are determined as the carriers to be pushed, where N is an integer greater than or equal to 1; or, the candidate carriers with a similarity greater than a preset similarity are determined as the carriers to be pushed.

6. The method according to any one of claims 1 to 5, characterized in that, Before comparing the target feature vector of the cargo information to be processed with the historical behavior feature vectors of each candidate carrier, the process also includes: Carriers that meet the target demand indicators contained in the cargo information to be processed are identified as candidate carriers.

7. A transportation capacity information push device, characterized in that, include: The vector determination module is used to determine the target feature vector of any source information to be processed based on the route code and price code of the source information to be processed. The vector comparison module is used to compare the target feature vector of the cargo information to be processed with the historical behavior feature vector of each candidate carrier; wherein, the historical behavior feature vector of any candidate carrier is determined based on the target feature vector of the historical cargo information corresponding to the candidate carrier. The information push module is used to select a carrier to be pushed from each candidate carrier based on the comparison results, and push the capacity information of the carrier to be pushed to the shipper corresponding to the cargo information to be processed. The vector determination module is specifically used for: The route code of the cargo information to be processed is input into the first transformation matrix to obtain a first vector representing the route characteristics; and the price code of the cargo information to be processed is input into the second transformation matrix to obtain a second vector representing the price characteristics. Based on the first vector and the second vector of the source information to be processed, determine the target feature vector of the source information to be processed; The first transformation matrix is ​​obtained in the following manner: The route codes of the first M-1 cargo information in each sample sequence are used as input, and the similarity between the output result and the route code of the Mth cargo information in the corresponding sample sequence is used as the optimization condition to train the first initial model, thereby obtaining the trained route model; wherein the first initial model includes a transformation matrix, a summation layer and a restoration layer; The transformation matrix in the route model is determined as the first transformation matrix; The second transformation matrix is ​​obtained in the following manner: The price codes of the first M-1 source information in each sample sequence are used as input, and the similarity between the output result and the price code of the Mth source information in the corresponding sample sequence is used as the optimization condition to train the second initial model, thereby obtaining the trained price model; wherein the second initial model includes a transformation matrix, a summation layer and a restoration layer; The transformation matrix in the price model is determined as the second transformation matrix; Each sample sequence contains M pieces of information about the source of goods.

8. An electronic device, characterized in that, It includes at least one processor and at least one memory, wherein the memory stores a computer program that, when executed by the processor, causes the processor to perform the method of any one of claims 1 to 6.

Images