A method and device for pushing product supply information

Abstract

The application discloses a kind of goods source information pushing method and equipment, for according to the demand and characteristics of user, accurately push goods source information to different users.The method comprises the following steps: according to the last transport task of idle user task, the carrying information of completion is determined from the goods source information to be taken in with the first goods source information set matched with the carrying information, and from the goods source information operated by the user, the second goods source information set matched with the carrying information is determined;Determine the first goods source feature corresponding to the first goods source information in the first goods source information set, and the second goods source feature corresponding to the second goods source information in the second goods source information set;According to the second goods source feature, determine the user characteristics of the user, push the first goods source information corresponding to the first goods source feature with the maximum similarity of the user characteristics to the user.

Description

Technical Field

[0001] This invention relates to the field of intelligent logistics technology, and in particular to a method and device for pushing cargo information. Background Technology

[0002] In the logistics sector, especially in long-haul truck freight, if a shipper (or dispatcher) wants to send their goods to a designated location, they need to post cargo information. When a driver (or carrier) sees the cargo information, if they want to take on the freight task in the cargo information, they can contact the shipper to reach an agreement (i.e., make the cargo information into an order or make the cargo information a sale).

[0003] Currently, platforms (or third parties) can help shippers and drivers (both are users on the platform) complete orders for cargo information. The platform can collect, summarize, and publish cargo information, saving shippers time in finding cargo information publishing channels and drivers time in finding cargo information, thereby increasing the speed of order completion for cargo information.

[0004] However, the resources available for displaying cargo information on the platform are currently limited. Carriers cannot possibly browse all the cargo information displayed. Therefore, how to accurately push cargo information to different carriers based on their needs and characteristics has become an urgent problem to be solved. Summary of the Invention

[0005] This invention provides a method and device for pushing product source information, which is used to accurately push product source information to different users according to their needs and characteristics.

[0006] In a first aspect, an embodiment of the present invention provides a method for pushing out supply information, comprising:

[0007] Based on the last completed transportation task information of users with available tasks, a first set of cargo information matching the transportation information is determined from the cargo information to be accepted, and a second set of cargo information matching the transportation information is determined from the cargo information operated by the user.

[0008] Determine the first source feature corresponding to the first source information in the first source information set, and the second source feature corresponding to the second source information in the second source information set;

[0009] Based on the second source characteristics, the user characteristics of the user are determined, and the first source information corresponding to the K first source characteristics with the highest similarity to the user characteristics is pushed to the user, where K is a positive integer.

[0010] The product supply push method provided in this embodiment of the invention takes into account the user's historical operation behavior on product supply information, and compares the user characteristics extracted from the product supply information that the user has operated on with the product supply characteristics of the product supply information to be accepted, thereby selecting product supply information that better meets the user's needs and characteristics for push, effectively improving the user experience.

[0011] As an optional implementation, determining the first source feature corresponding to the first source information in the first source information set, and the second source feature corresponding to the second source information in the second source information set, includes:

[0012] The route code in the first source of goods information is input into the route transformation matrix to obtain the first route feature, and the first route feature is determined as the first source of goods feature; and the route code in the second source of goods information is input into the route transformation matrix to obtain the second route feature, and the second route feature is determined as the second source of goods feature;

[0013] or,

[0014] The price code in the first source information is input into a price conversion matrix to obtain a first price feature, and the first price feature is determined as the first source feature; and the price code in the second source information is input into a price conversion matrix to obtain a second price feature, and the second price feature is determined as the second source feature;

[0015] or,

[0016] The route code in the first source information is input into the route conversion matrix to obtain the first route feature. The price code in the first source information is input into the price conversion matrix to obtain the first price feature. The first source feature is determined based on the first route feature and the first price feature.

[0017] Furthermore, the route code in the second source information is input into the route transformation matrix to obtain the second route feature, the price code in the second source information is input into the price transformation matrix to obtain the second price feature, and the second source feature is determined based on the second route feature and the second price feature.

[0018] As an optional implementation, determining the first source characteristics based on the first route characteristics and the first price characteristics includes:

[0019] The first source of goods characteristic is determined based on the average of the first route characteristic and the first price characteristic;

[0020] Determining the second source characteristics based on the second route characteristics and the second price characteristics includes:

[0021] The second source characteristics are determined based on the average of the second route characteristics and the second price characteristics.

[0022] As an optional implementation, the route transformation matrix is ​​obtained in the following manner:

[0023] The initial model is trained by taking the N-1 route codes that were viewed earliest by the sample users in each route sequence sample as input, and the similarity between the output result and the Nth route code that was viewed latest by the sample users in the corresponding route sequence sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0024] The transformation matrix in the route model is determined as the route transformation matrix;

[0025] The price conversion matrix is ​​obtained as follows:

[0026] The initial model is trained by taking the N-1 price codes with the earliest viewing time of the sample users in each price series sample as input, and the similarity between the output result and the Nth price code with the latest viewing time of the sample users in the corresponding price series sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0027] The transformation matrix in the price model is determined as the price transformation matrix.

[0028] As an optional implementation, the route sequence sample is determined based on the route information and viewing time in the cargo information viewed by the same sample user. The route sequence sample contains N sample route codes, and the difference between the latest viewing time and the earliest viewing time among the N sample route codes does not exceed M days, where N and M are both positive integers.

[0029] The price sequence sample is determined based on the price information and viewing time of the source information viewed by the same sample user. The price sequence sample contains N price codes, and the difference between the latest viewing time and the earliest viewing time among the N price codes does not exceed M days, where N and M are both positive integers.

[0030] As an optional implementation, determining the user's user characteristics based on the second source of goods characteristics includes:

[0031] The user characteristics are determined based on the average value of the second source characteristics; or,

[0032] The second source information in the second source information set is divided into multiple subsets, wherein the operation type corresponding to the second source information in each subset is the same. The user characteristics are determined according to the weight corresponding to each operation type and the second source characteristics corresponding to the second source information in each subset.

[0033] As an optional implementation, determining the user characteristics based on the weights corresponding to the subsets and the second source characteristics corresponding to the second source information in different subsets includes:

[0034] Determine the average value of the second source characteristics corresponding to the second source information in each subset;

[0035] Based on the weights corresponding to each operation type, the average values ​​of the subsets corresponding to each operation type are weighted and summed to obtain the user characteristics.

[0036] As an optional implementation, pushing the first source information corresponding to the K first source features with the highest similarity to the user features to the user includes:

[0037] Determine the cosine distance between the first source of goods characteristic and the user characteristic;

[0038] The information on the first source of goods corresponding to the K first source of goods features with the smallest cosine distance is pushed to the user.

[0039] Secondly, an embodiment of the present invention provides a device for pushing goods source information, comprising a processor and a memory, wherein the memory is used to store a program executable by the processor, and the processor is used to read the program in the memory and perform the following steps:

[0040] Based on the last completed transportation task information of users with available tasks, a first set of cargo information matching the transportation information is determined from the cargo information to be accepted, and a second set of cargo information matching the transportation information is determined from the cargo information operated by the user.

[0041] Determine the first source feature corresponding to the first source information in the first source information set, and the second source feature corresponding to the second source information in the second source information set;

[0042] Based on the second source characteristics, the user characteristics of the user are determined, and the first source information corresponding to the K first source characteristics with the highest similarity to the user characteristics is pushed to the user, where K is a positive integer.

[0043] As an optional implementation, the processor is specifically configured to execute:

[0044] The route code in the first source of goods information is input into the route transformation matrix to obtain the first route feature, and the first route feature is determined as the first source of goods feature; and the route code in the second source of goods information is input into the route transformation matrix to obtain the second route feature, and the second route feature is determined as the second source of goods feature;

[0045] or,

[0046] The price code in the first source information is input into a price conversion matrix to obtain a first price feature, and the first price feature is determined as the first source feature; and the price code in the second source information is input into a price conversion matrix to obtain a second price feature, and the second price feature is determined as the second source feature;

[0047] or,

[0048] The route code in the first source information is input into the route conversion matrix to obtain the first route feature. The price code in the first source information is input into the price conversion matrix to obtain the first price feature. The first source feature is determined based on the first route feature and the first price feature.

[0049] Furthermore, the route code in the second source information is input into the route transformation matrix to obtain the second route feature, the price code in the second source information is input into the price transformation matrix to obtain the second price feature, and the second source feature is determined based on the second route feature and the second price feature.

[0050] As an optional implementation, the processor is specifically configured to execute:

[0051] The first source of goods characteristic is determined based on the average of the first route characteristic and the first price characteristic;

[0052] The second source characteristics are determined based on the average of the second route characteristics and the second price characteristics.

[0053] As an optional implementation, the processor is specifically configured to obtain the route transformation matrix in the following manner:

[0054] The initial model is trained by taking the N-1 route codes that were viewed earliest by the sample users in each route sequence sample as input, and the similarity between the output result and the Nth route code that was viewed latest by the sample users in the corresponding route sequence sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0055] The transformation matrix in the route model is determined as the route transformation matrix;

[0056] The processor is specifically configured to obtain the price conversion matrix in the following manner:

[0057] The initial model is trained by taking the N-1 price codes with the earliest viewing time of the sample users in each price series sample as input, and the similarity between the output result and the Nth price code with the latest viewing time of the sample users in the corresponding price series sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0058] The transformation matrix in the price model is determined as the price transformation matrix.

[0059] As an optional implementation method,

[0060] The route sequence sample is determined based on the route information and viewing time in the cargo information viewed by the same sample user. The route sequence sample contains N sample route codes, and the difference between the latest viewing time and the earliest viewing time among the N sample route codes does not exceed M days, where N and M are both positive integers.

[0061] The price sequence sample is determined based on the price information and viewing time of the source information viewed by the same sample user. The price sequence sample contains N price codes, and the difference between the latest viewing time and the earliest viewing time among the N price codes does not exceed M days, where N and M are both positive integers.

[0062] As an optional implementation, the processor is specifically configured to execute:

[0063] The user characteristics are determined based on the average value of the second source characteristics; or,

[0064] The second source information in the second source information set is divided into multiple subsets, wherein the operation type corresponding to the second source information in each subset is the same. The user characteristics are determined according to the weight corresponding to each operation type and the second source characteristics corresponding to the second source information in each subset.

[0065] As an optional implementation, the processor is specifically configured to execute:

[0066] Determine the average value of the second source characteristics corresponding to the second source information in each subset;

[0067] Based on the weights corresponding to each operation type, the average values ​​of the subsets corresponding to each operation type are weighted and summed to obtain the user characteristics.

[0068] As an optional implementation, the processor is specifically configured to execute:

[0069] Determine the cosine distance between the first source of goods characteristic and the user characteristic;

[0070] The information on the first source of goods corresponding to the K first source of goods features with the smallest cosine distance is pushed to the user.

[0071] Thirdly, embodiments of the present invention also provide a device for pushing out supply information, the device comprising:

[0072] The cargo source set determination unit is used to determine a first cargo source information set that matches the last transportation information completed by users with available tasks, from the cargo source information to be accepted, and to determine a second cargo source information set that matches the transportation information from the cargo source information operated by the user.

[0073] The cargo source feature determination unit is used to determine the first cargo source feature corresponding to the first cargo source information in the first cargo source information set, and the second cargo source feature corresponding to the second cargo source information in the second cargo source information set;

[0074] The source-determining unit is used to determine the user's user characteristics based on the second source characteristics, and to push the first source information corresponding to the K first source characteristics with the highest similarity to the user characteristics to the user, where K is a positive integer.

[0075] As an optional implementation, the unit for determining the characteristics of the goods is specifically used for:

[0076] The route code in the first source of goods information is input into the route transformation matrix to obtain the first route feature, and the first route feature is determined as the first source of goods feature; and the route code in the second source of goods information is input into the route transformation matrix to obtain the second route feature, and the second route feature is determined as the second source of goods feature;

[0077] or,

[0078] The price code in the first source information is input into a price conversion matrix to obtain a first price feature, and the first price feature is determined as the first source feature; and the price code in the second source information is input into a price conversion matrix to obtain a second price feature, and the second price feature is determined as the second source feature;

[0079] or,

[0080] The route code in the first source information is input into the route conversion matrix to obtain the first route feature. The price code in the first source information is input into the price conversion matrix to obtain the first price feature. The first source feature is determined based on the first route feature and the first price feature.

[0081] Furthermore, the route code in the second source information is input into the route transformation matrix to obtain the second route feature, the price code in the second source information is input into the price transformation matrix to obtain the second price feature, and the second source feature is determined based on the second route feature and the second price feature.

[0082] As an optional implementation, the unit for determining the characteristics of the goods is specifically used for:

[0083] The first source of goods characteristic is determined based on the average of the first route characteristic and the first price characteristic;

[0084] The second source characteristics are determined based on the average of the second route characteristics and the second price characteristics.

[0085] As an optional implementation, the cargo source characteristic determination unit is specifically used to obtain the route transformation matrix in the following manner:

[0086] The initial model is trained by taking the N-1 route codes that were viewed earliest by the sample users in each route sequence sample as input, and the similarity between the output result and the Nth route code that was viewed latest by the sample users in the corresponding route sequence sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0087] The transformation matrix in the route model is determined as the route transformation matrix;

[0088] The unit for determining the source characteristics is specifically used to obtain the price conversion matrix in the following manner:

[0089] The initial model is trained by taking the N-1 price codes with the earliest viewing time of the sample users in each price series sample as input, and the similarity between the output result and the Nth price code with the latest viewing time of the sample users in the corresponding price series sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0090] The transformation matrix in the price model is determined as the price transformation matrix.

[0091] As an optional implementation method,

[0092] The route sequence sample is determined based on the route information and viewing time in the cargo information viewed by the same sample user. The route sequence sample contains N sample route codes, and the difference between the latest viewing time and the earliest viewing time among the N sample route codes does not exceed M days, where N and M are both positive integers.

[0093] The price sequence sample is determined based on the price information and viewing time of the source information viewed by the same sample user. The price sequence sample contains N price codes, and the difference between the latest viewing time and the earliest viewing time among the N price codes does not exceed M days, where N and M are both positive integers.

[0094] As an optional implementation, the unit for determining the delivery source is specifically used for:

[0095] The user characteristics are determined based on the average value of the second source characteristics; or,

[0096] The second source information in the second source information set is divided into multiple subsets, wherein the operation type corresponding to the second source information in each subset is the same. The user characteristics are determined according to the weight corresponding to each operation type and the second source characteristics corresponding to the second source information in each subset.

[0097] As an optional implementation, the unit for determining the delivery source is specifically used for:

[0098] Determine the average value of the second source characteristics corresponding to the second source information in each subset;

[0099] Based on the weights corresponding to each operation type, the average values ​​of the subsets corresponding to each operation type are weighted and summed to obtain the user characteristics.

[0100] As an optional implementation, the unit for determining the delivery source is specifically used for:

[0101] Determine the cosine distance between the first source of goods characteristic and the user characteristic;

[0102] The information on the first source of goods corresponding to the K first source of goods features with the smallest cosine distance is pushed to the user.

[0103] Fourthly, embodiments of the present invention also provide a computer storage medium having a computer program stored thereon, which, when executed by a processor, is used to implement the steps of the method described in the first aspect above.

[0104] These or other aspects of this application will become more apparent in the following description of embodiments. Attached Figure Description

[0105] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0106] Figure 1 A flowchart illustrating the implementation of a method for pushing goods source information according to an embodiment of the present invention;

[0107] Figure 2 A schematic diagram of the structure of an initial model provided in an embodiment of the present invention;

[0108] Figure 3 A flowchart illustrating a specific implementation of a method for pushing goods source information according to an embodiment of the present invention;

[0109] Figure 4 This is a schematic diagram of a device for pushing out product information, provided in an embodiment of the present invention.

[0110] Figure 5 This is a schematic diagram of a device for pushing out product information, provided in an embodiment of the present invention. Detailed Implementation

[0111] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0112] In this embodiment of the invention, the term "and / or" describes the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following associated objects have an "or" relationship.

[0113] The application scenarios described in the embodiments of this invention are for the purpose of more clearly illustrating the technical solutions of the embodiments of this invention, and do not constitute a limitation on the technical solutions provided by the embodiments of this invention. Those skilled in the art will understand that with the emergence of new application scenarios, the technical solutions provided by the embodiments of this invention are also applicable to similar technical problems. In the description of this invention, unless otherwise stated, "multiple" means two or more.

[0114] In the logistics sector, especially in long-haul trucking, shippers (or dispatchers) need to post cargo information if they wish to send their goods to a designated location. Drivers (or users), upon seeing this information, can contact the shipper to reach an agreement if they wish to accept the freight task listed. However, this method is inefficient. Therefore, platforms (or third parties) have emerged to help shippers and drivers complete cargo order placements. These platforms collect, aggregate, and post cargo information, saving shippers time searching for posting channels and drivers time, thus increasing the speed of order placement.

[0115] Currently, the resources available for displaying freight information on the platform are limited, and drivers cannot possibly browse all of them. This necessitates the platform pushing freight information to drivers. However, if the pushed freight information is unsuitable for drivers, it will only worsen their experience and waste their time. Therefore, how to push freight information based on drivers' needs and characteristics has become an urgent problem to be solved.

[0116] Example 1: This embodiment of the invention provides a method for pushing cargo information. The core idea is to extract user characteristics of different users within a historical period, thereby filtering cargo information similar to the user's user characteristics from the cargo information to be accepted, and then pushing it to the user. Since the feature information of cargo information that the user has operated on in a historical period and the cargo features of cargo information to be accepted are extracted, the cargo information corresponding to a preset number of cargo features most similar to the user's characteristics is pushed to the user, so that the cargo information pushed to the user is the cargo information of the cargo that the user is most likely to want to accept. This embodiment can realize efficient and personalized delivery of cargo information to users according to their needs and characteristics, effectively improving the user experience.

[0117] like Figure 1 As shown in the figure, this embodiment provides a method for pushing goods source information. The implementation process of this method is as follows:

[0118] Step 100: Based on the last completed transportation task information of users with idle tasks, determine a first set of cargo information matching the transportation information from the cargo information to be accepted, and determine a second set of cargo information matching the transportation information from the cargo information operated by the user.

[0119] It should be noted that the users in this embodiment include, but are not limited to, carriers, shippers, and other entities that have a need to view cargo information. In practice, a user may be only a carrier, or only a shipper, or may have a dual identity, being both a user and a shipper. This embodiment does not impose too many restrictions on the scope of the objects included as users.

[0120] In some embodiments, the cargo information to be accepted is cargo information that has not yet reached the loading time (or shipping announcement time) and there is no cargo information accepted by users.

[0121] In some embodiments, the transport information in this embodiment includes, but is not limited to, some or all of the user's registered vehicle model, the unloading location where the last transport task was completed, and the unloading time.

[0122] In some embodiments, the cargo information in this embodiment includes, but is not limited to, some or all of the following: route information, loading location information, unloading location information, loading time information, unloading time information, required vehicle type information, cargo price information, and other information related to the freight task.

[0123] In this embodiment, the first source information set includes one or more first source information items, and the second source information set includes one or more second source information items. It should be noted that the terms "first" and "second" in the first and second source information sets are used only to distinguish the source information. The information categories contained in the first and second source information are the same as those contained in the aforementioned source information. The first and second source information may contain the same information content or different information content; this embodiment does not impose excessive limitations on this. The terms "first" and "second" in the first and second source information sets are used only to distinguish the source information sets.

[0124] In some embodiments, the first resource information in this embodiment includes some or all of the following: route information, loading location information, unloading location information, loading time information, unloading time information, required vehicle type information, cargo price information, and other information related to the freight task. It may also be information encoded from the aforementioned information. Similarly, the second resource information in this embodiment includes some or all of the following: route information, loading location information, unloading location information, loading time information, unloading time information, required vehicle type information, cargo price information, and other information related to the freight task. It may also be information encoded from the aforementioned information. Optionally, the first cargo information includes at least one of a route code obtained by encoding the route information and a price code obtained by encoding the price information. Optionally, the first cargo information includes a route code, a price code, required vehicle type information, and loading location information.

[0125] Optionally, the methods for encoding route information and cargo price information include, but are not limited to, unique hot coding.

[0126] In practice, when a user has no pending transportation tasks, it means that the driver can currently accept transportation tasks from the new cargo information. At this time, cargo information can be pushed to the driver. First, based on the user's registered vehicle model, the unloading location and unloading time of the last completed transportation task, the first matching cargo information can be selected from the pending cargo information to form a first cargo information set. For example, the first cargo information can be selected that is the same as the user's registered vehicle model, the unloading location of the current transportation task is the same as the loading location in the first cargo information, and the unloading time of the current transportation task is earlier than the loading time in the first cargo information. Then, the selected first cargo information is combined into a first cargo information set.

[0127] In practice, users on logistics platforms engage in various behaviors related to cargo information. In this embodiment, these behaviors include, but are not limited to, at least one of viewing, order-grabbing, and transportation operations. For example, a user can view cargo information, transport cargo information, or grab orders for cargo information. This cargo information represents the cargo information that the user has interacted with. This embodiment can find cargo information corresponding to user actions based on these actions and compile this cargo information into a filter set. For instance, it can compile cargo information viewed, ordered, and transported by the user within a historical timeframe into a filter set, and further filter from this cargo information to obtain a second set of cargo information matching the transportation information.

[0128] In some embodiments, since the cargo information operated by the user may be different from the unloading location where the user finally completed the transportation task, in order to push cargo information to the user more accurately, this embodiment further filters out cargo information whose loading location is the same as the unloading location in the carrier information where the transportation task was finally completed from the cargo information operated by the user, and forms a second cargo information set from these cargo information.

[0129] In implementation, this embodiment filters out the first type of cargo information where the unloading location in the last completed transportation task by a user with available tasks is the same as the loading location in the cargo information to be accepted, and the unloading time in the carrier information is later than the loading time. It also filters out the second type of cargo information where the unloading location in the carrier information is the same as the loading location in the cargo information operated by the user. Thus, the first cargo feature of the first cargo information (cargo information to be accepted) is extracted, and the second cargo feature of the second cargo information (cargo information generated by the user's actions) is extracted. By comparing the first and second cargo features, cargo information to be accepted that better matches the user's actions is selected.

[0130] Step 101: Determine the first source feature corresponding to the first source information in the first source information set, and the second source feature corresponding to the second source information in the second source information set;

[0131] In some embodiments, this embodiment extracts the first source feature of the first source information and the second source feature of the second source information in any of the following ways:

[0132] Method 1) The cargo information includes route codes.

[0133] The route code in the first source of goods information is input into the route transformation matrix to obtain the first route feature, and the first route feature is determined as the first source of goods feature;

[0134] The route code in the second source information is input into the route transformation matrix to obtain the second route feature, and the second route feature is determined as the second source feature.

[0135] Method 2) The source information includes a price code.

[0136] Input the price code in the first source information into the price conversion matrix to obtain the first price feature, and determine the first price feature as the first source feature;

[0137] Input the price code in the second source information into the price conversion matrix to obtain the second price feature, and determine the second price feature as the second source feature.

[0138] Method 3) The source information includes route code and price code.

[0139] The route code in the first source information is input into the route conversion matrix to obtain the first route feature, and the price code in the first source information is input into the price conversion matrix to obtain the first price feature;

[0140] The first source of goods characteristic is determined based on the first route characteristic and the first price characteristic; optionally, the first source of goods characteristic is determined based on the average of the first route characteristic and the first price characteristic.

[0141] Input the route code from the second source information into the route transformation matrix to obtain the second route feature; input the price code from the second source information into the price transformation matrix to obtain the second price feature.

[0142] The second source characteristics are determined based on the second route characteristics and the second price characteristics. Optionally, the second source characteristics are determined based on the average of the second route characteristics and the second price characteristics.

[0143] In implementation, route codes can be obtained by encoding route information in cargo information, and price codes can be obtained by encoding price information. This facilitates the calculation of route characteristics corresponding to route codes and price characteristics corresponding to price codes. It also facilitates the evaluation of the characteristics of routes to be accepted and the characteristics of routes that users have already transported and / or expect to transport from the perspective of transportation routes, thereby finding cargo information corresponding to more matching routes to recommend to users. Similarly, it can evaluate the characteristics of prices to be accepted and the characteristics of prices that users have already transported and / or expect to transport from the perspective of transportation prices, thereby finding cargo information corresponding to more matching prices to recommend to users. Furthermore, it can comprehensively consider whether route characteristics and price characteristics meet the needs and characteristics of users, thereby making cargo information recommendations more accurate.

[0144] In some embodiments, the route code and price code in this embodiment can be understood as a vector or matrix. In practice, the product of the route code and the route transformation matrix (which is also a vector or matrix) can be determined as the output route feature. Similarly, the product of the price code and the price transformation matrix (which is also a vector or matrix) can be determined as the output price feature.

[0145] In some embodiments, the vector dimensions of the route features and price features in this embodiment are the same.

[0146] In some embodiments, to determine the route transformation matrix and the price transformation matrix, training samples can first be obtained through the following steps, including at least one of route sequence samples and price sequence samples:

[0147] Step (1): Users who view the product information more than P times within a set time period can be selected and formed into the first set;

[0148] Specifically, the time period can be set in advance, such as within six months prior to the current time. The value of P can also be preset, such as setting P=5. Since the platform can record relevant data such as users' historical viewing behavior of supply information, after determining the set time period and the value of P, the platform can filter users who have viewed supply information more than P times within the set time period from historical data and use them as sample users. For example, users who have viewed supply information more than 5 times in the past six months can be grouped into a first set. It should be noted that the same supply information can be viewed by different users. That is, one supply information can correspond to one user or multiple users. This embodiment does not impose too many limitations on this.

[0149] Step (2): Based on the source information viewed by each user in the first set, generate multiple sequence samples. This can be understood as each user corresponding to multiple sequence samples. Each sequence sample contains N source information. The source information in each sequence sample is arranged in the order of viewing time. The number of source information in each sequence sample is fixed and the same. For example, each sequence sample contains N source information. These N source information are arranged in the order of viewing time from early to late. The difference between the viewing time of the first source information and the Nth source information in each sequence sample does not exceed M days.

[0150] Optionally, the sequence samples in this embodiment include, but are not limited to, at least one of route sequence samples and price sequence samples. The route sequence samples are determined based on the route codes in the cargo information viewed by each user, and the price sequence samples are determined based on the price codes in the cargo information viewed by each user.

[0151] It should be noted that the source information contained in any sequence sample can be the same or different, and this embodiment does not impose too many restrictions on this.

[0152] To facilitate understanding, the following example illustrates how to generate multiple sequence samples based on the source information and corresponding viewing time in the first set:

[0153] Assume N=5, M=7, and there are currently 2 users in the first set, user A and user B;

[0154] From user A's viewing history, sorted by the time of viewing the source information, it can be seen that he viewed a total of 7 source information: source 1, source 2, ..., source 6, and source 7. In practice, it is possible to obtain all of user A's viewing history, or to obtain user A's viewing history within a certain period of time, such as user A's viewing history within the same set time period mentioned above. This method can also be used to obtain the viewing history of other users.

[0155] During implementation, User A viewed Source 1 on August 5, 2021, Sources 2 through 5 on August 6, 2021, Source 6 on August 11, 2021, and Source 7 on August 12, 2021.

[0156] Since N=5, a sequence sample needs to contain 5 consecutively viewed source information entries, and the time difference between viewing the first and fifth source information entries in a sequence sample cannot exceed 7 days. Therefore, source 1 to source 7 can form 3 sequence samples: source 1 to source 5 can be considered as one sequence sample (sequence sample 1), source 2 to source 6 can be considered as one sequence sample (sequence sample 2), and source 3 to source 7 can be considered as one sequence sample (sequence sample 3). Thus, based on user A's viewing history, 3 sequence samples can be obtained. User A corresponds to 3 sequence samples.

[0157] Similarly, from User B's viewing history, sorted by the time of viewing the source information, he viewed a total of 6 source information: source 8, source 9, ..., source 12, and source 13. In practice, User B viewed source 8 to source 9 on August 6, 2021, source 10 to source 11 on August 7, 2021, source 12 on August 8, 2021, and source 13 on August 20, 2021.

[0158] Since N=5, a sequence sample needs to contain 5 consecutively viewed source information items, and the time difference between viewing the first and fifth source information items in a sequence sample cannot exceed 7 days. Therefore, source items 8 to 13 can form 1 sequence sample, that is, source items 8 to 12 can be used as a sequence sample (i.e., sequence sample 4). Although source items 9 to 13 can meet the requirement for the number of source information items in a sequence sample, since the time difference between user B viewing source item 9 and user B viewing source item 13 exceeds 7 days, source items 9 to 13 cannot be used as a sequence sample.

[0159] Using the above method, this embodiment can acquire users who have viewed cargo information P times within a preset time period and form a first set. Then, based on the cargo information viewed by each user, multiple sequence samples are generated. In practice, based on the route code in the cargo information, multiple route sequence samples can be generated for each user; based on the price code in the cargo information, multiple price sequence samples can be generated for each user. In this embodiment, the generation of route sequence samples and price sequence samples corresponds to the same first set; that is, the first set is first acquired, and then corresponding route sequence samples and / or price sequence samples are generated based on the cargo information corresponding to each user in the first set.

[0160] Secondly, after obtaining route sequence samples and price sequence samples in the above manner, the initial model is trained using the obtained training samples to obtain the route conversion matrix and price conversion matrix, respectively.

[0161] In some embodiments, the route transformation matrix is ​​obtained in the following manner:

[0162] The initial model is trained by taking the N-1 route codes with the earliest viewing time of the sample users in each route sequence sample as input, and the similarity between the output result and the Nth route code with the latest viewing time of the sample users in the corresponding route sequence sample as the optimization condition, so as to obtain the trained route model; the transformation matrix in the route model is determined as the route transformation matrix.

[0163] In some embodiments, such as Figure 2 As shown, the initial model in this embodiment includes a transformation matrix 200, a summation layer 201, and a restoration layer 202. The route information (e.g., location 1-location 2, location 3-location 4, etc.) in the cargo source information can be pre-encoded to obtain a route code (vector).

[0164] The following example, using route sequence samples including cargo sources 1-5, illustrates the process of training the initial model:

[0165] In the route sequence sample, the route information for sources 1, 2, 3, 4, and 5 is arranged sequentially from front to back according to the viewing time, corresponding to route codes 1, 2, 3, 4, and 5 respectively. Route codes 1 through 4 are then input into the transformation matrix, i.e., multiplied by the transformation matrix respectively, resulting in four product vectors; the vectors in the route codes have the same dimension.

[0166] The four product vectors are input into the summation layer for summation. The resulting summation vector is then input into the decomposition layer to convert the dimension of the summation vector. There are many methods for using the decomposition layer to reduce the dimension of the vector, such as multiplying with a specific matrix. Finally, the output of the decomposition layer is compared with route code 5 for similarity. Based on the comparison results, the elements in the transformation matrix are continuously adjusted to improve the similarity between the output and route code 5 and to increase the number of outputs that are identical to route code 5. When the optimization conditions are met, training stops, and the trained route model is obtained.

[0167] In some embodiments, the route sequence sample is determined based on the route information and viewing time in the cargo information viewed by the same sample user. The route sequence sample contains N sample route codes, and the difference between the latest viewing time and the earliest viewing time among the N sample route codes does not exceed M days, where N and M are both positive integers.

[0168] In some embodiments, the price conversion matrix is ​​obtained in the following manner:

[0169] The initial model is trained by taking the N-1 earliest price codes viewed by sample users in each price series sample as input, and the similarity between the output result and the Nth price code viewed by sample users in the corresponding price series sample as optimization condition, to obtain the trained price model; the transformation matrix in the price model is determined as the price transformation matrix.

[0170] The initial model includes a transformation matrix, a summation layer, and a restoration layer. It should be noted that the process of training the initial model using price series samples and the process of training the initial model using route series samples in this embodiment are based on the same principle.

[0171] The following example, using price series samples including source 1 to source 5, illustrates the process of training the initial model:

[0172] The price sequence sample is arranged in chronological order of viewing time, from front to back, with the price information of source 1, source 2, source 3, source 4, and source 5 corresponding to price codes 1, 2, 3, 4, and 5 respectively. Price codes 1 to 4 are then input into the transformation matrix, i.e., multiplied by the transformation matrix respectively, resulting in four product vectors; the vectors in the price codes have the same dimension.

[0173] The four product vectors are input into the summation layer for summation. The resulting summation vector is then input into the decomposition layer to transform the dimension of the summation vector. There are many methods for using the decomposition layer to reduce the dimension of the vector, such as multiplying with a specific matrix. Finally, the output of the decomposition layer is compared with the price code 5 for similarity. Based on the comparison results, the elements in the transformation matrix are continuously adjusted to improve the similarity between the output and the price code 5 and to increase the number of outputs that are identical to the price code 5. When the optimization conditions are met, training stops, and the trained price model is obtained.

[0174] In some embodiments, the price sequence sample is determined based on the price information in the source information viewed by the same sample user and the viewing time. The price sequence sample contains N price codes, and the difference between the latest viewing time and the earliest viewing time among the N price codes does not exceed M days, where N and M are both positive integers.

[0175] Since the source information contains price-related information, the price can be vectorized. In implementation, the corresponding vector can be generated based on the price using existing methods. This implementation provides a vectorization method, such as setting intervals, with each interval corresponding to a code (the code can use existing methods, such as unique hot codes). Then, the corresponding price code is obtained according to the interval in which the price is located. Optionally, for example, the interval can be set with 100 intervals between prices from 0 to 20,000 yuan, and a corresponding unique hot code can be set for each interval. When the price in the source information falls into a certain interval, the unique hot code corresponding to this interval can be used as the price code.

[0176] It should be noted that during the initial model training process, the summation layer and the reduction layer are fixed, and the training objective is achieved by adjusting the transformation matrix.

[0177] Step 102: Determine the user's user characteristics based on the second source characteristics, and push the first source information corresponding to the K first source characteristics with the highest similarity to the user characteristics to the user, where K is a positive integer.

[0178] In some embodiments, after determining the second source of goods characteristics, this embodiment can determine the user characteristics based on the second source of goods characteristics, specifically through any of the following methods:

[0179] Method a: Determine the user characteristics based on the average value of the second source characteristics;

[0180] In practice, the second source information set includes one or more second source information pieces. Each second source information piece corresponds to a second source feature. When the second source information set includes one second source information piece, the second source feature of that second source information piece is determined as the user feature. When the second source information set includes multiple second source information pieces, the second source features corresponding to the multiple second source information pieces are averaged, and the average value is determined as the user feature.

[0181] Method b: Divide the second source information in the second source information set into multiple subsets, wherein the operation type corresponding to the second source information in each subset is the same, and determine the user characteristics based on the weight corresponding to each operation type and the second source characteristics corresponding to the second source information in each subset.

[0182] In some embodiments, the operation type includes, but is not limited to, at least one of viewing type, order grabbing type, and carrier type.

[0183] In implementation, due to different user actions on cargo information within a historical period, second cargo information with the same operation type in the second cargo information set is divided into the same subset based on the operation type. For example, the second cargo information set is split into three subsets. The operation type corresponding to the second cargo information in the first subset is viewing, meaning that the second cargo information in this subset has been viewed by the user. The operation type corresponding to the second cargo information in the second subset is order grabbing, meaning that the second cargo information in this subset has been order grabbing by the user. The operation type corresponding to the second cargo information in the third subset is transportation, meaning that the second cargo information in this subset has been transportation by the user.

[0184] In some embodiments, the user characteristics are determined based on the weights corresponding to the subsets and the second source characteristics corresponding to the second source information in different subsets, in the following manner:

[0185] First, determine the average value of the second source characteristics corresponding to the second source information in each subset; where the second source characteristics are understood as a vector or matrix, and the average value is also understood as a vector or matrix.

[0186] Secondly, based on the weights corresponding to each operation type, the average values ​​of the subsets corresponding to each operation type are weighted and summed to obtain the user features. The user features can also be understood as a vector or matrix.

[0187] In practice, the weights for viewing type, order-grabbing type, and transportation type are pre-set as a, b, and c, respectively. Then, the second source information in the subset corresponding to the viewing type has been viewed by the user, and the average value of this subset is A. The second source information in the subset corresponding to the order-grabbing type has been ordered by the user, and the average value of this subset is B. The second source information in the subset corresponding to the transportation type has been transported by the user, and the average value of this subset is C. Therefore, the user characteristic of this user is a×A+b×B+c×C.

[0188] In some embodiments, this embodiment pushes the first source information corresponding to the K first source features with the highest similarity to the user features to the user in the following manner:

[0189] Determine the cosine distance between the first source feature and the user feature; wherein, since the first source feature and the user feature are two vectors with the same dimension, the similarity between the two vectors is determined by calculating the cosine distance between the two vectors.

[0190] The information of the first source of goods corresponding to the K first source features with the smallest cosine distance is pushed to the user. The smaller the cosine distance between two vectors, the greater the similarity between the two vectors.

[0191] like Figure 3 As shown in the figure, the specific implementation process of the method for pushing goods source information provided in this embodiment is as follows:

[0192] Step 300: Obtain the registered vehicle model, unloading location, and unloading time of the user who is currently available to complete the transportation task.

[0193] Step 301: Determine the first source of cargo information that matches the registered vehicle model, unloading location, and unloading time from the cargo information to be accepted, and form a first source of cargo information set;

[0194] Step 302: Determine the second cargo information that matches the registered vehicle model, unloading location, and unloading time from the cargo information operated by the user, and form a second cargo information set;

[0195] Step 303: Input the route code in the first source information into the route conversion matrix to obtain the first route feature; input the price code in the first source information into the price conversion matrix to obtain the first price feature; determine the first source feature based on the average of the first route feature and the first price feature.

[0196] Step 304: Input the route code in the second source information into the route conversion matrix to obtain the second route feature; input the price code in the second source information into the price conversion matrix to obtain the second price feature; determine the second source feature based on the average of the second route feature and the second price feature.

[0197] Step 305: Divide the second source information in the second source information set into multiple subsets, where the operation type corresponding to the second source information in each subset is the same. Determine the average value of the second source features corresponding to the second source information in each subset. Based on the weights corresponding to each operation type, perform a weighted summation of the average values ​​of the subsets corresponding to each operation type to obtain the user features.

[0198] Step 306: Determine the cosine distance between the first source of goods feature and the user feature, and push the first source of goods information corresponding to the K first source of goods features with the smallest cosine distance to the user.

[0199] This embodiment can push commodity information based on user behavior, thereby allowing users to see the commodity information they wish to accept, improving user experience, and reducing the time spent by users searching for commodity information and the system resources consumed.

[0200] Example 2: Based on the same inventive concept, this embodiment of the invention also provides a device for pushing supply information. Since this device is the same as the device in the method of this embodiment of the invention, and the principle of the device in solving the problem is similar to that of the method, the implementation of this device can refer to the implementation of the method, and the repeated parts will not be described again.

[0201] like Figure 4 As shown, the device includes a processor 400 and a memory 401. The memory 401 stores programs executable by the processor 400. The processor 400 reads the programs from the memory 401 and performs the following steps:

[0202] The processor 400 includes a processor 400 and a memory 401. The memory 401 stores programs executable by the processor 400. The processor 400 reads the programs from the memory 401 and performs the following steps:

[0203] Based on the last completed transportation task information of users with available tasks, a first set of cargo information matching the transportation information is determined from the cargo information to be accepted, and a second set of cargo information matching the transportation information is determined from the cargo information operated by the user.

[0204] Determine the first source feature corresponding to the first source information in the first source information set, and the second source feature corresponding to the second source information in the second source information set;

[0205] Based on the second source characteristics, the user characteristics of the user are determined, and the first source information corresponding to the K first source characteristics with the highest similarity to the user characteristics is pushed to the user, where K is a positive integer.

[0206] As an optional implementation, the processor 400 is specifically configured to perform:

[0207] The route code in the first source of goods information is input into the route transformation matrix to obtain the first route feature, and the first route feature is determined as the first source of goods feature; and the route code in the second source of goods information is input into the route transformation matrix to obtain the second route feature, and the second route feature is determined as the second source of goods feature;

[0208] or,

[0209] The price code in the first source information is input into a price conversion matrix to obtain a first price feature, and the first price feature is determined as the first source feature; and the price code in the second source information is input into a price conversion matrix to obtain a second price feature, and the second price feature is determined as the second source feature;

[0210] or,

[0211] The route code in the first source information is input into the route conversion matrix to obtain the first route feature. The price code in the first source information is input into the price conversion matrix to obtain the first price feature. The first source feature is determined based on the first route feature and the first price feature.

[0212] Furthermore, the route code in the second source information is input into the route transformation matrix to obtain the second route feature, the price code in the second source information is input into the price transformation matrix to obtain the second price feature, and the second source feature is determined based on the second route feature and the second price feature.

[0213] As an optional implementation, the processor 400 is specifically configured to perform:

[0214] The first source of goods characteristic is determined based on the average of the first route characteristic and the first price characteristic;

[0215] The second source characteristics are determined based on the average of the second route characteristics and the second price characteristics.

[0216] As an optional implementation, the processor 400 is specifically configured to obtain the route transformation matrix in the following manner:

[0217] The initial model is trained by taking the N-1 route codes that were viewed earliest by the sample users in each route sequence sample as input, and the similarity between the output result and the Nth route code that was viewed latest by the sample users in the corresponding route sequence sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0218] The transformation matrix in the route model is determined as the route transformation matrix;

[0219] The processor 400 is specifically configured to obtain the price conversion matrix in the following manner:

[0220] The initial model is trained by taking the N-1 price codes with the earliest viewing time of the sample users in each price series sample as input, and the similarity between the output result and the Nth price code with the latest viewing time of the sample users in the corresponding price series sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0221] The transformation matrix in the price model is determined as the price transformation matrix.

[0222] As an optional implementation method,

[0223] The route sequence sample is determined based on the route information and viewing time in the cargo information viewed by the same sample user. The route sequence sample contains N sample route codes, and the difference between the latest viewing time and the earliest viewing time among the N sample route codes does not exceed M days, where N and M are both positive integers.

[0224] The price sequence sample is determined based on the price information and viewing time of the source information viewed by the same sample user. The price sequence sample contains N price codes, and the difference between the latest viewing time and the earliest viewing time among the N price codes does not exceed M days, where N and M are both positive integers.

[0225] As an optional implementation, the processor 400 is specifically configured to perform:

[0226] The user characteristics are determined based on the average value of the second source characteristics; or,

[0227] The second source information in the second source information set is divided into multiple subsets, wherein the operation type corresponding to the second source information in each subset is the same. The user characteristics are determined according to the weight corresponding to each operation type and the second source characteristics corresponding to the second source information in each subset.

[0228] As an optional implementation, the processor 400 is specifically configured to perform:

[0229] Determine the average value of the second source characteristics corresponding to the second source information in each subset;

[0230] Based on the weights corresponding to each operation type, the average values ​​of the subsets corresponding to each operation type are weighted and summed to obtain the user characteristics.

[0231] As an optional implementation, the processor 400 is specifically configured to perform:

[0232] Determine the cosine distance between the first source of goods characteristic and the user characteristic;

[0233] The information on the first source of goods corresponding to the K first source of goods features with the smallest cosine distance is pushed to the user.

[0234] Example 3: Based on the same inventive concept, this embodiment of the invention also provides a device for pushing out source information. Since this device is the same as the device in the method of this embodiment of the invention, and the principle of the device in solving the problem is similar to that of the method, the implementation of this device can refer to the implementation of the method, and the repeated parts will not be described again.

[0235] like Figure 5 As shown, the device includes:

[0236] The cargo collection unit 500 is used to determine a first cargo information set that matches the last transportation information of the user with idle tasks from the cargo information to be accepted, and to determine a second cargo information set that matches the transportation information from the cargo information operated by the user.

[0237] The cargo source feature determination unit 501 is used to determine the first cargo source feature corresponding to the first cargo source information in the first cargo source information set, and the second cargo source feature corresponding to the second cargo source information in the second cargo source information set;

[0238] The source determination unit 502 is used to determine the user characteristics of the user based on the second source characteristics, and push the first source information corresponding to the K first source characteristics with the highest similarity to the user characteristics to the user, wherein K is a positive integer.

[0239] As an optional implementation, the unit 501 for determining the characteristics of the goods source is specifically used for:

[0240] The route code in the first source of goods information is input into the route transformation matrix to obtain the first route feature, and the first route feature is determined as the first source of goods feature; and the route code in the second source of goods information is input into the route transformation matrix to obtain the second route feature, and the second route feature is determined as the second source of goods feature;

[0241] or,

[0242] The price code in the first source information is input into a price conversion matrix to obtain a first price feature, and the first price feature is determined as the first source feature; and the price code in the second source information is input into a price conversion matrix to obtain a second price feature, and the second price feature is determined as the second source feature;

[0243] or,

[0244] The route code in the first source information is input into the route conversion matrix to obtain the first route feature. The price code in the first source information is input into the price conversion matrix to obtain the first price feature. The first source feature is determined based on the first route feature and the first price feature.

[0245] Furthermore, the route code in the second source information is input into the route transformation matrix to obtain the second route feature, the price code in the second source information is input into the price transformation matrix to obtain the second price feature, and the second source feature is determined based on the second route feature and the second price feature.

[0246] As an optional implementation, the unit 501 for determining the characteristics of the goods source is specifically used for:

[0247] The first source of goods characteristic is determined based on the average of the first route characteristic and the first price characteristic;

[0248] The second source characteristics are determined based on the average of the second route characteristics and the second price characteristics.

[0249] As an optional implementation, the cargo source characteristic determination unit 501 is specifically used to obtain the route transformation matrix in the following manner:

[0250] The initial model is trained by taking the N-1 route codes that were viewed earliest by the sample users in each route sequence sample as input, and the similarity between the output result and the Nth route code that was viewed latest by the sample users in the corresponding route sequence sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0251] The transformation matrix in the route model is determined as the route transformation matrix;

[0252] The source characteristic determination unit 501 is specifically used to obtain the price conversion matrix in the following manner:

[0253] The initial model is trained by taking the N-1 price codes with the earliest viewing time of the sample users in each price series sample as input, and the similarity between the output result and the Nth price code with the latest viewing time of the sample users in the corresponding price series sample as the optimization condition. The initial model includes a transformation matrix, a summation layer and a restoration layer.

[0254] The transformation matrix in the price model is determined as the price transformation matrix.

[0255] As an optional implementation method,

[0256] The route sequence sample is determined based on the route information and viewing time in the cargo information viewed by the same sample user. The route sequence sample contains N sample route codes, and the difference between the latest viewing time and the earliest viewing time among the N sample route codes does not exceed M days, where N and M are both positive integers.

[0257] The price sequence sample is determined based on the price information and viewing time of the source information viewed by the same sample user. The price sequence sample contains N price codes, and the difference between the latest viewing time and the earliest viewing time among the N price codes does not exceed M days, where N and M are both positive integers.

[0258] As an optional implementation, the unit 502 for determining the delivery source is specifically used for:

[0259] The user characteristics are determined based on the average value of the second source characteristics; or,

[0260] The second source information in the second source information set is divided into multiple subsets, wherein the operation type corresponding to the second source information in each subset is the same. The user characteristics are determined according to the weight corresponding to each operation type and the second source characteristics corresponding to the second source information in each subset.

[0261] As an optional implementation, the unit 502 for determining the delivery source is specifically used for:

[0262] Determine the average value of the second source characteristics corresponding to the second source information in each subset;

[0263] Based on the weights corresponding to each operation type, the average values ​​of the subsets corresponding to each operation type are weighted and summed to obtain the user characteristics.

[0264] As an optional implementation, the unit 502 for determining the delivery source is specifically used for:

[0265] Determine the cosine distance between the first source of goods characteristic and the user characteristic;

[0266] The information on the first source of goods corresponding to the K first source of goods features with the smallest cosine distance is pushed to the user.

[0267] Based on the same inventive concept, embodiments of the present invention also provide a computer storage medium storing a computer program, which, when executed by a processor, performs the following steps:

[0268] Based on the last completed transportation task information of users with available tasks, a first set of cargo information matching the transportation information is determined from the cargo information to be accepted, and a second set of cargo information matching the transportation information is determined from the cargo information operated by the user.

[0269] Determine the first source feature corresponding to the first source information in the first source information set, and the second source feature corresponding to the second source information in the second source information set;

[0270] Based on the second source characteristics, the user characteristics of the user are determined, and the first source information corresponding to the K first source characteristics with the highest similarity to the user characteristics is pushed to the user, where K is a positive integer.

[0271] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention 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 and optical storage) containing computer-usable program code.

[0272] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will 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 and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 Devices that specify the functions in one or more boxes.

[0273] 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 an instruction device, which is implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0274] 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.

[0275] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for pushing information of a source of goods, characterized in that, The method comprises: According to the last completed transportation task of the idle user, the first set of source information is determined to match the transportation information, and the second set of source information is determined to match the transportation information from the user's operation source information; the transportation information includes the user's registered vehicle type, the last completed transportation task unloading location and unloading time; the source information to be taken is the source information that has not arrived at the loading time and has not been taken by the user; the source information includes route information, loading location information, unloading location information, loading time information, unloading time information, required transportation vehicle type information, source price information, and other information related to freight transportation task; the operation of the source information includes viewing operation, grabbing operation, and transportation operation; Determine the first source feature corresponding to the first source information in the first source information set, and the second source feature corresponding to the second source information in the second source information set, including: inputting the route code in the first source information into the route conversion matrix to obtain the first route feature, and determining the first route feature as the first source feature; and inputting the route code in the second source information into the route conversion matrix to obtain the second route feature, and determining the second route feature as the second source feature; or, input the price code in the first source information into the price conversion matrix to obtain the first price feature, and determine the first price feature as the first source feature; and input the price code in the second source information into the price conversion matrix to obtain the second price feature, and determine the second price feature as the second source feature; or, input the route code in the first source information into the route conversion matrix to obtain the first route feature, input the price code in the first source information into the price conversion matrix to obtain the first price feature, and determine the first source feature according to the first route feature and the first price feature; and input the route code in the second source information into the route conversion matrix to obtain the second route feature, input the price code in the second source information into the price conversion matrix to obtain the second price feature, and determine the second source feature according to the second route feature and the second price feature; The route conversion matrix is obtained by: taking the N-1 route encodings with the earliest sample user viewing time in each route sequence sample as input, taking the similarity between the output result and the Nth route encoding with the latest sample user viewing time in the corresponding route sequence sample as the optimization condition, training an initial model to obtain a trained route model, wherein the initial model includes a conversion matrix, a summation layer, and a restoration layer; determining the conversion matrix in the route model as the route conversion matrix; the route sequence sample is determined according to the route information and the viewing time of the information source viewed by the same sample user, the route sequence sample contains N sample route encodings, the difference between the latest viewing time and the earliest viewing time in the N sample route encodings is not more than M days, and N and M are positive integers; The price conversion matrix is obtained by: taking the N-1 price encodings with the earliest sample user viewing time in each price sequence sample as input, taking the similarity between the output result and the Nth price encoding with the latest sample user viewing time in the corresponding price sequence sample as the optimization condition, training an initial model to obtain a trained price model, wherein the initial model includes a conversion matrix, a summation layer, and a restoration layer; determining the conversion matrix in the price model as the price conversion matrix; the price sequence sample is determined according to the price information and the viewing time of the information source viewed by the same sample user, the price sequence sample contains N price encodings, the difference between the latest viewing time and the earliest viewing time in the N price encodings is not more than M days, and N and M are positive integers; The user feature of the user is determined according to the second route feature and the second price feature, and the first information source corresponding to the K first information source features with the greatest similarity to the user feature is pushed to the user, wherein K is a positive integer.

2. The method of claim 1, wherein, The first information source feature is determined according to the first route feature and the first price feature, including: The first information source feature is determined according to the average value of the first route feature and the first price feature. The second information source feature is determined according to the second route feature and the second price feature, including: The second information source feature is determined according to the average value of the second route feature and the second price feature.

3. The method of claim 1, wherein, The user feature of the user is determined according to the second route feature and the second price feature, including: The user feature is determined according to the average value of the second information source feature; or The second information source information in the second information source information set is divided into a plurality of sub-sets, wherein the operation types corresponding to the second information source information in each sub-set are the same, the user feature is determined according to the weight corresponding to each operation type and the second information source feature corresponding to the second information source information in each sub-set.

4. The method of claim 3, wherein, The user feature is determined according to the weight corresponding to the sub-set and the second information source feature corresponding to the second information source information in different sub-sets, including: The average value of the second information source feature corresponding to the second information source information in each sub-set is determined. According to the weight corresponding to each operation type, the average values of the sub-sets corresponding to each operation type are weighted and summed to obtain the user feature.

5. The method according to any one of claims 1 to 4, characterized in that, The first cargo information corresponding to the K first cargo source features with the largest similarity to the user feature is pushed to the user, including: Determining the cosine distance between the first cargo source feature and the user feature; The first cargo information corresponding to the K first cargo source features with the smallest cosine distance is pushed to the user.

6. A device for pushing information of a source of goods, characterized by The device comprises a processor and a memory for storing programs executable by the processor, the processor being configured to read the programs in the memory and perform the steps of any one of claims 1-5.

7. A computer storage medium having stored thereon a computer program, characterized in that The program is executed by the processor to implement the steps of any one of claims 1-5.

Images