CN111523918A - Commodity clustering method, commodity clustering device, commodity clustering equipment and storage medium - Google Patents

Commodity clustering method, commodity clustering device, commodity clustering equipment and storage medium Download PDF

Info

Publication number
CN111523918A
CN111523918A CN201910107025.4A CN201910107025A CN111523918A CN 111523918 A CN111523918 A CN 111523918A CN 201910107025 A CN201910107025 A CN 201910107025A CN 111523918 A CN111523918 A CN 111523918A
Authority
CN
China
Prior art keywords
commodity
network
commodities
inventory
clustering
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910107025.4A
Other languages
Chinese (zh)
Other versions
CN111523918B (en
Inventor
陈伟
韩昊
孙凯
王玉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Geekplus Technology Co Ltd
Original Assignee
Beijing Geekplus Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Geekplus Technology Co Ltd filed Critical Beijing Geekplus Technology Co Ltd
Priority to CN201910107025.4A priority Critical patent/CN111523918B/en
Priority to US17/419,724 priority patent/US12051038B2/en
Priority to PCT/CN2019/128557 priority patent/WO2020140818A1/en
Priority to EP19907152.3A priority patent/EP3905154A4/en
Publication of CN111523918A publication Critical patent/CN111523918A/en
Application granted granted Critical
Publication of CN111523918B publication Critical patent/CN111523918B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Data Mining & Analysis (AREA)
  • Human Resources & Organizations (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Warehouses Or Storage Devices (AREA)

Abstract

The embodiment of the invention discloses a method, a device, equipment and a storage medium for commodity clustering, wherein the method comprises the following steps: mapping commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes appearing together in a historical order as an association weight between the two network nodes, and constructing a commodity association network between the commodities; determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm; and clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes. The technical scheme of the embodiment of the invention solves the problem that the commodities with obvious connection among the commodities cannot be seen from the description content of the commodities can not be well clustered, realizes the clustering of different commodities, improves the clustering effect among the commodities and greatly improves the subsequent commodity processing efficiency.

Description

Commodity clustering method, commodity clustering device, commodity clustering equipment and storage medium
Technical Field
The embodiment of the invention relates to the technical field of logistics storage, in particular to a method, a device, equipment and a storage medium for commodity clustering.
Background
In warehouse management, in order to increase warehouse operation efficiency and save cost, a large number of commodities, orders and storage positions are generally required to be clustered according to existing information, and the key of the clustering process is that different commodities are required to be classified into different clusters according to similarity or correlation.
In the related technology, commodity clustering mainly realizes clustering of different commodities according to description contents of the different commodities, each commodity needs detailed text description contents during clustering, and the clustering method has a good clustering effect on commodities with similar description contents. However, in the actual use process, there are a lot of commodities, and although the obvious connection between the commodities is not seen from the description content of the commodities, there are mutual auxiliary relations in the practical process, such as towels and soap boxes, so that the commodities are often clustered together. It can be seen that, for the commodities in which the obvious connection between the commodities cannot be seen from the description contents of the commodities in the above case, the commodity clustering scheme in the related art cannot solve the problem of achieving good clustering.
Disclosure of Invention
In view of the foregoing problems, embodiments of the present invention provide a method, an apparatus, a device, and a storage medium for clustering commodities, so as to cluster different commodities and improve a clustering effect among the commodities.
In a first aspect, an embodiment of the present invention provides a method for clustering commodities, where the method includes:
mapping commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes appearing together in a historical order as an association weight between the two network nodes, and constructing a commodity association network between the commodities;
determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm;
and clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes.
In a second aspect, an embodiment of the present invention further provides a device for clustering commodities, where the device includes:
the building module is used for mapping the commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes which commonly appear in the historical order into the association weight between the two network nodes, and building a commodity association network between the commodities;
the determining module is used for determining the characteristic values of the commodities corresponding to the network nodes in the commodity association network according to a graph embedding algorithm;
and the clustering module is used for clustering each commodity according to the characteristic value of the commodity corresponding to each network node to obtain a plurality of commodity classes.
In a third aspect, an embodiment of the present invention further provides an apparatus, where the apparatus includes:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method of merchandise clustering according to any embodiment of the invention.
In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for clustering commodities according to any embodiment of the present invention.
According to the technical scheme provided by the embodiment of the invention, the commodities are mapped into the network nodes, the frequency of the common appearance of the commodities corresponding to any two network nodes in the historical orders is mapped into the association weight between the two network nodes, the commodity association network between the commodities is constructed, the characteristic values of the commodities corresponding to the network nodes in the commodity association network are determined according to a graph embedding algorithm, and the characteristic that the obvious association between the commodities cannot be seen from the description content of the commodities can be extracted, so that the commodities can be clustered according to the characteristic values of the commodities corresponding to the network nodes, a plurality of commodity classes are obtained, the clustering of different commodities is realized, the clustering effect among the commodities is improved, and the subsequent commodity sorting efficiency is improved.
The above summary of the present invention is merely an overview of the technical solutions of the present invention, and the present invention can be implemented in accordance with the content of the description in order to make the technical means of the present invention more clearly understood, and the above and other objects, features, and advantages of the present invention will be more clearly understood.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1a is a schematic system structure diagram of an inventory system provided in an embodiment of the present invention;
FIG. 1b is a schematic view of a shelf according to an embodiment of the present invention;
FIG. 1c is a schematic structural diagram of a robot provided in an embodiment of the present invention;
fig. 2 is a schematic flow chart of a method for clustering commodities, provided in an embodiment of the present invention;
fig. 3 is a schematic network structure diagram of a commodity association network provided in an embodiment of the present invention;
FIG. 4 is a schematic flow chart illustrating another method for clustering commodities, provided in an embodiment of the present invention;
FIG. 5 is a schematic flow chart illustrating a method for determining a storage location of a commodity according to a commodity clustering result according to an embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating logical partitioning of an inventory container bit according to an embodiment of the present invention;
fig. 7 is a schematic structural diagram of an apparatus for clustering commodities, provided in an embodiment of the present invention;
fig. 8 is a schematic structural diagram of an apparatus provided in an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. In addition, it should be noted that, for convenience of description, only a part of structures related to the present invention, not all of the structures, are shown in the drawings.
Before discussing exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart may describe the operations (or steps) as a sequential process, many of the operations can be performed in parallel, concurrently or simultaneously. In addition, the order of the operations may be re-arranged. The process may be terminated when its operations are completed, but may have additional steps not included in the figure. The processes may correspond to methods, functions, procedures, subroutines, and the like.
Fig. 1a is a schematic system structure diagram of an inventory system provided in an embodiment of the present invention. Referring to FIG. 1a, the inventory system 100 may include: the robot 110, the control system 120, the inventory receptacle area 130, and the workstation 140, the inventory receptacle area 130 is provided with a plurality of inventory receptacles 131, various inventory items may be placed on the inventory receptacles 131, as well as shelves found in supermarkets on which various merchandise items are placed, alternatively, the inventory receptacles 131 may also be provided with carrying devices such as bins or trays, in which various inventory items are accommodated, the plurality of inventory receptacles 131 being arranged in an array. Generally, a plurality of workstations 140 may be provided at one side of the inventory receptacle area 130.
The control system 120 is in wireless communication with the robot 110, and the control system 120 is operable by an operator via the console 160, and the robot 110 is operable to carry inventory receptacles under the control of the control system 120. Where the inventory receptacles may include, but are not limited to, removable inventory receptacles, the robot 110 may be a self-propelled robot. Taking the inventory container 131 as a movable inventory container, for example, the movable inventory container may be a movable shelf, and the robot 110 may travel along the empty space (a portion of the aisle through which the robot 110 travels) in the movable shelf array, move to the bottom of the movable shelf, lift the movable shelf using the lifting mechanism, and transport to the assigned work station 140.
In one example, the robot 110 may have a lifting mechanism or a hook structure and have a positioning navigation function, the robot 110 can travel to the bottom of the inventory receptacle 131 and lift the entire inventory receptacle 131 with the lifting mechanism or pull the entire inventory receptacle 131 with the hook structure, so that the entire inventory receptacle 131 can move up and down with the lifting mechanism having the lifting function or pull with the hook mechanism.
In another example, the robot 110 can travel forward according to the two-dimensional code information captured by the camera and can travel to under the inventory receptacle 131 prompted by the control system 120 according to the route determined by the control system 120. The robot 110 carries the inventory containers 131 to the workstation 140, and a worker 141 or other automated equipment (e.g., robotic arm) at the workstation 140 performs various types of inventory operations on the inventory containers 131, including but not limited to: picking, inventory or restocking, etc. Taking a picking operation as an example, a worker 141 or other automated device picks items from inventory receptacles 131 and places them into totes 150 for packing.
Taking a shelf as an example, fig. 1b is a schematic structural diagram of a shelf provided in the embodiment of the present invention. As shown in fig. 1b, shelf 131 includes a plurality of compartments on which various items 136 may be placed directly, and four floor support posts 1362. In particular embodiments, the items 136 may be suspended from hooks or bars within or on the shelf where the items 136 can be placed on the interior or exterior surfaces of the shelf in any suitable manner.
The interlayer of the goods shelf can also be provided with a plurality of bins which can be separated from the goods shelf or integrated with the goods shelf, and one or more articles can be placed in the bins. In addition, the goods shelf can be a bidirectional opening goods shelf, two articles can be placed along the depth direction of the interlayer, namely, one article is placed along each opening direction, or two bins are arranged along the depth direction of the interlayer, namely, one bin is arranged along each opening direction. The shelf may also be a one-way open shelf (shown in fig. 1b as a one-way open shelf), and one article may be placed along the depth direction of the partition, i.e. only one article is placed along the opening direction, or one bin may be arranged along the depth direction of the partition, i.e. only one bin is arranged along the opening direction.
Fig. 1c is a schematic structural diagram of a robot provided in the embodiment of the present invention. As shown in fig. 1c, in one example, the self-driven robot 110 may include a driving mechanism 1101, by which the self-driven robot 110 can move within the work space, and the self-driven robot 110 may further include a lifting mechanism 1102 for carrying a shelf, and the self-driven robot 110 may move to below the target shelf 131, lift the target shelf 131 using the lifting mechanism 1102, and carry to the assigned work station 140. The lifting mechanism 1102 lifts the entire target shelf 131 from the ground when lifted, so that the self-driven robot 110 carries the target shelf 131, and the lifting mechanism 1102 lowers the target shelf 131 on the ground. The target recognition unit 1103 on the self-propelled robot 110 can effectively recognize the target shelf 131 when the self-propelled robot 110 lifts the target shelf 131.
In addition, if the navigation is based on two-dimensional code navigation, the self-propelled robot 110 further includes a navigation recognition component (not shown in fig. 1 c) for recognizing the two-dimensional code mark on the paving floor. The self-driven robot 110 may adopt other navigation modes such as inertial navigation and SLAM navigation besides two-dimensional code navigation, and may also combine two or more navigation modes such as two-dimensional code navigation and inertial navigation, SLAM navigation and two-dimensional code navigation. Of course, the self-driven robot 110 further includes a control module (not shown in fig. 1 c) for controlling the whole self-driven robot 110 to implement the functions of movement, navigation, and the like. In one example, the self-propelled robot 110 includes at least two cameras, up and down, that can travel forward based on two-dimensional code information (and other ground markings as well) captured by the camera down, and can travel to under the target shelf 131 prompted by the control system 120 based on the route determined by the control system 120.
As shown in fig. 1b, the two-dimensional code 1361 is disposed at the center of the bottom of the target shelf 131, and when the self-driven robot 110 travels below the target shelf 131, the two-dimensional code 1361 is correctly photographed by the upward camera, so that the self-driven robot 10 is ensured to be located right below the target shelf 131, and thus the self-driven robot 110 can stably lift and transport the target shelf 131.
The control system 120 is a software system with data storage and information processing capability running on a server, and can be connected with a robot, a hardware input system and other software systems through wireless or wired connection. The control system 120 may include one or more servers, which may be a centralized control architecture or a distributed computing architecture. The server has a processor 1201 and a memory 1202, and may have an order pool 1203 in the memory 1202.
In the inventory system shown in fig. 1, as the area of the warehouse increases, the moving distance of the robot in the warehouse increases, and the picking efficiency decreases; warehousing and warehousing goods in a warehouse are a real-time dynamic process, and the position of a shelf in the warehouse needs to be adjusted in real time; when the order is selected, the order needs to be subjected to wave grouping, so that the ex-warehouse efficiency can be effectively improved; in the warehouse-crossing sorting scheme, the sorting time and the recommendation of the station where the sorting is carried out on the commodities which are subjected to warehouse-crossing operation according to the relevance of the commodities are required, and the warehouse-crossing sorting efficiency can be improved. The above-mentioned various operation operations in the warehouse, especially for the goods sorting operation, are implemented by means of the clustering result of the goods. Therefore, it is necessary to improve the clustering method of the commodities and determine the clustering result of the commodities.
The following describes in detail a method, an apparatus, a device, and a storage medium for clustering commodities provided in an embodiment of the present invention.
Fig. 2 is a schematic flow chart of a method for clustering commodities, which is provided in an embodiment of the present invention, and the embodiment of the present invention is applicable to a scene in which commodities are clustered according to commodity relevance, in particular, in a warehouse management center, in order to increase the operation efficiency of a warehouse, a large number of commodities need to be clustered. The method can be executed by a commodity clustering device, the device can be realized in a software and/or hardware mode, the device can be integrated on any equipment with a network communication function, the equipment can be terminal equipment or a server, the terminal equipment can be a mobile phone, a tablet computer, a computer and the like, and the server can be a background server for data processing or other servers. As shown in fig. 2, the method for clustering commodities in the embodiment of the present invention may include:
s201, the commodities are mapped into network nodes, the frequency of the commodities corresponding to any two network nodes appearing together in the historical order is mapped into the association weight between the two network nodes, and a commodity association network between the commodities is constructed.
In the embodiment, the types of the commodities are wide, and the commodities can be tangible commodities such as clothes, food and the like; and may also be intangible goods such as financial products and the like. In addition, the product may have a certain association attribute with the product, and the association attribute may be an attribute inherent to the product itself, a property that the product is different from other products in different fields, or an association between the product and the product defined by a behavior of the user. For example, in practical applications, the association attribute between the goods may be a co-occurrence attribute in which the goods and the goods appear together, that is, an attribute in which two goods appear together in the same order. Based on the co-occurrence attributes of the two commodities, an association network between the commodities can be constructed. The association relationship between the commodities included in the commodity association network can be reflected from the commodity association network.
In this embodiment, if two commodities appear in the same historical order together, it indicates that the two commodities have a co-appearing association attribute; if the two commodities do not appear in the same historical order together, the two commodities do not have the associated attribute appearing together. In addition, since the frequency of two commodities appearing together in the same historical order may be one time or multiple times, different frequencies may reflect the degree of association between the two commodities.
In this embodiment, fig. 3 is a network structure intention of a product association network provided in this embodiment of the present invention. Referring to fig. 3, when a commodity association network between commodities is constructed, each commodity may be mapped to network nodes, and each network node may represent one dimension. If any two commodities appear together in the same historical order, adding an edge between network nodes corresponding to the two commodities so as to associate the two commodities; if the two commodities do not appear in the same historical order together, no edge is added between the network nodes corresponding to the two commodities, namely the two commodities are not correlated.
In this embodiment, referring to fig. 3, considering that the frequency of two commodities appearing in the same historical order together may be one time or multiple times, when constructing the commodity association network, the frequency of the commodities appearing in the historical order together corresponding to any two network nodes may also be referred to at the same time. Optionally, the frequency of the common occurrence of the commodities corresponding to any two network nodes in the historical order may be mapped as the association weight between the two network nodes. The frequency of the common appearance of the commodities corresponding to the two network nodes in the historical order is in direct proportion to the weight of the edge added between the network nodes corresponding to the two commodities. The more times that two commodities appear in a historical order together, the greater the weight of an edge between network nodes corresponding to the two commodities in the constructed commodity association network; the smaller the number of times that two commodities appear together in one historical order, the smaller the weight of the edge between the network nodes corresponding to the two commodities in the commodity association network. By adopting the mode, the commodity association network which is more suitable for the actual situation can be constructed according to the actual commodity-commodity association situation, so that the real association degree among the commodities can be reflected through the commodity association network.
S202, determining characteristic values of the commodities corresponding to the network nodes in the commodity association network according to a graph embedding algorithm.
In this embodiment, after the commodity association network is constructed, each commodity in the commodity association network can be used as an object to be evaluated. Aiming at each commodity contained in the commodity association network, the commodity association network can be processed through a graph embedding algorithm, and the characteristic value of the commodity corresponding to each network node is extracted from the commodity association network. The characteristic value of the commodity corresponding to the network node may refer to a characteristic vector for representing an association relationship between the commodity and other commodities in the commodity association network. Optionally, after the commodity association network is constructed, a graph embedding algorithm may be used to process each network node included in the commodity association network, extract association information between each network node and other network nodes from the commodity association network, and use the association information as a feature value of the commodity corresponding to each network node.
In this embodiment, the graph embedding algorithm may include: deep walking algorithm, large-scale information network embedding algorithm and the like. Determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm may include: and calculating the characteristic values of the commodities corresponding to the network nodes in the commodity association network by adopting a deep walking algorithm or a large-scale information network embedding algorithm.
In an optional manner of this embodiment, calculating the feature value of the commodity corresponding to each network node in the commodity-associated network by using the deep migration algorithm may include the following steps S2021a to S2021 b:
s2021a, during each random walk, uniformly and randomly sampling a network node from the article-associated network as a starting point of the current random walk, uniformly and randomly sampling an adjacent point with associated weight for a previous node during the walk, ending the current random walk until the walk reaches a preset maximum length, and finally obtaining a plurality of node sequences.
S2021b, training the node sequences through machine learning to obtain the characteristic value vector of the commodity corresponding to each network node.
In the embodiment, the implicit representation information of each network node in the commodity association network can be learned through a deep walking algorithm. The implicit representation information can be understood as implicit association information between commodities corresponding to each network node reflected in the commodity association network.
In this embodiment, a random walk generator may be used to uniformly select at least one network node from the commodity association network, and a deep walk algorithm may be used to perform a random walk process on each selected network node. Optionally, when random walk is performed each time, a random walk generator may be used to uniformly and randomly sample one network node from the commodity associated network as a starting point of the random walk, and each walk uniformly and randomly samples one adjacent point with associated weight to a last visited network node until the maximum length is reached, so as to obtain a walking network node combination of the random walk as a node sequence of the random walk. According to the mode, the random walk generator can be used for uniformly sampling a plurality of network nodes from the commodity associated network to be respectively used as the starting point of random walk each time, and a plurality of node sequences can be finally obtained through a plurality of random walk processes.
In this embodiment, after obtaining the plurality of network node sequences, the obtained plurality of node sequences may be trained through a preset machine learning model, so as to map each network node in the node sequences obtained by random walk to a continuous vector space, and further obtain a feature value corresponding to each network node in the commodity association network. A plurality of node sequences can be obtained from the commodity association network in a random walk mode through a deep walk algorithm, and then association feature vectors existing among all network nodes are learned from the node sequences through the best training of the node sequences, so that feature values of commodities corresponding to all network nodes in the commodity association network can be obtained.
Illustratively, referring to fig. 3, the commodity association network may include a plurality of network nodes and association weights between the network nodes. The commodity association network can be represented by a commodity association graph G (V, E), wherein V represents the commodity type in the commodity association network, E represents the connection of the midpoint of V, different network nodes are named by different symbols V, and a storage structure of a two-dimensional array is used for representing whether a connecting edge exists between the two network nodes and the existence is represented as1, otherwise 0. The specific process is as follows: in each random walk process, a network node V can be randomly and uniformly selected from a commodity association network G (V, E) by adopting a deep random walk algorithmiAnd will select the network node viAnd performing random walk as a starting point of the random walk, so that a network node combination of the random walk can be obtained and used as a node sequence of the random walk. By uniformly sampling a plurality of network nodes from the commodity associated network to respectively serve as the starting point of random walk each time, a plurality of node sequences can be finally obtained through a plurality of random walk processes. After a plurality of node sequences with fixed length are constructed and generated, the constructed and generated node sequences can be trained by adopting a Skip-gram model, low-dimensional feature vectors corresponding to each network node are obtained by learning and training from the middle network node, and the low-dimensional feature vectors corresponding to each network node are used as feature values of commodities corresponding to each network node in the commodity association network. For example, the plurality of node sequences obtained by random walk may specifically be: 1-4-13-1-11-5-7 …, 33-34-10-3-20-25 …, 28-24-30-27-9-21 …, and further training the node sequences to obtain low-dimensional feature vectors corresponding to each network node in the commodity association network, so that feature values of commodities corresponding to the network nodes can be obtained. Referring to fig. 3, the characteristic value of the commodity corresponding to the network node 1 may be: 0.016579, -0.0336, 0.3452167, 0.04698 …; the characteristic value of the commodity corresponding to the network node 2 may be: -0.00703, 0.26589, -0.351422, 0.043923 ….
In the present embodiment, with the above technical solution, when extracting feature values of a commodity corresponding to each network node from a commodity association network, the characteristics of adaptivity of a deep walking algorithm (after a network node is newly added, it is not necessary to learn again), rationality (probability of classifying network nodes with similar feature vectors into the same class is high), low latitude (classification efficiency of the commodity can be accelerated by generating a low-dimensional feature vector), and continuity (order representation can be performed in a continuous space, and classification effect can be increased) are fully utilized, so that feature values of the commodity corresponding to each network node can be better extracted from the commodity association network.
Although the association information between the network nodes can be extracted from the commodity association network as much as possible through the deep migration algorithm to obtain the characteristic value of the commodity corresponding to each network node, the deep migration algorithm only considers the first-order proximity of the commodity, and the information of hidden deeper multi-order proximity in the commodity association network cannot be effectively obtained. In view of the above, the embodiment of the present invention may employ a large-scale information network embedding algorithm to extract information hiding deeper multiple levels of proximity from the commodity association network.
In an optional manner of this embodiment, calculating the feature value of the commodity corresponding to each network node in the commodity-associated network by using the large-scale information network embedding algorithm may include the following steps S2022a to S2022 c:
s2022a, determining first-order similarity among network nodes in the commodity association network, and optimizing the first-order similarity among the network nodes to obtain first eigenvalue vectors of commodities corresponding to the network nodes.
S2022b, determining second-order similarity among network nodes in the commodity association network, and optimizing the second-order similarity among the network nodes to obtain a second characteristic value vector of the commodity corresponding to each network node.
S2022c, splicing the first characteristic value vector and the second characteristic value vector to obtain characteristic value vectors of the commodities corresponding to the network nodes.
In this embodiment, a first-order similarity and a second-order similarity may be introduced at the same time, and hidden multi-order proximity information is extracted from the commodity association network by combining the first-order similarity and the second-order similarity, so as to extract a feature value of the commodity corresponding to each network node from the commodity association network.
In this embodiment, the first-order similarity in the commodity association network may refer to a local pairwise similarity between two network nodes, and a weight of an edge connecting between two network nodes may be used to represent the first-order similarity between the two network nodes. If no edge exists between the two network nodes, the first-order similarity between the two network nodes is 0.The second-order similarity in the commodity association network may refer to the similarity of the neighbor network structure in the commodity association network, for example, in the commodity association network, p is useduRepresenting a first order of similarity between network node u and other adjacent network nodes, denoted by pvThe first-order similarity between the network node v and other adjacent network nodes is represented, and the second-order similarity between the corresponding network node v and the corresponding network node u is puAnd pvThe similarity between them. If no other network node is connected to network node u and network node v, then the second order similarity between network node v and network node u is 0.
In this embodiment, in order to ensure the accuracy of the obtained first-order similarity, after the first-order similarity between each network node in the commodity association network is determined, the first-order similarity between each network node may be optimized, so that the first eigenvalue vector of the commodity corresponding to each network node may be obtained; in addition, the second-order similarity between the network nodes in the commodity association network can be determined according to the first-order similarity between the network nodes, and the second-order similarity between the network nodes is optimized, so that the second characteristic value vector of the commodity corresponding to each network node can be obtained.
In this embodiment, after the first eigenvalue vector and the second eigenvalue vector of the commodity corresponding to each network node are obtained, the first eigenvalue vector and the second eigenvalue vector may be subjected to a stitching process to obtain a third eigenvalue vector corresponding to each network node, and the third eigenvalue vector may be used as the eigenvalue vector of the commodity corresponding to each network node. Further, it is considered that the first eigenvalue vector and the second eigenvalue vector do not necessarily both satisfy the condition of minimum similarity. In other words, even if the first-order similarity represented by the first eigenvalue vector alone satisfies the minimum similarity condition and the second-order similarity represented by the second eigenvalue vector alone satisfies the minimum similarity condition, the similarity represented by the third eigenvalue vector, which does not represent concatenation, satisfies the minimum similarity condition. For this purpose, weight proportions may be set for the first eigenvalue vector and the second eigenvalue vector, respectively, and the spliced third eigenvalue vector may be balanced by the weight proportions. For example, taking the first eigenvalue vector as a, the second eigenvalue vector as B, and the third eigenvalue vector after concatenation as C ═ a; b ], setting a weight β for the first eigenvalue vector a, and setting a weight γ for the second eigenvalue vector B, to obtain a weight-balanced third eigenvalue vector C' ═ β a; and gamma B ], and taking the third characteristic value vector C' after weight balance as the characteristic value vector of the commodity corresponding to each network node to obtain the characteristic value of the commodity corresponding to each network node.
In this embodiment, in order to better perform the concatenation processing on the first eigenvalue vector and the second eigenvalue vector, the first eigenvalue vector and the second eigenvalue vector may be normalized first, and the first eigenvalue vector and the second eigenvalue vector after the normalization processing may be concatenated to obtain a third eigenvalue vector after the concatenation processing. Optionally, when the first eigenvalue vector and the second eigenvalue vector after the normalization processing are spliced, the weight ratio of the first eigenvalue vector and the second eigenvalue vector may be balanced, so as to obtain accurate eigenvalue vectors of the commodities corresponding to each network node.
In this embodiment, the product association network may be represented by a product association graph G ═ V, E, (V, E) where V represents a product type in the product association network, E represents a connection of midpoints of V, and each pair of edges is an ordered pair E ═ u, V) and has a weight w greater than zerouvIt indicates how many orders of all orders contain both u and v commodities. Each network node v in the commodity association graph can be mapped to a low-dimensional space R by adopting a large-scale network algorithmdIn (1), a function f is learnedG:V→RdWherein d is<<V | so that in space RdThe first order similarity and the second order similarity are retained. The following is an exemplary description of the first-order similarity optimization process and the second-order similarity optimization process.
(1) The case of performing optimization processing on the first-order similarity:
specifically, two network nodes V are defined for each edge (i, j) in the commodity association graph G ═ (V, E)i，vjThe connection probability between network nodes may specifically be:
Figure BDA0001967001440000111
wherein u isiIs viThe low-dimensional vector of (b) represents a distribution in a V × V space defined by the product correlation diagram G ═ V, E, or an empirical distribution
Figure BDA0001967001440000112
Wherein,
Figure BDA0001967001440000113
to preserve the first order similarity, the following objective function may be reduced:
Figure BDA0001967001440000121
where d is the distance between the two distributions.
By reducing the KL divergence of the two probability distributions, replacing the distance function with the KL divergence and removing the constant, one can obtain:
Figure BDA0001967001440000122
finding a reduced form of the above
Figure BDA0001967001440000123
Each point in the d-dimensional space can be represented and thus a first vector of eigenvalues can be obtained.
(2) The case of performing optimization processing on the second-order similarity:
specifically, given a hypothetical directional commodity association graph G ═ V, E, it is assumed that any network node shares multiple connections with other network nodes, and each network node has two cases: the network node itself and external nodes of other network nodes, when two vectors are introduced
Figure BDA0001967001440000124
V representing network nodes respectivelyiAnd as external nodesV isiFor each directed edge (i, j), environment v is first definedjGenerating a network node viProbability of (c):
Figure BDA0001967001440000125
where | V | is the number of network nodes or environments. For each network node viOf the above formula p2(vj,vi) A distribution of conditions over the environment is determined.
To preserve the second order similarity, the conditional distribution should be determined from the low dimensional representation to approximate the empirical distribution
Figure BDA0001967001440000126
The simplest approach is to reduce the following objective function:
Figure BDA0001967001440000127
wherein d (·,) represents two distributed distances, and λ is introduced due to different importance of network nodes in the commodity association graph G ═ (V, E)iTo indicate the importance of the network node i, which can be measured by degree or algorithm.
Empirical distribution
Figure BDA0001967001440000128
Wherein wijRepresents the weight of the edge (i, j), diIs the degree of departure of the node i, and introduces KL divergence as a distance function and takes lambda as the distance function for simplifying the calculationiIs set to degree diAnd removing constants to obtain the following objective function:
Figure BDA0001967001440000129
by learning
Figure BDA00019670014400001210
And
Figure BDA00019670014400001211
to reduce the objective function of the above equation, a d-dimensional vector can be used
Figure BDA00019670014400001212
Represents each node viThereby, a second eigenvalue vector can be obtained.
After the optimization processing of the first-order similarity and the second-order similarity, a first eigenvalue vector and a second eigenvalue vector can be obtained, and the eigenvalue vectors of the commodities corresponding to the network nodes can be obtained by splicing the first eigenvalue vector and the second eigenvalue vector.
In this embodiment, the above-mentioned objective function O is jointly trained by combining the first-order similarity and the second-order similarity1And an objective function O2Taking into account the calculation of the conditional probability p2All nodes need to be accumulated, the cost is very high, so that negative sampling is introduced, a plurality of negative edges are sampled according to the noise distribution of the edges (i, j) between every two nodes, and the following functions are specially appointed to the edges between every two nodes:
Figure BDA0001967001440000131
where σ (x) — 1/(1+ exp (-x)), the first term constructs an observation edge, the second term constructs a negative edge drawn by a noise distribution, and K is the number of negative edges. Order to
Figure BDA0001967001440000132
Wherein d isvIs the out degree of node v. In addition, to optimize the above equation, an asynchronous random gradient Algorithm (ASGD) may be employed for optimization.
And S203, clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes.
In this embodiment, after the characteristic values of the commodities corresponding to the network nodes in the commodity association network, the commodities corresponding to the network nodes in the commodity association network may be clustered through a preset clustering algorithm according to the characteristic values of the commodities corresponding to the network nodes. According to the clustering result of the commodities corresponding to each network node, the commodities corresponding to each network node included in the commodity association network can be allocated to different commodity classes, so that a plurality of commodity classes can be obtained. The preset clustering algorithm can be used for clustering and analyzing commodities corresponding to each network node in the commodity association network. Optionally, the preset clustering algorithm may include, but is not limited to, K-means clustering (K-means), spectral clustering, and the like.
In this embodiment, the clustering in this embodiment may be understood as dividing the commodity set including a plurality of commodities into different classes or clusters according to a certain criterion, so that the similarity of the commodities in the same class or cluster is as large as possible, and the difference of the commodities in different classes or clusters is also as large as possible.
In an optional manner of this embodiment, after determining the feature value of the commodity corresponding to each network node in the commodity association network according to the graph embedding algorithm, the method may further include:
and taking one characteristic value as a dimension, and performing dimension reduction processing on the characteristic value of the commodity corresponding to each network node in the commodity association network.
In this embodiment, after the feature values of the commodities corresponding to the network nodes in the commodity-associated network are determined according to the graph embedding algorithm, the dimension of the feature value of the commodity corresponding to each network node may still be relatively high, and invalid data may exist in the feature value of the commodity corresponding to each network node. If the dimension of the characteristic value of the commodity corresponding to the network node is high or invalid data exists, the complexity of data processing is increased when clustering is performed according to the characteristic value of the commodity corresponding to each network node, so that the clustering efficiency of the commodity is influenced. Therefore, a preset dimension reduction algorithm can be adopted to take the characteristic value of the commodity corresponding to each network node as a dimension, and dimension reduction is carried out on the characteristic value of the commodity corresponding to each network node in the commodity association network. The preset dimension reduction algorithm may include Principal Component Analysis (PCA), isometry mapping (Isomap), laplacian feature mapping (LE), Local Linear Embedding (LLE), t-distribution random neighborhood embedding (t-SNE), and other algorithms.
According to the technical scheme of the embodiment of the invention, the commodities are mapped into the network nodes, the frequency of common appearance of the commodities corresponding to any two network nodes in the historical orders is mapped into the association weight between the two network nodes, the commodity association network between the commodities is constructed, the characteristic values of the commodities corresponding to the network nodes in the commodity association network are determined according to a graph embedding algorithm, and the characteristic that the obvious association between the commodities cannot be seen from the description content of the commodities can be extracted, so that the commodities can be clustered according to the characteristic values of the commodities corresponding to the network nodes, a plurality of commodity classes are obtained, clustering of different commodities is realized, the clustering effect among the commodities is improved, and the subsequent commodity sorting efficiency is greatly improved.
Fig. 4 is a schematic flow chart of another method for clustering commodities, provided in the embodiment of the present invention, where the embodiment of the present invention performs optimization based on the above-described embodiment, and the embodiment of the present invention may be combined with various alternatives in one or more of the above-described embodiments. As shown in fig. 4, the method for clustering commodities in the embodiment of the present invention may include:
s401, the commodities are mapped into network nodes, the frequency of the commodities corresponding to any two network nodes appearing together in the historical order is mapped into the association weight between the two network nodes, and a commodity association network between the commodities is constructed.
S402, determining characteristic values of the commodities corresponding to the network nodes in the commodity association network according to a graph embedding algorithm.
And S403, clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes.
S404, according to the obtained multiple commodity classes, determining a commodity storage position mode, a group ordering wave mode, a shelf adjusting mode and/or a warehouse-crossing sorting mode in the warehouse.
In this embodiment, each item may be provided with a unique SKU, and SKUs of items belonging to the same item class may be located in the same SKU cluster. The SKU (Stock Keeping Unit) is a Unit of Stock in and out metering, and may be a Unit of a member, a box, a tray, or the like. The SKUs referred to in the embodiments of the present application may be referred to as short names of unified serial numbers of inventory items, and each inventory corresponds to a unique SKU number. SKUs may be understood as a uniform or unique identification number of items in stock, and the identity of each item may be identified by its corresponding SKU.
The technical scheme of the embodiment of the invention solves the problem that the commodities with obvious connection among the commodities cannot be seen from the description content of the commodities can not be well clustered, realizes the clustering of different commodities, and improves the clustering effect among the commodities, thereby realizing the guidance of a commodity storage position mode, a group ordering wave mode, a shelf adjusting mode and/or a warehouse-crossing sorting mode in a warehouse according to the clustering result.
The following respectively describes the operation steps of determining a goods storage position mode, a order group wave mode, a shelf adjusting mode and/or a cross-warehouse sorting mode in the warehouse according to the obtained multiple goods categories.
In an application scenario scheme of this embodiment, a detailed scheme for determining a commodity storage position mode according to a commodity clustering result is provided in this embodiment. Fig. 5 is a schematic flow chart illustrating a method for determining a storage position of a commodity according to a commodity clustering result according to an embodiment of the present invention. The present embodiment is optimized based on the above embodiments. As shown in fig. 5, the step of determining the storage position mode of the commodity according to the commodity clustering result provided in the embodiment of the present invention may include:
s501, dividing an inventory system of a warehouse into a plurality of logic partitions according to the obtained plurality of commodity classes; any inventory container and any station in the inventory system have logical partitions to which they belong.
In this embodiment, compared with the conventional scheme of processing orders based on physical partitions, the technical solution of this embodiment processes orders based on logical partitions. In other words, according to the scheme of the present embodiment, when processing an order, the order processing is not performed according to the physical partition, but according to the logical partition. The logical partition management is not to perform partition management on the inventory system according to the actual position in the traditional sense, but performs partition management on the inventory system from the logical point of view.
In this embodiment, a fixed stock container position is provided in the stock system, and the stock container can be placed in the corresponding stock container position. Each inventory receptacle location may correspond to only one inventory receptacle at any one time and each inventory receptacle may only be placed on the corresponding inventory receptacle location. Of course, the corresponding relationship between the stock container position and the stock container may be adjusted adaptively according to the position of the stock container in the stock system.
In this embodiment, a plurality of logical partitions may be included in the inventory system, each logical partition in the inventory system may be associated with at least one station in the inventory system (in an alternative, one logical partition is associated with one station in the inventory system) and a plurality of inventory receptacles and a plurality of inventory receptacle slots, the sum of all logical partition associated inventory receptacles is at least a portion of, and even all, all inventory receptacles in the inventory system, and the sum of all logical partition associated inventory receptacle slots is at least a portion of, and even all, all inventory receptacle slots in the inventory system. The inventory containers may include shelves or other types of containers on which trays, bins, or other items holding devices may be placed.
In an optional manner of this embodiment, dividing the inventory system of the warehouse into a plurality of logical partitions according to the obtained plurality of commodity classes may include the following steps S501a to S501 d:
s501a, determining a plurality of logic partitions according to the obtained commodity classes.
In this embodiment, one logical partition may be associated with each commodity class, or a plurality of commodity classes may be associated with each logical partition. Preferably one for each commodity class. After the plurality of commodity classes are obtained, the dividing amount for performing the logical division on the inventory system can be determined according to the obtained plurality of commodity classes, so that each commodity class is ensured to correspond to at least one logical division.
S501b, according to the coincidence degree between the inventory goods in the inventory container in the inventory system and the goods in each goods class, the inventory container in the inventory system is divided into logical partitions corresponding to the goods class with high coincidence degree.
In this embodiment, each inventory container of the inventory system may store one or more inventory items, and the inventory items stored in different inventory containers may have a certain difference, that is, the inventory items stored in different inventory containers may be the same or different, or may be partially the same and partially different. For this reason, the commodities stored in each inventory container in the inventory system may be respectively matched with the commodities in each commodity class, and the inventory containers in the inventory system may be divided into logical partitions corresponding to the commodity classes with high degrees of coincidence according to the degrees of coincidence between the inventory commodities in the inventory containers in the inventory system and the commodities in each commodity class.
In this embodiment, the inventory container having the highest product coincidence degree may be divided into logical partitions corresponding to the matched product classes according to the coincidence degree between the product on the inventory container and the product included in the product class. Optionally, the inventory containers may be sorted according to the coincidence degree between the commodities on the inventory container and the commodities included in the commodity class, and the inventory containers with the preset number sorted in the front are divided into the logical partitions corresponding to the commodity class matched with the inventory containers.
Illustratively, three inventory receptacles and two commodity classes are included in the inventory system. The three inventory containers are respectively a first inventory container, a second inventory container and a third inventory container. The first inventory container is stored with a first commodity, a second commodity, a third commodity and a fourth commodity; the second inventory container stores a second commodity, a third commodity and a fourth commodity; the third inventory receptacle stores the first product, the fifth product, the sixth product, and the seventh product. The two commodity classes are respectively a commodity class of the first logic partition and a commodity class of the second logic partition. Wherein the commodity class of the first logical partition includes: the commodity class of the second logical partition comprises: a first commodity, a fifth commodity, a sixth commodity, and a seventh commodity.
It is easy to see that the commodities stored in the first inventory container are all the same as the commodities included in the commodity class of the first logical partition, that the commodities stored in the second inventory container have 3 identical commodities compared with the commodities included in the commodity class of the first logical partition, and that the commodities stored in the third inventory container have only 1 identical commodity compared with the commodities included in the commodity class of the first logical partition. In this case, the commodities in the respective stock containers are sorted according to the degree of coincidence between the commodities in the commodity class of the first logical partition, and then the commodities are sequentially sorted into a first stock container, a second stock container, and a third stock container. Similarly, the commodities in the inventory containers are sorted according to the degree of coincidence between the commodities in the commodity class of the second logical partition, and then are sequentially the third inventory container, the second inventory container and the first inventory container.
Based on the above analysis, a first inventory container having the highest degree of overlap with the commodities included in the commodity class of the first logical partition may be classified into the first logical partition, and a third inventory container having the highest degree of overlap with the commodities included in the commodity class of the second logical partition may be classified into the second logical partition.
S501c, dividing the inventory receptacle bit in the inventory system into different logical partitions according to the number of inventory receptacles associated with each logical partition.
In this embodiment, after dividing each inventory receptacle in the inventory system into logical partitions to which each inventory receptacle belongs, each logical partition may be associated with at least one inventory receptacle. In view of the fact that each inventory receptacle location may correspond to only one inventory receptacle at any one time, and that each inventory receptacle may only be placed on the corresponding inventory receptacle location. For this reason, after each inventory container in the inventory system is divided into logical partitions to which each inventory container belongs, the inventory container in the inventory system also needs to be divided into logical partitions. The inventory receptacle bits in the inventory system may be divided into different logical partitions specifically according to the number of inventory receptacles associated with each logical partition. In order to ensure that the inventory receptacles of the same logical partition are placed in the same position range as much as possible, the inventory receptacle positions in the inventory system may be sequentially divided into different logical partitions according to the number of inventory receptacles associated with each logical partition and according to a preset orientation sequence. The preset orientation sequence may be understood as an orientation sequence from left to right, from right to left, from front to back, or from back to front.
Illustratively, fig. 6 is a schematic diagram of logical partitioning of an inventory container bit provided in an embodiment of the present invention. Referring to fig. 6, taking the example of dividing the inventory system into the first logical partition, the second logical partition, and the third logical partition, the number of inventory receptacles associated with the first logical partition is 20, the number of inventory receptacles associated with the second logical partition is 30, and the number of inventory receptacles associated with the third logical partition is 40, and according to the number of inventory receptacles associated with each logical partition, the method may divide 20 inventory receptacle bits into the first logical partition, divide 30 inventory receptacle bits into the second logical partition, and divide 40 inventory receptacle bits into the third logical partition in order from left to right, so as to divide the inventory receptacle bits in the inventory system into different logical partitions.
And S501d, dividing the workstations in the inventory system into the logical partitions which are close to the workstations according to the distances between the workstations and the inventory container positions associated with the logical partitions.
In this embodiment, a plurality of stations may exist in the inventory system, and based on the positional relationship between each station and the inventory receptacle location associated with each logical partition, the distances from different inventory receptacle locations to the same station may be different, and the distances from different stations to the same inventory receptacle location may also be different. Therefore, the workstations in the inventory system can be divided into the logical partitions which are close to the workstations according to the distance between the workstations and the inventory container positions associated with the logical partitions, so that the distance between each workstation and the inventory container position of each logical partition is ensured to be as short as possible. Therefore, when the robot needs to be controlled to convey the stock containers in the stock container positions to the station, the station and the stock containers can be guaranteed to belong to the same associated logic partition, and the distance between the station and the stock containers is further guaranteed to be as short as possible, so that the conveying distance of the robot is reduced.
In this embodiment, optionally, for each workstation located in the inventory system, the distance between each workstation and the inventory receptacle associated with each logical partition is calculated, the logical partition with the closest distance between the workstation located in the inventory system and the inventory receptacle associated with the logical partition is determined, and the workstation located in the inventory system is partitioned into the logical partition with the closest distance. Optionally, for each station located in the inventory system, a distance between each station and the inventory receptacle associated with each logical partition is calculated, the obtained distances are sorted, a logical partition with a top-ranked preset number is determined, and the station located in the inventory system is sorted into the logical partition with the top-ranked preset number.
In this embodiment, optionally, dividing the workstations in the inventory system into the logical partitions that are close to the workstations according to the distance between the workstations and the inventory receptacle associated with each logical partition includes: according to the position relation between the stock container positions associated with the logic partitions and the stations, calculating a first transportation distance required by each stock container position in the stock container positions associated with the logic partitions to reach the stations, and calculating a second transportation distance required by each station to reach the stock container positions associated with the logic partitions; and dividing the work stations in the inventory system into logical partitions which are close to the work stations according to the first transportation distance and the second transportation distance.
S502, when a target order is processed, determining a logic partition to which the target order belongs from a plurality of logic partitions as a target logic partition; at least one of the plurality of inventory containers associated with the target logical partition contains inventory goods required by the target order.
In this embodiment, one logical partition may be associated with at least one workstation, and one logical partition may be associated with a plurality of inventory receptacles. The target order may include information about the inventory items needed. At least one inventory container of the plurality of inventory containers associated with the target logical partition contains inventory items required for the target order. Illustratively, taking the inventory system shown in FIG. 1 as an example, referring to FIG. 1, the inventory system may include stations in the workstation 140, inventory receptacles in the inventory receptacle zone 130, and the robot 110. Each logical partition in the inventory system may be associated with at least one workstation located at workstation 140 and each logical partition associated with a plurality of inventory receptacles located in inventory receptacle area 130. Wherein inventory receptacles located in the inventory receptacle area 130 may be used to store inventory items.
And S503, distributing the target order to one station associated with the target logic partition as a target station.
In this embodiment, since each logical partition may be associated with at least one workstation in the inventory system, after the target logical partition to which the target order belongs is determined, the number of workstations associated with the target logical partition may include a plurality of workstations, and any one workstation may be selected as the target workstation.
And S504, the control robot carries the target inventory container containing the inventory items required by the target order in the plurality of inventory containers associated with the target logic partition to the target station.
In this embodiment, after the target order is assigned to the target workstation associated with the target logical partition, the robot may be controlled to transport the target inventory container (i.e., the inventory container containing the inventory item required for the target order among the plurality of inventory containers associated with the target logical partition) to the target workstation. At the corresponding target station, a worker or automated equipment may grasp the inventory items required for the target order from the target inventory container and place them in the order container to wait for packaging. Of course, task operations such as article replenishment and article stocking may be performed in addition to the operation of article sorting.
By adopting the technical scheme of the embodiment, the shelving strategy of the commodities in the warehouse area can be guided according to the obtained commodity classes, the logical partitions of the warehouse area are realized, and only specific types of commodities are stored in each logical partition, so that the probability of the robot for carrying the inventory containers remotely is reduced, the average carrying distance of the robot for carrying the inventory containers is reduced, and the carrying efficiency is improved.
In another application scenario scheme of this embodiment, a detailed scheme for determining an order wave manner according to a product clustering result is provided in this embodiment. The step of determining the order wave mode according to the commodity clustering result provided by the embodiment of the invention can comprise the following steps:
according to the obtained multiple commodity classes, different orders corresponding to different commodities in the same commodity class are combined in the same order wave task, so that the order wave task is distributed to the same workstation to execute a picking task.
In this embodiment, taking the first commodity and the second commodity in the same commodity class, where the first order includes the first commodity, and the second order includes the second commodity as an example, after obtaining a plurality of commodity classes, according to a clustering result between the first commodity and the second commodity, it is known that the first commodity and the second commodity are in the same commodity class, and the first order and the second order can be combined into one wave number and assigned to the same workstation to execute a task.
The technical scheme has the advantages that the first commodity and the second commodity are placed on the same stock container or two adjacent stock containers, when the workstation processes the first order and the second order in a repeating task, the stock containers can be hit and picked as few as possible, the picking efficiency is improved, and the carrying times of the stock containers are reduced.
In another application scenario scheme of this embodiment, a detailed scheme for determining a shelf adjustment manner according to a commodity clustering result is provided in this embodiment. The step of determining the shelf adjustment mode according to the commodity clustering result provided by the embodiment of the invention may include:
according to the obtained multiple commodity classes, placing each commodity contained in the same commodity class on an inventory container meeting a preset distance condition; or according to the obtained multiple commodity classes, stripping multiple commodities which are placed on the inventory container meeting the preset distance condition and are not contained in the same commodity class.
In this embodiment, the inventory containers with the preset distance condition are the same inventory container, or a plurality of inventory containers with the distance between the inventory containers smaller than or equal to the preset distance threshold.
The technical scheme has the advantages that the warehouse entering and the warehouse exiting of the warehouse commodities are a real-time dynamic process, the clustering result of each commodity can be used for providing a guiding principle for adjusting the inventory containers, the position of the inventory in the warehouse can be adjusted in time, and the conveying distance and the conveying efficiency of the robot in conveying the inventory containers are reduced.
In another application scenario scheme of this embodiment, a detailed scheme for determining an over-library sorting manner according to a commodity clustering result is provided in this embodiment. The step of determining the cross-warehouse sorting mode according to the commodity clustering result provided by the embodiment of the invention can comprise the following steps:
dividing each store into different store combinations according to the contact ratio between the commodities contained in each commodity class and the commodities required by each store order;
the same sorting time period is allocated to a plurality of stores belonging to the same store group so that the plurality of stores belonging to the same store group perform the article sorting operation within the same sorting time period.
In the embodiment, the idea of warehouse crossing is to cross a warehousing link, generally applied to goods sorting of stores, regularly distribute goods to the stores, directly send the goods from a supplier (the sent goods are all bulk goods placed on a pallet) when the goods need to be distributed, then directly send the goods to the stores after being sorted according to the stores, the goods do not need to be stored in the warehouse for a long time to wait for distribution to the stores, and the use cost required by the warehouse is saved. Therefore, the over-warehouse sorting scenario has a limited origin and is not as large as a warehouse.
In this embodiment, after obtaining a plurality of commodity classes, each store may be divided into different store groups according to the degree of coincidence between a commodity included in each commodity class and a commodity required for each store order, and the same sorting time period may be allocated to a plurality of stores belonging to the same store group, so that the plurality of stores belonging to the same store group perform a commodity sorting operation within the same sorting time period.
By adopting the technical scheme of the embodiment, a plurality of stores belonging to the same store combination can be arranged in the same time period for sorting, for example, the stores 1 and the stores 2 are sorted in the morning, so that trays can be used as few as possible during sorting, and under the condition that a warehouse-crossing sorting site is limited and can only accommodate limited sorting stations, goods sorting tasks of more stores can be processed as much as possible.
Fig. 7 is a schematic structural diagram of a device for clustering commodities, which is provided in an embodiment of the present invention, and the embodiment of the present invention is applicable to a scene in which commodities are clustered according to commodity relevance. The apparatus can be implemented in software and/or hardware, and the apparatus can be integrated on any device with network communication function.
As shown in fig. 7, the apparatus for clustering commodities in the embodiment of the present invention may include: a building module 701, a determining module 702 and a clustering module 703. Wherein:
a building module 701, configured to map the commodities into network nodes, map the frequency of common occurrence of the commodities corresponding to any two network nodes in a historical order into an association weight between the two network nodes, and build a commodity association network between the commodities;
a determining module 702, configured to determine, according to a graph embedding algorithm, feature values of a commodity corresponding to each network node in the commodity association network;
the clustering module 703 is configured to perform clustering processing on each commodity according to the feature value of the commodity corresponding to each network node, so as to obtain multiple commodity classes.
On the basis of the above embodiment, optionally, the apparatus may further include:
the application module 704 is configured to determine a commodity storage location mode, a order group wave mode, a shelf adjustment mode and/or a cross-warehouse sorting mode in the warehouse according to the obtained multiple commodity classes.
On the basis of the above embodiment, optionally, the application module 704 may include:
the logical partition dividing unit is used for dividing the inventory system of the warehouse into a plurality of logical partitions according to the obtained plurality of commodity classes; any inventory container and any station in the inventory system are provided with the logic partitions to which the inventory container and the station belong;
a target partition determining unit, configured to determine, as a target logical partition, a logical partition to which a target order belongs from among a plurality of logical partitions when processing the target order; wherein one of the logical partitions is associated with at least one workstation and a plurality of inventory receptacles, and at least one of the plurality of inventory receptacles associated with the target logical partition contains inventory items required by the target order;
the target order distribution unit is used for distributing the target order to a station associated with the target logic partition to serve as a target station;
and the target order processing unit is used for controlling the robot to convey the target inventory container containing the inventory items required by the target order from the plurality of inventory containers associated with the target logical partition to the target station.
On the basis of the above embodiment, optionally, the application module 704 may include:
the logic partition determining subunit is used for determining a plurality of logic partitions according to the obtained plurality of commodity classes;
the first dividing and dividing unit is used for dividing the inventory containers in the inventory system into logic partitions corresponding to the commodity classes with high coincidence degrees according to the coincidence degrees between the inventory commodities on the inventory containers in the inventory system and the commodities in the commodity classes;
the second division subunit is used for dividing the stock container positions in the stock system into different logical partitions according to the number of the stock containers associated with each logical partition;
and the third dividing subunit is used for dividing the workstations in the inventory system into the logical partitions which are close to the workstations according to the distances between the workstations and the inventory container positions associated with the logical partitions.
On the basis of the above embodiment, optionally, the application module 704 may include:
and the order combining unit is used for combining different orders corresponding to different commodities in the same commodity class in the same order wave task according to the obtained commodity classes so as to distribute the order wave task to the same workstation to execute the picking task.
On the basis of the above embodiment, optionally, the application module 704 may include:
the first goods shelf adjusting unit is used for placing each commodity contained in the same commodity class on the inventory container meeting the preset distance condition according to the obtained plurality of commodity classes; or,
and the second shelf adjusting unit is used for peeling off a plurality of commodities which are placed on the inventory container meeting the preset distance condition and are not contained in the same commodity class according to the obtained commodity classes.
On the basis of the above embodiment, optionally, the inventory containers with the preset distance condition are the same inventory container, or multiple inventory containers with an inter-inventory-container distance smaller than or equal to a preset distance threshold.
On the basis of the above embodiment, optionally, the application module 704 may include:
the store dividing unit is used for dividing each store into different store combinations according to the contact ratio between the commodities contained in each commodity class and the commodities required by each store order;
and the sorting time distribution unit is used for distributing the same sorting time period to a plurality of stores belonging to the same store combination so as to enable the stores belonging to the same store combination to carry out commodity sorting operation in the same sorting time period.
On the basis of the above embodiment, optionally, the apparatus may further include:
the dimension reduction processing module 705 is configured to perform dimension reduction processing on the feature value of the commodity corresponding to each network node in the commodity association network, with one feature value as one dimension.
On the basis of the foregoing embodiment, optionally, the determining module 702 may include:
and the characteristic value determining unit is used for calculating the characteristic values of the commodities corresponding to the network nodes in the commodity association network by adopting a deep walking algorithm or a large-scale information network embedding algorithm.
On the basis of the foregoing embodiment, optionally, the feature value determining unit is configured to:
when random walk is carried out each time, uniformly and randomly sampling a network node from the commodity association network as a starting point of the random walk, uniformly and randomly sampling an adjacent point with associated weight for a last visited node in the walk process, ending the random walk until the walk reaches a preset maximum length, and finally obtaining a plurality of node sequences;
and training the plurality of node sequences through machine learning to obtain the characteristic value vector of the commodity corresponding to each network node.
On the basis of the foregoing embodiment, optionally, the feature value determining unit is configured to:
determining first-order similarity among network nodes in the commodity association network, and optimizing the first-order similarity among the network nodes to obtain a first characteristic value vector of a commodity corresponding to each network node;
determining second-order similarity among network nodes in the commodity correlation network, and optimizing the second-order similarity among the network nodes to obtain a second characteristic value vector of the commodity corresponding to each network node;
and splicing the first characteristic value vector and the second characteristic value vector to obtain the characteristic value vector of the commodity corresponding to each network node.
The commodity clustering device provided by the embodiment of the invention can execute the commodity clustering method provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the commodity clustering method.
Fig. 8 is a schematic structural diagram of an apparatus provided in an embodiment of the present invention. FIG. 8 illustrates a block diagram of an exemplary device 812 suitable for use in implementing embodiments of the present invention. The device 812 shown in fig. 8 is only an example and should not bring any limitations to the functionality or scope of use of embodiments of the present invention.
As shown in FIG. 8, device 812 may take the form of a general purpose computing device. Components of device 812 may include, but are not limited to: one or more processors or processing units 816, a system memory 828, and a bus 818 that couples various system components including the system memory 828 and the processing unit 816.
Bus 818 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus using any of a variety of bus architectures. By way of example, such architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, micro-channel architecture (MAC) bus, enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.
Device 812 typically includes a variety of computer system readable media. Such media can be any available media that is accessible by device 812 and includes both volatile and nonvolatile media, removable and non-removable media.
The system memory 828 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM)830 and/or cache memory 832. Device 812 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 834 may be used to read from or write to non-removable, nonvolatile magnetic media (not shown in FIG. 8, often referred to as a "hard disk drive"). Although not shown in FIG. 8, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 818 by one or more data media interfaces. Memory 828 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.
A program/utility 840 having a set (at least one) of program modules 842, including but not limited to an operating system, one or more application programs, other program modules, and program data, may be stored in, for example, memory 828, each of which examples or some combination thereof may include an implementation of a network environment. Program modules 842 generally perform the functions and/or methodologies of the described embodiments of the invention.
Device 812 may also communicate with one or more external devices 814 (e.g., keyboard, pointing device, display 824, etc.), with one or more devices that enable a user to interact with device 812, and/or with any devices (e.g., network card, modem, etc.) that enable device 812 to communicate with one or more other computing devices. Such communication may occur via input/output (I/O) interfaces 822. Also, device 812 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN) and/or a public network such as the Internet) through network adapter 820. As shown, the network adapter 820 communicates with the other modules of the device 812 over the bus 818. It should be appreciated that although not shown in FIG. 8, other hardware and/or software modules may be used in conjunction with device 812, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.
The processing unit 816 executes various functional applications and data processing by executing programs stored in the system memory 828, for example, implementing a method for clustering commodities provided in the embodiment of the present invention, the method including:
mapping commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes appearing together in a historical order as an association weight between the two network nodes, and constructing a commodity association network between the commodities;
determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm;
and clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes.
Of course, those skilled in the art can understand that the processor may also implement the technical solution in the method for clustering commodities provided in any embodiment of the present invention.
An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements a method for clustering commodities, where the method includes:
mapping commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes appearing together in a historical order as an association weight between the two network nodes, and constructing a commodity association network between the commodities;
determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm;
and clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes.
Of course, the storage medium provided in the embodiments of the present invention includes computer-executable instructions, and the computer-executable instructions are not limited to the operations of the method for clustering commodities described above, and may also perform related operations in the method for clustering commodities provided in any embodiment of the present invention, and have corresponding functions and advantages.
Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + +, python, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
Further, the embodiment of the invention also discloses the following contents:
a1, a method for commodity clustering, the method comprising:
mapping commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes appearing together in a historical order as an association weight between the two network nodes, and constructing a commodity association network between the commodities;
determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm;
and clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes.
a2, the method according to claim a1, wherein the method further comprises, after clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes:
and determining a commodity storage position mode, a group ordering wave mode, a shelf adjusting mode and/or a warehouse-crossing sorting mode in the warehouse according to the obtained multiple commodity classes.
a3, the method of claim a2, wherein the determining the storage position of the commodity in the warehouse according to the obtained commodity classes comprises:
dividing an inventory system of a warehouse into a plurality of logic partitions according to the obtained plurality of commodity classes; any inventory container and any station in the inventory system are provided with the logic partitions to which the inventory container and the station belong;
when a target order is processed, determining a logical partition to which the target order belongs from a plurality of logical partitions as a target logical partition; wherein one of the logical partitions is associated with at least one workstation and a plurality of inventory receptacles, and at least one of the plurality of inventory receptacles associated with the target logical partition contains inventory items required by the target order;
distributing the target order to a station associated with the target logic partition to serve as a target station;
and the control robot carries the target inventory container containing the inventory item required by the target order in the plurality of inventory containers associated with the target logical partition to the target station.
a4, the method of claim a3, wherein the method comprises the steps of dividing the inventory system of the warehouse into a plurality of logical partitions according to the obtained commodity classes, the method comprises the steps of:
determining a plurality of logic partitions according to the obtained commodity classes;
according to the contact ratio between inventory commodities on inventory containers in an inventory system and commodities in each commodity class, dividing the inventory containers in the inventory system into logic partitions corresponding to the commodity classes with high contact ratios;
dividing inventory container positions in an inventory system into different logical partitions according to the number of inventory containers associated with each logical partition;
the workstations in the inventory system are partitioned into logical partitions that are in close proximity to the workstations according to the distance between the workstations and the inventory receptacle locations associated with each logical partition.
a5, the method of claim a2, wherein the determining the order wave manner in the warehouse according to the obtained multiple commodity classes comprises:
according to the obtained multiple commodity classes, different orders corresponding to different commodities in the same commodity class are combined in the same order wave task, so that the order wave task is distributed to the same workstation to execute a picking task.
a6, the method of claim a2, wherein the determining the shelf adjustment mode in the warehouse according to the obtained multiple commodity classes comprises:
according to the obtained multiple commodity classes, placing each commodity contained in the same commodity class on an inventory container meeting a preset distance condition; or,
and according to the obtained multiple commodity classes, stripping multiple commodities which are placed on the inventory container meeting the preset distance condition and are not contained in the same commodity class.
a7, the method of claim a6, wherein the inventory containers of the preset distance condition are the same inventory container, or a plurality of inventory containers with a distance between inventory containers less than or equal to a preset distance threshold.
a8, the method of claim a2, wherein determining the cross-bin sorting method according to the obtained plurality of commodity classes, comprises:
dividing each store into different store combinations according to the contact ratio between the commodities contained in each commodity class and the commodities required by each store order;
the same sorting time period is allocated to a plurality of stores belonging to the same store group so that the plurality of stores belonging to the same store group perform the article sorting operation within the same sorting time period.
a9, the method according to claim a1, further comprising, after determining the feature value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm:
and taking one characteristic value as a dimension, and performing dimension reduction processing on the characteristic values of the commodities corresponding to the network nodes in the commodity association network.
a10, the method according to claim a1, wherein the determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm comprises:
and calculating the characteristic value of the commodity corresponding to each network node in the commodity association network by adopting a deep walking algorithm or a large-scale information network embedding algorithm.
a11, the method of claim a10, wherein the calculating the characteristic value of the commodity corresponding to each network node in the commodity association network by using the deep walking algorithm comprises:
when random walk is carried out each time, uniformly and randomly sampling a network node from the commodity association network as a starting point of the random walk, uniformly and randomly sampling an adjacent point with associated weight for a last visited node in the walk process, ending the random walk until the walk reaches a preset maximum length, and finally obtaining a plurality of node sequences;
and training the plurality of node sequences through machine learning to obtain the characteristic value vector of the commodity corresponding to each network node.
a12, the method of claim a10, wherein the calculating the characteristic value of the commodity corresponding to each network node in the commodity association network by using a large-scale information network embedding algorithm comprises:
determining first-order similarity among network nodes in the commodity association network, and optimizing the first-order similarity among the network nodes to obtain a first characteristic value vector of a commodity corresponding to each network node;
determining second-order similarity among network nodes in the commodity correlation network, and optimizing the second-order similarity among the network nodes to obtain a second characteristic value vector of the commodity corresponding to each network node;
and splicing the first characteristic value vector and the second characteristic value vector to obtain the characteristic value vector of the commodity corresponding to each network node.
a13, an apparatus for clustering commodities, the apparatus comprising:
the building module is used for mapping the commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes which commonly appear in the historical order into the association weight between the two network nodes, and building a commodity association network between the commodities;
the determining module is used for determining the characteristic values of the commodities corresponding to the network nodes in the commodity association network according to a graph embedding algorithm;
and the clustering module is used for clustering each commodity according to the characteristic value of the commodity corresponding to each network node to obtain a plurality of commodity classes.
a14, the apparatus of claim a13, the apparatus further comprising:
and the application module is used for determining a commodity storage position mode, a group ordering wave mode, a shelf adjusting mode and/or a warehouse-crossing sorting mode in the warehouse according to the obtained multiple commodity classes.
a15, the apparatus of claim a14, the application module comprising:
the logical partition dividing unit is used for dividing the inventory system of the warehouse into a plurality of logical partitions according to the obtained plurality of commodity classes; any inventory container and any station in the inventory system are provided with the logic partitions to which the inventory container and the station belong;
a target partition determining unit, configured to determine, as a target logical partition, a logical partition to which a target order belongs from among a plurality of logical partitions when processing the target order; wherein one of the logical partitions is associated with at least one workstation and a plurality of inventory receptacles, and at least one of the plurality of inventory receptacles associated with the target logical partition contains inventory items required by the target order;
the target order distribution unit is used for distributing the target order to a station associated with the target logic partition to serve as a target station;
and the target order processing unit is used for controlling the robot to convey the target inventory container containing the inventory items required by the target order from the plurality of inventory containers associated with the target logical partition to the target station.
a16, the apparatus of claim a15, the application module comprising:
the logic partition determining subunit is used for determining a plurality of logic partitions according to the obtained plurality of commodity classes;
the first dividing and dividing unit is used for dividing the inventory containers in the inventory system into logic partitions corresponding to the commodity classes with high coincidence degrees according to the coincidence degrees between the inventory commodities on the inventory containers in the inventory system and the commodities in the commodity classes;
the second division subunit is used for dividing the stock container positions in the stock system into different logical partitions according to the number of the stock containers associated with each logical partition;
and the third dividing subunit is used for dividing the workstations in the inventory system into the logical partitions which are close to the workstations according to the distances between the workstations and the inventory container positions associated with the logical partitions.
a17, the apparatus of claim a14, the application module comprising:
and the order combining unit is used for combining different orders corresponding to different commodities in the same commodity class in the same order wave task according to the obtained commodity classes so as to distribute the order wave task to the same workstation to execute the picking task.
a18, the apparatus of claim a14, the application module comprising:
the first goods shelf adjusting unit is used for placing each commodity contained in the same commodity class on the inventory container meeting the preset distance condition according to the obtained plurality of commodity classes; or,
and the second shelf adjusting unit is used for peeling off a plurality of commodities which are placed on the inventory container meeting the preset distance condition and are not contained in the same commodity class according to the obtained commodity classes.
a19, the apparatus of claim a18, wherein the inventory containers of the preset distance condition are the same inventory container, or a plurality of inventory containers with a distance between inventory containers less than or equal to a preset distance threshold.
a20, the apparatus of claim a14, the application module comprising:
the store dividing unit is used for dividing each store into different store combinations according to the contact ratio between the commodities contained in each commodity class and the commodities required by each store order;
and the sorting time distribution unit is used for distributing the same sorting time period to a plurality of stores belonging to the same store combination so as to enable the stores belonging to the same store combination to carry out commodity sorting operation in the same sorting time period.
a21, the apparatus of claim a13, the apparatus further comprising:
and the dimension reduction processing module is used for taking one characteristic value as a dimension and carrying out dimension reduction processing on the characteristic value of the commodity corresponding to each network node in the commodity association network.
a22, the apparatus of claim a13, the means for determining comprising:
and the characteristic value determining unit is used for calculating the characteristic values of the commodities corresponding to the network nodes in the commodity association network by adopting a deep walking algorithm or a large-scale information network embedding algorithm.
a23, the apparatus of claim a22, the eigenvalue determination unit to:
when random walk is carried out each time, uniformly and randomly sampling a network node from the commodity association network as a starting point of the random walk, uniformly and randomly sampling an adjacent point with associated weight for a last visited node in the walk process, ending the random walk until the walk reaches a preset maximum length, and finally obtaining a plurality of node sequences;
and training the plurality of node sequences through machine learning to obtain the characteristic value vector of the commodity corresponding to each network node.
a24, the method of claim a22, the eigenvalue determination unit being used to:
determining first-order similarity among network nodes in the commodity association network, and optimizing the first-order similarity among the network nodes to obtain a first characteristic value vector of a commodity corresponding to each network node;
determining second-order similarity among network nodes in the commodity correlation network, and optimizing the second-order similarity among the network nodes to obtain a second characteristic value vector of the commodity corresponding to each network node;
and splicing the first characteristic value vector and the second characteristic value vector to obtain the characteristic value vector of the commodity corresponding to each network node.
a25, an apparatus comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method for merchandise clustering as in any one of claims a1-a12 above.
a26, a computer readable storage medium having stored thereon a computer program which, when executed by a processor, carries out a method of clustering items according to any of the preceding claims a1-a 12.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims (10)

1. A method of commodity clustering, the method comprising:
mapping commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes appearing together in a historical order as an association weight between the two network nodes, and constructing a commodity association network between the commodities;
determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm;
and clustering the commodities according to the characteristic values of the commodities corresponding to the network nodes to obtain a plurality of commodity classes.
2. The method of claim 1, wherein after clustering each commodity according to the characteristic value of the commodity corresponding to each network node to obtain a plurality of commodity classes, the method further comprises:
and determining a commodity storage position mode, a group ordering wave mode, a shelf adjusting mode and/or a warehouse-crossing sorting mode in the warehouse according to the obtained multiple commodity classes.
3. The method according to claim 1, after determining the feature value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm, further comprising:
and taking one characteristic value as a dimension, and performing dimension reduction processing on the characteristic values of the commodities corresponding to the network nodes in the commodity association network.
4. The method of claim 1, wherein determining the characteristic value of the commodity corresponding to each network node in the commodity association network according to a graph embedding algorithm comprises:
and calculating the characteristic value of the commodity corresponding to each network node in the commodity association network by adopting a deep walking algorithm or a large-scale information network embedding algorithm.
5. The method according to claim 4, wherein calculating the characteristic value of the commodity corresponding to each network node in the commodity association network by using a deep walking algorithm comprises:
when random walk is carried out each time, uniformly and randomly sampling a network node from the commodity association network as a starting point of the random walk, uniformly and randomly sampling an adjacent point with associated weight for a last visited node in the walk process, ending the random walk until the walk reaches a preset maximum length, and finally obtaining a plurality of node sequences;
and training the plurality of node sequences through machine learning to obtain the characteristic value vector of the commodity corresponding to each network node.
6. The method according to claim 4, wherein calculating the characteristic value of the commodity corresponding to each network node in the commodity association network by using a large-scale information network embedding algorithm comprises:
determining first-order similarity among network nodes in the commodity association network, and optimizing the first-order similarity among the network nodes to obtain a first characteristic value vector of a commodity corresponding to each network node;
determining second-order similarity among network nodes in the commodity correlation network, and optimizing the second-order similarity among the network nodes to obtain a second characteristic value vector of the commodity corresponding to each network node;
and splicing the first characteristic value vector and the second characteristic value vector to obtain the characteristic value vector of the commodity corresponding to each network node.
7. An apparatus for clustering items, the apparatus comprising:
the building module is used for mapping the commodities into network nodes, mapping the frequency of the commodities corresponding to any two network nodes which commonly appear in the historical order into the association weight between the two network nodes, and building a commodity association network between the commodities;
the determining module is used for determining the characteristic values of the commodities corresponding to the network nodes in the commodity association network according to a graph embedding algorithm;
and the clustering module is used for clustering each commodity according to the characteristic value of the commodity corresponding to each network node to obtain a plurality of commodity classes.
8. The apparatus of claim 7, further comprising:
and the application module is used for determining a commodity storage position mode, a group ordering wave mode, a shelf adjusting mode and/or a warehouse-crossing sorting mode in the warehouse according to the obtained multiple commodity classes.
9. An apparatus, characterized in that the apparatus comprises:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method of merchandise clustering as in any one of claims 1-6 above.
10. A computer-readable storage medium, on which a computer program is stored, which program, when being executed by a processor, is adapted to carry out the method of clustering products according to any one of the preceding claims 1 to 6.
CN201910107025.4A 2018-12-30 2019-02-02 Commodity clustering method, device, equipment and storage medium Active CN111523918B (en)

Priority Applications (4)

Application Number Priority Date Filing Date Title
CN201910107025.4A CN111523918B (en) 2019-02-02 2019-02-02 Commodity clustering method, device, equipment and storage medium
US17/419,724 US12051038B2 (en) 2018-12-30 2019-12-26 Order processing method, apparatus and device, and storage medium
PCT/CN2019/128557 WO2020140818A1 (en) 2018-12-30 2019-12-26 Order processing method, apparatus and device, and storage medium
EP19907152.3A EP3905154A4 (en) 2018-12-30 2019-12-26 Order processing method, apparatus and device, and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910107025.4A CN111523918B (en) 2019-02-02 2019-02-02 Commodity clustering method, device, equipment and storage medium

Publications (2)

Publication Number Publication Date
CN111523918A true CN111523918A (en) 2020-08-11
CN111523918B CN111523918B (en) 2023-09-19

Family

ID=71900658

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910107025.4A Active CN111523918B (en) 2018-12-30 2019-02-02 Commodity clustering method, device, equipment and storage medium

Country Status (1)

Country Link
CN (1) CN111523918B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112364181A (en) * 2020-11-27 2021-02-12 深圳市慧择时代科技有限公司 Insurance product matching degree determination method and device
CN113111182A (en) * 2021-04-15 2021-07-13 北京沃东天骏信息技术有限公司 Information recommendation method and device and computer-readable storage medium
CN113763080A (en) * 2020-08-26 2021-12-07 北京沃东天骏信息技术有限公司 Method and device for determining recommended article, electronic equipment and storage medium
CN113763014A (en) * 2021-01-05 2021-12-07 北京沃东天骏信息技术有限公司 Article co-occurrence relation determining method and device and judgment model obtaining method and device
CN116402438A (en) * 2023-03-29 2023-07-07 乐麦信息技术(杭州)有限公司 Warehouse management method and system
TWI842973B (en) * 2020-08-31 2024-05-21 大陸商中國銀聯股份有限公司 A method and device for determining similarity between objects

Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101251859A (en) * 2008-03-17 2008-08-27 北京科文书业信息技术有限公司 Commercial articles clustering system and method based on shopping behaviors
JP2009205365A (en) * 2008-02-27 2009-09-10 Nec Corp System, method and program for optimizing inventory management and sales of merchandise
CN102254028A (en) * 2011-07-22 2011-11-23 青岛理工大学 Personalized commodity recommendation method and system integrating attributes and structural similarity
US20140100715A1 (en) * 2012-10-04 2014-04-10 Amazon Technologies, Inc. Filling an order at an inventory pier
CN105469237A (en) * 2015-11-18 2016-04-06 北京京东尚科信息技术有限公司 Method and system for automatic logistics processing
CN106446943A (en) * 2016-09-19 2017-02-22 大连海事大学 Commodity correlation big data sparse network quick clustering method
CN106886524A (en) * 2015-12-15 2017-06-23 天津科技大学 A kind of community network community division method based on random walk
CN106919921A (en) * 2017-03-06 2017-07-04 重庆邮电大学 With reference to sub-space learning and the gait recognition method and system of tensor neutral net
CN107545381A (en) * 2016-06-24 2018-01-05 北京京东尚科信息技术有限公司 Manage the method and warehouse management system of warehouse inventory
CN107851218A (en) * 2015-02-10 2018-03-27 D-波系统公司 For the system of quantum processor architecture, device, article and method
CN107944896A (en) * 2016-10-13 2018-04-20 北京京东尚科信息技术有限公司 Determine the method and device of commodity association degree
CN108053268A (en) * 2017-12-29 2018-05-18 广州品唯软件有限公司 A kind of commercial articles clustering confirmation method and device
CN108182621A (en) * 2017-12-07 2018-06-19 合肥美的智能科技有限公司 The Method of Commodity Recommendation and device for recommending the commodity, equipment and storage medium
CN108388651A (en) * 2018-02-28 2018-08-10 北京理工大学 A kind of file classification method based on the kernel of graph and convolutional neural networks
CN108846609A (en) * 2018-06-15 2018-11-20 北京极智嘉科技有限公司 Picking method, device, server and medium based on order taking responsibility
CN108898459A (en) * 2018-06-25 2018-11-27 中国联合网络通信集团有限公司 A kind of Method of Commodity Recommendation and device
CN108932598A (en) * 2018-05-31 2018-12-04 广东工业大学 A kind of grand scale logistic allocator based on cloud platform
CN109214730A (en) * 2017-06-29 2019-01-15 北京京东尚科信息技术有限公司 Information-pushing method and device

Patent Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2009205365A (en) * 2008-02-27 2009-09-10 Nec Corp System, method and program for optimizing inventory management and sales of merchandise
CN101251859A (en) * 2008-03-17 2008-08-27 北京科文书业信息技术有限公司 Commercial articles clustering system and method based on shopping behaviors
CN102254028A (en) * 2011-07-22 2011-11-23 青岛理工大学 Personalized commodity recommendation method and system integrating attributes and structural similarity
US20140100715A1 (en) * 2012-10-04 2014-04-10 Amazon Technologies, Inc. Filling an order at an inventory pier
CN107851218A (en) * 2015-02-10 2018-03-27 D-波系统公司 For the system of quantum processor architecture, device, article and method
CN105469237A (en) * 2015-11-18 2016-04-06 北京京东尚科信息技术有限公司 Method and system for automatic logistics processing
CN106886524A (en) * 2015-12-15 2017-06-23 天津科技大学 A kind of community network community division method based on random walk
CN107545381A (en) * 2016-06-24 2018-01-05 北京京东尚科信息技术有限公司 Manage the method and warehouse management system of warehouse inventory
CN106446943A (en) * 2016-09-19 2017-02-22 大连海事大学 Commodity correlation big data sparse network quick clustering method
CN107944896A (en) * 2016-10-13 2018-04-20 北京京东尚科信息技术有限公司 Determine the method and device of commodity association degree
CN106919921A (en) * 2017-03-06 2017-07-04 重庆邮电大学 With reference to sub-space learning and the gait recognition method and system of tensor neutral net
CN109214730A (en) * 2017-06-29 2019-01-15 北京京东尚科信息技术有限公司 Information-pushing method and device
CN108182621A (en) * 2017-12-07 2018-06-19 合肥美的智能科技有限公司 The Method of Commodity Recommendation and device for recommending the commodity, equipment and storage medium
CN108053268A (en) * 2017-12-29 2018-05-18 广州品唯软件有限公司 A kind of commercial articles clustering confirmation method and device
CN108388651A (en) * 2018-02-28 2018-08-10 北京理工大学 A kind of file classification method based on the kernel of graph and convolutional neural networks
CN108932598A (en) * 2018-05-31 2018-12-04 广东工业大学 A kind of grand scale logistic allocator based on cloud platform
CN108846609A (en) * 2018-06-15 2018-11-20 北京极智嘉科技有限公司 Picking method, device, server and medium based on order taking responsibility
CN108898459A (en) * 2018-06-25 2018-11-27 中国联合网络通信集团有限公司 A kind of Method of Commodity Recommendation and device

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
卢桂馥;林忠;金忠;: "具有统计不相关性的核化图嵌入算法", 中国图象图形学报, vol. 16, no. 04, pages 618 - 624 *
洪文兴;翁洋;朱顺痣;李茂青;: "垂直电子商务网站的混合型推荐系统", 系统工程理论与实践, no. 05, pages 928 - 935 *
王召义;薛晨杰;: "基于K-MEANS聚类的电商网站用户行为分析", 温州大学学报(自然科学版), no. 03, pages 53 - 58 *
蔡莉;梁宇;朱扬勇;何婧;: "数据质量的历史沿革和发展趋势", 计算机科学, no. 04, pages 7 - 16 *
谢桂林;詹志强;李凯;: "基于聚类的因子分解机推荐算法研究", 软件, vol. 37, no. 10, pages 113 - 117 *

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113763080A (en) * 2020-08-26 2021-12-07 北京沃东天骏信息技术有限公司 Method and device for determining recommended article, electronic equipment and storage medium
TWI842973B (en) * 2020-08-31 2024-05-21 大陸商中國銀聯股份有限公司 A method and device for determining similarity between objects
CN112364181A (en) * 2020-11-27 2021-02-12 深圳市慧择时代科技有限公司 Insurance product matching degree determination method and device
CN112364181B (en) * 2020-11-27 2024-05-28 深圳市慧择时代科技有限公司 Insurance product matching degree determining method and apparatus
CN113763014A (en) * 2021-01-05 2021-12-07 北京沃东天骏信息技术有限公司 Article co-occurrence relation determining method and device and judgment model obtaining method and device
CN113111182A (en) * 2021-04-15 2021-07-13 北京沃东天骏信息技术有限公司 Information recommendation method and device and computer-readable storage medium
WO2022218181A1 (en) * 2021-04-15 2022-10-20 北京沃东天骏信息技术有限公司 Information recommendation method and device and computer storable medium
CN116402438A (en) * 2023-03-29 2023-07-07 乐麦信息技术(杭州)有限公司 Warehouse management method and system

Also Published As

Publication number Publication date
CN111523918B (en) 2023-09-19

Similar Documents

Publication Publication Date Title
CN110197309B (en) Order processing method, device, equipment and storage medium
CN111523918B (en) Commodity clustering method, device, equipment and storage medium
Zeng et al. Robotic pick-and-place of novel objects in clutter with multi-affordance grasping and cross-domain image matching
US11182743B2 (en) Order processing method and device, server, and storage medium
CN111557013B (en) Order grouping in warehouse order fulfillment operations
CN109117999B (en) Seeding bit distribution method, device, server and medium
WO2020140818A1 (en) Order processing method, apparatus and device, and storage medium
WO2020238657A1 (en) Goods sorting method and goods sorting system
CN109733783B (en) Goods shelving method and device, electronic equipment and storage medium
CN109658027A (en) A kind of processing method of order taking responsibility, device, server and medium
US12083681B2 (en) Robot congestion management
EP4242945A1 (en) Warehouse management system and method
CN110390449B (en) Warehouse replenishment method and device
WO2023020105A1 (en) Supplies inventory method and apparatus, and device and storage medium
CN113646789A (en) Robot dwell time minimization in warehouse order execution operations
CN114229312B (en) Cargo handling method, device, server and storage medium
KR102687433B1 (en) Optimized tote recommendation process in warehouse fulfillment operations
CN113269494A (en) Inventory scheduling method, equipment and system
CN109118150B (en) Commodity volume estimation method and device, computer equipment and storage medium
US10878374B1 (en) Techniques for utilizing multi-functional storage containers
CN111382969A (en) Order processing method, device, equipment and storage medium
Chen et al. Warehouses without aisles: Layout design of a multi-deep rack climbing robotic system
WO2024093758A1 (en) Transportation scheduling method and apparatus for vertical warehouse
KR20210001706A (en) Method and system for managing a storage based on a cloud service
CN113570179A (en) Container determination method and device, electronic equipment and computer-readable storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
CB02 Change of applicant information
CB02 Change of applicant information

Address after: 100020 No. 1 Building, No. 36 Chuangyuan Road, Chaoyang District, Beijing 101

Applicant after: Beijing jizhijia Technology Co.,Ltd.

Address before: 100020 No. 1 Building, No. 36 Chuangyuan Road, Chaoyang District, Beijing 101

Applicant before: Beijing Geekplus Technology Co.,Ltd.

GR01 Patent grant
GR01 Patent grant