CN116628138A - Logistics order text mining method and system applied to deep learning - Google Patents

Logistics order text mining method and system applied to deep learning Download PDF

Info

Publication number
CN116628138A
CN116628138A CN202310721340.2A CN202310721340A CN116628138A CN 116628138 A CN116628138 A CN 116628138A CN 202310721340 A CN202310721340 A CN 202310721340A CN 116628138 A CN116628138 A CN 116628138A
Authority
CN
China
Prior art keywords
logistics
time interval
target time
estimation
description text
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.)
Pending
Application number
CN202310721340.2A
Other languages
Chinese (zh)
Inventor
周仁富
陈荣
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.)
Jiangsu Saifei Information Technology Co ltd
Original Assignee
Jiangsu Saifei Information 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 Jiangsu Saifei Information Technology Co ltd filed Critical Jiangsu Saifei Information Technology Co ltd
Priority to CN202310721340.2A priority Critical patent/CN116628138A/en
Publication of CN116628138A publication Critical patent/CN116628138A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33Querying
    • G06F16/3331Query processing
    • G06F16/334Query execution
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q10/00Administration; Management
    • G06Q10/08Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/083Shipping
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION 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
    • G06Q30/00Commerce
    • G06Q30/06Buying, selling or leasing transactions
    • G06Q30/0601Electronic shopping [e-shopping]
    • G06Q30/0633Lists, e.g. purchase orders, compilation or processing
    • G06Q30/0635Processing of requisition or of purchase orders
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Tourism & Hospitality (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Operations Research (AREA)
  • Human Resources & Organizations (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Quality & Reliability (AREA)

Abstract

The invention provides a logistics order text mining method and system applied to deep learning, and relates to the technical field of artificial intelligence. In the invention, a logistics transportation state description vector of a target time interval is excavated, and a logistics order text to be analyzed corresponding to the target time interval is determined; performing key information mining operation on the to-be-analyzed logistics order text to output logistics article data description vectors corresponding to the target time intervals; carrying out aggregation operation of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval so as to output an aggregation multi-dimensional description vector corresponding to the target time interval; and carrying out logistics object estimation operation based on the aggregate multidimensional description vector, and outputting logistics object estimation data corresponding to the target time interval. Based on the above, the reliability of the logistics item estimation can be improved to a certain extent.

Description

Logistics order text mining method and system applied to deep learning
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a logistics order text mining method and system applied to deep learning.
Background
The logistics order data comprises more effective information, such as information recorded on logistics articles, logistics users and the like. Therefore, more effective information can be obtained by mining the logistics order data, for example, the logistics order data can be mined based on an artificial intelligence technology (artificial intelligence, artificial Intelligence, AI for short, is a theory, a method, a technology and an application system for simulating, extending and expanding human intelligence by using a digital computer or a digital computer control calculation, sensing environment, acquiring knowledge and obtaining an optimal result by using the knowledge. However, in the prior art, there is a problem of poor reliability in analyzing the physical distribution order data, specifically, a problem of poor reliability in estimating physical distribution articles.
Disclosure of Invention
In view of the above, the present invention is directed to a method and a system for mining a physical distribution order text applied to deep learning, so as to improve the reliability of physical distribution object estimation to a certain extent.
In order to achieve the above purpose, the embodiment of the present invention adopts the following technical scheme:
a method of logistics order text mining for deep learning, the method of logistics order text mining comprising:
excavating a logistics transportation state description vector of a target time interval, and determining a logistics order text to be analyzed corresponding to the target time interval, wherein the logistics order text to be analyzed comprises an object description text segment for describing a logistics object, and the logistics object refers to a logistics transported object;
performing key information mining operation on the to-be-analyzed logistics order text to output logistics article data description vectors corresponding to the target time interval;
carrying out aggregation operation of different dimensionalities on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval so as to output an aggregation multi-dimensional description vector corresponding to the target time interval;
And carrying out logistics object estimation operation based on the aggregate multi-dimensional description vector, and outputting logistics object estimation data corresponding to the target time interval, wherein the logistics object estimation data is used for reflecting the estimated transportation quantity or the estimated transaction quantity of the objects in the appointed time interval after the target time interval.
In some preferred embodiments, in the above method for mining a logistics order text applied to deep learning, the step of performing a key information mining operation on the logistics order text to be analyzed to output a logistics article data description vector corresponding to the target time interval includes:
extracting an object description text fragment set configured for the target time interval;
performing a comparison analysis operation on the object description text fragments in the object description text fragment set and the object description text fragments in the to-be-analyzed logistics order text to output important object description text fragments corresponding to the target time interval, wherein the important object description text fragments refer to the object description text fragments belonging to the object description text fragment set and the to-be-analyzed logistics order text at the same time;
According to the to-be-analyzed logistics order text, determining the fragment distribution state of the important object description text fragment so as to form a fragment distribution state characterization parameter corresponding to the important object description text fragment;
performing feature space mapping operation on the important object description text fragment to form a fragment feature space mapping vector corresponding to the important object description text fragment;
and marking the data merging result of the fragment characteristic space mapping vector corresponding to the important object description text fragment and the fragment distribution state characterization parameter of the important object description text fragment pair so as to be marked as the logistics article data description vector corresponding to the target time interval.
In some preferred embodiments, in the above method for text mining a logistic order applied to deep learning, the step of extracting the set of object description text fragments configured for the target time interval includes:
analyzing first matching relation characterization parameters between a plurality of undetermined object description text fragments corresponding to the target time interval and a predetermined specified object description text fragment;
Determining the plurality of undetermined object description text fragments according to the first matching relation characterization parameters so as to output object description text fragments for optimizing fragment sets;
performing optimization operation on the historical object description text fragment set according to the object description text fragment for optimizing the fragment set to form an optimized historical object description text fragment set;
analyzing a second matching relation characterization parameter between the object description text fragment and the appointed object description text fragment in the optimized historical object description text fragment set;
determining the object description text fragments in the optimized historical object description text fragment set according to the second matching relation characterization parameters so as to output a plurality of determined object description text fragments;
and determining an object description text fragment set corresponding to the target time interval according to the plurality of determined object description text fragments.
In some preferred embodiments, in the above-described method for mining a logistics order text for deep learning, before the step of extracting the set of object description text fragments configured for the target time interval, the method for mining a logistics order text for deep learning further includes:
Determining the estimation data of the example logistics articles corresponding to the example time interval, and determining the identification data of the example logistics articles corresponding to the example time interval;
determining that the extraction to the set of object description text fragments configured for the target time interval is required to be performed in a case where the degree of distinction between the exemplary physical distribution item estimation data and the exemplary physical distribution item identification data exceeds a pre-configured reference degree of distinction;
in the case where the degree of distinction between the exemplary physical distribution item estimation data and the exemplary physical distribution item identification data does not exceed the reference degree of distinction configured in advance, it is determined that the step of extracting the set of object description text fragments configured for the target time interval does not need to be performed.
In some preferred embodiments, in the above method for mining a logistics order text applied to deep learning, the step of performing feature space mapping operation on the important object description text segment to form a segment feature space mapping vector corresponding to the important object description text segment includes:
performing feature space mapping operation on the important object description text fragments to form fragment feature space mapping results corresponding to the important object description text fragments;
And performing depth mapping operation on the fragment characteristic space mapping result corresponding to the important object description text fragment to form a fragment characteristic space mapping vector corresponding to the important object description text fragment.
In some preferred embodiments, in the above-mentioned method for mining a logistics order text applied to deep learning, the deep mapping operation is performed by using a deep mapping network model, where the deep mapping network model includes a plurality of linear integration units connected in sequence;
the step of performing depth mapping operation on the segment feature space linear integration vector corresponding to the important object description text segment to form a segment feature space mapping vector corresponding to the important object description text segment includes:
performing linear integration operation on the segment feature space linear integration vector corresponding to the important object description text segment by using a first linear integration unit in the plurality of linear integration units which are sequentially connected in sequence so as to output a corresponding linear integration vector;
loading the linear integration vector corresponding to the first linear integration unit into a linear integration unit connected with the next linear integration unit, and performing linear integration operation by utilizing the next linear integration unit to output the linear integration vector corresponding to the next linear integration unit, wherein the linear integration vectors are sequentially performed to load the linear integration vector corresponding to the last but one linear integration unit into the last linear integration unit;
And determining a segment feature space mapping vector corresponding to the important object description text segment based on a linear integration vector obtained by the linear integration operation of the last linear integration unit.
In some preferred embodiments, in the above-mentioned method for mining a logistics order text for deep learning, before the step of performing aggregation operations of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics item data description vector corresponding to the target time interval to output an aggregated multi-dimensional description vector corresponding to the target time interval, the method for mining a logistics order text for deep learning further includes:
mapping operation of a preset space is carried out on the logistics transportation state description vector of the target time interval so as to form a corresponding mapped logistics transportation state description vector;
carrying out mapping operation of a preset space on the segment distribution state characterization parameters in the logistics object data description vector to form a mapped logistics object data description vector, wherein the mapped logistics object data description vector comprises a mapping vector of the segment distribution state characterization parameters and a segment feature space mapping vector corresponding to an important object description text segment included in the logistics order text to be analyzed;
The step of performing aggregation operations of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval to output an aggregation multi-dimensional description vector corresponding to the target time interval includes:
performing aggregation operation of different dimensions on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregation multi-dimensional description vector corresponding to the target time interval;
the step of performing aggregation operations of different dimensions on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval includes:
performing cascading combination operation on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval; or, performing superposition operation on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval.
In some preferred embodiments, in the above method for mining a logistics order text applied to deep learning, the step of performing a logistics item estimation operation based on the aggregate multi-dimensional description vector and outputting logistics item estimation data corresponding to the target time interval includes:
carrying out logistics object estimation operation on the aggregate multidimensional description vector according to a plurality of logistics object estimation neural networks to form logistics object estimation data corresponding to each logistics object estimation neural network;
and combining the logistics object estimation data corresponding to each logistics object estimation neural network according to the network estimation reliability parameters of each logistics object estimation neural network so as to output the logistics object estimation data corresponding to the target time interval.
In some preferred embodiments, in the above-mentioned method for mining a physical distribution order text applied to deep learning, before the step of merging physical distribution item estimation data corresponding to each physical distribution item estimation neural network according to the network estimation reliability parameter of each physical distribution item estimation neural network to output physical distribution item estimation data corresponding to the target time interval, the method for mining a physical distribution order text applied to deep learning further includes:
Performing a preset operation for an a-th logistic article estimation neural network of the plurality of logistic article estimation neural networks, wherein the a-th logistic article estimation neural network belongs to any one logistic article estimation neural network of the plurality of logistic article estimation neural networks, and the preset operation comprises:
performing logistics object estimation operation on the aggregate multidimensional description vector of the example time interval by using the a-th logistics object estimation neural network to output example logistics object estimation data corresponding to the a-th logistics object estimation neural network;
according to the exemplary logistics object identification data corresponding to the exemplary time interval and the exemplary logistics object estimation data corresponding to the a-th logistics object estimation neural network, analyzing network analysis indexes corresponding to the a-th logistics object estimation neural network;
analyzing the network estimation reliability parameters of the a-th logistics object estimation neural network according to the network analysis indexes corresponding to the a-th logistics object estimation neural network and the network analysis indexes corresponding to the plurality of logistics object estimation neural networks;
the step of analyzing the network analysis index corresponding to the a-th physical distribution item estimation neural network according to the physical distribution item identification data corresponding to the example time interval and the example physical distribution item estimation data corresponding to the a-th physical distribution item estimation neural network includes:
Calculating expected deviation distances for the exemplary logistics object identification data corresponding to the exemplary time interval and the exemplary logistics object estimation data corresponding to the a-th logistics object estimation neural network so as to output corresponding first expected deviation distances;
performing interval mapping operation of parameters on the first expected deviation distance to form a network analysis index corresponding to the a-th logistics object estimation neural network;
the step of analyzing the network estimation reliability parameter of the a-th physical distribution item estimation neural network according to the network analysis index corresponding to the a-th physical distribution item estimation neural network and the network analysis indexes corresponding to the multiple physical distribution item estimation neural networks respectively comprises the following steps:
extracting a pre-configured reference parameter;
calculating the expected deviation distance of the reference parameter and the network analysis index corresponding to the a-th logistics object estimation neural network so as to output a corresponding second expected deviation distance;
and performing interval mapping operation of parameters on the second expected deviation distance according to network analysis indexes corresponding to the plurality of logistics object estimation neural networks to form network estimation reliability parameters of the a-th logistics object estimation neural network.
The embodiment of the invention also provides a logistics order text mining system applied to deep learning, which comprises a processor and a memory, wherein the memory is used for storing a computer program, and the processor is used for executing the computer program so as to realize the logistics order text mining method applied to deep learning.
The logistics order text mining method and system applied to deep learning provided by the embodiment of the invention can be used for firstly mining the logistics transportation state description vector of the target time interval and determining the logistics order text to be analyzed corresponding to the target time interval; performing key information mining operation on the to-be-analyzed logistics order text to output logistics article data description vectors corresponding to the target time intervals; carrying out aggregation operation of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval so as to output an aggregation multi-dimensional description vector corresponding to the target time interval; and carrying out logistics object estimation operation based on the aggregate multidimensional description vector, and outputting logistics object estimation data corresponding to the target time interval. Based on the foregoing, the basis for performing the logistics article estimation operation is the aggregate multi-dimensional description vector, and the aggregate multi-dimensional description vector is obtained by aggregating the information of the two aspects of the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval, so that the basis for the logistics article estimation operation is more reliable, the result of the logistics article estimation operation is also more reliable, thereby improving the reliability of the logistics article estimation to a certain extent and improving the defects in the prior art.
In order to make the above objects, features and advantages of the present invention more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
Fig. 1 is a block diagram of a logistics order text mining system applied to deep learning according to an embodiment of the present invention.
Fig. 2 is a schematic flow chart of steps included in a method for mining a logistics order text for deep learning according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of each module included in the logistic order text mining device applied to deep learning according to an embodiment of the present invention.
Description of the embodiments
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in FIG. 1, the embodiment of the invention provides a logistics order text mining system applied to deep learning. Wherein the logistic order text mining system applied to deep learning can comprise a memory and a processor.
In detail, the memory and the processor are electrically connected directly or indirectly to realize transmission or interaction of data. For example, electrical connection may be made to each other via one or more communication buses or signal lines. The memory may store at least one software functional module (computer program) that may exist in the form of software or firmware. The processor may be configured to execute the executable computer program stored in the memory, so as to implement the logistic order text mining method applied to deep learning provided by the embodiment of the present invention.
It should be appreciated that in some possible embodiments, the Memory may be, but is not limited to, random access Memory (Random Access Memory, RAM), read Only Memory (ROM), programmable Read Only Memory (Programmable Read-Only Memory, PROM), erasable Read Only Memory (Erasable Programmable Read-Only Memory, EPROM), electrically erasable Read Only Memory (Electric Erasable Programmable Read-Only Memory, EEPROM), and the like.
It should be appreciated that in some possible embodiments, the processor may be a general purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), a System on Chip (SoC), etc.; but also Digital Signal Processors (DSPs), application Specific Integrated Circuits (ASICs), field Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
It should be appreciated that in some possible embodiments, the logistic order text mining system applied to deep learning may be a server with data processing capabilities.
With reference to fig. 2, the embodiment of the invention further provides a logistics order text mining method applied to deep learning, which can be applied to the logistics order text mining system applied to deep learning. The method steps defined by the flow related to the logistics order text mining method applied to the deep learning can be realized by the logistics order text mining system applied to the deep learning.
The specific flow shown in fig. 2 will be described in detail.
Step S110, a logistics transportation state description vector of a target time interval is mined, and a logistics order text to be analyzed corresponding to the target time interval is determined.
In the embodiment of the invention, the logistics order text mining system applied to deep learning can mine the logistics transportation state description vector of the target time interval and determine the logistics order text to be analyzed corresponding to the target time interval. The to-be-analyzed logistics order text comprises an object description text segment (such as data for describing names, attributes and the like of logistics objects) for describing the logistics objects, wherein the logistics objects refer to articles transported by logistics. In addition, the target time interval may be the latest time interval, and the specific time length is not limited and is configured according to actual requirements.
And step S120, carrying out key information mining operation on the to-be-analyzed logistics order text so as to output the logistics article data description vector corresponding to the target time interval.
In the embodiment of the invention, the logistic order text mining system applied to deep learning can perform key information mining operation on the logistic order text to be analyzed so as to output the logistic article data description vector corresponding to the target time interval.
Step S130, performing aggregation operations of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval, so as to output an aggregation multidimensional description vector corresponding to the target time interval.
In the embodiment of the invention, the logistic order text mining system applied to deep learning can perform different-dimension aggregation operation on the logistic transportation state description vector corresponding to the target time interval and the logistic article data description vector corresponding to the target time interval so as to output an aggregation multidimensional description vector corresponding to the target time interval.
And step S140, carrying out logistics object estimation operation based on the aggregate multi-dimensional description vector, and outputting logistics object estimation data corresponding to the target time interval.
In the embodiment of the invention, the logistics order text mining system applied to deep learning can perform logistics object estimation operation based on the aggregate multidimensional description vector and output logistics object estimation data corresponding to the target time interval. The logistics article estimation data are used for reflecting the estimated traffic volume or the estimated transaction volume of the articles in the appointed time interval after the target time interval, namely, the estimated traffic volume or the estimated transaction volume, and the traffic volume and the transaction volume belong to equal or positively correlated data. The specified time interval may refer to a time interval subsequent to the target time interval, for example, a separation point between the specified time interval and the target time interval is not the current time, and in addition, the interval length of the specified time interval may be less than or equal to the interval length of the target time interval.
Based on the foregoing (i.e. step S110-step S140), the basis for performing the physical distribution item estimation operation is the aggregate multi-dimensional description vector, and the aggregate multi-dimensional description vector is obtained by aggregating the information of the physical distribution transportation state description vector corresponding to the target time interval and the physical distribution item data description vector corresponding to the target time interval, so that the basis for the physical distribution item estimation operation is more reliable, and therefore, the result of the physical distribution item estimation operation is also more reliable, thereby improving the reliability of physical distribution item estimation to a certain extent and improving the defects in the prior art.
It should be understood that, in some possible embodiments, step S110 in the foregoing description, that is, the step of mining the logistics transportation status description vector of the target time interval, may further include the following specific implementation matters:
acquiring logistics transportation state information of each logistics transportation road in a target area in the target time interval, and performing feature space mapping operation on the logistics transportation state information to form transportation state mapping vectors corresponding to each logistics transportation road, wherein the logistics transportation state information can comprise information such as traffic flow, speed, accidents, congestion states and the like of the corresponding logistics transportation roads, and the target area can refer to a city area, a province area or a country;
Combining the transportation state mapping vectors corresponding to each logistics transportation road to form a corresponding vector combining and distributing array;
determining relevant state information between every two logistics transportation roads, wherein the relevant state information is used for representing whether the two corresponding logistics transportation roads are connected or not, for example, when the two logistics transportation roads are connected, the relevant state information is represented by a first numerical value, such as 1, and when the two logistics transportation roads are not connected, the relevant state information is represented by a second numerical value, such as 0;
constructing a relevant state distribution array based on relevant state information between every two logistics transportation roads, wherein in the relevant state distribution array, the same row of array parameters represent relevant state information of one logistics transportation road and each logistics transportation road (all logistics transportation roads) respectively, and the same row of array parameters represent relevant state information of one logistics transportation road and each logistics transportation road respectively, so that the relevant state information between one logistics transportation road and the relevant state information between the logistics transportation road and the relevant state information can be represented by a first numerical value, namely the relevant state information represents that the relevant state information is communicated with the relevant state information; and, the correlation state distribution array may be a symmetric square matrix;
Loading the vector merging distribution array and the relevant state distribution array to load the vector merging distribution array and the relevant state distribution array into a target filtering neural network;
utilizing each filtering unit in a plurality of filtering units which are sequentially connected and included in the target filtering neural network, utilizing the filtering unit to carry out filtering operation on the data to be processed corresponding to the filtering unit so as to output the output data corresponding to the filtering unit, wherein the data to be processed corresponding to each filtering unit comprises an array distribution to be processed and a relevant state distribution array, the array distribution to be processed corresponding to the first filtering unit is the vector merging distribution array, the array distribution to be processed corresponding to each other filtering unit except the first filtering unit is the aggregation result of the array distribution to be processed of the previous filtering unit and the output data corresponding to the previous filtering unit, and the aggregation result refers to the result of the aggregation operation on the array distribution to be processed and the output data; for example, when the filtering operation is performed on the data to be processed, the cascade combination operation may be performed on the array to be processed and the related state distribution array included in the processing to be processed, and then the filtering operation is performed on the cascade combination distribution array obtained by the cascade combination operation, so as to obtain corresponding output data;
And carrying out aggregation operation on the array distribution to be processed of the last filtering unit and the output data corresponding to the last filtering unit to form the logistics transportation state description vector of the target time interval.
It should be understood that, in some possible embodiments, the step of aggregating the to-be-processed array distribution of the last filtering unit and the output data corresponding to the last filtering unit to form the logistics transportation state description vector of the target time interval may further include the following specific implementation matters:
and carrying out superposition operation or weighted superposition operation on the array distribution to be processed of the last filtering unit and output data corresponding to the last filtering unit to form a logistics transportation state description vector of the target time interval, wherein a weighting coefficient can be used as a network parameter of the neural network to be formed in an optimized mode.
It should be understood that, in some possible embodiments, step S120 in the foregoing description, that is, the step of performing the key information mining operation on the to-be-analyzed logistics order text to output the logistics item data description vector corresponding to the target time interval, may further include the following specific implementation matters:
Extracting an object description text fragment set configured for the target time interval;
performing a comparison analysis operation on the object description text segments in the object description text segment set and the object description text segments in the to-be-analyzed logistics order text to output important object description text segments corresponding to the target time interval, wherein the important object description text segments refer to the object description text segments belonging to the object description text segment set and the to-be-analyzed logistics order text at the same time, namely, coincident object description text segments, that is, the object description text segments belonging to the object description text segment set in the object description text segments included in the to-be-analyzed logistics order text can be used as important object description text segments;
according to the to-be-analyzed logistics order text, carrying out fragment distribution state determining operation on the important object description text fragments to form fragment distribution state characterization parameters corresponding to the important object description text fragments, wherein the fragment distribution state characterization parameters can be used for reflecting the quantity proportion of the important object description text fragments in all the object description text fragments included in the to-be-analyzed logistics order text, namely, the quantity proportion can be positively correlated with the fragment distribution state characterization parameters;
Performing feature space mapping operation on the important object description text fragment to form a fragment feature space mapping vector corresponding to the important object description text fragment;
and marking the segment feature space mapping vector corresponding to the important object description text segment and the data merging result of the segment distribution state characterization parameters of the important object description text segment pair so as to be marked as a logistics item data description vector corresponding to the target time interval, namely, the logistics item data description vector can comprise information of two dimensions of the segment feature space mapping vector and the segment distribution state characterization parameters.
It should be appreciated that, in some possible embodiments, the step of extracting the set of object description text segments configured for the target time interval may further include the following specific implementation matters:
analyzing first matching relation characterization parameters between a plurality of undetermined object description text fragments corresponding to the target time interval and predetermined specified object description text fragments, wherein the specified object description text fragments can refer to description texts of one logistics object or one type of logistics objects or description texts of multiple types of logistics objects or multiple types of logistics objects, so that the logistics object estimation data can refer to estimated transportation quantity or estimated transaction quantity of the corresponding logistics object, one type of logistics objects, multiple types of logistics objects or multiple types of logistics objects; in addition, the first matching relationship characterization parameter may refer to a text similarity between the text segment of the object description to be specified and the text segment of the object description to be specified, and the specific calculation manner may refer to the related prior art;
According to the first matching relation characterization parameters, determining the plurality of undetermined object description text fragments to output object description text fragments for optimizing a fragment set, wherein one or more undetermined object description text fragments with the largest first matching relation characterization parameters can be used as the object description text fragments for optimizing the fragment set, or undetermined object description text fragments corresponding to each first matching relation characterization parameter larger than preset matching relation characterization parameters can be used as the object description text fragments for optimizing the fragment set, and specific values of the preset matching relation characterization parameters are not limited and can be configured according to actual requirements;
performing an optimization operation on the historical object description text fragment set according to the object description text fragment for optimizing the fragment set to form an optimized historical object description text fragment set, and exemplarily, the object description text fragment for optimizing the fragment set may be distributed into the historical object description text fragment set to form an optimized historical object description text fragment set, where the historical object description text fragment set may refer to a previously formed object description text fragment set, for example, the object description text fragment set also belongs to the historical object description text fragment set for a logistics item estimation operation performed for a period of time later;
Analyzing a second matching relation characterization parameter between the object description text segment in the optimized historical object description text segment set and the specified object description text segment, wherein the second matching relation characterization parameter can refer to text similarity between the object description text segment in the optimized historical object description text segment set and the specified object description text segment, and a specific calculation mode can refer to the related prior art;
determining the object description text segments in the optimized historical object description text segment set according to the second matching relation characterization parameters to output a plurality of determined object description text segments, wherein one or more object description text segments with the maximum second matching relation characterization parameters can be used as determined object description text segments, or object description text segments corresponding to each second matching relation characterization parameter larger than the reference matching relation characterization parameters can be used as determined object description text segments, and the specific values of the reference matching relation characterization parameters are not limited and can be configured according to actual requirements;
And determining an object description text fragment set corresponding to the target time interval according to the determined object description text fragments, wherein the object description text fragment set can comprise the determined object description text fragments.
It should be appreciated that, in some possible embodiments, before the step of extracting the set of object description text segments configured for the target time interval, the method for mining a logistics order text applied to deep learning may further include the following specific implementation matters:
determining the estimation data of the example logistics articles corresponding to the example time interval, and determining the identification data of the example logistics articles corresponding to the example time interval, wherein the example time interval can be a time interval before the target time interval;
in the case where the degree of distinction between the exemplary physical distribution item estimation data and the exemplary physical distribution item identification data exceeds a pre-configured reference degree of distinction, it is determined that the step of extracting the set of object description text fragments configured for the target time interval needs to be performed, that is, the accuracy of the physical distribution item estimation operation has been relatively low, and thus, it is necessary to redetermine the set of object description text fragments, the specific value of the reference degree of distinction being not limited;
In the case where the degree of distinction between the exemplary physical distribution item estimation data and the exemplary physical distribution item identification data does not exceed the pre-configured reference degree of distinction, it is determined that the step of extracting the set of object description text fragments configured for the target time interval does not need to be performed, that is, the accuracy of the physical distribution item estimation operation is relatively high, the set of object description text fragments does not need to be redetermined, and the set of historical object description text fragments can be directly utilized.
It should be understood that, in some possible embodiments, the step of performing feature space mapping operation on the important object description text segment to form a segment feature space mapping vector corresponding to the important object description text segment may further include the following specific implementation matters:
performing feature space mapping operation on the important object description text fragments to form fragment feature space mapping results corresponding to the important object description text fragments, namely converting discrete text data into continuous vectors to be represented, so as to realize feature space mapping operation;
And performing depth mapping operation on the segment feature space mapping result corresponding to the important object description text segment to form a segment feature space mapping vector corresponding to the important object description text segment, namely performing depth feature mining analysis, and improving the expression capability of the segment feature space mapping vector.
It should be understood that, in some possible embodiments, the depth mapping operation is performed by using a depth mapping network model, where the depth mapping network model includes a plurality of linear integration units sequentially connected, for example, an output terminal of a first linear integration unit is connected to an input terminal of a second linear integration unit, and an output terminal of the second linear integration unit is connected to an input terminal of a third linear integration unit, and based on this, the step of performing depth mapping operation on a segment feature space linear integration vector corresponding to the important object description text segment to form a segment feature space mapping vector corresponding to the important object description text segment may further include the following specific implementation matters described below:
performing linear integration operation on the segment feature space linear integration vector corresponding to the important object description text segment by using a first linear integration unit in the plurality of linear integration units which are sequentially connected to output a corresponding linear integration vector, wherein the linear integration unit may include a plurality of neurons, the plurality of neurons may be sequentially connected and connected in parallel, and for each neuron, the neuron may include a first parameter and a second parameter, the first parameter is used for weighting input data, and the second parameter is used for superposing weighted results, so that the superposed results may be used as output data of the neuron;
Loading the linear integration vector corresponding to the first linear integration unit into a linear integration unit connected with the next linear integration unit, and performing linear integration operation by using the next linear integration unit to output the linear integration vector corresponding to the next linear integration unit, wherein the linear integration vectors are sequentially performed to load the linear integration vector corresponding to the last but one linear integration unit (namely corresponding output data) into the last linear integration unit;
and determining a segment feature space mapping vector corresponding to the important object description text segment based on a linear integration vector obtained by the linear integration operation of the last linear integration unit.
It should be understood that, in some possible embodiments, the step of loading the linear integration vector corresponding to the first linear integration unit into the linear integration unit connected to the subsequent linear integration unit and performing the linear integration operation by using the subsequent linear integration unit to output the linear integration vector corresponding to the subsequent linear integration unit may further include the following specific implementation matters:
and loading the linear integration vector corresponding to the first linear integration unit into a linear integration unit (namely a second linear integration unit) connected with the next linear integration unit, performing linear integration operation by using the next linear integration unit, and then, aggregating the output data of the linear integration operation and the linear integration vector corresponding to the first linear integration unit, such as superposition, cascade combination and the like, so as to output the linear integration vector corresponding to the next linear integration unit, that is, aggregating the input data of each linear integration unit and the output data of the linear integration operation on the input data so as to output the corresponding linear integration vector.
It should be understood that, in some possible embodiments, before the step of performing the aggregation operation of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics item data description vector corresponding to the target time interval to output the aggregate multi-dimensional description vector corresponding to the target time interval, that is, before step S130 in the foregoing description, the logistics order text mining method applied to deep learning may further include the following specific implementation matters:
mapping the logistics transportation state description vector in the target time interval in a preset space to form a corresponding mapped logistics transportation state description vector, wherein the preset space can be a feature space corresponding to a fragment feature space mapping vector (as described above) corresponding to an important object description text fragment included in the logistics order text to be analyzed;
and carrying out mapping operation of a preset space on the segment distribution state characterization parameters in the logistics object data description vector to form a mapped logistics object data description vector, wherein the mapped logistics object data description vector comprises a mapping vector of the segment distribution state characterization parameters and a segment feature space mapping vector corresponding to the important object description text segment included in the logistics order text to be analyzed, so that the space where the vectors are located is ensured to be consistent.
Based on the foregoing, in some possible embodiments, the step of performing aggregation operations of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics item data description vector corresponding to the target time interval to output an aggregate multi-dimensional description vector corresponding to the target time interval may further include the following specific implementation contents:
and carrying out aggregation operation of different dimensions on the mapped logistics transportation state description vector and the mapped logistics article data description vector, for example, carrying out cascading combination operation so as to output an aggregation multi-dimensional description vector corresponding to the target time interval.
Based on the foregoing, in some possible embodiments, the step of performing aggregation operations of different dimensions on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval may further include the following specific implementation details:
performing cascading combination operation on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval; or, performing superposition operation on the mapped logistics transportation state description vector and the mapped logistics article data description vector, which may be weighted superposition, so as to output an aggregate multi-dimensional description vector corresponding to the target time interval.
It should be understood that, in some possible embodiments, step S140 in the foregoing description, that is, the step of performing the logistics object estimation operation based on the aggregate multi-dimensional description vector, and outputting the logistics object estimation data corresponding to the target time interval, may further include the following specific implementation matters:
performing a physical distribution item estimation operation on the aggregate multidimensional description vector according to a plurality of physical distribution item estimation neural networks (for example, the aggregate multidimensional description vector may be subjected to full connection processing to obtain a corresponding full connection vector, then the full connection vector is subjected to activation processing based on an activation function, where the activation function may be a softmax function), so as to form physical distribution item estimation data corresponding to each physical distribution item estimation neural network, that is, each physical distribution item estimation neural network performs a physical distribution item estimation operation on the aggregate multidimensional description vector, so as to obtain a plurality of physical distribution item estimation data corresponding to a plurality of physical distribution item estimation neural networks;
and combining the logistics item estimation data corresponding to each logistics item estimation neural network according to the network estimation reliability parameters of each logistics item estimation neural network to output the logistics item estimation data corresponding to the target time interval, for example, the network estimation reliability parameters can be used as weighting coefficients, and the logistics item estimation data corresponding to the logistics item estimation neural network can be weighted and summed to obtain the logistics item estimation data corresponding to the target time interval.
It should be understood that, in some possible embodiments, before the step of merging the physical distribution item estimation data corresponding to each physical distribution item estimation neural network according to the network estimation reliability parameter of each physical distribution item estimation neural network to output the physical distribution item estimation data corresponding to the target time interval, the physical distribution order text mining method applied to deep learning may further include the following specific implementation matters:
performing a preset operation on an a-th physical distribution item estimation neural network among the plurality of physical distribution item estimation neural networks, the a-th physical distribution item estimation neural network belonging to any physical distribution item estimation neural network among the plurality of physical distribution item estimation neural networks (corresponding operation is performed for each physical distribution item estimation neural network), the preset operation including:
performing logistics object estimation operation on the aggregate multidimensional description vector of the example time interval by using the a-th logistics object estimation neural network to output example logistics object estimation data corresponding to the a-th logistics object estimation neural network;
According to the exemplary logistics object identification data corresponding to the exemplary time interval and the exemplary logistics object estimation data corresponding to the a-th logistics object estimation neural network, analyzing network analysis indexes corresponding to the a-th logistics object estimation neural network;
and analyzing the network estimation reliability parameters of the a-th logistics item estimation neural network according to the network analysis indexes corresponding to the a-th logistics item estimation neural network and the network analysis indexes corresponding to the plurality of logistics item estimation neural networks.
It should be understood that, in some possible embodiments, the step of analyzing the network analysis index corresponding to the a-th physical distribution item estimation neural network according to the exemplary physical distribution item identification data corresponding to the exemplary time interval and the exemplary physical distribution item estimation data corresponding to the a-th physical distribution item estimation neural network may further include the following specific implementation matters described below:
performing a calculation operation of an expected deviation distance for the exemplary physical distribution item identification data corresponding to the exemplary time interval and the exemplary physical distribution item estimation data corresponding to the a-th physical distribution item estimation neural network to output a corresponding first expected deviation distance, and for example, an average value between the exemplary physical distribution item identification data and the exemplary physical distribution item estimation data may be calculated first, then, differences between the exemplary physical distribution item identification data and the exemplary physical distribution item estimation data and the average value may be calculated respectively, and finally, a sum of squares of the differences may be calculated respectively, and a positive correlation value of the sum of squares may be taken as the first expected deviation distance, for example, a ratio between the sum of squares and the first expected deviation distance is equal to 2;
And performing interval mapping operation of parameters on the first expected deviation distance to form a network analysis index corresponding to the a-th logistics object estimation neural network, for example, the first expected deviation distance can be mapped to an interval 0-1 to obtain the network analysis index.
It should be understood that, in some possible embodiments, the step of analyzing the network estimation reliability parameter of the a-th physical distribution item estimation neural network according to the network analysis index corresponding to the a-th physical distribution item estimation neural network and the network analysis index corresponding to each of the plurality of physical distribution item estimation neural networks may further include the following specific implementation matters described below:
extracting a pre-configured reference parameter, for example, the reference parameter may be equal to 1, 2, etc.;
calculating the expected deviation distance of the network analysis index corresponding to the reference parameter and the a-th logistics object estimation neural network so as to output a corresponding second expected deviation distance, such as a calculation process of the first expected deviation distance;
and performing interval mapping operation of parameters, such as mapping to intervals 0-1, on the second expected deviation distance according to network analysis indexes corresponding to the plurality of logistics object estimation neural networks to form network estimation reliability parameters of the a-th logistics object estimation neural network.
With reference to fig. 3, the embodiment of the invention further provides a logistics order text mining device applied to deep learning, which can be applied to the logistics order text mining system applied to deep learning. The logistics order text mining device applied to deep learning can comprise:
the logistics data determining module is used for mining a logistics transportation state description vector of a target time interval and determining a logistics order text to be analyzed corresponding to the target time interval, wherein the logistics order text to be analyzed comprises an object description text segment for describing a logistics object, and the logistics object refers to a logistics transportation object;
the key information mining module is used for carrying out key information mining operation on the to-be-analyzed logistics order text so as to output logistics article data description vectors corresponding to the target time interval;
the description vector aggregation module is used for conducting aggregation operations of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval so as to output an aggregation multi-dimensional description vector corresponding to the target time interval;
The logistics article estimation module is used for carrying out logistics article estimation operation based on the aggregate multidimensional description vector and outputting logistics article estimation data corresponding to the target time interval, wherein the logistics article estimation data are used for reflecting estimated transportation quantity or estimated transaction quantity of articles in a specified time interval after the target time interval.
In summary, the method and the system for mining the logistics order text applied to deep learning provided by the invention can firstly mine the logistics transportation state description vector of the target time interval and determine the logistics order text to be analyzed corresponding to the target time interval; performing key information mining operation on the to-be-analyzed logistics order text to output logistics article data description vectors corresponding to the target time intervals; carrying out aggregation operation of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval so as to output an aggregation multi-dimensional description vector corresponding to the target time interval; and carrying out logistics object estimation operation based on the aggregate multidimensional description vector, and outputting logistics object estimation data corresponding to the target time interval. Based on the foregoing, the basis for performing the logistics article estimation operation is the aggregate multi-dimensional description vector, and the aggregate multi-dimensional description vector is obtained by aggregating the information of the two aspects of the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval, so that the basis for the logistics article estimation operation is more reliable, the result of the logistics article estimation operation is also more reliable, thereby improving the reliability of the logistics article estimation to a certain extent and improving the defects in the prior art.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. The logistics order text mining method applied to deep learning is characterized by comprising the following steps of:
excavating a logistics transportation state description vector of a target time interval, and determining a logistics order text to be analyzed corresponding to the target time interval, wherein the logistics order text to be analyzed comprises an object description text segment for describing a logistics object, and the logistics object refers to a logistics transported object;
performing key information mining operation on the to-be-analyzed logistics order text to output logistics article data description vectors corresponding to the target time interval;
carrying out aggregation operation of different dimensionalities on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval so as to output an aggregation multi-dimensional description vector corresponding to the target time interval;
And carrying out logistics object estimation operation based on the aggregate multi-dimensional description vector, and outputting logistics object estimation data corresponding to the target time interval, wherein the logistics object estimation data is used for reflecting the estimated transportation quantity or the estimated transaction quantity of the objects in the appointed time interval after the target time interval.
2. The method for mining logistics order text applied to deep learning as set forth in claim 1, wherein the step of performing a key information mining operation on the logistics order text to be analyzed to output the logistics article data description vector corresponding to the target time interval comprises the steps of:
extracting an object description text fragment set configured for the target time interval;
performing a comparison analysis operation on the object description text fragments in the object description text fragment set and the object description text fragments in the to-be-analyzed logistics order text to output important object description text fragments corresponding to the target time interval, wherein the important object description text fragments refer to the object description text fragments belonging to the object description text fragment set and the to-be-analyzed logistics order text at the same time;
According to the to-be-analyzed logistics order text, determining the fragment distribution state of the important object description text fragment so as to form a fragment distribution state characterization parameter corresponding to the important object description text fragment;
performing feature space mapping operation on the important object description text fragment to form a fragment feature space mapping vector corresponding to the important object description text fragment;
and marking the data merging result of the fragment characteristic space mapping vector corresponding to the important object description text fragment and the fragment distribution state characterization parameter of the important object description text fragment pair so as to be marked as the logistics article data description vector corresponding to the target time interval.
3. The method for text mining of a logistic order applied to deep learning according to claim 2, wherein the step of extracting the set of object description text fragments configured for the target time interval comprises:
analyzing first matching relation characterization parameters between a plurality of undetermined object description text fragments corresponding to the target time interval and a predetermined specified object description text fragment;
Determining the plurality of undetermined object description text fragments according to the first matching relation characterization parameters so as to output object description text fragments for optimizing fragment sets;
performing optimization operation on the historical object description text fragment set according to the object description text fragment for optimizing the fragment set to form an optimized historical object description text fragment set;
analyzing a second matching relation characterization parameter between the object description text fragment and the appointed object description text fragment in the optimized historical object description text fragment set;
determining the object description text fragments in the optimized historical object description text fragment set according to the second matching relation characterization parameters so as to output a plurality of determined object description text fragments;
and determining an object description text fragment set corresponding to the target time interval according to the plurality of determined object description text fragments.
4. The method for deep learning-applied logistics order text mining of claim 3, wherein prior to the step of extracting the set of object description text fragments configured for the target time interval, the method for deep learning-applied logistics order text mining further comprises:
Determining the estimation data of the example logistics articles corresponding to the example time interval, and determining the identification data of the example logistics articles corresponding to the example time interval;
determining that the extraction to the set of object description text fragments configured for the target time interval is required to be performed in a case where the degree of distinction between the exemplary physical distribution item estimation data and the exemplary physical distribution item identification data exceeds a pre-configured reference degree of distinction;
in the case where the degree of distinction between the exemplary physical distribution item estimation data and the exemplary physical distribution item identification data does not exceed the reference degree of distinction configured in advance, it is determined that the step of extracting the set of object description text fragments configured for the target time interval does not need to be performed.
5. The method for mining logistics order text applied to deep learning as set forth in claim 2, wherein the step of performing feature space mapping operation on the important object description text segment to form a segment feature space mapping vector corresponding to the important object description text segment comprises:
performing feature space mapping operation on the important object description text fragments to form fragment feature space mapping results corresponding to the important object description text fragments;
And performing depth mapping operation on the fragment characteristic space mapping result corresponding to the important object description text fragment to form a fragment characteristic space mapping vector corresponding to the important object description text fragment.
6. The method for text mining of a logistics order for deep learning of claim 5, wherein said deep mapping operation is performed using a deep mapping network model comprising a plurality of linear integration units connected in sequence;
the step of performing depth mapping operation on the segment feature space linear integration vector corresponding to the important object description text segment to form a segment feature space mapping vector corresponding to the important object description text segment includes:
performing linear integration operation on the segment feature space linear integration vector corresponding to the important object description text segment by using a first linear integration unit in the plurality of linear integration units which are sequentially connected in sequence so as to output a corresponding linear integration vector;
loading the linear integration vector corresponding to the first linear integration unit into a linear integration unit connected with the next linear integration unit, and performing linear integration operation by utilizing the next linear integration unit to output the linear integration vector corresponding to the next linear integration unit, wherein the linear integration vectors are sequentially performed to load the linear integration vector corresponding to the last but one linear integration unit into the last linear integration unit;
And determining a segment feature space mapping vector corresponding to the important object description text segment based on a linear integration vector obtained by the linear integration operation of the last linear integration unit.
7. The method for text-mining a logistic order for deep learning according to claim 1, wherein before the step of performing aggregation operations of different dimensions on the logistic transportation state description vector corresponding to the target time interval and the logistic item data description vector corresponding to the target time interval to output an aggregated multi-dimensional description vector corresponding to the target time interval, the method for text-mining a logistic order for deep learning further comprises:
mapping operation of a preset space is carried out on the logistics transportation state description vector of the target time interval so as to form a corresponding mapped logistics transportation state description vector;
carrying out mapping operation of a preset space on the segment distribution state characterization parameters in the logistics object data description vector to form a mapped logistics object data description vector, wherein the mapped logistics object data description vector comprises a mapping vector of the segment distribution state characterization parameters and a segment feature space mapping vector corresponding to an important object description text segment included in the logistics order text to be analyzed;
The step of performing aggregation operations of different dimensions on the logistics transportation state description vector corresponding to the target time interval and the logistics article data description vector corresponding to the target time interval to output an aggregation multi-dimensional description vector corresponding to the target time interval includes:
performing aggregation operation of different dimensions on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregation multi-dimensional description vector corresponding to the target time interval;
the step of performing aggregation operations of different dimensions on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval includes:
performing cascading combination operation on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval; or, performing superposition operation on the mapped logistics transportation state description vector and the mapped logistics article data description vector to output an aggregate multi-dimensional description vector corresponding to the target time interval.
8. The method for text mining of a logistic order for deep learning according to any one of claims 1 to 7, wherein the step of performing a logistic article estimation operation based on the aggregate multi-dimensional description vector and outputting logistic article estimation data corresponding to the target time interval includes:
carrying out logistics object estimation operation on the aggregate multidimensional description vector according to a plurality of logistics object estimation neural networks to form logistics object estimation data corresponding to each logistics object estimation neural network;
and combining the logistics object estimation data corresponding to each logistics object estimation neural network according to the network estimation reliability parameters of each logistics object estimation neural network so as to output the logistics object estimation data corresponding to the target time interval.
9. The method for text-mining a physical distribution order for deep learning according to claim 8, wherein before the step of merging physical distribution item estimation data corresponding to each physical distribution item estimation neural network according to the network estimation reliability parameter of each physical distribution item estimation neural network to output physical distribution item estimation data corresponding to the target time interval, the method for text-mining a physical distribution order for deep learning further comprises:
Performing a preset operation for an a-th logistic article estimation neural network of the plurality of logistic article estimation neural networks, wherein the a-th logistic article estimation neural network belongs to any one logistic article estimation neural network of the plurality of logistic article estimation neural networks, and the preset operation comprises:
performing logistics object estimation operation on the aggregate multidimensional description vector of the example time interval by using the a-th logistics object estimation neural network to output example logistics object estimation data corresponding to the a-th logistics object estimation neural network;
according to the exemplary logistics object identification data corresponding to the exemplary time interval and the exemplary logistics object estimation data corresponding to the a-th logistics object estimation neural network, analyzing network analysis indexes corresponding to the a-th logistics object estimation neural network;
analyzing the network estimation reliability parameters of the a-th logistics object estimation neural network according to the network analysis indexes corresponding to the a-th logistics object estimation neural network and the network analysis indexes corresponding to the plurality of logistics object estimation neural networks;
the step of analyzing the network analysis index corresponding to the a-th physical distribution item estimation neural network according to the physical distribution item identification data corresponding to the example time interval and the example physical distribution item estimation data corresponding to the a-th physical distribution item estimation neural network includes:
Calculating expected deviation distances for the exemplary logistics object identification data corresponding to the exemplary time interval and the exemplary logistics object estimation data corresponding to the a-th logistics object estimation neural network so as to output corresponding first expected deviation distances;
performing interval mapping operation of parameters on the first expected deviation distance to form a network analysis index corresponding to the a-th logistics object estimation neural network;
the step of analyzing the network estimation reliability parameter of the a-th physical distribution item estimation neural network according to the network analysis index corresponding to the a-th physical distribution item estimation neural network and the network analysis indexes corresponding to the multiple physical distribution item estimation neural networks respectively comprises the following steps:
extracting a pre-configured reference parameter;
calculating the expected deviation distance of the reference parameter and the network analysis index corresponding to the a-th logistics object estimation neural network so as to output a corresponding second expected deviation distance;
and performing interval mapping operation of parameters on the second expected deviation distance according to network analysis indexes corresponding to the plurality of logistics object estimation neural networks to form network estimation reliability parameters of the a-th logistics object estimation neural network.
10. A logistic order text mining system for deep learning, comprising a processor and a memory, the memory being adapted to store a computer program, the processor being adapted to execute the computer program to implement the method of any one of claims 1 to 9.
CN202310721340.2A 2023-06-16 2023-06-16 Logistics order text mining method and system applied to deep learning Pending CN116628138A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310721340.2A CN116628138A (en) 2023-06-16 2023-06-16 Logistics order text mining method and system applied to deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310721340.2A CN116628138A (en) 2023-06-16 2023-06-16 Logistics order text mining method and system applied to deep learning

Publications (1)

Publication Number Publication Date
CN116628138A true CN116628138A (en) 2023-08-22

Family

ID=87602617

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310721340.2A Pending CN116628138A (en) 2023-06-16 2023-06-16 Logistics order text mining method and system applied to deep learning

Country Status (1)

Country Link
CN (1) CN116628138A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117910906A (en) * 2024-02-19 2024-04-19 广东康利达物联科技有限公司 Data visualization method and system applied to intelligent logistics

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117910906A (en) * 2024-02-19 2024-04-19 广东康利达物联科技有限公司 Data visualization method and system applied to intelligent logistics

Similar Documents

Publication Publication Date Title
CN115098705B (en) Network security event analysis method and system based on knowledge graph reasoning
CN116126945B (en) Sensor running state analysis method and system based on data analysis
CN116109121B (en) User demand mining method and system based on big data analysis
CN116628138A (en) Logistics order text mining method and system applied to deep learning
CN116126947B (en) Big data analysis method and system applied to enterprise management system
CN117077018B (en) Data processing method, device and storage medium based on machine learning
CN112508456A (en) Food safety risk assessment method, system, computer equipment and storage medium
CN114579584A (en) Data table processing method and device, computer equipment and storage medium
CN115730659A (en) Vehicle safety analysis method and system applied to AI (Artificial Intelligence)
CN116501972B (en) Content pushing method and AI intelligent pushing system based on big data online service
CN117253368A (en) Traffic flow abnormality detection method, device, computer equipment and storage medium
CN117290420A (en) Enterprise financial management method and system for financial data extraction
CN116737800A (en) Big data mining method and system applied to supply chain platform service
CN112131274A (en) Method, device and equipment for detecting time series abnormal points and readable storage medium
CN116070149A (en) Data analysis method and system based on artificial intelligence and cloud platform
CN116955451A (en) Intelligent mapping data management method and system
CN116662415B (en) Intelligent matching method and system based on data mining
CN112862536B (en) Data processing method, device, equipment and storage medium
CN115906170B (en) Security protection method and AI system applied to storage cluster
CN116975300B (en) Information mining method and system based on big data set
CN117251790B (en) Intelligent logistics data acquisition method and system
CN117787729B (en) Intelligent park management method and related device based on identification analysis
CN116955459A (en) Sensor operation behavior estimation network optimization method and system based on artificial intelligence
CN114625747A (en) Wind control updating method and system based on information security
CN116662574A (en) Big data acquisition method and system for anti-fraud AI prediction model

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