CN109978465A - Source of goods recommended method, device, electronic equipment, storage medium - Google Patents
Source of goods recommended method, device, electronic equipment, storage medium Download PDFInfo
- Publication number
- CN109978465A CN109978465A CN201910255022.5A CN201910255022A CN109978465A CN 109978465 A CN109978465 A CN 109978465A CN 201910255022 A CN201910255022 A CN 201910255022A CN 109978465 A CN109978465 A CN 109978465A
- Authority
- CN
- China
- Prior art keywords
- source
- configuration file
- goods
- cargo
- model
- 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
Links
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
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
Abstract
The present invention provides a kind of source of goods recommended method, device, electronic equipment, storage medium, and source of goods recommended method includes: real-time to obtain primitive character；The model configuration file and aspect configuration file of housebroken source of goods Matching Model are called, the model configuration file and the aspect configuration file generate after source of goods Matching Model training；Feature vector is converted by the primitive character according to the aspect configuration file；Source of goods Matching Model object is established according to the model configuration file；And described eigenvector is inputted into the source of goods Matching Model object to obtain the output of the source of goods Matching Model.Method and device provided by the invention improves consistency of the feature when model training and model use.
Description
Technical field
The present invention relates to Internet technical field more particularly to a kind of source of goods recommended method, device, electronic equipments, storage
Medium.
Background technique
With the arrival of big data era, for logistics platform, efficient matchings look for goods driver and cargo that can greatly help
Information search cost is reduced, market efficiency is promoted.Platform has data edge, can be calculated according to the mass historical data of accumulation
Driver's history preference, and attempt to be that driver recommends and cargo similar in its preference.However answering due to long haul reality
Polygamy only guesses that the current preference of driver has large error sometimes by historical data.Therefore more real-time department is utilized as far as possible
Machine behavioral data guesses that its preference becomes extremely important.
Technology mainly passes through the cargo characteristics that driver most often pays close attention in history and recommends cargo for it at present.Disadvantage is currently
Cargo interested to driver not necessarily its most often pay close attention in history, influence factor such as driver position, it is expected that
Search pattern, same headstock draw different length lorry, the three months past driver 60% deliver cargo on a route simultaneously
Cannot suppose that this searches goods concern also is this route etc., it is more difficult to obtain effective information only by the statistics of historical data.
On the other hand, the information of freight source clicked only by this in source of goods details page scene is due to the more difficulty of source of goods intrinsic dimensionality itself
Accurately recommended with directly passing through simple computation.
Therefore, the matched Accurate Prediction of the source of goods of logistics platform is still a problem to be solved.
It is some in the prior art, the source of goods of logistics platform can be improved by the model prediction of the machine learning of open source
Accuracy rate with prediction.
Current model prediction is mainly realized by two modes in line computation:
1) it is calculated after the language by model entirety transcription for online service support；
But which has the drawback that development amount is big.Model iteration speed increasingly accelerate and there are many
In the case where mature open source scientific algorithm library, the inefficient development plan of artificial transcription model is difficult to meet business need at last
It asks.Service logic is coupled with calculating logic.Machine learning model even if service logic is not changed, under any business scenario
Or parameter, when needing to change, integrity service requires issue again on a large scale.
2) using part machine learning library provide deployment interface and calculated.
And the scheme for calling directly existing open source machine learning bank interface, problem are mainly that current different business needs
To utilize different frame and language.When calling various frames respectively, it is easy to that on-line off-line model is caused not united
One, the problems such as repeating compatibility setting.
It can be seen that in the prior art, how to reduce model the development cost of line computation, on-line off-line model it is consistent
Property, to improve the matched accuracy of the source of goods when being applied to source of goods matching prediction, it is a problem to be solved.
Summary of the invention
The present invention provides a kind of source of goods recommended method, device, electronics and sets to overcome defect existing for above-mentioned the relevant technologies
Standby, storage medium, so overcome caused by the limitation and defect due to the relevant technologies at least to a certain extent one or
Multiple problems.
According to an aspect of the present invention, a kind of source of goods recommended method is provided, comprising:
Primitive character is obtained in real time；
Call the model configuration file and aspect configuration file of housebroken source of goods Matching Model, the model configuration file
And the aspect configuration file generates after source of goods Matching Model training, the model configuration file uniquely corresponds to a feature
Configuration file；
Feature vector is converted by the primitive character according to the aspect configuration file；
Source of goods Matching Model object is established according to the model configuration file；And
Described eigenvector is inputted into the source of goods Matching Model object to obtain the output of the source of goods Matching Model.
Optionally, the aspect configuration file according to the source of goods Matching Model training when, the primitive character of acquisition with it is defeated
Enter being converted into for the feature vector of the source of goods Matching Model.
Optionally, the real-time acquisition primitive character includes:
Driver's account is obtained in real time to the clicking operation of cargo；
Obtain the cargo feature of this click；
Driver's account is obtained to the history clicking operation of cargo；
Obtain the cargo feature of each cargo in history clicking operation；
Obtain the cargo feature of each cargo to be sorted；
Driver's feature of driver's account is obtained,
Wherein, the primitive character include at least cargo feature that this clicks, in history clicking operation each cargo goods
Object feature, the cargo feature of each cargo to be sorted, driver's feature, the source of goods Matching Model export the driver and treat sequence
The click probability of each cargo.
Optionally, described eigenvector includes at least: the primitive character include this click cargo feature with wait arrange
In first similarity of the cargo feature of each cargo of sequence, the history clicking operation statistical value of the cargo feature of each cargo with
Second similarity of the cargo feature of each cargo to be sorted,
The aspect configuration file includes at least the calculation method of first similarity and second similarity.
Optionally, each cargo to be sorted is generated according to the screening conditions that driver's account inputs.
Optionally, described that described eigenvector is inputted into the source of goods Matching Model object to obtain the source of goods matching mould
After the output of type further include:
Each cargo to be sorted is ranked up by the output of the source of goods Matching Model.
Optionally, after the real-time acquisition primitive character, and the model for calling housebroken source of goods Matching Model
Before configuration file and aspect configuration file further include:
The primitive character is cleaned, and the primitive character after cleaning is serialized；
After the model configuration file and aspect configuration file for calling housebroken source of goods Matching Model, and described
It converts the primitive character to before feature vector according to the aspect configuration file further include:
Primitive character after serializing is subjected to unserializing.
Optionally, the model configuration file is pressed default Naming conventions name and is referred to when for calling according to the Naming conventions
The analysis mode shown parses the model configuration file to establish source of goods Matching Model object.
Optionally, when for training the primitive character of the source of goods Matching Model to change, by the primitive character after change
The training source of goods Matching Model, updates the model configuration file and aspect configuration file.
According to another aspect of the invention, a kind of source of goods recommendation apparatus is also provided, comprising:
Module is obtained, for obtaining primitive character in real time；
Calling module, for calling the model configuration file and aspect configuration file of housebroken source of goods Matching Model, institute
It states model configuration file and the aspect configuration file to generate after source of goods Matching Model training, the model configuration file
A unique corresponding aspect configuration file；
Conversion module, for converting feature vector for the primitive character according to the aspect configuration file；
Module is established, for establishing source of goods Matching Model object according to the model configuration file；And
Output module obtains the source of goods matching for described eigenvector to be inputted the source of goods Matching Model object
The output of model.
According to another aspect of the invention, a kind of electronic equipment is also provided, the electronic equipment includes: processor；Storage
Medium, is stored thereon with computer program, and the computer program executes step as described above when being run by the processor.
According to another aspect of the invention, a kind of storage medium is also provided, computer journey is stored on the storage medium
Sequence, the computer program execute step as described above when being run by processor.
Compared with prior art, present invention has an advantage that
By the generation and calling of model configuration file and aspect configuration file, unified model training and mould when model prediction
The consistency of type and feature, so that the calculating section in each business under logistics scene is decomposed lotus root, sufficiently benefit with business department
With open source resources, guarantee on-line off-line model and feature unification, raising product iteration efficiency.
Detailed description of the invention
Its example embodiment is described in detail by referring to accompanying drawing, above and other feature of the invention and advantage will become
It is more obvious.
Fig. 1 shows the flow chart of source of goods recommended method according to an embodiment of the present invention.
Fig. 2 shows the schematic diagrames of source of goods Matching Model according to an embodiment of the present invention training.
Fig. 3 shows the schematic diagram of source of goods Matching Model prediction according to an embodiment of the present invention.
Fig. 4 shows the module map of source of goods recommendation apparatus according to an embodiment of the present invention.
Fig. 5 schematically shows a kind of computer readable storage medium schematic diagram in exemplary embodiment of the present.
Fig. 6 schematically shows a kind of electronic equipment schematic diagram in exemplary embodiment of the present.
Specific embodiment
Example embodiment is described more fully with reference to the drawings.However, example embodiment can be with a variety of shapes
Formula is implemented, and is not understood as limited to example set forth herein；On the contrary, thesing embodiments are provided so that the present invention will more
Fully and completely, and by the design of example embodiment comprehensively it is communicated to those skilled in the art.Described feature, knot
Structure or characteristic can be incorporated in any suitable manner in one or more embodiments.
In addition, attached drawing is only schematic illustrations of the invention, it is not necessarily drawn to scale.Identical attached drawing mark in figure
Note indicates same or similar part, thus will omit repetition thereof.Some block diagrams shown in the drawings are function
Energy entity, not necessarily must be corresponding with physically or logically independent entity.These function can be realized using software form
Energy entity, or these functional entitys are realized in one or more hardware modules or integrated circuit, or at heterogeneous networks and/or place
These functional entitys are realized in reason device device and/or microcontroller device.
Flow chart shown in the drawings is merely illustrative, it is not necessary to including all steps.For example, the step of having
It can also decompose, and the step of having can merge or part merges, therefore, the sequence actually executed is possible to according to the actual situation
Change.
Fig. 1 shows the flow chart of source of goods recommended method according to an embodiment of the present invention.Source of goods recommended method includes as follows
Step:
Step S110: primitive character is obtained in real time.
Step S120: the model configuration file and aspect configuration file of housebroken source of goods Matching Model, the mould are called
Type configuration file and the aspect configuration file generate after source of goods Matching Model training, and the model configuration file is unique
A corresponding aspect configuration file.
Step S130: feature vector is converted for the primitive character according to the aspect configuration file.
Step S140: source of goods Matching Model object is established according to the model configuration file.
Step S150: described eigenvector is inputted into the source of goods Matching Model object to obtain the source of goods Matching Model
Output.
Generation and tune in source of goods recommended method provided by the invention, through model configuration file and aspect configuration file
With model when the training of, unified model and model prediction and the consistency of feature, thus by each business under logistics scene
Calculating section decomposes lotus root with business department, makes full use of open source resources, the on-line off-line model of guarantee and feature are unified, improve product to change
For efficiency.
In some embodiments of the invention, it when the aspect configuration file is according to source of goods Matching Model training, obtains
The primitive character taken is converted into the feature vector of the input source of goods Matching Model.
Specifically, the aspect configuration file may include the processing of the screening of primitive character, primitive character.The spy
Sign vector may include multiple input feature vectors, and each feature can be converted by one or more primitive characters and be obtained.The present invention can
To realize more variation patterns, it will not be described here.
In some embodiments of the invention, after the step S110 obtains primitive character in real time, and the step
S120 calls the model configuration file of housebroken source of goods Matching Model and aspect configuration file further includes before following steps: right
The primitive character is cleaned, and the primitive character after cleaning is serialized.The step S120 calls the housebroken source of goods
After the model configuration file and aspect configuration file of Matching Model, and the step S130 will according to the aspect configuration file
The primitive character further includes following steps before being converted into feature vector: the primitive character after serializing is carried out antitone sequence
Change.
Specifically, the serializing of features described above is by Feature Conversion be the form that can store or transmit process.It examines
The calling for considering step S120 needs between the systems to transmit cleaned primitive character, therefore, passes through serializing and inverted sequence
Columnization come reduce network connection during data transmit load.
In some embodiments of the invention, the model configuration file presses default Naming conventions name, when for calling,
The model configuration file is parsed according to the analysis mode of Naming conventions instruction to establish source of goods Matching Model object.
Specifically, it is contemplated that model can be by different programming languages training (such as Python, java, spark), can be with
The model come is trained to different language in advance and does same Naming conventions.For example, configuring text using the model of Python training
Part have it is unified and can identified Naming conventions (such as increase mark, setting naming order etc., the present invention not as
Limit)；There are unified, can be identified and different from Python Naming conventions using the model configuration file of java training；Benefit
There are unified, can be identified and different from Python, java Naming conventions with the model configuration file of spark training.By
This, can distinguish different language according to different Naming conventions, so as to according to different language to model configuration file
It is parsed, to establish source of goods Matching Model object.
In some embodiments of the invention, it when for training the primitive character of the source of goods Matching Model to change, presses
The primitive character training source of goods Matching Model after change, updates the model configuration file and aspect configuration file.
Specifically, as follows in most complicated model iteration situation: needing to increase new feature field while deleting old
Field, and according to the feature field of latest edition training new edition source of goods Matching Model.Firstly, increasing individual features word in user terminal
Segment record, accumulation data are stored in offline database.After off-line data amount is enough, spy is converted to using the feature field of latest edition
Vector training new edition source of goods Matching Model is levied, and generates the model configuration file and aspect configuration file of update in the process.
After off-line training, while newest pairs of model configuration file and aspect configuration file being uploaded and being loaded.It gets in stocks online
It calls latest edition aspect configuration file with the conversion of more new feature in source matching business, and calls latest edition through the source of goods of off-line training
Matching Model, to carry out model prediction.
In a specific implementation of the invention, above-mentioned steps can be realized by a computing platform, which mentions
For three external interfaces, respectively serialize (serializing interface), predict (calculating interface) and saveModel (model
Pond interface).Wherein, the characteristic sequenceization that serializing interface provides can reduce the load during network connection.Calculate interface
Most important computing function is provided.Model basin interface can be called indirectly by the front end model basin configured pool administration page that platform provides
(specifically, model basin can save multiple model configuration files and associated aspect configuration file, for calling).Separately
Outside, which provides the standalone tool of feature cleaning (for example, the library pip of python and two kinds of branch of jar packet of java can be provided
Hold), to be cleaned (such as may include the cleaning steps such as preliminary screening, uniform format, normalization) to primitive character.It is counting
Platform interior is calculated, including unserializing module (obtaining the feature for mode input for unserializing), module is kept (to be used for
Preservation of the model in redis database and the loading in cache pool), Web page module (is matched the front end for providing unified
Set interface), the intrinsic calls interface such as configuration module (the various intermediate configurations for supporting platform interior to call).Each model configuration file
A unique corresponding aspect configuration file matches model configuration file and feature when the line upper mounting plate uploads model configuration file
File is set to be stored in pairs in redis database；In online calling model, model configuration file and aspect configuration file in pairs from
It is read in platform cache pool.
Based on above-mentioned platform, source of goods Matching Model include the following steps:
The source of goods matches scene, real-time characteristic can be obtained from user terminal, and construct primitive character (primitive character field
precRaw)；The primitive character after cleaning after after primitive character field preRaw is screened using the standalone tool of feature cleaning
Field precFeature；The primitive character field precFeature obtained after cleaning is turned to using serializing interface sequence
precFeatureByte；Using the primitive character field precFeatureByte of serializing as input, specified calculating needs to call
Pairs of model configuration file and aspect configuration file, calling platform provide calculating interface；Platform receives source of goods matching
The computation requests of scene；By the primitive character field precFeatureByte unserializing of serializing at primitive character field
precFeature；According to specified aspect configuration file, after being by primitive character field precFeature character string parsing
The input object (feature vector) of format needed for continuous model；According to specified model configuration file, source of goods Matching Model is established
Object；Input object is inputted in source of goods Matching Model object, calculate and calculated result is returned into respective calls side.
Of the invention some in the specific implementation, it may include walking as follows that the step S110 obtains primitive character in real time
It is rapid: to obtain driver's account in real time to the clicking operation of cargo；Obtain the cargo feature of this click；Driver's account is obtained to goods
The history clicking operation of object；Obtain the cargo feature of each cargo in history clicking operation；Obtain the cargo of each cargo to be sorted
Feature；Obtain driver's feature of driver's account, wherein the primitive character include at least this click cargo feature,
The cargo feature of each cargo, the cargo feature of each cargo to be sorted, driver's feature in history clicking operation, the source of goods matching
Model exports the click probability that the driver treats each cargo of sequence.
Specifically, the source of goods in the present embodiment recommends that cargo details page touching can be carried out in driver's account click cargo
Hair.According to the secondary click of driver's account, the available cargo feature that this is clicked, the vehicle commander as needed for the cargo, cargo weight
Amount, transportation range, if contacted, click distance delivery availability etc. digitization feature.In view of according to details page field
What scape occurred single clicks on many cargo features, and driver is interested may to only have Partial Feature, therefore is gone through by driver
The cargo feature that history is clicked, can also counting driver's more possible interested feature, (such as feature frequency of occurrence is greater than in advance
Determining threshold decision is interested feature, or carries out standard deviation calculating to the cargo feature that driver's history is clicked, if standard deviation
It is smaller, it is believed that system is not limited thereto to the relatively stable present invention of certain feature preferences in driver).According to identified interested spy
The cargo in each source of standardization is levied, such as this is clicked and cargo that history is clicked all is extracted such as goods weight,
Demand vehicle commander, the features such as transportation range.Specifically, what each cargo to be sorted can be inputted according to driver's account
Screening conditions generate, and screening conditions for example may include route, starting point terminating point, cargo type, goods weight etc., the present invention
System is not limited thereto.The screening conditions can be driver's account and click the screening conditions inputted when browsing cargo before the cargo.
The cargo feature of each cargo to be sorted can be that feature normalization above-mentioned processing extracts such as goods weight, demand vehicle commander,
The features such as transportation range.Driver's feature for example can be driver's gender, driver's age, driver's driving age etc., and the present invention is not with this
For limitation.
Further, in this embodiment, described eigenvector includes at least: the primitive character includes what this was clicked
The cargo of each cargo in first similarity of cargo feature and the cargo feature of each cargo to be sorted, the history clicking operation
Second similarity of the cargo feature of the statistical value of feature and each cargo to be sorted, the aspect configuration file include at least institute
State the calculation method of the first similarity and second similarity.
Specifically, the primitive character includes the cargo feature of this cargo feature and each cargo to be sorted clicked
The first similarity can be calculated according to such as under type: n/N, wherein n is this cargo feature for clicking and cargo to be sorted
The identical quantity of cargo feature, N be cargo feature sum.In a change case, the primitive character includes this point
First similarity of the cargo feature of the cargo feature and each cargo to be sorted hit can be calculated according to such as under type: 1- ∑ |
ai-bi|/ai, wherein i is 1 integer for arriving N, aiFor i-th of cargo feature that this is clicked, biIt is i-th of cargo to be sorted
Cargo feature.Above is only the calculation for schematically describing the first similarity, and system is not limited thereto in the present invention.
Specifically, the statistical value for stating the cargo feature of each cargo in history clicking operation can be according to driver's history
Hundreds of clicks obtain standardized feature, and carry out mean value, standard deviation, median, the statistical dispositions such as accounting.Can by mean value,
Median etc. carries out the calculating of the second similarity, and the second similarity can be calculated using mode identical with the first similarity (will
The cargo feature that mean value, median are clicked as this in the first similarity calculation), system is not limited thereto in the present invention.
Specifically, the cargo characteristic processing of cargo calculates except similarity, partial cargo feature can also directly be made
For mode input, partial cargo feature for example can be positive rating, the other informations such as cargo clicking rate, the present invention not as
Limitation.
In the present embodiment, described eigenvector is inputted the source of goods Matching Model object to obtain by the step S150
It further include following steps after the output of the source of goods Matching Model: by the output of the source of goods Matching Model to described wait sort
Each cargo be ranked up.
In the present embodiment, above-mentioned each feature is to generate in real time on line, and database is recorded and carries out data accumulation,
Model training is carried out, positive and negative label is that cargo is clicked by driver, is only browsed and is not clicked on by driver with cargo.It is trained collection
Test set divides, and logistic regression can be used as basic model, model is online as a comparison by GBDT (gradient boosted tree), this hair
Bright is not with secondary for limitation.Source of goods Matching Model is also possible to other existing machine learning models, the present invention not as
Limitation.
Referring to Fig. 2 and Fig. 3 to combine the various embodiments described above, as shown in Fig. 2, in model training, to this click
Operation 201, history clicking operation 202 and cargo to be sorted 203 carry out feature extraction 210 obtain this click cargo feature 221,
History clicks each cargo characteristic statistics value 222, cargo feature 223 to be sorted.This is set to click cargo feature 221, history click
Each cargo characteristic statistics value 222, cargo feature 223 to be sorted (and can be generated via characteristic processing 230 during characteristic processing
Corresponding aspect configuration file), it obtains this and clicks cargo feature and the first similarity of cargo feature 241 to be sorted, history point
Hit each cargo characteristic statistics value and the second similarity of cargo feature 242 to be sorted, driver's feature 243, the other spies of cargo to be sorted
Sign 244.It is clicked using repeatedly this accumulative click cargo feature and the first similarity of cargo feature 241 to be sorted, history each
Cargo characteristic statistics value and the second similarity of cargo feature 242 to be sorted, driver's feature 243, cargo other feature 244 to be sorted
It carries out model training and generates model configuration file 250.
When model prediction, referring to Fig. 3, to this clicking operation 301, history clicking operation 302 and cargo to be sorted 303 into
Row feature extraction 310 (identical as step 210, can also to be stored in aspect configuration file in training) obtains this and clicks cargo
Feature 321, history click each cargo characteristic statistics value 322, cargo feature 323 to be sorted.Make this click cargo feature 321,
History clicks each cargo characteristic statistics value 322, cargo feature 323 to be sorted and carries out characteristic processing 330 by aspect configuration file, obtains
This click cargo feature and the first similarity of cargo feature 341 to be sorted, history click each cargo characteristic statistics value with to
Sort the second similarity of cargo feature 342, driver's feature 343, cargo other feature 344 to be sorted.It calls and feature configuration text
Uniquely corresponding model configuration file establishes source of goods Matching Model object to part, clicks cargo feature and cargo to be sorted using this
The first similarity of feature 341, history click each cargo characteristic statistics value and the second similarity of cargo feature 342 to be sorted, driver
Feature 343, cargo other feature 344 to be sorted input source of goods Matching Model object, obtain model output 350.Model exports
Driver's account treats the click probability of sequence cargo, and the click probability row for the treatment of from high to low of sequence cargo is treated by driver's account
Sequence cargo is ranked up, and is shown to driver's account.
Pass through the generation and calling of model configuration file and aspect configuration file as a result, unified model is trained and model is pre-
The consistency of model and feature when survey, thus by each business under logistics scene calculating section and business department decompose lotus root,
It makes full use of open source resources, guarantee on-line off-line model and feature unification, raising product iteration efficiency.
It is above only one or more specific implementations provided by the invention, the present invention is not for limitation.
Fig. 4 shows the module map of source of goods recommendation apparatus according to an embodiment of the present invention.Source of goods recommendation apparatus 400 includes obtaining
Modulus block 410, conversion module 430, establishes module 440 and output module 450 at calling module 420.
Module 410 is obtained for obtaining primitive character in real time；
Calling module 420 is used to call the model configuration file and aspect configuration file of housebroken source of goods Matching Model,
The model configuration file and the aspect configuration file generate after source of goods Matching Model training, the model configuration text
Part uniquely corresponds to an aspect configuration file；
Conversion module 430 is used to convert feature vector for the primitive character according to the aspect configuration file；
Module 440 is established for establishing source of goods Matching Model object according to the model configuration file；And
Output module 450 is used to described eigenvector inputting the source of goods Matching Model object to obtain the source of goods
Output with model.
Generation and tune in source of goods recommendation apparatus provided by the invention, through model configuration file and aspect configuration file
With model when the training of, unified model and model prediction and the consistency of feature, thus by each business under logistics scene
Calculating section decomposes lotus root with business department, makes full use of open source resources, the on-line off-line model of guarantee and feature are unified, improve product to change
For efficiency.
Fig. 4 is only to show schematically source of goods recommendation apparatus 400 provided by the invention, without prejudice to present inventive concept
Under the premise of, the fractionation of module, increases all within protection scope of the present invention merging.Source of goods recommendation apparatus provided by the invention
400 can be realized that the present invention is not limited thereto by software, hardware, firmware, plug-in unit and any combination between them.
In an exemplary embodiment of the present invention, a kind of computer readable storage medium is additionally provided, meter is stored thereon with
Source of goods recommended method described in any one above-mentioned embodiment may be implemented when being executed by such as processor in calculation machine program, the program
The step of.In some possible embodiments, various aspects of the invention are also implemented as a kind of form of program product,
It includes program code, and when described program product is run on the terminal device, said program code is for setting the terminal
Standby the step of executing described in this specification above-mentioned source of goods recommended method part various illustrative embodiments according to the present invention.
Refering to what is shown in Fig. 5, describing the program product for realizing the above method of embodiment according to the present invention
700, can using portable compact disc read only memory (CD-ROM) and including program code, and can in terminal device,
Such as it is run on PC.However, program product of the invention is without being limited thereto, in this document, readable storage medium storing program for executing can be with
To be any include or the tangible medium of storage program, the program can be commanded execution system, device or device use or
It is in connection.
Described program product can be using any combination of one or more readable mediums.Readable medium can be readable letter
Number medium or readable storage medium storing program for executing.Readable storage medium storing program for executing for example can be but be not limited to electricity, magnetic, optical, electromagnetic, infrared ray or
System, device or the device of semiconductor, or any above combination.The more specific example of readable storage medium storing program for executing is (non exhaustive
List) include: electrical connection with one or more conducting wires, portable disc, hard disk, random access memory (RAM), read-only
Memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disc read only memory
(CD-ROM), light storage device, magnetic memory device or above-mentioned any appropriate combination.
The computer readable storage medium may include in a base band or the data as the propagation of carrier wave a part are believed
Number, wherein carrying readable program code.The data-signal of this propagation can take various forms, including but not limited to electromagnetism
Signal, optical signal or above-mentioned any appropriate combination.Readable storage medium storing program for executing can also be any other than readable storage medium storing program for executing
Readable medium, the readable medium can send, propagate or transmit for by instruction execution system, device or device use or
Person's program in connection.The program code for including on readable storage medium storing program for executing can transmit with any suitable medium, packet
Include but be not limited to wireless, wired, optical cable, RF etc. or above-mentioned any appropriate combination.
The program for executing operation of the present invention can be write with any combination of one or more programming languages
Code, described program design language include object oriented program language-Java, C++ etc., further include conventional
Procedural programming language-such as " C " language or similar programming language.Program code can be fully in tenant
It calculates and executes in equipment, partly executed in tenant's equipment, being executed as an independent software package, partially in tenant's calculating
Upper side point is executed on a remote computing or is executed in remote computing device or server completely.It is being related to far
Journey calculates in the situation of equipment, and remote computing device can pass through the network of any kind, including local area network (LAN) or wide area network
(WAN), it is connected to tenant and calculates equipment, or, it may be connected to external computing device (such as utilize ISP
To be connected by internet).
In an exemplary embodiment of the present invention, a kind of electronic equipment is also provided, which may include processor,
And the memory of the executable instruction for storing the processor.Wherein, the processor is configured to via described in execution
Executable instruction is come the step of executing source of goods recommended method described in any one above-mentioned embodiment.
Person of ordinary skill in the field it is understood that various aspects of the invention can be implemented as system, method or
Program product.Therefore, various aspects of the invention can be embodied in the following forms, it may be assumed that complete hardware embodiment, complete
The embodiment combined in terms of full Software Implementation (including firmware, microcode etc.) or hardware and software, can unite here
Referred to as circuit, " module " or " system ".
The electronic equipment 500 of this embodiment according to the present invention is described referring to Fig. 6.The electronics that Fig. 6 is shown
Equipment 500 is only an example, should not function to the embodiment of the present invention and use scope bring any restrictions.
As shown in fig. 6, electronic equipment 500 is showed in the form of universal computing device.The component of electronic equipment 500 can wrap
It includes but is not limited to: at least one processing unit 510, at least one storage unit 520, (including the storage of the different system components of connection
Unit 520 and processing unit 510) bus 530, display unit 540 etc..
Wherein, the storage unit is stored with program code, and said program code can be held by the processing unit 510
Row, so that various according to the present invention described in the execution of the processing unit 510 this specification above-mentioned source of goods recommended method part
The step of illustrative embodiments.For example, the processing unit 510 can execute step as shown in Figure 1 to Figure 3.
The storage unit 520 may include the readable medium of volatile memory cell form, such as random access memory
Unit (RAM) 5201 and/or cache memory unit 5202 can further include read-only memory unit (ROM) 5203.
The storage unit 520 can also include program/practical work with one group of (at least one) program module 5205
Tool 5204, such program module 5205 includes but is not limited to: operating system, one or more application program, other programs
It may include the realization of network environment in module and program data, each of these examples or certain combination.
Bus 530 can be to indicate one of a few class bus structures or a variety of, including storage unit bus or storage
Cell controller, peripheral bus, graphics acceleration port, processing unit use any bus structures in a variety of bus structures
Local bus.
Electronic equipment 500 can also be with one or more external equipments 600 (such as keyboard, sensing equipment, bluetooth equipment
Deng) communication, the equipment that also tenant can be enabled interact with the electronic equipment 500 with one or more communicates, and/or with make
Any equipment (such as the router, modulation /demodulation that the electronic equipment 500 can be communicated with one or more of the other calculating equipment
Device etc.) communication.This communication can be carried out by input/output (I/O) interface 550.Also, electronic equipment 500 can be with
By network adapter 560 and one or more network (such as local area network (LAN), wide area network (WAN) and/or public network,
Such as internet) communication.Network adapter 560 can be communicated by bus 530 with other modules of electronic equipment 500.It should
Understand, although not shown in the drawings, other hardware and/or software module can be used in conjunction with electronic equipment 500, including but unlimited
In: microcode, device driver, redundant processing unit, external disk drive array, RAID system, tape drive and number
According to backup storage system etc..
Through the above description of the embodiments, those skilled in the art is it can be readily appreciated that example described herein is implemented
Mode can also be realized by software realization in such a way that software is in conjunction with necessary hardware.Therefore, according to the present invention
The technical solution of embodiment can be embodied in the form of software products, which can store non-volatile at one
Property storage medium (can be CD-ROM, USB flash disk, mobile hard disk etc.) in or network on, including some instructions are so that a calculating
Equipment (can be personal computer, server or network equipment etc.) executes the above-mentioned source of goods of embodiment according to the present invention
Recommended method.
Compared with prior art, present invention has an advantage that
By the generation and calling of model configuration file and aspect configuration file, unified model training and mould when model prediction
The consistency of type and feature, so that the calculating section in each business under logistics scene is decomposed lotus root, sufficiently benefit with business department
With open source resources, guarantee on-line off-line model and feature unification, raising product iteration efficiency.
Those skilled in the art after considering the specification and implementing the invention disclosed here, will readily occur to of the invention its
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the invention, these modifications, purposes or
Person's adaptive change follows general principle of the invention and including the undocumented common knowledge in the art of the present invention
Or conventional techniques.The description and examples are only to be considered as illustrative, and true scope and spirit of the invention are by appended
Claim is pointed out.
Claims (12)
1. a kind of source of goods recommended method characterized by comprising
Primitive character is obtained in real time；
Call the model configuration file and aspect configuration file of housebroken source of goods Matching Model, the model configuration file and institute
It states aspect configuration file to generate after source of goods Matching Model training, the model configuration file uniquely corresponds to a feature configuration
File；
Feature vector is converted by the primitive character according to the aspect configuration file；
Source of goods Matching Model object is established according to the model configuration file；And
Described eigenvector is inputted into the source of goods Matching Model object to obtain the output of the source of goods Matching Model.
2. source of goods recommended method as described in claim 1, which is characterized in that the aspect configuration file is according to the source of goods
When with model training, the primitive character of acquisition and being converted into for the feature vector of the input source of goods Matching Model.
3. source of goods recommended method as claimed in claim 2, which is characterized in that the real-time acquisition primitive character includes:
Driver's account is obtained in real time to the clicking operation of cargo；
Obtain the cargo feature of this click；
Driver's account is obtained to the history clicking operation of cargo；
Obtain the cargo feature of each cargo in history clicking operation；
Obtain the cargo feature of each cargo to be sorted；
Driver's feature of driver's account is obtained,
Wherein, the primitive character includes at least that cargo feature that this clicks, the cargo of each cargo is special in history clicking operation
Sign, the cargo feature of each cargo to be sorted, driver's feature, the source of goods Matching Model export each goods that the driver treats sequence
The click probability of object.
4. source of goods recommended method as claimed in claim 3, which is characterized in that
Described eigenvector includes at least: the primitive character includes cargo feature and each cargo to be sorted that this is clicked
The statistical value of the cargo feature of each cargo and each goods to be sorted in first similarity of cargo feature, the history clicking operation
Second similarity of the cargo feature of object,
The aspect configuration file includes at least the calculation method of first similarity and second similarity.
5. source of goods recommended method as claimed in claim 3, which is characterized in that each cargo to be sorted is according to the driver
The screening conditions of account input generate.
6. source of goods recommended method as claimed in claim 3, which is characterized in that described that described eigenvector is inputted the source of goods
After output of the Matching Model object to obtain the source of goods Matching Model further include:
Each cargo to be sorted is ranked up by the output of the source of goods Matching Model.
7. source of goods recommended method as described in claim 1, which is characterized in that
After the real-time acquisition primitive character, and the model configuration file for calling housebroken source of goods Matching Model and spy
Before sign configuration file further include:
The primitive character is cleaned, and the primitive character after cleaning is serialized；
After the model configuration file and aspect configuration file for calling housebroken source of goods Matching Model, and it is described according to institute
Aspect configuration file is stated to convert the primitive character to before feature vector further include:
Primitive character after serializing is subjected to unserializing.
8. source of goods recommended method as described in claim 1, which is characterized in that the model configuration file presses default Naming conventions
Name when for calling, parses the model configuration file according to the analysis mode of Naming conventions instruction to establish the source of goods
With model object.
9. source of goods recommended method as described in claim 1, which is characterized in that when for training the original of the source of goods Matching Model
When beginning feature changes, by the primitive character training source of goods Matching Model after change, the model configuration file and spy are updated
Levy configuration file.
10. a kind of source of goods recommendation apparatus characterized by comprising
Module is obtained, for obtaining primitive character in real time；
Calling module, for calling the model configuration file and aspect configuration file of housebroken source of goods Matching Model, the mould
Type configuration file and the aspect configuration file generate after source of goods Matching Model training, and the model configuration file is unique
A corresponding aspect configuration file；
Conversion module, for converting feature vector for the primitive character according to the aspect configuration file；
Module is established, for establishing source of goods Matching Model object according to the model configuration file；And
Output module, for described eigenvector to be inputted the source of goods Matching Model object to obtain the source of goods Matching Model
Output.
11. a kind of electronic equipment, which is characterized in that the electronic equipment includes:
Processor；
Memory is stored thereon with computer program, is executed when the computer program is run by the processor as right is wanted
Seek 1 to 9 described in any item steps.
12. a kind of storage medium, which is characterized in that be stored with computer program, the computer program on the storage medium
Step as described in any one of claim 1 to 9 is executed when being run by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910255022.5A CN109978465B (en) | 2019-03-29 | 2019-03-29 | Goods source recommendation method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910255022.5A CN109978465B (en) | 2019-03-29 | 2019-03-29 | Goods source recommendation method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109978465A true CN109978465A (en) | 2019-07-05 |
| CN109978465B CN109978465B (en) | 2021-08-03 |
Family
ID=67081966
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910255022.5A Active CN109978465B (en) | 2019-03-29 | 2019-03-29 | Goods source recommendation method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109978465B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110598989A (en) * | 2019-08-14 | 2019-12-20 | 江苏满运软件科技有限公司 | Goods source quality evaluation method, device, equipment and storage medium |
| CN110796415A (en) * | 2019-10-30 | 2020-02-14 | 江苏满运软件科技有限公司 | Goods source determining method, device, equipment and storage medium |
| CN111105044A (en) * | 2019-12-20 | 2020-05-05 | 江苏满运软件科技有限公司 | Discrete feature processing method and device, computer equipment and storage medium |
| CN112668969A (en) * | 2020-12-25 | 2021-04-16 | 江苏满运物流信息有限公司 | User tag processing method, system, electronic device and storage medium |
| CN113763069A (en) * | 2020-06-28 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Method, device and equipment for generating article information display page and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101980495A (en) * | 2010-08-24 | 2011-02-23 | 无锡开创信息技术有限公司 | Payload characteristic identification-based QQ China game service identification method |
| CN102867183A (en) * | 2012-08-23 | 2013-01-09 | 安科智慧城市技术（中国）有限公司 | Method and device for detecting littered objects of vehicle and intelligent traffic monitoring system |
| CN105302926A (en) * | 2014-05-28 | 2016-02-03 | 国家电网公司 | Data model modeling method of intelligent substation management core |
| CN106611169A (en) * | 2016-12-31 | 2017-05-03 | 中国科学技术大学 | Dangerous driving behavior real-time detection method based on deep learning |
| CN107844634A (en) * | 2017-09-30 | 2018-03-27 | 平安科技（深圳）有限公司 | Polynary universal model platform modeling method, electronic equipment and computer-readable recording medium |
| CN108319633A (en) * | 2017-11-17 | 2018-07-24 | 腾讯科技（深圳）有限公司 | A kind of image processing method, device and server, system, storage medium |
| CN108733639A (en) * | 2018-04-09 | 2018-11-02 | 中国平安人寿保险股份有限公司 | A kind of configuration parameter regulation means, device, terminal device and storage medium |
| CN109190581A (en) * | 2018-09-17 | 2019-01-11 | 金陵科技学院 | Image sequence target detection recognition methods |
| CN109242044A (en) * | 2018-09-30 | 2019-01-18 | 江苏满运软件科技有限公司 | Training method, device, storage medium and the electronic equipment of vehicle and goods matching model |
| CN109447334A (en) * | 2018-10-19 | 2019-03-08 | 江苏满运软件科技有限公司 | The Method of Data with Adding Windows of information of freight source, device, electronic equipment, storage medium |
-
2019
- 2019-03-29 CN CN201910255022.5A patent/CN109978465B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101980495A (en) * | 2010-08-24 | 2011-02-23 | 无锡开创信息技术有限公司 | Payload characteristic identification-based QQ China game service identification method |
| CN102867183A (en) * | 2012-08-23 | 2013-01-09 | 安科智慧城市技术（中国）有限公司 | Method and device for detecting littered objects of vehicle and intelligent traffic monitoring system |
| CN105302926A (en) * | 2014-05-28 | 2016-02-03 | 国家电网公司 | Data model modeling method of intelligent substation management core |
| CN106611169A (en) * | 2016-12-31 | 2017-05-03 | 中国科学技术大学 | Dangerous driving behavior real-time detection method based on deep learning |
| CN107844634A (en) * | 2017-09-30 | 2018-03-27 | 平安科技（深圳）有限公司 | Polynary universal model platform modeling method, electronic equipment and computer-readable recording medium |
| CN108319633A (en) * | 2017-11-17 | 2018-07-24 | 腾讯科技（深圳）有限公司 | A kind of image processing method, device and server, system, storage medium |
| CN108733639A (en) * | 2018-04-09 | 2018-11-02 | 中国平安人寿保险股份有限公司 | A kind of configuration parameter regulation means, device, terminal device and storage medium |
| CN109190581A (en) * | 2018-09-17 | 2019-01-11 | 金陵科技学院 | Image sequence target detection recognition methods |
| CN109242044A (en) * | 2018-09-30 | 2019-01-18 | 江苏满运软件科技有限公司 | Training method, device, storage medium and the electronic equipment of vehicle and goods matching model |
| CN109447334A (en) * | 2018-10-19 | 2019-03-08 | 江苏满运软件科技有限公司 | The Method of Data with Adding Windows of information of freight source, device, electronic equipment, storage medium |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110598989A (en) * | 2019-08-14 | 2019-12-20 | 江苏满运软件科技有限公司 | Goods source quality evaluation method, device, equipment and storage medium |
| CN110598989B (en) * | 2019-08-14 | 2022-09-23 | 江苏满运软件科技有限公司 | Goods source quality evaluation method, device, equipment and storage medium |
| CN110796415A (en) * | 2019-10-30 | 2020-02-14 | 江苏满运软件科技有限公司 | Goods source determining method, device, equipment and storage medium |
| CN110796415B (en) * | 2019-10-30 | 2022-09-16 | 贵阳货车帮科技有限公司 | Goods source determining method, device, equipment and storage medium |
| CN111105044A (en) * | 2019-12-20 | 2020-05-05 | 江苏满运软件科技有限公司 | Discrete feature processing method and device, computer equipment and storage medium |
| CN111105044B (en) * | 2019-12-20 | 2022-09-23 | 江苏满运软件科技有限公司 | Discrete feature processing method and device, computer equipment and storage medium |
| CN113763069A (en) * | 2020-06-28 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Method, device and equipment for generating article information display page and storage medium |
| CN112668969A (en) * | 2020-12-25 | 2021-04-16 | 江苏满运物流信息有限公司 | User tag processing method, system, electronic device and storage medium |
| CN112668969B (en) * | 2020-12-25 | 2022-07-08 | 江苏满运物流信息有限公司 | User tag processing method, system, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109978465B (en) | 2021-08-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111428008B (en) | Method, apparatus, device and storage medium for training a model | |
| CN109978465A (en) | Source of goods recommended method, device, electronic equipment, storage medium | |
| US8719192B2 (en) | Transfer of learning for query classification | |
| US11030547B2 (en) | System and method for intelligent incident routing | |
| EP3896618A1 (en) | Method for generating user interactive information processing model and method for processing user interactive information | |
| CA3052527C (en) | Target document template generation | |
| US8589396B2 (en) | Cross-guided data clustering based on alignment between data domains | |
| CN110968695A (en) | Intelligent labeling method, device and platform based on active learning of weak supervision technology | |
| CN108305050B (en) | Method, device, equipment and medium for extracting report information and service demand information | |
| CN109447334B (en) | Data dimension reduction method and device for goods source information, electronic equipment and storage medium | |
| CN108776696B (en) | Node configuration method and device, storage medium and electronic equipment | |
| CN102597979A (en) | Productive distribution for result optimization within a hierarchical architecture | |
| CN106796608A (en) | Contextual search character string synonym is automatically generated | |
| CN111400471B (en) | Question recommendation method, system, electronic device and storage medium | |
| CN110019742A (en) | Method and apparatus for handling information | |
| CN115455161A (en) | Conversation processing method, conversation processing device, electronic equipment and storage medium | |
| CN110717333B (en) | Automatic generation method and device for article abstract and computer readable storage medium | |
| CN112749238A (en) | Search ranking method and device, electronic equipment and computer-readable storage medium | |
| CN110348581B (en) | User feature optimizing method, device, medium and electronic equipment in user feature group | |
| CN111861004A (en) | Method, system, apparatus and storage medium for automatic commission prediction of daily income production | |
| CN112785111A (en) | Production efficiency prediction method, device, storage medium and electronic equipment | |
| CN116048463A (en) | Intelligent recommendation method and device for content of demand item based on label management | |
| CN115860872A (en) | Target object determination method and device, electronic equipment and storage medium | |
| CN115017385A (en) | Article searching method, device, equipment and storage medium | |
| CN113935802A (en) | Information processing method, device, equipment and 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |