CN106779126A - Malice accounts for the processing method and system of an order - Google Patents
Malice accounts for the processing method and system of an order Download PDFInfo
- Publication number
- CN106779126A CN106779126A CN201611263344.7A CN201611263344A CN106779126A CN 106779126 A CN106779126 A CN 106779126A CN 201611263344 A CN201611263344 A CN 201611263344A CN 106779126 A CN106779126 A CN 106779126A
- Authority
- CN
- China
- Prior art keywords
- order
- malice
- model
- accounts
- account
- 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
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
- G06Q10/02—Reservations, e.g. for tickets, services or events
- G06Q10/025—Coordination of plural reservations, e.g. plural trip segments, transportation combined with accommodation
Landscapes
- Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Engineering & Computer Science (AREA)
- Operations Research (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides the processing method and system that a kind of malice accounts for an order.Wherein, the method includes:From ticketing data storehouse extract it is lower one in the behavioural characteristic and the sequence information of one this order by under lower before single operation under execution and/or during single operation under performing;According to behavioural characteristic and sequence information, the malice degree of order is judged;In the case where the malice degree for determining order exceedes pre-set threshold value, malice is taken to account for a risk prevention instruments for order.By the present invention, can according to it is lower one behavioural characteristic and one sequence information lower come judge an order whether be malice account for an order, and be judged as taking malice to account for a risk prevention instruments in the case that malice accounts for an order, solve the problems, such as that the malice in correlation technique accounts for a risk prevention instruments and takes precautions against that effect is poor, influence freight space sale because that cannot judge that malice is accounted for caused by an order as granularity with order, improve identification and strick precaution effect that malice accounts for an order.
Description
Technical field
The present invention relates to sale of ticket field, the processing method and system of an order are accounted in particular to a kind of malice.
Background technology
On the electronic passenger ticket official website of sale air ticket, user can directly place an order, and ticket sale system can be user's house seat
Half an hour or a hour (domestic half an hour, one hour of international air ticket), to after the designated time, if the user is also not
Pay, seat can be released.This is a kind of marketing set up on the basis of sincerity.Made a reservation with the free reservation that restaurant provides and taken
Unlike business, specially for the freight space that a user retains has obvious " tying up effect " in sale of ticket, other users are seen
Less than the freight space.On a flight, if the excessive order without willingness to pay, it is formed malice and accounts for seat.Set up
Electronic passenger ticket marketing method on the basis of sincerity, user and agent's folk prescription penalty cost are almost nil, even if promise breaking, user
Or agent is also an option that other airlines.It is that (a small number of course lines are in relative sufficient competition market plus domestic Civil Aviation Industry
Close to excessive competition), malice superior eigenvalue is more frequent in sale of ticket system.
One example conveniently explanation malice can account for a present situation.On July 4th, 2014, the hot topic from Beijing to Erdos
Tourism route, East Airways MU3721, by certain monarch, within 10 minutes, continuously places an order 6 times, every time booking number up to 9, and
The people did not buy East Airways's air ticket within the week where July 4 in East Airways's electronic passenger ticket official website.In tourism
In the busy season, maliciously accounting for seat can show everyday, and this forces East Airways B2C official websites to make transaction limits:Reduction single transaction middle deck bit quantity,
The number for buying the freight space seat is allowed to be reduced to 3 from 9;One trading account has only allowed 2 within one day and has not propped up
Pay order;The excessive account implementation of single number is abandoned to promise breaking recently to close down (i.e. black and white lists).
As above transaction limits, although gather effect, but for malice occupant, still show slightly extensive.Such as, following table has counted flight
MU3711 (Beijing-Erdos) takes off 11 points of a that morning nearly hour in 2014-08-28, maliciously account for seat another
The performance of evolution:
Table 1:Malice accounts for seat statistics displaying table to short frequency soon
Order number | User_id | Place an order the time | Booking number | IP | City |
711140828610304 | 46257683 | 2014-8-28 11:16 | 3 | 125.64.211.105 | Yibin City |
711140828610661 | 46257683 | 2014-8-28 11:19 | 3 | 125.64.211.105 | Yibin City |
711140828611214 | 46258032 | 2014-8-28 11:23 | 3 | 125.64.211.105 | Yibin City |
711140828611307 | 46258032 | 2014-8-28 11:23 | 3 | 125.64.211.105 | Yibin City |
711140828611535 | 42252814 | 2014-8-28 11:25 | 3 | 125.64.211.105 | Yibin City |
711140828611599 | 46149931 | 2014-8-28 11:25 | 3 | 125.64.211.105 | Yibin City |
711140828611602 | 42252814 | 2014-8-28 11:25 | 3 | 125.64.211.105 | Yibin City |
711140828611689 | 42373924 | 2014-8-28 11:26 | 3 | 125.64.211.105 | Yibin City |
711140828611782 | 42373924 | 2014-8-28 11:27 | 3 | 125.64.211.105 | Yibin City |
711140828612066 | 46149931 | 2014-8-28 11:29 | 3 | 125.64.211.105 | Yibin City |
711140828612245 | 46258332 | 2014-8-28 11:30 | 3 | 125.64.211.105 | Yibin City |
711140828612320 | 46258332 | 2014-8-28 11:31 | 3 | 125.64.211.105 | Yibin City |
711140828612647 | 46258586 | 2014-8-28 11:33 | 3 | 125.64.211.105 | Yibin City |
711140828612926 | 46258586 | 2014-8-28 11:35 | 3 | 125.64.211.105 | Yibin City |
711140828613060 | 46259026 | 2014-8-28 11:36 | 3 | 125.64.211.105 | Yibin City |
711140828613471 | 46259026 | 2014-8-28 11:39 | 3 | 125.64.211.105 | Yibin City |
711140828613623 | 46259294 | 2014-8-28 11:41 | 3 | 125.64.211.105 | Yibin City |
711140828613705 | 46259294 | 2014-8-28 11:41 | 3 | 125.64.211.105 | Yibin City |
711140828613872 | 46259628 | 2014-8-28 11:42 | 3 | 125.64.211.105 | Yibin City |
711140828613962 | 46259628 | 2014-8-28 11:43 | 3 | 125.64.211.105 | Yibin City |
The malice of table 1 accounts for seat statistics and seems to show such some weak points in trading rules:(1) though per pre- in transaction
Booking is reduced to 3 from 9, but fast and many small pen malice order also can overcome the disadvantages that the inconvenience that this regulation is brought seemingly;(2)
Though being closed down because abandoning single excessive account, free login mechanism only increases extra machine input.One big generation
Reason, can possess thousands of official register accounts.Still more, transaction system will not close down an account often drawn a bill,
Using the account often drawn a bill, it is also possible to when and rationally account for seat every now and then;(3) though defining non-pay invoice number, with numerous
Account produce malice order, only increase the small input cost of agent.
There is different place with eTerm in the ticket sale system based on ecommerce:Blank screen marketing system accounts for seat for malice
ID Process identifiers (PID or OFFICE) can directly seal transaction port, and ecommerce trade company-individual (B2C) website is common
Enjoy a client numbers, such as SHA777, transaction port is sealed as blank screen transaction system can may influence website to conclude the business.
It is general " first to pay the bill, then account for seat and draw a bill " using " completing to represent electronic passenger ticket order number after paying " mode of doing business,
Also can preferably administer malice and account for a behavior.It is just the lower single reservation freight space after single completion pays under this way requirement.Such as
Really lower one does not complete to pay, and will can't see order number, and transaction system is also the lower single reservation freight space, therefore malice accounts for seat
Do not occur.If lower, one is taken a long time after completion pays, transaction system house seat again, it is possible that without freight space feelings
Condition, can bring a series of transaction issues.Such as, transaction system is probably due to without this freight space, consequently, it is possible into " Building K " state;
Or require that user increases payment;Or give user's apology reimbursement.This can bring poor Consumer's Experience, in transaction system
When busy, or even more customer complaint can be caused.
Malice accounts for seat and directly affects normal sale on occupied flight:(1) high performance-price ratio freight space sale is not gone out.Cost performance is high
Freight space (form false sale) because being preempted, really have a user of purchase intention, inquiry less than, it is impossible to buy.(2) user is reduced
Range of choice.The freight space information being preempted, air ticket search engine (including take journey, go where, search engine etc. skill dragon) in or
Because freight space distribution state becomes non-availability, possibly cannot manifest, so as to reduce user's range of choice;(3) freight space is reduced to go out
Sell chance.Or because of reasons such as prices, that is, enable searched engine and retrieve, can also come behind search result set, so as to reduce
The sales opportunnities of the flight freight space.Because on e-commerce website, sorting forward, the probability being undoubtedly easily selected by a user can increase
Plus.
Term is explained
It refers in electronic passenger ticket automatic selling system, to try to be the first and place an order that malice accounts for seat, but a kind of behavior without willingness to pay.
I.e. automatic ticket-selling system generates sealing (having reserved freight space) for the order, but it is lower one have no willingness to pay, until transaction system
System cancels this order automatically because of time-out, the shared freight space of release.
Two sub-models are a kind of Mathematical Modelings for answering yes/no.
The content of the invention
The invention provides the processing method and system that a kind of malice accounts for an order, at least to solve malice in correlation technique
Account for the problem that seat is screened and prevention method is present.
According to an aspect of the invention, there is provided a kind of malice accounts for the processing method of an order, including:
From ticketing data storehouse extract it is lower one in the behavioural characteristic before single operation under execution and/or during single operation under performing and
The sequence information of one this order by under lower;
According to the behavioural characteristic and the sequence information, the malice degree of the order is judged;
In the case where the malice degree for determining the order exceedes pre-set threshold value, malice is taken to account for for the order
Seat risk prevention instrumentses.
Alternatively, the behavioural characteristic and the sequence information include at least one of:
Single ID correspondences are abandoned single rate, lower single number of repetition, order people's name number of repetition, order people Email and are repeated down
Number of times, order people's phone number of repetition, order people regular guest number of repetition, the ID numbers of repetition of PNR, PNR names number of repetition,
Account for a number, the last time abandon single time, apart from the departure time, whether account for popular flight, whether have wait pay invoice.
Alternatively, the behavioural characteristic and the sequence information also include at least one of:
According to log information extract it is lower one satisfy the need the search of line price it is time-consuming, according to log information extract it is lower one
Fill in and seize the opportunity the time-consuming of people's information.
Alternatively, according to the behavioural characteristic and the sequence information, judging the malice degree of the order includes:
Set up the malice degree model based on two sub-models;
The successful order that will be concluded the business in History Order accounts for an order as non-malicious, by Fail Transaction in History Order and not
The order that trial carried out delivery operation accounts for an order as malice, accounts for an order by the non-malicious and the malice accounts for seat
Order distinguishes corresponding behavioural characteristic and sequence information, and the malice degree model is trained;
Using trained malice degree model, the malice degree of one this order by under lower is judged.
Alternatively, two sub-model includes one below:
Decision-tree model, neural network model, Bayesian model, Multiple regression model, supporting vector machine model.
Alternatively, taking malice to account for a risk prevention instruments for the order includes:
Operational phase according to residing for the order determines to take precautions against point;
Malice is taken to account for a risk prevention instruments in the point of taking precautions against.
Alternatively, the point of taking precautions against includes at least one of:
When seizing the opportunity the submission of people's information, confirm when seizing the opportunity people and line information, generate order number and be prepared to enter into flow of payments
Cheng Shi.
Alternatively, the malice accounts for a risk prevention instruments includes at least one of:
Picture checking, mobile phone short message verification, shorten the time of payment, actively abandon order, refusal transaction, order pays successfully
Non- pay invoice then discharges seat in providing order number and short time again afterwards.
According to another aspect of the present invention, there is provided a kind of malice accounts for the processing system of an order, including:
Abstraction module, for extracting one before single operation under execution and/or single operation under execution from ticketing data storehouse
When behavioural characteristic and one this order by under lower sequence information;
Judge module, for according to the behavioural characteristic and the sequence information, judging the malice degree of the order;
Module is taken precautions against, in the case of exceeding pre-set threshold value in the malice degree for determining the order, for described
Order takes malice to account for a risk prevention instruments.
Alternatively, the behavioural characteristic and the sequence information include at least one of:
Single ID correspondences are abandoned single rate, lower single number of repetition, order people's name number of repetition, order people Email and are repeated down
Number of times, order people's phone number of repetition, order people regular guest number of repetition, the ID numbers of repetition of PNR, PNR names number of repetition,
Account for a number, the last time abandon single time, apart from the departure time, whether account for popular flight, whether have wait pay invoice.
Alternatively, the behavioural characteristic and the sequence information also include at least one of:
According to log information extract it is lower one satisfy the need the search of line price it is time-consuming, according to log information extract it is lower one
Fill in and seize the opportunity the time-consuming of people's information.
Alternatively, the judge module includes:
Modeling unit, for setting up the malice degree model based on two sub-models;
Training unit, for will be concluded the business in History Order, successful order accounts for an order as non-malicious, by History Order
The order that middle Fail Transaction and not attempting carried out delivery operation accounts for an order as malice, and an order is accounted for by the non-malicious
An order being accounted for the malice and distinguishing corresponding behavioural characteristic and sequence information, the malice degree model is trained;
Judging unit, for use trained malice degree model, judge it is described it is lower one ordered described in by under
Single malice degree.
Alternatively, two sub-model includes one below:
Decision-tree model, neural network model, Bayesian model, Multiple regression model, supporting vector machine model.
Alternatively, the strick precaution module includes:
Determining unit, determines to take precautions against point for the operational phase according to residing for the order;
Unit is taken precautions against, for taking malice to account for a risk prevention instruments in the point of taking precautions against.
Alternatively, the point of taking precautions against includes at least one of:
When seizing the opportunity the submission of people's information, confirm when seizing the opportunity people and line information, generate order number and be prepared to enter into flow of payments
Cheng Shi.
Alternatively, the malice accounts for a risk prevention instruments includes at least one of:
Picture checking, mobile phone short message verification, shorten the time of payment, actively abandon order, refusal transaction, order pays successfully
Non- pay invoice then discharges seat in providing order number and short time again afterwards.
By the present invention, using from ticketing data storehouse extract it is lower one before single operation under execution and/or perform the behaviour that places an order
The sequence information of behavioural characteristic and one this order by under lower when making;According to behavioural characteristic and sequence information, judge to order
Single malice degree;In the case where the malice degree for determining order exceedes pre-set threshold value, malice is taken to account for seat for order
The mode of risk prevention instrumentses, such that it is able to according to it is lower one behavioural characteristic and one sequence information lower be judging an order
It is no and to be judged as taking malice to account for a risk prevention instruments in the case that malice accounts for an order for malice accounts for an order, solve phase
Malice in the technology of pass accounts for a risk prevention instruments and takes precautions against effect because that cannot judge that malice is accounted for caused by an order as granularity with order
Difference, the problem of the sale of influence freight space, improve malice and account for the identification of an order and take precautions against effect.
Brief description of the drawings
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, this hair
Bright schematic description and description does not constitute inappropriate limitation of the present invention for explaining the present invention.In the accompanying drawings:
Fig. 1 is the structural representation of the processing system that malice according to embodiments of the present invention accounts for an order;
Fig. 2 is the workflow diagram of the processing system that malice according to embodiments of the present invention accounts for an order;
Fig. 3 is that malice according to embodiments of the present invention accounts for the processing system of an order and the strick precaution Contrast on effect of legacy system
Figure;
Fig. 4 is that malice according to embodiments of the present invention accounts for the processing system of an order and interacts timing diagram with electronic passenger ticket;
Fig. 5 is the processing system and electronic passenger ticket transaction system that malice according to the preferred embodiment of the invention accounts for an order
Data flow figure;
Fig. 6 is the network structure of the processing system that malice according to the preferred embodiment of the invention accounts for an order;
Fig. 7 is the schematic diagram of malice degree model selection interface according to the preferred embodiment of the invention;
Fig. 8 is the schematic diagram of the configuration interface of selective protection flight according to the preferred embodiment of the invention;
Fig. 9 is the schematic diagram of precautionary measures configuration interface according to the preferred embodiment of the invention.
Specific embodiment
Describe the present invention in detail below with reference to accompanying drawing and in conjunction with the embodiments.It should be noted that not conflicting
In the case of, the feature in embodiment and embodiment in the application can be mutually combined.
Although the embodiment of the present invention is described and illustrates as a example by order is maliciously accounted in civil aviaton's ticketing and is taken precautions against,
But the present invention is not limited to apply in civil aviaton's ticketing, can also be real using design of the invention in other ticketing systems
The now strick precaution treatment of malice order.
Present embodiments provide a kind of malice and account for the processing scheme of an order and a behavior is accounted for preventing malice.This programme is base
Carry out preventing malice in behavioural characteristic and account for seat.The technical scheme is based on two sub-models, is comprehensively containing existing strick precaution dimension
Meanwhile, extracted it is more can under reflection single behavioural characteristic dimension, from more extensively, the more comprehensive visual field goes to consider one and orders
It is that malice accounts for seat that whether list constitutes malice accounts on seat, and much degree.And then can accomplish based on order this most granule
Degree maliciously accounts for a behavior to take precautions against, hit.
Result of the test shows that Behavior-based control feature, the easily extension malice of the present embodiment construction account for the treatment of an order
System, can not only recognize that is used for accounting for a malice order for seat with degree of precision, and malice accounts for the processing system of an order and exists
(strick precaution in advance is can be described as again, now transaction system does not produce sealing also), the order can be just identified in 0.1s before placing an order
Malice degree.Also, malice accounts for the ready-made supporting precautionary measures of processing system of an order, can be according to different malice journeys
Degree, using the different precautionary measures, can effectively reduce malice order.
Fig. 1 is the structural representation of the processing system that malice according to embodiments of the present invention accounts for an order, as shown in figure 1,
For existing electronic passenger ticket transaction system, the processing system for accounting for an order with malice by way of messaging is interacted:
Transaction node sends lower single information and order contents to processing system that malice accounts for an order, maliciously accounts for the treatment of an order
System sends diagnostic result to transaction node with the precautionary measures.Transaction system with the precautionary measures accounts for seat according to malice
The processing system returning result of order determines to enable the corresponding precautionary measures.
Fig. 2 is the workflow diagram of the processing system that malice according to embodiments of the present invention accounts for an order, and Fig. 2 shows malice
The processing system for accounting for an order is one kind diagnostic system based on model.Its technical scheme thinking is:Evade and judge a generation
Reason people two sides sex chromosome mosaicism, region be directly facing an order, and whether according to one behavioural characteristic lower, it is malice order (malice to distinguish
Order is the order for accounting for seat).One electronic passenger ticket transaction system, completes an order, it usually needs route via price
A series of processes such as people (PNR) information are seized the opportunity in inquiry, typing.Each step in this serial procedures, can be according to detailed daily record
Information, analyzes one behavioural characteristic lower.If finding these differences, more can successfully distinguish malice and account for an order and just
Normal order.
The realization that malice accounts for the processing system technical scheme of an order will be introduced centered on Fig. 1 and Fig. 2 below:
Feature extraction
One order generating process is described from different dimensions, so as to can also investigate a malice degree for order.Such as,
Malice occupant, substantially belongs to the type that is skilled in technique that places an order.General their stronger purposes, or for the boat for being about to take off
Class, or for popular course line in popular flight, or for cost performance freight space high.The speed of their search patterns, fills in and multiplies
Machine people's information rate, payment speed will be significantly faster than that average speed.
The process of feature extraction, it is also desirable in view of the difficulty and cheap property that extract.In practice, the reality according to oneself
Situation, can employ such as the feature extraction of table 2:
Table 2:Characteristic dimension table
Dimension | Obtain data source | Function |
Single ID correspondences abandon single rate down | The next day under line, calculates | Place an order the conventional quality of people |
Single number of repetition down | Abandon crt_id in single set | It is lower that one has a kind of how black description |
Order people's name number of repetition | Single contacts list down | How black the name have |
Order people's email numbers of repetition | Single contacts list down | How black the email have |
Order people's phone number of repetition | Single telephone number down | How black the phone have |
Order people regular guest's number of repetition | Single user management table down | How black this be uniquely labeled with |
The ID numbers of repetition of PNR | Seize the opportunity people's table | How black the ID of the PNR have |
PNR name numbers of repetition | Seize the opportunity people's table | How black the name of the PNR have |
Account for a number | Order table | One singly accounts for how many seat |
The last time abandons single time | Abandon single table | Black freshness |
Apart from the departure time | Order table and travel schedule | Urgency level |
Whether popular flight is accounted for | Travel schedule | Maliciously accounted for a flight set |
Whether wait pay invoice is had | Order table | One quality instantly lower |
Feature extraction should be not limited to feature cited by above-mentioned table 2.Suggestion has to be realized searching for time-consuming with extraction condition, increase
Dimension, fills in and seizes the opportunity time-consuming dimension of people etc..
Malice degree model is selected and set up
According to the thinking of the technical scheme, the malice degree for judging an order is accomplished that.And answer an order
It is normal order, or for accounting for the order of seat, can be described with two sub-models.If describing one using two sub-models
Order is normal order, or malice order, and it is -1 or+1 this problem that can simply answer.Such as, represented just with "+1 "
Normal order (positive example), a malice order (negative example) is represented with " -1 ".
If using table 2 it is described as feature, can be from existing transaction data base, it is easy to find.Such as,
From the order of the Successful Transaction of nearly month, some positive examples can be obtained, because these orders have completed transaction.Can also
From the order store of nearly month, the order of those Fail Transactions is found out as counter-example.When counter-example is found, can deduct
Those repeatedly pay unsuccessful situation.In practice, we have selected those and do not click on abandoning singly (maliciously for " immediate payment " button
Occupant did not accounted for paying), as counter-example.
There is positive counter-example, there is feature again, it is possible to training pattern.There are many existing disaggregated models to can be used for two
The modeling of sub-model, such as, decision-tree model (C5.0), neural network model (BP), Bayesian model (NB), logistic is returned
Return model (LG), supporting vector machine model (SVM), etc..How preference pattern is, it is necessary to skilled for model according to developer
Usage degree.Such as, how the language that model translation is supported into existing system, how to be introduced into existing engineering.At us
Practice in, we have selected supporting vector machine model, and the model is introduced into the transaction system based on J2EE platforms.
Transaction system adds the precautionary measures
The model that above-mentioned modeling procedure is obtained, can quantitatively consider a malice degree for order.If it find that certain
Order is malice, it would be desirable to which transaction system can make strick precaution according to malice degree.Answer two such problem:Take precautions against point
Where is settingHow to take precautions against
(1) point is taken precautions against to set:Typically can complete the step of needing according to an order to set.Malice accounts for an order
Processing system need according to it is lower one fill in people's information is seized the opportunity to analyze the malice degree of the order.Therefore take precautions against point and set, most
People's information early can be seized the opportunity in lower single submission to start.General submission is seized the opportunity after people's information, in addition it is also necessary to which lower one reaffirms.Again
Secondary confirmation can also set strick precaution.Reaffirm and finish, be put into the payment stage.Into the stage of payment, before generation order number,
Strick precaution can also be set.The present embodiment combing strick precaution point as shown in table 3 is set:
Table 3:Electronic passenger ticket transaction system is taken precautions against point and is set
Transaction step | Transaction step explanation | Take precautions against possibility |
Searching of line | Not yet start transaction | Need more excavation IP information |
Fill in and seize the opportunity people's information | It has been logged in that, submission information starts | When submitting to, can set up defences |
People and line information are seized the opportunity in confirmation | Confirm submission information | When clicking on confirmation, can set up defences |
Generation order number, waits to be paid | Payment button is not clicked on | It is prepared to enter into paying, can sets up defences |
The selection means of payment | Click on payment button | Etc. completion to be paid, without setting up defences |
Pay and complete | Payment has been completed | Pay and complete, without setting up defences |
Table 3 gives developer's opportunity and sets, i.e., how to trigger the precautionary measures and come into force.
(2) precautionary measures.The purpose of the precautionary measures is to allow malice occupant to beat a retreat in the face of difficulties, and reaches not cargo reservation or few
The purpose of cargo reservation.Present embodiments provide effective precautionary measures as shown in table 4 below:
Table 4:Electronic passenger ticket transaction system is taken precautions against point and is set
Measure " first pay the bill, then to order number " described in table 4, this strick precaution is not to improve " first pay the bill, then draw a bill "
Foot part, first or for this it is lower one stay seat, but stay the time at seat very short only 3 minutes, and, order number does not also show
To it is lower one.According to the statistics of our newest payment deadlines, general 5 minutes etc, more than 96% it is lower one all may be used
To complete delivery operation.It is skilled it is lower one, it is general just to complete to pay less than 20 seconds, stay 3 minutes to one it is skilled it is lower one,
It is very sufficient.
Actively abandon order, be allow it is lower one oneself abandon current shared seat.It is lower one check current order information
When, transaction system should actively add the button of " abandoning order ".One corrects oneself because of operation at any time under can so facilitating
What error was caused accounts for seat.Actively abandon order, it is also possible to accelerate the timely release of shared freight space.
There is precautionary measures electronic passenger ticket transaction system, by according to one behavior characteristic information lower, be diagnosed to be and currently order
Whether single is malice order.If malice order, the strick precaution point that can be configured according to transaction system, prevention method, camera bullet
Go out the precautionary measures such as mobile phone note verification code, refusal transaction.Electronic passenger ticket transaction system with the precautionary measures, increased malice
The transaction cost of occupant, forms a kind of relative tight mechanism of exchange, can reach and allow what malice occupant can not do at will
Purpose.
Fig. 3 illustrates 2015 tourist seasons malice and accounts for the processing system of an order after the experiment of East Airways B2C official websites is reached the standard grade,
A most serious top 10 flight is accounted on the same day with 2014 (processing system for maliciously accounting for an order is not reached the standard grade), accounts for order of seats number totalling right
Than.As can be seen from Figure 3 after employing the processing system that malice accounts for an order, maliciously account for a situation and obtained largely delaying
Solution.Fig. 4 illustrates malice and accounts for the processing system of an order and interacts timing diagram with electronic passenger ticket.
Above-described embodiment is described and illustrated below by preferred embodiment.
Behavioural characteristic is extracted and conversion is realized
The extraction of general behavior feature is converted and included under line and part on line.It will be to find one that partly extraction is under line
Suitable model service.And part on line, then be by an original order, be immediately converted to one can by Model Identification to
Amount.
Extracted under line and converted.According to the dimension of the extraction of table 2, how table 5 is optionally illustrated by these characteristics
Word, to form vector.The feature instructed according to table 5 digitizes formula, it is possible to produce the vector with distinguishing characteristic,
Model can easily be recognized and normal order and abandons list.
Table 5:Feature digitlization sample table under line
How the exemplary illustration of table 5, change into vector by the order that a History Order number is 711140718412498
Process.The formula of table 5 is also indicated that, as long as finding corresponding numerator value, just an order rapidly can be changed into vector, because
For denominator can regard constant as, store in internal memory.Order number in table 5 is a kind of sign, is ordered for unique sign one
Unidirectional amount.
Extracted on line and converted
Extraction process on line is needed according to transaction flow, and where combing can find corresponding feature and (look for if going out
To the denominator value of each dimension of the requirement of table 5), and this feature is converged, it is sent to the processing system service that malice accounts for an order
Device.The feature of needs can be just drawn into according to table 6:
Table 6:Special combing based on order status
Malice accounts for the numerator value of the processing system server according to each dimension of an order, just can be by the order according to table 5
Order vector is changed into, the two sub-models treatment for training is given.
Online part order number is different with table 5, order for processing may without regular order number, it is necessary to
Others sign.Such as, a random number is added using system exchange hour (being accurate to microsecond) during we put into practice, as any
The sign of one order.
Model is obtained, trained, checking
Positive example and counter-example are obtained
According to table 5 by an order vectorization after, be Successful Transaction according to the order, still abandon list, stick respectively "+
1 " and after " -1 " label, training file train.dat and test file test.dat is formed, it is possible to be put into model
Practise training.Data form in Training document can be as follows:
-1 1:0.29294 2:0.40746 3:0.44291 4:0.24736 5:0.00000 6:0.31343 7:
0.00811 8:0.05598 9:0.07370 10:0.35315 11:0.44291 12:0.25109 13:0.00000 14:
0.00000
Wherein " -1 " is exactly counter-example, and numbering 1 to 14 is exactly vector dimension, corresponding numerical value, is exactly each dimension according to table 5
The value that formula is calculated.
Model training
Positive example and counter-example data formation training text file train.dat are put into model acquisition program and be trained.
According to following commands steps, SVM model trainings can be just completed:
svm_learn-e 0.01train.dat bmodel_0920
As above in ordering, svm_learn is that eXecute UML program (has many publicity patterns can be with the various calculations of Free Acquisition
Method model), it is responsible for for train.dat changing into a kind of two sub-models bmodel_0920.
According to such as issuing orders as follows:
svm_classify bmodel_0920test.dat
The result that two sub-model bmodel_0920 test test.dat can be obtained.
If the result after being trained using BP algorithm of neural network, is exactly BP models.This model will be used for distinguishing just
Normal order and malice account for the model of an order.Universal model is to need to know its accuracy before application is put into.That is the model
Order of accuarcy is how high.
Model is verified and evaluated
The test data of universal model is sightless relative to training pattern.In order to reach such test effect, can
Directly to carry out test model using newest forms data of abandoning on line.Commented according to the model accuracy analysis method in data mining theories
The model accuracy that valency is obtained.Model accuracy can be just completed according to table 7 and table 8 to evaluate.
Table 7:The processing system that malice accounts for an order recognizes successfully order statistical form
In practice, the investigation method of table 8 may be more efficient:
Table 8:Single statistical form is abandoned in the processing system identification that malice accounts for an order
Inspection project | Item value | Index meaning |
Check data | 2015-1-3~4 | Investigate the time period |
Abandon and only count | 2,181 | Abandon single total number |
Success recognizes number | 43 | Model Identification total number out |
Success order is sentenced into and abandons odd number | 0 | Misjudgement situation is counted |
Single precision is abandoned in identification | 100%=43/ (43+0) | The precision of Model Identification malice order |
Accuracy | 0.95% | Recognize the accuracy of malice order |
Table 7 and table 8 are, by reifications such as learned classification of assessment model accuracy and recall rates, to be tested from two angles
The accuracy of model is set up in card and examination.Table 7 is surveyed under environment in pressure, is investigated and is maliciously accounted for whether an order strick precaution model is deposited on line
Situation is killed in a large amount of mistakes.Test result shows that the model maliciously accounted on the processing system line of an order is (after parameter adjustment
Model) it is very high for malice order accuracy of identification.
The precautionary measures are chosen and are implemented
For the precautionary measures described by table 4.Realization can be just implemented to according to table 9:
Table 9:The electronic passenger ticket transaction system precautionary measures are implemented
The last a kind of measure of table 9, it may be necessary to change existing transaction flow.But ours practice have shown that, these change
Amount is all little.By taking the implementation of the United Airlines, Inc as an example, we have only used the modification for just completing All Activity flow in 3 days, in the implementation of the United Airlines, Inc
During, we have just reached good only with mobile phone note verification code, shortening time of payment and rejection three kinds of measures of transaction
Effect.
On-line implement, checking and adjustment
Malice account for the processing system of an order according to it is as described above realize completing after, concluded the business system with original electronic passenger ticket
System constitutes interactive timing diagram and data flow figure as shown in Figure 5 as shown in Figure 4.Malice accounts for the processing system technology of an order
Scheme can be adjusted with different transaction applications.The processing system technical scheme implementation process that malice accounts for an order is as follows:
Hardware machine
Fig. 6 shows that maliciously accounting for the processing system technical scheme of an order needs to operate on independent server.Malice is accounted for
The processing system of seat order can use the hardware configuration mode of life or death.Malice accounts for the processing system of an order to hardware without special
It is required that, it is only necessary in two machine double-core 2G internal memories, installing linux or windows systems can just run.
Database
Inquiry is reviewed for convenience of data accumulation and daily record, maliciously accounting for the processing system of an order needs to connect database.Dislike
The processing system that meaning accounts for an order easily can come into contacts with various databases very much, because implementation process employs MyBatis
Connectivity Technical of Database.The MyBatis that can prevent SQL injection from attacking, supports that SQL database is grasped using standard SQL language description
Make.The processing system that malice accounts for an order can conveniently access more classical database, such as, and oracle, mysql, informix,
Db2, sqlserver, teradata, sybase, BDB etc..
Connector realizes deployment
The connector of the processing system that malice accounts for an order is the communication based on Web.Malice accounts for the processing system of an order
System according to httpclient agreements are used, the processing system that malice accounts for an order is sent the data to the text of JSON forms
System server, realizes that malice accounts for the processing system of an order and existing transaction system is in communication with each other.
This communication can just be completed using class HttpClient, and this class only needs to introduce
Org.apache.http.client.HttpClient, calls ready-made execute methods just to realize.
The service and self-management service that malice accounts for the processing system of an order can be deployed in popular web server
On, such as, websphere, tomcat, jetty, resin, jboss etc. issue container, or based on the only of servlet
Vertical process.The service of the processing system that malice accounts for an order is realized based on struts2.3+json+httpclient modes.
The processing system client that malice accounts for an order interacts realization with transaction system
As shown in Fig. 2 schematic diagrames, maliciously accounting for the processing system of an order needs a kind of client to complete and existing electronics
Passenger ticket transaction system is interacted.The client is exactly that connector is a kind of to be extended, and is completed using HttpClient+JSON.
The client needs to complete two kinds of function:First, the sequence information of existing transaction system is sent into evil
Meaning accounts for the processing system of an order, while being also required to tell that malice accounts for the processing system of an order by the change status information of order
System.Because the processing system that malice accounts for an order manages any air ticket order by the way of order life-cycle, i.e.,
One order of complete monitoring from submission, to generation order, to etc. it is to be paid, completed to paying, or as abandoning list, and mistake herein
In range monitoring, ceaselessly learn and model in more new line.Second, malice to be accounted for processing system diagnostic result and the configuration of an order
Good Precaution Tactics take back transaction system.Facilitate transaction system that the corresponding precautionary measures are made in subsequent step in time.
Interactive mode and information extraction mode as shown in table 10 below is sorted out in this preferred embodiment:
Table 10:Interaction implementation based on order status
According to the instruction of table 10, required for just completing to be extracted in transaction system the processing system for maliciously accounting for an order
Information.Mark emphatically be that malice accounts for the processing system of an order and transaction system is interacted it is most crucial once.This is returned
Mutually complete the transmission of diagnostic result and Precaution Tactics.And the management of follow-up four states, malice can be allowed to account for the place of an order
Reason system grasps newest order messages and state in time, completes the on-line study and renewal of itself.Malice accounts for the place of an order
Why reason system easily just can accurately judge that a new malice accounts for a user in the second single cross, and the logic of dependence is exactly order
The strict tracking and judgement of state.
Model preparation
According to the method described in table 2 and table 4, can quickly from the History Order storehouse of transaction system, find positive example and
Counter-example., it is necessary to maintain one kind of positive example and counter-example balanced during this.It is proposed that the ratio control of positive example and counter-example is existed
5:1 or so.
If using artificial neural network BP model, the model for finally obtaining needs developer's energy reduced model data.Otherwise,
Need to encapsulate the model with Java localizationization technologies, the input of singly test vector is received to facilitate.It is proposed that
Developer conscientiously understands the learning matrix value that newff (one of method that BP is realized in matlab) method is returned.So as to energy side
Just input database and J2EE platforms.
Such as, for supporting vector machine model (SVM), we have conscientiously parsed the model obtained after training in practice,
Allow the model to be presented in database table in the way of learning parameter, clearly indicate each and extract the important of dimension values
Property and correlation.May certify that the model after any SVM training may finally be converted into a hyperplane equation.
If SVM models are difficult to be developed personnel and are understood, decision tree and Bayesian model are also good selection, training
The model for going out can quickly take parameter.
Model analyzing can be brought into many benefits into learning parameter.Can intuitively watch, can be stored in table, Ke Yifang
Just it is put into internal memory, it is also possible to easily human intervention model.
Checking
Using table 8 and the method for table 7, can very intuitively on the line of observation model accuracy.The model on adjustment line
In precision process, practice informs us:The last degree of accuracy of table 8 and table 7 can not possibly be all very high, i.e. malice order is correctly recognized
The degree of accuracy and the correct recognition accuracy of normal order are all very high, and this is probably unpractical, and this meets ROC, and (model accuracy is recalled
Rate is checked) theoretical explanation.
Dull and rush season for sale of ticket uses spring, summer, autumn, winter all different model, as shown in Figure 7.
Normal operation
Malice account for an order processing system normally put into operation after, the flight that can wherein account for a related frequency is managed
Reason, as shown in Figure 8.
The condition of selection protection flight can be the flight that basis currently abandons before the ranking that single state is calculated 10, also may be used
Being that visiting rate exceedes the flight of threshold value (typically visiting rate exceedes classical threshold value, illustrate that when the time comes exactly the flight sells air ticket
Gold opportunity).
Selective protection flight can alleviate two misgivings:The transition that wrong may be killed to model is worried, and pair can
Poor user can be brought to experience filtering.And, in the precautionary measures, can not select strong refusal to conclude the business as far as possible, such as Fig. 9 institutes
Show.
According to the configuration shown in Fig. 9, even if there is malice occupant in protection flight, transaction system is that ejection mobile phone is tested
Card code, when this it is lower one be conscientiously input into correct mobile phone identifying code after, still can complete transaction.This mode is relative
" refusal service " is loose.Malice based on two sub-models accounts for the processing system preferable precision ejection checking of an order
Code, this is a kind of warning for frequent malice occupant and deters.
In the precautionary measures simultaneously, it would however also be possible to employ various precautionary measures come into force.Such as, specific aim uses and " first pays the bill, then give
The precautionary measures of order number ".
Such scheme is compared with " first pay the bill, then account for seat " precautionary measures, maliciously account for the processing system of an order have it is following
Distinguishing feature:(1) take precautions against in advance." can first stay seat, then pay the bill " without changing existing habit of transaction.Because malice is accounted for
The processing system of seat order it is lower one fill in after completion seizes the opportunity people's information, can just analyze the malice degree of the order.(2)
Combination features are analyzed.The processing system that malice accounts for an order be it is a set of based on lower single inquiry, filling in order content based on
Analytical technology.One IP for often maliciously accounting for seat, it is also possible to normally place an order.Simply in normal placing an order, normally under single table
Existing behavioural characteristic shows different when maliciously being accounted for it.(3) on-line study.For the malice without any historical record
Occupant, the second list can just judge that it maliciously accounts for a behavior.Have for one account for seat record it is lower one, first single can just sentence
It is disconnected.(4) take precautions against point and set flexible.The processing system energy ex ante analysis order malice degree of an order is accounted for due to malice, so that can
To allow electronic passenger ticket transaction system to flexibly set strick precaution point.Such as, take precautions against point can it is lower one filled in and seized the opportunity people's information
Afterwards, after one has filled in information instantly, if the order malice degree is high, mobile phone note verification code can immediately be ejected defeated
Enter the page, it is desirable to lower single input validation code, and 99% it is normally lower one, be that can't see such identifying code page.When
So can also be directed to method of commerce of such malicious user using " first pay the bill, then stay seat ", but will not for it is vast it is normal under it is alone
Family.
Additionally, the processing system that malice accounts for an order is quantitatively to answer in a much degree of order with the method for classification
Malice order, target is (- 1 ,+1).If we are by target serialization, this recurrence way for being based on two sub-models, it is also possible to
For judging any one agent for percentage contribution in the transaction of whole civil aviaton's electronic passenger ticket, and handed in whole electronic passenger ticket
Weight order in easily.And according to this order and percentage contribution, arrange different access access authorization for resource.
Such as, if the thinking of the processing system for accounting for an order using malice carries out Civil Aviation System internal gain model management, give
Different agency's marking, then, agent is given different preferential degree according to different agent's score values.This it is preferential can be with AV
It is preferential in resource access times, it is preferential in freight rate inquiry times.We can find according to the method shown in table 2 completely
Some can reflect the dimension of agent's identity, such as, agent's amount of drawing a bill of nearly month abandons single amount, Building K number of times, special behaviour
Make number of times, violation number of times etc. as data vector dimension.For desired value, we can vouch ratio according to it, abandon single rate comprehensive
Altogether, artificial marking.Then normalization using proposed model, just can quickly obtain learning model to (0,1), instruct
We carry out giving a mark to all agents.
The preferred embodiments of the present invention are the foregoing is only, is not intended to limit the invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.It is all within the spirit and principles in the present invention, made any repair
Change, equivalent, improvement etc., should be included within the scope of the present invention.
Claims (16)
1. a kind of malice accounts for the processing method of an order, it is characterised in that including:
From ticketing data storehouse extract it is lower one in the behavioural characteristic before single operation under execution and/or during single operation under performing and described
The sequence information of one this order by under lower;
According to the behavioural characteristic and the sequence information, the malice degree of the order is judged;
In the case where the malice degree for determining the order exceedes pre-set threshold value, take malice to account for seat for the order and prevent
Model means.
2. method according to claim 1, it is characterised in that the behavioural characteristic and the sequence information include it is following extremely
It is one of few:
Down single ID correspondence abandon single rate, lower single number of repetition, order people's name number of repetition, order people Email numbers of repetition,
Order people's phone number of repetition, order people regular guest number of repetition, the ID numbers of repetition of PNR, PNR names number of repetition, account for seat
Number, the last time abandon single time, apart from the departure time, whether account for popular flight, whether have wait pay invoice.
3. method according to claim 1, it is characterised in that the behavioural characteristic and the sequence information also include following
At least one:
According to log information extract it is lower one satisfy the need the search of line price it is time-consuming, according to log information extract it is lower one fill in
Seize the opportunity the time-consuming of people's information.
4. method according to claim 1, it is characterised in that according to the behavioural characteristic and the sequence information, judges
The malice degree of the order includes:
Set up the malice degree model based on two sub-models;
The successful order that will be concluded the business in History Order accounts for an order as non-malicious, by Fail Transaction in History Order and does not attempt
The order for carrying out delivery operation accounts for an order as malice, accounts for an order by the non-malicious and the malice accounts for an order
Corresponding behavioural characteristic and sequence information, are trained to the malice degree model respectively;
Using trained malice degree model, the malice degree of one this order by under lower is judged.
5. method according to claim 4, it is characterised in that two sub-model includes one below:
Decision-tree model, neural network model, Bayesian model, Multiple regression model, supporting vector machine model.
6. method according to any one of claim 1 to 5, it is characterised in that take malice to account for seat for the order
Risk prevention instrumentses include:
Operational phase according to residing for the order determines to take precautions against point;
Malice is taken to account for a risk prevention instruments in the point of taking precautions against.
7. method according to claim 6, it is characterised in that the strick precaution point includes at least one of:
When seizing the opportunity the submission of people's information, when confirming when seizing the opportunity people and line information, generating order number and be prepared to enter into payment flow.
8. method according to claim 6, it is characterised in that the malice account for a risk prevention instruments include it is following at least it
One:
Picture checking, mobile phone short message verification, shorten the time of payment, actively abandon order, refusal transaction, order pay successfully after again
Non- pay invoice then discharges seat in providing order number and short time.
9. a kind of malice accounts for the processing system of an order, it is characterised in that including:
Abstraction module, for from ticketing data storehouse extract it is lower one before single operation under execution and/or during single operation under performing
The sequence information of behavioural characteristic and one this order by under lower;
Judge module, for according to the behavioural characteristic and the sequence information, judging the malice degree of the order;
Module is taken precautions against, in the case of exceeding pre-set threshold value in the malice degree for determining the order, for the order
Malice is taken to account for a risk prevention instruments.
10. system according to claim 9, it is characterised in that the behavioural characteristic and the sequence information include following
At least one:
Down single ID correspondence abandon single rate, lower single number of repetition, order people's name number of repetition, order people Email numbers of repetition,
Order people's phone number of repetition, order people regular guest number of repetition, the ID numbers of repetition of PNR, PNR names number of repetition, account for seat
Number, the last time abandon single time, apart from the departure time, whether account for popular flight, whether have wait pay invoice.
11. systems according to claim 9, it is characterised in that the behavioural characteristic and the sequence information also include with
It is at least one lower:
According to log information extract it is lower one satisfy the need the search of line price it is time-consuming, according to log information extract it is lower one fill in
Seize the opportunity the time-consuming of people's information.
12. systems according to claim 9, it is characterised in that the judge module includes:
Modeling unit, for setting up the malice degree model based on two sub-models;
Training unit, for will be concluded the business in History Order, successful order accounts for an order as non-malicious, will be handed in History Order
The order for easily failing and not attempting to carry out delivery operation accounts for an order as malice, and an order and institute are accounted for by the non-malicious
State malice and account for the corresponding behavioural characteristic of order difference and a sequence information, the malice degree model is trained;
Judging unit, for using trained malice degree model, judges one this order by under lower
Malice degree.
13. systems according to claim 12, it is characterised in that two sub-model includes one below:
Decision-tree model, neural network model, Bayesian model, Multiple regression model, supporting vector machine model.
14. system according to any one of claim 9 to 13, it is characterised in that the strick precaution module includes:
Determining unit, determines to take precautions against point for the operational phase according to residing for the order;
Unit is taken precautions against, for taking malice to account for a risk prevention instruments in the point of taking precautions against.
15. systems according to claim 14, it is characterised in that the strick precaution point includes at least one of:
When seizing the opportunity the submission of people's information, when confirming when seizing the opportunity people and line information, generating order number and be prepared to enter into payment flow.
16. systems according to claim 14, it is characterised in that the malice account for a risk prevention instruments include it is following at least it
One:
Picture checking, mobile phone short message verification, shorten the time of payment, actively abandon order, refusal transaction, order pay successfully after again
Non- pay invoice then discharges seat in providing order number and short time.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611263344.7A CN106779126A (en) | 2016-12-30 | 2016-12-30 | Malice accounts for the processing method and system of an order |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611263344.7A CN106779126A (en) | 2016-12-30 | 2016-12-30 | Malice accounts for the processing method and system of an order |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106779126A true CN106779126A (en) | 2017-05-31 |
Family
ID=58954006
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611263344.7A Pending CN106779126A (en) | 2016-12-30 | 2016-12-30 | Malice accounts for the processing method and system of an order |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106779126A (en) |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107403472A (en) * | 2017-08-10 | 2017-11-28 | 中国民航信息网络股份有限公司 | The method and apparatus that modification operation to electronic passenger ticket is monitored |
CN107403325A (en) * | 2017-08-10 | 2017-11-28 | 中国民航信息网络股份有限公司 | Air ticket order reliability evaluation method and device |
CN107481019A (en) * | 2017-07-28 | 2017-12-15 | 上海携程商务有限公司 | Order fraud recognition methods, system, storage medium and electronic equipment |
CN108229749A (en) * | 2018-01-16 | 2018-06-29 | 厦门快商通信息技术有限公司 | Bad booking behavior management method based on deep learning |
CN108390853A (en) * | 2018-01-10 | 2018-08-10 | 北京思特奇信息技术股份有限公司 | A kind of malice IP processing methods and system |
CN108564423A (en) * | 2017-12-28 | 2018-09-21 | 携程旅游网络技术(上海)有限公司 | Malice occupy-place recognition methods, system, equipment and the storage medium of ticketing service order |
CN109064255A (en) * | 2018-07-05 | 2018-12-21 | 厦门微芽互娱科技有限公司 | Product is prescribed a time limit method of commerce, medium, terminal device and system |
CN109345332A (en) * | 2018-08-27 | 2019-02-15 | 中国民航信息网络股份有限公司 | A kind of intelligent detecting method of Airline reservation malicious act |
CN109377301A (en) * | 2018-08-27 | 2019-02-22 | 中国民航信息网络股份有限公司 | A kind of Feature Extraction Method based on Airline reservation behavioral data |
CN109886631A (en) * | 2019-02-27 | 2019-06-14 | 深圳市丰巢科技有限公司 | Courier sends monitoring and managing method, device, equipment and the medium of part behavior |
CN110443675A (en) * | 2019-06-27 | 2019-11-12 | 北京三快在线科技有限公司 | Determine method, apparatus, electronic equipment and the storage medium of order life-cycle |
CN111970269A (en) * | 2020-08-14 | 2020-11-20 | 中国民航信息网络股份有限公司 | Server access behavior identification method and device and server |
CN112163932A (en) * | 2020-09-30 | 2021-01-01 | 中国民航信息网络股份有限公司 | Malicious seat occupying order identification method and device and electronic equipment |
CN112907263A (en) * | 2021-03-22 | 2021-06-04 | 北京太火红鸟科技有限公司 | Abnormal order quantity detection method, device, equipment and storage medium |
CN117892145A (en) * | 2024-01-18 | 2024-04-16 | 朗臻(广州)信息技术有限责任公司 | Air ticket order safety detection method, system, equipment and storage medium |
CN117892145B (en) * | 2024-01-18 | 2024-09-03 | 朗臻(广州)信息技术有限责任公司 | Air ticket order safety detection method, system, equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104809589A (en) * | 2015-05-08 | 2015-07-29 | 北京嘀嘀无限科技发展有限公司 | Order processing method and device |
CN105447137A (en) * | 2015-11-23 | 2016-03-30 | 浪潮软件股份有限公司 | Algorithm for retrieving same master-slave relation data from big data based on relational database |
CN105468742A (en) * | 2015-11-25 | 2016-04-06 | 小米科技有限责任公司 | Malicious order recognition method and device |
CN105763548A (en) * | 2016-02-06 | 2016-07-13 | 北京祥云天地科技有限公司 | User login identification method based on behavior model and equipment and system thereof |
CN106251202A (en) * | 2016-07-29 | 2016-12-21 | 北京小米移动软件有限公司 | Maliciously order recognition methods and device |
-
2016
- 2016-12-30 CN CN201611263344.7A patent/CN106779126A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104809589A (en) * | 2015-05-08 | 2015-07-29 | 北京嘀嘀无限科技发展有限公司 | Order processing method and device |
CN105447137A (en) * | 2015-11-23 | 2016-03-30 | 浪潮软件股份有限公司 | Algorithm for retrieving same master-slave relation data from big data based on relational database |
CN105468742A (en) * | 2015-11-25 | 2016-04-06 | 小米科技有限责任公司 | Malicious order recognition method and device |
CN105763548A (en) * | 2016-02-06 | 2016-07-13 | 北京祥云天地科技有限公司 | User login identification method based on behavior model and equipment and system thereof |
CN106251202A (en) * | 2016-07-29 | 2016-12-21 | 北京小米移动软件有限公司 | Maliciously order recognition methods and device |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107481019A (en) * | 2017-07-28 | 2017-12-15 | 上海携程商务有限公司 | Order fraud recognition methods, system, storage medium and electronic equipment |
CN107403325A (en) * | 2017-08-10 | 2017-11-28 | 中国民航信息网络股份有限公司 | Air ticket order reliability evaluation method and device |
CN107403325B (en) * | 2017-08-10 | 2021-05-11 | 中国民航信息网络股份有限公司 | Method and device for evaluating reliability of air ticket order |
CN107403472A (en) * | 2017-08-10 | 2017-11-28 | 中国民航信息网络股份有限公司 | The method and apparatus that modification operation to electronic passenger ticket is monitored |
CN108564423A (en) * | 2017-12-28 | 2018-09-21 | 携程旅游网络技术(上海)有限公司 | Malice occupy-place recognition methods, system, equipment and the storage medium of ticketing service order |
CN108390853B (en) * | 2018-01-10 | 2021-03-30 | 北京思特奇信息技术股份有限公司 | Malicious IP processing method and system |
CN108390853A (en) * | 2018-01-10 | 2018-08-10 | 北京思特奇信息技术股份有限公司 | A kind of malice IP processing methods and system |
CN108229749A (en) * | 2018-01-16 | 2018-06-29 | 厦门快商通信息技术有限公司 | Bad booking behavior management method based on deep learning |
CN109064255A (en) * | 2018-07-05 | 2018-12-21 | 厦门微芽互娱科技有限公司 | Product is prescribed a time limit method of commerce, medium, terminal device and system |
CN109377301A (en) * | 2018-08-27 | 2019-02-22 | 中国民航信息网络股份有限公司 | A kind of Feature Extraction Method based on Airline reservation behavioral data |
CN109345332A (en) * | 2018-08-27 | 2019-02-15 | 中国民航信息网络股份有限公司 | A kind of intelligent detecting method of Airline reservation malicious act |
CN109886631A (en) * | 2019-02-27 | 2019-06-14 | 深圳市丰巢科技有限公司 | Courier sends monitoring and managing method, device, equipment and the medium of part behavior |
CN110443675A (en) * | 2019-06-27 | 2019-11-12 | 北京三快在线科技有限公司 | Determine method, apparatus, electronic equipment and the storage medium of order life-cycle |
CN111970269A (en) * | 2020-08-14 | 2020-11-20 | 中国民航信息网络股份有限公司 | Server access behavior identification method and device and server |
CN111970269B (en) * | 2020-08-14 | 2022-04-08 | 中国民航信息网络股份有限公司 | Server access behavior identification method and device and server |
CN112163932A (en) * | 2020-09-30 | 2021-01-01 | 中国民航信息网络股份有限公司 | Malicious seat occupying order identification method and device and electronic equipment |
CN112907263A (en) * | 2021-03-22 | 2021-06-04 | 北京太火红鸟科技有限公司 | Abnormal order quantity detection method, device, equipment and storage medium |
CN117892145A (en) * | 2024-01-18 | 2024-04-16 | 朗臻(广州)信息技术有限责任公司 | Air ticket order safety detection method, system, equipment and storage medium |
CN117892145B (en) * | 2024-01-18 | 2024-09-03 | 朗臻(广州)信息技术有限责任公司 | Air ticket order safety detection method, system, equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106779126A (en) | Malice accounts for the processing method and system of an order | |
US10885561B2 (en) | Transaction facilitating marketplace platform | |
TWI712981B (en) | Risk identification model training method, device and server | |
US20180365228A1 (en) | Tree kernel learning for text classification into classes of intent | |
US20210004795A1 (en) | Anomaly and fraud detection using duplicate event detector | |
US20160358268A1 (en) | Methods and systems for automatically detecting fraud and compliance issues in expense reports and invoices | |
Abbott | Applied predictive analytics: Principles and techniques for the professional data analyst | |
CN108399509A (en) | Determine the method and device of the risk probability of service request event | |
CN108985347A (en) | Training method, the method and device of shop classification of disaggregated model | |
US20190073660A1 (en) | Classification by natural language grammar slots across domains | |
CN110084468B (en) | Risk identification method and device | |
US11282078B2 (en) | Transaction auditing using token extraction and model matching | |
Abrahams et al. | Audience targeting by B-to-B advertisement classification: A neural network approach | |
CN110659961A (en) | Method and device for identifying off-line commercial tenant | |
CN111882420A (en) | Generation method of response rate, marketing method, model training method and device | |
US11250345B1 (en) | Methods for identifying transactions with user location information | |
Restrepo-Amariles | Algorithmic decision systems: automation and machine learning in the public administration | |
US10803508B1 (en) | Method, medium, and system for deal simulation between merchants and purchasers | |
CN113393299A (en) | Recommendation model training method and device, electronic equipment and storage medium | |
Hatim et al. | E-FoodCart: An online food ordering service | |
CN112990989B (en) | Value prediction model input data generation method, device, equipment and medium | |
CN105719145A (en) | Method and device for acquiring arrival time of goods | |
Macanovic et al. | The moral embeddedness of cryptomarkets: text mining feedback on economic exchanges on the dark web | |
Gupta | Applied analytics through case studies using Sas and R: implementing predictive models and machine learning techniques | |
CN111373395A (en) | Artificial intelligence system and method based on hierarchical clustering |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170531 |