WO2020114352A1 - 计算设备的防误操作处理 - Google Patents

计算设备的防误操作处理 Download PDF

Info

Publication number
WO2020114352A1
WO2020114352A1 PCT/CN2019/122406 CN2019122406W WO2020114352A1 WO 2020114352 A1 WO2020114352 A1 WO 2020114352A1 CN 2019122406 W CN2019122406 W CN 2019122406W WO 2020114352 A1 WO2020114352 A1 WO 2020114352A1
Authority
WO
WIPO (PCT)
Prior art keywords
misoperation
user
operating environment
terminal interface
probability
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.)
Ceased
Application number
PCT/CN2019/122406
Other languages
English (en)
French (fr)
Inventor
李陆启
李一山
纪伟
周熠
张鑫淼
杨飞宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Sankuai Online Technology Co Ltd
Original Assignee
Beijing Sankuai Online Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Sankuai Online Technology Co Ltd filed Critical Beijing Sankuai Online Technology Co Ltd
Publication of WO2020114352A1 publication Critical patent/WO2020114352A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/03—Arrangements for converting the position or the displacement of a member into a coded form
    • G06F3/041—Digitisers, e.g. for touch screens or touch pads, characterised by the transducing means
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/03—Arrangements for converting the position or the displacement of a member into a coded form
    • G06F3/033—Pointing devices displaced or positioned by the user, e.g. mice, trackballs, pens or joysticks; Accessories therefor
    • G06F3/0354—Pointing devices displaced or positioned by the user, e.g. mice, trackballs, pens or joysticks; Accessories therefor with detection of two-dimensional [2D] relative movements between the device, or an operating part thereof, and a plane or surface, e.g. 2D mice, trackballs, pens or pucks

Definitions

  • the present disclosure relates to the field of Internet technology, and in particular, to an operation processing method and apparatus, and a computing device.
  • mobile terminals are widely used in people's life and work. When people use mobile terminals, they usually operate them by touching the screen or clicking with a mouse. According to statistics, up to 50% of mobile ad clicks are accidentally clicked, and in addition to mobile ads, other types of functions and pages in mobile touch screen applications also have a large number of unintentional wrong clicks, for example, the user's attention when operating It is not on the guide element, just the wrong touch will lead to a mistake; for example, when the new content replaces the old content instantaneously, the user does not have enough time to judge, and intends to click on the old element, but clicks on the newly appeared element, causing the mistake.
  • the purpose of the present disclosure is to provide an operation processing method and apparatus and a computing device, so as to overcome, at least to some extent, one or more problems due to limitations and defects of the related art.
  • an operation processing method includes: reading values of a plurality of operating environment factors based on user operations detected by a terminal interface; The value is input into a pre-trained operation recognition model to obtain a probability value that the user operation belongs to a misoperation.
  • the operation recognition model includes weight coefficients corresponding to the plurality of operating environment factors; and The range of values enables the terminal interface to trigger a corresponding subsequent response.
  • an operation processing device which includes: an environment detection module configured to read values of a plurality of operation environment factors based on user operations detected by a terminal interface; an operation recognition module , Set to input values of the plurality of operating environment factors into a pre-trained operation recognition model to obtain a probability value that the user operation belongs to a misoperation, and the operation recognition model includes corresponding to the plurality of operation environment factors A weighting factor; and an operation response module, which is set to cause the terminal interface to trigger a corresponding subsequent response according to the numerical range that the probability value falls within.
  • a storage medium storing a computer program, which when executed by a processor of a computer, causes the computer to execute the method as described in the first aspect above.
  • a computing device including: a processor; a memory storing instructions executable by the processor; wherein the processor is configured to execute as described in the first aspect above Methods.
  • an operation recognition model is established based on the user's historical operation to output the user's misoperation probability in real time, and the subsequent response is controlled accordingly, which can improve the recognition accuracy rate and accurately classify to provide more accurate data and save Traffic waste and improve user experience.
  • Fig. 1 is a flow chart of an operation processing method according to an exemplary embodiment.
  • Fig. 2 is a flow chart of an operation processing method according to an exemplary embodiment.
  • Fig. 3 is a schematic diagram of an interface for receiving user feedback according to an exemplary embodiment.
  • FIG. 4 is an exemplary flowchart of step S220 in the embodiment shown in FIG. 2.
  • Fig. 5 is a flow chart of an operation processing method according to an exemplary embodiment.
  • Fig. 6 is a flow chart showing an operation processing method according to another exemplary embodiment.
  • Fig. 7 is a schematic diagram of a calculation process of an operation process according to another exemplary embodiment.
  • FIG. 8 is a structural block diagram of an operation processing device according to an embodiment of the present disclosure.
  • FIG. 9 is a schematic diagram of a computing device according to an embodiment of the present disclosure.
  • a method and system for preventing false clicks of advertisements based on pressure sensing technology receives the pressure of the user's finger on the advertisement bar and judges whether it reaches the set value; wherein, when the pressure is greater than the first set value, the preset is triggered The first event is to enter the advertisement interface linked to the advertisement bar or display a full-screen advertisement interface to the user.
  • the judgment basis of the false click of the advertisement is the finger pressure of a single user, and in comparing the finger pressure with the set value, the diversity of different user groups and objective environments is not considered, so it may be There will be some inaccurate judgments of operation errors. In addition, the problem of misplacement caused by the replacement of old and new content cannot be solved.
  • Fig. 1 is a flow chart of an operation processing method according to an exemplary embodiment. As shown in FIG. 1, the method of this embodiment includes the following steps S110-S130. In one embodiment, the method of this embodiment may be executed on a user terminal (eg, mobile phone, tablet computer).
  • a user terminal eg, mobile phone, tablet computer.
  • step S110 based on the user operation detected on the terminal interface, the values of a plurality of operating environment factors are read.
  • the operating environment factor is the environmental data collected by the user terminal when the user operates the terminal interface.
  • the operating environment factors may include, for example, but not limited to, display layout categories, hardware environment categories, user information categories, and guide element content categories.
  • the display layout class refers to the parameters involved in the display of the operation object in the terminal interface.
  • the display layout class may include, for example, but not limited to, the visible percentage of the user operation object (the percentage of the visible area of the operation object in the area of the operation page), and the user operation position (for example The distance between the coordinates of the user's operation position and the center point of the operation object), the display duration of the operation object (the time interval between the loading and display of the operation object until the operation occurs), the duration of the user interface rest (the interval between the scrollable list scrolling and the time when the user operation occurs) 2.
  • the display time of "in-situ content replacement" (the time interval from when the new page content is loaded and the original content is replaced until the user operation occurs), etc.
  • the hardware environment category refers to the hardware-related parameters of the current terminal, which may include, for example, but not limited to CPU (Central Processing Unit) occupancy rate, available memory, stuck duration, etc.
  • the user information category refers to parameters related to the current user that can affect their operating preferences, which may include, for example, but not limited to, the user's age, gender, job category, education, etc.
  • the guide element content class refers to the relevant parameters of the current operation object, which may include, for example, but not limited to the theme, nature, etc. of the operation object.
  • step S120 the values of a plurality of operating environment factors are input into a pre-trained operation recognition model to obtain a probability value that a user operation belongs to an erroneous operation.
  • the operation recognition model includes weight coefficients corresponding to the plurality of operating environment factors.
  • the operation recognition model can be used to calculate the probability value that the user operation is a misoperation.
  • the operation recognition model may further include a linear function between the probability that each operating environment factor causes a misoperation and the operating environment factor.
  • the probability of misoperation caused by each operating environment factor can be obtained first according to the linear function, and the probability of misoperation caused by each operating environment factor and its corresponding The weight coefficients are weighted and summed, and the sum value is the probability value that the user operation is a misoperation.
  • step S130 the terminal interface triggers a corresponding subsequent response according to the numerical range that the probability value falls within.
  • the misoperation attribute of the current user operation may be determined according to the numerical range in which the probability value falls. Corresponding subsequent responses can be distinguished based on misoperation attributes.
  • the terminal interface is stopped from responding to the user operation.
  • the terminal interface is caused to generate a prompt for receiving user feedback, and confirm whether to continue the response to the user operation based on the received user feedback.
  • the probability value falls within the third numerical range, it is confirmed that the user operation is not a misoperation, and the terminal interface continues to respond to the user operation.
  • multiple operating environment factors and misoperation attributes of the current user operation may also be used as training data of the operation recognition model to update the operation recognition model in real time to improve its recognition accuracy.
  • Fig. 2 is a flowchart of an operation processing method according to an exemplary embodiment. As shown in Fig. 2, the method of this embodiment includes the following steps S210-S250.
  • step S210 sample data for training an operation recognition model is collected, the sample data includes multiple historical operations of the detected user, historical values of operating environment factors corresponding to multiple historical operations, and multiple historical operations Misuse attribute.
  • step S130 can be obtained in real time to update the sample data for training the operation recognition model.
  • the misoperation attribute can indicate whether the corresponding user operation is a misoperation.
  • the method for acquiring the misoperation attribute may be, for example: calculating the return operation time value based on the historical operation, and the return operation time value represents the time elapsed from the detection of the operation triggered by the user to trigger the terminal interface to the detection of the return operation; When the returned operation time value is less than the preset time value, a prompt interface for receiving user feedback is generated; and based on the user feedback received on the prompt interface, the misoperation attribute of the historical operation is determined.
  • the prompt interface for receiving user feedback may be as shown in FIG. 3, for example.
  • the operation that the user triggers the terminal interface switching is, for example, the operation of clicking a certain link.
  • a return operation is detected to return to the clicked page when the user triggered the terminal interface switching operation.
  • the above two operations The interval between them is the return operation time value.
  • the value of the preset time may be calculated and obtained based on actual statistical data or mechanism data, and the embodiments of the present disclosure have no limitation on this.
  • the method for acquiring the misoperation attribute may, for example, determine whether the historical operation meets the following conditions: the historical operation includes the first operation, the second operation, and the third operation in sequence; the first page, the first Two pages and a first page; the first operation is detected when the in-situ replacement of the display element occurs on the first page, the terminal interface switches to the second page in response to the first operation, and the second operation is detected on the second page, The terminal interface returns to the first page in response to the second operation, detects a third operation for the display element before in-situ replacement occurs on the first page; and when the historical operation meets the above conditions, it is determined that the first operation is a misoperation.
  • in-place replacement refers to that the original display content of a certain display area is replaced with new display content on the page. In-situ replacement may occur, for example, but not limited to, during the rolling display of advertisements or during the rotation of different advertisements.
  • the first page is the terminal interface, which contains the display elements that the user intends to click, such as the advertisement page;
  • the first operation is the operation that the user clicks on the advertisement page of the in-place replacement area when the in-situ replacement occurs, that is, the advertisement opened by the user is not The originally displayed advertisement on the first page;
  • the second page is a jump page in response to the first operation, that is, a new advertisement page after in-situ replacement;
  • the second operation may be a user's return operation, which is used to return to the first page.
  • the third operation may be, for example, a click operation performed by the user on the display element before the in-situ replacement occurs on the first page, that is, a click operation performed on the advertisement page that the user intends to click.
  • step S220 a weight coefficient corresponding to a plurality of operating environment factors is calculated based on the sample data.
  • the weight coefficients are included in the operation recognition model.
  • the calculation method can be, for example, a neural network to learn the sample data. Multiple operation environment factors in the sample data are used as the input of the neural network, and the probability of misoperation is used as the expected output. After iterative training, to Obtain weight coefficients corresponding to multiple operating environment factors.
  • step S230 based on the user operation detected on the terminal interface, the values of a plurality of operating environment factors are read.
  • step S240 the values of a plurality of operating environment factors are input into a pre-trained operation recognition model to obtain a probability value that a user operation belongs to a misoperation.
  • the operation recognition model includes weight coefficients corresponding to the plurality of operating environment factors.
  • step S250 the terminal interface triggers a corresponding subsequent response according to the numerical range that the probability value falls within.
  • Steps S230-S250 have been explained in detail in FIG. 1 and will not be repeated here.
  • step S220 may be implemented as the exemplary process shown in FIG. 4, including the following steps S410-S430.
  • step S410 cluster analysis is performed on a plurality of operating environment factors according to the degree of influence caused by misoperation.
  • Clustering is a process of dividing a collection of physical or abstract objects into multiple classes composed of similar objects.
  • a clustering algorithm can be used to classify the samples according to the values of multiple operating environment factors.
  • step S420 a linear fitting is performed on each operating environment factor, and the probability of misoperation caused by the operating environment factor is calculated.
  • Linear fitting can characterize the linear function between each operating environment factor and the probability that the operating environment factor causes a misoperation.
  • the linear function obtained by linear fitting can be used to determine the probability of malfunction due to the value of the current operating environment factor.
  • step S430 based on the back-propagation BP neural network, the probability of erroneous operation caused by multiple operating environment factors is used as input to perform forward calculation, and reverse correction is performed according to the erroneous operation attribute, and iteratively obtains corresponding to multiple operating environment factors Weight coefficient.
  • BP backpropagation neural network
  • the BP neural network algorithm includes two processes of forward propagation of signals and back propagation of errors.
  • the weight coefficient may be represented by, for example, the weight value between the input neuron and the hidden layer neuron in the BP neural network.
  • Fig. 5 is a flow chart of an operation processing method according to an exemplary embodiment. As shown in the figure, the method of this embodiment includes the following steps S510-S520.
  • step S510 the values of a plurality of operating environment factors are respectively input into a linear function to obtain the probability of misoperation caused by each operating environment factor.
  • the linear function may be acquired by step S420, for example.
  • step S520 based on the weighted sum of the probability of erroneous operation caused by each operating environment factor and the weighting coefficient, the probability value that the user operation belongs to the erroneous operation is obtained.
  • the probability of misoperation is expressed as P 1 , P 2 , ..., P n
  • their corresponding weight coefficients are expressed as W 1 , W 2 , respectively. .., W n
  • the formula for calculating the probability that the user operation is a misoperation can be expressed as:
  • Fig. 6 is a flow chart showing an operation processing method according to another exemplary embodiment. As shown in the figure, the method of this embodiment includes the following steps S610-S630.
  • step S610 sample data is collected.
  • the sample data can be obtained, for example, by collecting and extracting multiple historical operations of the user.
  • the sample data may include, for example, user information type data, display layout type data, hardware environment type data, and guide element content type data.
  • the value of the user's erroneous response time can also be collected, that is, after clicking the guidance factor, click the back button to return to the original page N seconds after the page jump occurs. Take N seconds as the value of the delayed reaction time.
  • the value of the reaction time to the misoperation can be used, for example, to determine the misoperation attribute of the current user operation. The method for determining the misoperation attribute has been described in detail in the embodiment of step S210, and will not be repeated here.
  • step S620 the weight coefficients of multiple environmental factors are calculated to obtain an operation recognition model.
  • the weight coefficients of environmental factors can be calculated by the BP neural network. Taking the probability of misoperation caused by multiple operating environment factors as input, the weight coefficient of the BP neural network is reversely corrected according to the misoperation attribute, and by learning the sample data, the weight value of a set of input layers to the hidden layer can be obtained , That is, the weight coefficient desired in this step.
  • step S630 the operation recognition model is delivered to the user terminal to perform real-time judgment on the user operation.
  • the input of the operation recognition model is the probability of misoperation caused by the operating environment factor. By weighting and summing it with the weight coefficient, the probability value that the user operation belongs to misoperation can be obtained.
  • the misoperation recognition process can be shown in FIG. 7 for example.
  • different subsequent responses can be selected through probability values.
  • a threshold of 0.5 is set. When the probability value is greater than 0.5, the user operation is considered to be a misoperation, and the user can choose to terminate the response to the user operation; when the probability value is less than 0.5, the user operation is not considered to be a misoperation, and the user can continue to respond to the current User actions.
  • an intermediate value range can be set. When the probability value falls within the value range, it is considered that the misoperation attribute of the user operation is not very certain.
  • a prompt interface for receiving user feedback may be generated. The interface may be, for example, FIG. 3 As shown.
  • the operation processing method of this embodiment by collecting a plurality of operation environment factors during the operation of the user, and using a pre-trained operation to identify the probability value of the model misoperation. Triggering the subsequent response of the terminal interface according to the probability value can improve the accuracy of identifying whether the user operation is a misoperation, prevent entering the other interface due to the user's misoperation, thereby reducing the terminal's misoperation rate, improving the terminal's processing efficiency, and avoiding errors
  • the waste of traffic generated by operations improves the conversion rate of advertisements and operations, makes advertising billing and sharing more accurate, and improves the user experience. ; Further, by continuously updating the sample data of historical operations, the operation recognition model can be updated in real time to adapt to the real-time update of webpage content and user data.
  • the device of this embodiment includes a bomb environment detection module 801, an operation recognition module 802, and an operation response module 803.
  • the environment detection module 801 is configured to read the values of multiple operation environment factors based on the user operation detected by the terminal interface.
  • the operation recognition module 802 is configured to input values of a plurality of operation environment factors into a pre-trained operation recognition model to obtain a probability value that a user operation is a misoperation.
  • the operation recognition model includes weight coefficients corresponding to the plurality of operation environment factors.
  • the operation response module 803 is configured to cause the terminal interface to trigger a corresponding subsequent response according to the numerical range that the probability value falls within.
  • the environment detection module 801 can also collect sample data for training the operation recognition model, the sample data includes multiple historical operations of the detected user, historical values of operating environment factors corresponding to the multiple historical operations, and Misoperation attributes corresponding to multiple historical operations. Further, a weight coefficient corresponding to multiple operating environment factors may be calculated based on the sample data to apply the weight coefficient to the operation recognition module 802.
  • the operation response module 803 may calculate the return operation time value according to the historical operation.
  • the return operation time value represents the time elapsed from the detection of the operation triggered by the user terminal interface switch to the detection of the return operation;
  • a prompt interface for receiving user feedback is generated; and based on the user feedback received on the prompt interface, the misoperation attribute of the historical operation is determined.
  • the operation recognition module 802 may be further configured to determine whether the historical operation meets the following conditions: the historical operation includes the first operation, the second operation, and the third operation in sequence; the first page and the second page are displayed in sequence on the terminal interface And the first page; the first operation is detected when the in-situ replacement of the display element occurs on the first page, the terminal interface switches to the second page in response to the first operation, the second operation is detected on the second page, the terminal interface In response to the second operation returning to the first page, a third operation for the display element before in-situ replacement on the first page is detected; and when the historical operation meets the above conditions, it is determined that the first operation is a misoperation to obtain the Misoperation attributes of historical operations.
  • the weighting coefficient of the operation recognition module 802 may be obtained by first performing cluster analysis on multiple operating environment factors according to the degree of influence caused by misoperation; linearly fitting each operating environment factor to calculate the operating environment The probability of misoperation caused by factors; and based on the back propagation BP neural network, the probability of misoperation caused by multiple operating environment factors is used as the input for forward calculation, and the reverse correction is made according to the attribute of misoperation, and multiple operating environments are obtained through iteration The weighting factor corresponding to the factor.
  • the operation recognition module 802 is also provided with a probability of misoperation caused by each operating environment factor.
  • the operation recognition module 802 may first input the values of multiple operating environment factors into a linear function to obtain the probability of misoperation caused by each operating environment factor; and based on each The probability of misoperation caused by the operating environment factor and the weighted sum of the weight coefficients are used to obtain the probability value that the user operation belongs to misoperation.
  • the operation response module 803 may be set to confirm that the user operation is a misoperation when the probability value falls within the first numerical range, and cause the terminal interface to stop responding to the user operation; when the probability value falls within the second numerical range , Make the terminal interface generate a prompt for receiving user feedback, and confirm whether to continue to respond to the user operation based on the received user feedback; and when the probability value falls within the third numerical range, confirm that the user operation is not a misuse, and make The terminal interface continues to respond to user operations.
  • the operation processing device of this embodiment by collecting a plurality of operation environment factors during the operation of the user, and using a pre-trained operation to identify the probability value of the model misoperation. Triggering the subsequent response of the terminal interface according to the probability value can improve the accuracy of identifying whether the user operation is a misoperation, prevent entering the other interface due to the user's misoperation, thereby reducing the terminal's misoperation rate, improving the terminal's processing efficiency, and avoiding errors
  • the waste of traffic generated by operations improves the conversion rate of advertisements and operations, makes advertising billing and sharing more accurate, and improves the user experience. ; Further, by continuously updating the sample data of historical operations, the operation recognition model can be updated in real time to adapt to the real-time update of webpage content and user data.
  • This embodiment provides an operation processing apparatus based on the same inventive concept as the above method embodiment. This embodiment can be used to implement the operation processing method provided in the above embodiment.
  • modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory.
  • the features and functions of the two or more modules or units described above may be embodied in one module or unit.
  • the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.
  • the components displayed as modules or units may or may not be physical units, that is, they may be located in one place, or may be distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the wooden disclosure scheme. Those of ordinary skill in the art can understand and implement without paying creative labor.
  • a computer-readable storage medium is also provided, on which a computer program is stored, and when the program is executed by a processor, the steps of the method described in any one of the foregoing embodiments may be implemented.
  • the computer-readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, or the like.
  • a computing device is also provided.
  • the device may be a mobile terminal such as a mobile phone or a tablet computer, or may be a terminal device such as a desktop computer or a server, which is not limited in this example embodiment.
  • 9 shows a schematic diagram of a computing device 90 according to an example embodiment of the present disclosure.
  • the device 90 may be provided as a mobile terminal.
  • the device 90 includes a processing component 91, which further includes one or more processors, and memory resources represented by the memory 92 for storing instructions executable by the processing component 91, such as application programs.
  • the application programs stored in the memory 92 may include one or more modules each corresponding to a set of instructions.
  • the processing component 91 is configured to execute instructions to perform the above-mentioned operation processing method.
  • the memory can be used to store software programs and modules, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory.
  • the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory.
  • the memory may further include memories remotely provided with respect to the processor, and these remote memories may be connected to a computer terminal or a mobile terminal through a network. Examples of the above network include but are not limited to the Internet, intranet, local area network, mobile communication network, and combinations thereof.
  • the device 90 may also include a power component 93 configured to perform power management of the device 90, a wired or wireless network interface 94 configured to connect the device 90 to the network, and an input/output (I/O) interface 95.
  • the device 90 may operate based on an operating system stored in the memory 92, such as Android, IOS, or the like.

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • User Interface Of Digital Computer (AREA)

Abstract

一种操作处理方法、装置、电子设备及计算机可读介质。该方法包括：基于终端界面检测到的用户操作，读取多个操作环境因子的取值（S110）；将所述多个操作环境因子的取值输入预先训练的操作识别模型，得到所述用户操作属于误操作的概率值（S120），所述操作识别模型包括与所述多个操作环境因子对应的权重系数；以及根据所述概率值落入的数值范围，使所述终端界面触发相应的后续响应（S130）。

Description

计算设备的防误操作处理 技术领域
本公开涉及互联网技术领域，具体而言，涉及一种操作处理方法和装置以及计算设备。
背景技术
目前移动终端广泛应用于人们的生活工作，人们在使用移动终端时，通常通过触摸屏幕或鼠标点击进行对其的操作。据统计，高达50％的移动广告点击是不慎点击的，而除了移动广告，移动触屏应用中其他类型的功能和页面也存在大量非主观意愿的失误点击，例如，用户在操作时注意力并不在引导元素上，只是误触将导致误点；又例如，新内容瞬间替换旧内容时用户无足够时间判断，意图点击旧元素，却点到了新出现的元素，造成的误点。
在所述背景技术部分公开的上述信息仅用于加强对本公开的背景的理解，因此它可以包括不构成对本领域普通技术人员已知的现有技术的信息。
发明内容
有鉴于此，本公开的目的是提供一种操作处理方法和装置以及计算设备，进而至少在一定程度上克服由于相关技术的限制和缺陷而导致的一个或者多个问题。
本公开的其他特性和优点将通过下面的详细描述变得显然，或部分地通过本公开的实践而习得。
根据本公开实施例的第一方面，提出一种操作处理方法，该方法包括：基于终端界面检测到的用户操作，读取多个操作环境因子的取值；将所述多个操作环境因子的取值输入预先训练的操作识别模型，得到所述用户操作属于误操作的概率值，所述操作识别模型包括与所述多个操作环境因子对应的权重系数；以及根据所述概率值落入的数值范围，使所述终端界面触发相应的后续响应。
根据本公开实施例的第二方面，提供一种操作处理装置，该装置包括：环境检测模块，设置为基于终端界面检测到的用户操作，读取多个操作环境因子的取值；操作识别模块，设置为将所述多个操作环境因子的取值输入预先训练的操作识别模型，得到所述 用户操作属于误操作的概率值，所述操作识别模型包括与所述多个操作环境因子对应的权重系数；以及操作响应模块，设置为根据所述概率值落入的数值范围，使所述终端界面触发相应的后续响应。
根据本公开实施例的第三方面，提供一种存储有计算机程序的存储介质，所述计算机程序在由计算机的处理器运行时，使所述计算机执行如以上第一方面所述的方法。
根据本公开实施例的第四方面，提供一种计算设备，包括：处理器；存储器，存储有可由所述处理器执行的指令；其中所述处理器被配置为执行如以上第一方面所述的方法。
根据本公开实施例提供的技术方案，基于用户历史操作建立操作识别模型，以实时输出用户误操作概率，并据此控制后续响应，能够提高识别准确率，精准分类以提供更准确的数据，节省流量浪费并提升用户体验。
应当理解的是，以上的一般描述和后文的细节描述仅是示例性和解释性的，并不能限制本公开。
附图说明
通过参照附图详细描述其示例实施例，本公开的上述和其它目标、特征及优点将变得更加显而易见。下面描述的附图仅仅是本公开的一些实施例，对于本领域的普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是根据一示例性实施例示出的一种操作处理方法流程图。
图2是根据一示例性实施例示出的一种操作处理方法流程图。
图3是根据一示例性实施例示出的用于接收用户反馈的界面示意图。
图4是图2所示实施例中步骤S220的示例性流程图。
图5是根据一示例性实施例示出的一种操作处理方法流程图。
图6是根据另一示例性实施例示出的一种操作处理方法流程图。
图7是根据另一示例性实施例示出的一种操作处理的计算过程示意图。
图8是根据本公开一实施例的操作处理装置结构框图。
图9是根据本公开一实施例的计算设备的示意图。
具体实施方式
现在将参考附图更全面地描述示例实施例。然而，示例实施例能够以多种形式实施，且不应被理解为限于在此阐述的实施例；相反，提供这些实施例使得本公开将全面和完整，并将示例实施例的构思全面地传达给本领域的技术人员。在图中相同的附图标记表示相同或类似的部分，因而将省略对它们的重复描述。
此外，所描述的特征、结构或特性可以以任何合适的方式结合在一个或更多实施例中。在下面的描述中，提供许多具体细节从而给出对本公开的实施例的充分理解。然而，本领域技术人员将意识到，可以实践本公开的技术方案而没有特定细节中的一个或更多，或者可以采用其它的方法、组元、装置、步骤等。在其它情况下，不详细示出或描述公知方法、装置、实现或者操作以避免模糊本公开的各方面。
附图中所示的方框图仅仅是功能实体，不一定必须与物理上独立的实体相对应。即，可以采用软件形式来实现这些功能实体，或在一个或多个硬件模块或集成电路中实现这些功能实体，或在不同网络和/或处理器装置和/或微控制器装置中实现这些功能实体。
附图中所示的流程图仅是示例性说明，不是必须包括所有的内容和操作/步骤，也不是必须按所描述的顺序执行。例如，有的操作/步骤还可以分解，而有的操作/步骤可以合并或部分合并，因此实际执行的顺序有可能根据实际情况改变。
应理解，虽然本文中可能使用术语第一、第二、第三等来描述各种组件，但这些组件不应受这些术语限制。这些术语乃用以区分一组件与另一组件。因此，下文论述的第一组件可称为第二组件而不偏离本公开概念的教示。如本文中所使用，术语“及/或”包括相关联的列出项目中的任一个及一或多者的所有组合。
本领域技术人员可以理解，附图只是示例实施例的示意图，附图中的模块或流程并不一定是实施本公开所必须的，因此不能用于限制本公开的保护范围。
在一种基于压力感应技术的广告误点击防止方法和系统中，接收广告条上的用户手指压力大小并判断其是否达到设定值；其中，当压力大于第一设定值时，触发预设的第一事件，该第一事件为进入到广告条所链接的广告界面或者向用户展示全屏广告界面。通过该方法，可以相对精确地模拟用户点击广告的真实意思表示，且相对精准地防止各种误碰，从而给用户更好的用户体验。
然而，在该方法中，广告误点击的判断依据是单一用户手指压力大小，而且在将手指压力大小和设定值进行对比的环节中，没有考虑不同用户群体、客观环境的多样性， 因此可能会存在部分误点操作判断不准确。此外，也无法解决新旧内容替换造成误点的问题。
因此，需要一种新的操作处理方法和装置以及计算设备。
图1是根据一示例性实施例示出的一种操作处理方法流程图。如图1所示，本实施例的方法包括以下步骤S110-S130。在一个实施例中，本实施例的方法可在用户终端(例如手机、平板电脑)来执行。
在步骤S110中，基于在终端界面检测到的用户操作，读取多个操作环境因子的取值。
用户操作可例如包括但不限于点击广告、游戏链接、商品链接等等。操作环境因子是在用户对终端界面进行操作时，通过用户终端收集的环境数据。操作环境因子可例如包括但不限于展示布局类、硬件环境类、用户信息类以及引导元素内容类。展示布局类指操作对象在终端界面中展示时涉及的参数，展示布局类可例如包括但不限于用户操作对象的可见百分比(操作对象可见区域占操作页面面积大小的百分比)、用户操作位置(例如用户操作位置坐标与操作对象中心点的距离)、操作对象显示持续时间(操作对象加载显示到操作发生的时间间隔)、用户界面静止持续时间(可滑动列表滚动停顿到用户操作发生的时间间隔)、“内容原位替代”的展示时间(新页面内容加载完成，替换掉原有内容时，到用户操作发生的时间间隔)等。硬件环境类指当前终端中与硬件相关的参数，其可例如包括但不限于CPU(Central Processing Unit)占用率、可用内存、卡顿持续时间等。用户信息类指与当前用户相关的可影响其操作偏好的参数，其可例如包括但不限于用户的年龄、性别、工作类别、学历等。引导元素内容类指当前操作对象的相关参数，其可例如包括但不限于操作对象的专题、性质等等。
在步骤S120中，将多个操作环境因子的取值输入预先训练的操作识别模型，得到用户操作属于误操作的概率值，操作识别模型包括与多个操作环境因子对应的权重系数。
操作识别模型可用于计算用户操作属于误操作的概率值。在一个实施例中，操作识别模型还可包括每个操作环境因子造成误操作的概率与该操作环境因子之间的线性函数。在一个实施例中，当将操作环境因子输入操作识别模型时，可首先根据线性函数得到每一操作环境因子造成误操作的概率，并通过将每一操作环境因子造成误操作的概率与其对应的权重系数进行加权求和，该和值即为用户操作属于误操作的概率值。
在步骤S130中，根据概率值落入的数值范围，使终端界面触发相应的后续响应。
在一个实施例中，可根据概率值落入的数值范围确定当前用户操作的误操作属性。相应的后续响应可根据误操作属性进行区分。在一个实施例中，在概率值落入第一数值范围时，确认用户操作属于误操作，并使终端界面停止对用户操作的响应。在一个实施例中，在概率值落入第二数值范围时，使终端界面生成用于接收用户反馈的提示，并基于接收的用户反馈确认是否继续对用户操作的响应。在一个实施例中，在概率值落入第三数值范围时，确认用户操作不属于误操作，并使终端界面继续对用户操作的响应。
在一个实施例中，还可将当前用户操作的多个操作环境因子及其误操作属性作为操作识别模型的训练数据，以实时更新操作识别模型，以提高其识别准确度。
图2是根据一示例性实施例示出的一种操作处理方法流程图，如图2所示，本实施例的方法包括以下步骤S210-S250。
在步骤S210中，采集用于训练操作识别模型的样本数据，样本数据包括检测到的用户的多个历史操作、对应于多个历史操作的操作环境因子历史取值、和对应于多个历史操作的误操作属性。
如前所述，可实时获取步骤S130的结果，以更新用于训练操作识别模型的样本数据。
误操作属性可说明其对应的用户操作是否属于误操作。在一个实施例中，误操作属性的获取方法可例如：根据历史操作计算返回操作时间值，返回操作时间值表示从检测到用户触发终端界面切换的操作至检测到返回操作所经过的时间；在返回操作时间值小于预设时间值时，生成用于接收用户反馈的提示界面；以及基于在提示界面接收的用户反馈，确定历史操作的误操作属性。用于接收用户反馈的提示界面可例如图3所示。用户触发终端界面切换的操作例如为其点击某一链接的操作，当该操作引发跳转后，又检测到返回操作，以返回用户触发终端界面切换的操作时的被点击页面，上述两次操作之间的间隔时间即为返回操作时间值。预设时间的取值可根据实际统计数据或机理数据推算获得，本公开的实施例对此并无限制。
在一个实施例中，误操作属性的获取方法可例如判断历史操作是否符合以下条件：历史操作包括依次进行的第一操作、第二操作和第三操作；在终端界面依次显示第一页面、第二页面和第一页面；在第一页面上发生显示元素的原位替换时检测到第一操作，终端界面响应于第一操作切换为第二页面，在第二页面上检测到第二操作，终端界面响应于第二操作返回第一页面，检测到针对第一页面上发生原位替换前的显示元素的第三 操作；以及在历史操作符合以上条件时，确定第一操作属于误操作。若第一操作属于误操作，则可确定该历史操作属于误操作。这里的原位替换指页面中，某一展示区域的原有展示内容被替换为新的展示内容。原位替换可例如但不限于发生在广告的滚动展示过程中或者不同广告的轮流展示过程中。第一页面为终端界面，其中包含有用户意图点击的显示元素，例如广告页面；第一操作为用户在发生原位替换时点击原位替换发生区域的广告页面的操作，即用户打开的广告并非第一页面中原有展示的广告；第二页面为响应于第一操作的跳转页面，即原位替换后新的广告页面；第二操作可为用户的返回操作，用于返回第一页面。第三操作可例如为用户在第一页面中发生原位替换前的显示元素上进行的点击操作，即对其意图点击的广告页面进行的点击操作。
在步骤S220中，基于样本数据计算得到与多个操作环境因子对应的权重系数。
权重系数包含于操作识别模型中，其计算方法可例如使用神经网络对样本数据进行学习，样本数据中的多个操作环境因子作为神经网络的输入，误操作概率作为期望输出，经过迭代训练，以获取多个操作环境因子对应的权重系数。
在步骤S230中，基于在终端界面检测到的用户操作，读取多个操作环境因子的取值。
在步骤S240中，将多个操作环境因子的取值输入预先训练的操作识别模型，得到用户操作属于误操作的概率值，操作识别模型包括与多个操作环境因子对应的权重系数。
在步骤S250中，根据概率值落入的数值范围，使终端界面触发相应的后续响应。
步骤S230-S250已在图1中进行了详细解释，此处不再赘述。
在一个实施例中，步骤S220可实施为图4所示的示例性流程，包括以下步骤S410-S430。
在步骤S410中，按照造成误操作的影响程度，对多个操作环境因子进行聚类分析。
聚类为一种将物理或抽象对象的集合分成由类似的对象组成的多个类的过程。在本实施例中，使用聚类算法可根据多个操作环境因子的取值对样本进行分类。
在步骤S420中，对每个操作环境因子进行线性拟合，计算该操作环境因子造成误操作的概率。
线性拟合可刻画每一操作环境因子与该操作环境因子造成误操作的概率之间的线性函数。通过线性拟合得到的线性函数，可通过当前操作环境因子的取值确定其造成误操 作的概率。
在步骤S430中，基于反向传播BP神经网络，以多个操作环境因子造成误操作的概率作为输入进行正向计算，根据误操作属性进行反向修正，通过迭代得到多个操作环境因子对应的权重系数。
BP(back propagation)神经网络是一种按照误差逆向传播算法训练的多层前馈神经网络。BP神经网络算法包括信号的前向传播和误差的反向传播两个过程。权重系数可例如由BP神经网络中输入神经元与隐含层神经元之间的权重值表示。
图5是根据一示例性实施例示出的一种操作处理方法流程图。如图所示，本实施例的方法包括以下步骤S510-S520。
在步骤S510中，将多个操作环境因子的取值分别输入线性函数，得到每个操作环境因子造成误操作的概率。
在一个实施例中，线性函数可例如通过步骤S420获取。
在步骤S520中，基于每个操作环境因子造成误操作的概率与权重系数的加权求和，得到用户操作属于误操作的概率值。
在一个实施例中，假设有n个操作环境因子，其造成误操作概率分别表示为P 1，P 2，...，P n，其对应的权重系数分别表示为W 1，W 2，...，W n，则计算用户操作属于误操作的概率值的公式可表示为：
Figure PCTCN2019122406-appb-000001
图6是根据另一示例性实施例示出的一种操作处理方法流程图。如图所示，本实施例的方法包括以下步骤S610-S630。
在步骤S610中，采集样本数据。
样本数据可例如通过收集并提取用户的多个历史操作来获取。样本数据可例如包括用户信息类数据、展示布局类数据、硬件环境类数据以及引导元素内容类数据。
在一个实施例中，还可收集用户的误点反应时间值，即点击引导因素后，在页面发生跳转后的N秒点击返回按钮以回到原页面。将N秒作为误点反应时间值。对误点反应时间值可例如用于判断当前用户操作的误操作属性。误操作属性的判断方法在步骤S210的实施例中已有详细介绍，在此不再赘述。
在步骤S620中，计算多个环境因子的权重系数，以获得操作识别模型。其中，环境因子的权重系数可通过BP神经网络进行计算。将多个操作环境因子造成误操作的概率作为输入，根据误操作属性对BP神经网络的权重系数进行反向修正，通过对样本数据的学习，可得到一组输入层对隐含层的权重值，即本步骤希望得到的权重系数。
在步骤S630中，将操作识别模型下发至用户终端，以对用户操作进行实时判断。
操作识别模型的输入为操作环境因子造成误操作的概率，通过将其与权重系数进行加权求和，可得到用户操作属于误操作的概率值。误操作识别过程可例如图7所示。
进一步地，可通过概率值选择不同的后续响应。例如设定一阈值0.5，在概率值大于0.5时，认为用户操作属于误操作，可选择终止对用户操作的响应；在概率值小于0.5时，认为用户操作不属于误操作，可选择继续响应当前用户操作。进一步地，可设定一中间数值范围，当概率值落入该取值范围时，认为用户操作的误操作属性不太确定，可例如生成用于接收用户反馈的提示界面，界面可例如图3所示。
根据本实施例的操作处理方法，通过收集用户操作过程中的多个操作环境因子，并使用预先训练的操作识别模型误操作的概率值。以根据概率值触发终端界面的后续响应，可以提高识别用户操作是否为误操作的准确率，防止因用户误操作而进入其他界面，从而降低终端的误操作率，提高终端的处理效率，避免误操作产生的流量浪费，提升广告、运营转化率，使广告计费和分成更精准，同时改善用户体验。；进一步地，通过不断更新历史操作的样本数据，可实时更新操作识别模型，以适应网页内容、用户数据的实时更新。
下述为本公开装置实施例，可以用于执行本公开方法实施例。对于本公开装置实施例中未披露的细节，请参照本公开方法实施例。
图8是根据本公开一实施例的操作处理装置结构框图，如图所示，本实施例的装置包括弹环境检测模块801、操作识别模块802和操作响应模块803。
环境检测模块801设置为基于终端界面检测到的用户操作，读取多个操作环境因子的取值。
操作识别模块802设置为将多个操作环境因子的取值输入预先训练的操作识别模型，得到用户操作属于误操作的概率值，操作识别模型包括与多个操作环境因子对应的权重系数。
操作响应模块803设置为根据概率值落入的数值范围，使终端界面触发相应的后续 响应。
可选地，还可通过环境检测模块801采集用于训练操作识别模型的样本数据，样本数据包括检测到的用户的多个历史操作、对应于多个历史操作的操作环境因子历史取值、和对应于多个历史操作的误操作属性。进一步地，可基于样本数据计算得到与多个操作环境因子对应的权重系数，以将权重系数应用于操作识别模块802。
可选地，操作响应模块803可根据历史操作计算返回操作时间值，返回操作时间值表示从检测到用户触发终端界面切换的操作至检测到返回操作所经过的时间；在返回操作时间值小于预设时间值时，生成用于接收用户反馈的提示界面；以及基于在提示界面接收的用户反馈，确定历史操作的误操作属性。
可选地，操作识别模块802还可设置为判断历史操作是否符合以下条件：历史操作包括依次进行的第一操作、第二操作和第三操作；在终端界面依次显示第一页面、第二页面和第一页面；在第一页面上发生显示元素的原位替换时检测到第一操作，终端界面响应于第一操作切换为第二页面，在第二页面上检测到第二操作，终端界面响应于第二操作返回第一页面，检测到针对第一页面上发生原位替换前的显示元素的第三操作；以及在历史操作符合以上条件时，确定第一操作属于误操作，以得到该历史操作的误操作属性。
可选地，操作识别模块802的权重系数的获取可例如首先按照造成误操作的影响程度，对多个操作环境因子进行聚类分析；对每个操作环境因子进行线性拟合，计算该操作环境因子造成误操作的概率；以及基于反向传播BP神经网络，以多个操作环境因子造成误操作的概率作为输入进行正向计算，根据误操作属性进行反向修正，通过迭代得到多个操作环境因子对应的权重系数。
可选地，操作识别模块802还设置有每个操作环境因子造成误操作的概率。在计算每个操作环境因子造成误操作的概率时，操作识别模块802可首先将多个操作环境因子的取值分别输入线性函数，得到每个操作环境因子造成误操作的概率；以及基于每个操作环境因子造成误操作的概率与权重系数的加权求和，得到用户操作属于误操作的概率值。
可选地，操作响应模块803可设置为在概率值落入第一数值范围时，确认用户操作属于误操作，并使终端界面停止对用户操作的响应；在概率值落入第二数值范围时，使终端界面生成用于接收用户反馈的提示，并基于接收的用户反馈确认是否继续对用户操 作的响应；以及在概率值落入第三数值范围时，确认用户操作不属于误操作，并使终端界面继续对用户操作的响应。
根据本实施例的操作处理装置，通过收集用户操作过程中的多个操作环境因子，并使用预先训练的操作识别模型误操作的概率值。以根据概率值触发终端界面的后续响应，可以提高识别用户操作是否为误操作的准确率，防止因用户误操作而进入其他界面，从而降低终端的误操作率，提高终端的处理效率，避免误操作产生的流量浪费，提升广告、运营转化率，使广告计费和分成更精准，同时改善用户体验。；进一步地，通过不断更新历史操作的样本数据，可实时更新操作识别模型，以适应网页内容、用户数据的实时更新。
本实施例基于与上述方法实施例同样的发明构思，提供了一种操作处理装置，本实施例能够用于实现上述实施例中提供的在操作处理方法。
应当注意，尽管在上文详细描述中提及了用于动作执行的设备的若干模块或者单元，但是这种划分并非强制性的。实际上，根据本公开的实施方式，上文描述的两个或更多模块或者单元的特征和功能可以在一个模块或者单元中具体化。反之，上文描述的一个模块或者单元的特征和功能可以进一步划分为由多个模块或者单元来具体化。作为模块或单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现木公开方案的目的。本领域普通技术人员在不付出创造性劳动的情况下，即可以理解并实施。
通过以上实施方式的描述，本领域的技术人员易于理解，上文描述的示例实施方式可以通过软件实现，也可以通过软件结合必要的硬件的方式来实现。
例如，在一个示例实施方式中，还提供一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时可以实现上述任意一个实施例中所述方法的步骤。所述方法的具体步骤可参考前述实施例中的详细描述，此处不再赘述。所述计算机可读存储介质可以是ROM、随机存取存储器(RAM)、CD-ROM、磁带、软盘和光数据存储设备等。
在另一个示例实施方式中，还提供一种计算设备，该设备可以是手机、平板电脑等移动终端，也可以是台式计算机、服务器等终端设备，本示例实施方式中对此不作限制。图9示出根据本公开示例实施方式中一种计算设备90的示意图。例如，设备90可以被提供为一移动终端。参照图9，设备90包括处理组件91，其进一步包括一个或多个处 理器，以及由存储器92所代表的存储器资源，用于存储可由处理组件91的执行的指令，例如应用程序。存储器92中存储的应用程序可以包括一个或一个以上的每一个对应于一组指令的模块。此外，处理组件91被配置为执行指令，以执行上述操作处理方法。
其中，存储器可用于存储软件程序以及模块，处理器通过运行存储在存储器内的软件程序以及模块，从而执行各种功能应用以及数据处理。存储器可包括高速随机存储器，还可以包括非易失性存储器，如一个或者多个磁性存储装置、闪存、或者其他非易失性固态存储器。在一些实例中，存储器可进一步包括相对于处理器远程设置的存储器，这些远程存储器可以通过网络连接至计算机终端或移动终端。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
装置90还可以包括一个电源组件93被配置为执行装置90的电源管理，一个有线或无线网络接口94被配置为将装置90连接到网络，和一个输入输出(I/O)接口95。装置90可以操作基于存储在存储器92的操作系统，例如Android、IOS或类似。
本领域技术人员在考虑说明书及实践这里公开的发明后，将容易想到本公开的其它实施方案。本申请旨在涵盖本公开的任何变型、用途或者适应性变化，这些变型、用途或者适应性变化遵循本公开的一般性原理并包括本公开未公开的本技术领域中的公知常识或惯用技术手段。说明书和实施例仅被视为示例性的，本公开的真正范围和精神由所附的权利要求指出。
虽然已参照几个典型实施例描述了本公开，但应当理解，所用的术语是说明和示例性、而非限制性的术语。由于本公开能够以多种形式具体实施而不脱离申请的精神或实质，所以应当理解，上述实施例不限于任何前述的细节，而应在随附权利要求所限定的精神和范围内广泛地解释，因此落入权利要求或其等效范围内的全部变化和改型都应为随附权利要求所涵盖。

Claims (10)

  1. 一种操作处理方法，包括：
    基于在终端界面检测到的用户操作，读取多个操作环境因子的取值；
    将所述多个操作环境因子的取值输入预先训练的操作识别模型，得到所述用户操作属于误操作的概率值，所述操作识别模型包括与所述多个操作环境因子对应的权重系数；以及
    根据所述概率值落入的数值范围，使所述终端界面触发相应的后续响应。
  2. 如权利要求1所述的方法，还包括：
    采集用于训练所述操作识别模型的样本数据，所述样本数据包括检测到的所述用户的多个历史操作、对应于所述多个历史操作的操作环境因子历史取值、和对应于所述多个历史操作的误操作属性；以及
    基于所述样本数据计算得到与所述多个操作环境因子对应的所述权重系数。
  3. 如权利要求2所述的方法，其特征在于，采集用于训练所述操作识别模型的样本数据，包括：根据所述历史操作计算返回操作时间值，所述返回操作时间值表示从检测到用户触发终端界面切换的操作至检测到返回操作所经过的时间；
    在所述返回操作时间值小于预设时间值时，生成用于接收用户反馈的提示界面；以及
    基于在所述提示界面接收的用户反馈，确定所述历史操作的误操作属性。
  4. 如权利要求2所述的方法，其特征在于，采集用于训练所述操作识别模型的样本数据，包括：
    判断所述历史操作是否符合以下条件：所述历史操作包括依次进行的第一操作、第二操作和第三操作；在所述终端界面依次显示第一页面、第二页面和所述第一页面；在所述第一页面上发生显示元素的原位替换时检测到所述第一操作，所述终端界面响应于所述第一操作切换为所述第二页面，在所述第二页面上检测到所述第二操作，所述终端界面响应于所述第二操作返回所述第一页面，检测到针对所述第一页面上发生原位替换前的显示元素的第三操作；以及
    在所述历史操作符合以上条件时，确定所述第一操作属于误操作，以得到所述历史操作的误操作属性。
  5. 如权利要求2所述的方法，其特征在于，基于所述样本数据计算得到与所述多个操作环境因子对应的所述权重系数，包括：
    按照造成误操作的影响程度，对所述多个操作环境因子进行聚类分析；
    对每个操作环境因子进行线性拟合，计算该操作环境因子造成误操作的概率；以及
    基于反向传播BP神经网络，以所述多个操作环境因子造成误操作的概率作为输入进行正向计算，根据所述误操作属性进行反向修正，通过迭代得到所述多个操作环境因子对应的所述权重系数。
  6. 如权利要求1所述的方法，其特征在于，所述操作识别模型还包括每个操作环境因子造成误操作的概率与该操作环境因子之间的线性函数，
    将所述多个操作环境因子的取值输入预先训练的操作识别模型，得到所述用户操作属于误操作的概率值，包括：
    将所述多个操作环境因子的取值分别输入所述线性函数，得到每个操作环境因子造成误操作的概率；以及
    基于所述每个操作环境因子造成误操作的概率与所述权重系数的加权求和，得到所述用户操作属于误操作的概率值。
  7. 如权利要求1所述的方法，其特征在于，根据所述概率值落入的数值范围，使所述终端界面触发相应的后续响应，包括以下任一步骤：
    在所述概率值落入第一数值范围时，确认所述用户操作属于误操作，并使所述终端界面停止对所述用户操作的响应；
    在所述概率值落入第二数值范围时，使所述终端界面生成用于接收用户反馈的提示，并基于接收的用户反馈确认是否继续对所述用户操作的响应；以及
    在所述概率值落入第三数值范围时，确认所述用户操作不属于误操作，并使所述终端界面继续对所述用户操作的响应。
  8. 一种操作处理装置，其特征在于，包括：
    环境检测模块，设置为基于终端界面检测到的用户操作，读取多个操作环境因子的取值；
    操作识别模块，设置为将所述多个操作环境因子的取值输入预先训练的操作识别模型，得到所述用户操作属于误操作的概率值，所述操作识别模型包括与所述多个操作环境因子对应的权重系数；以及
    操作响应模块，设置为根据所述概率值落入的数值范围，使所述终端界面触发相应的后续响应。
  9. 一种存储有计算机程序的存储介质，所述计算机程序在由计算机的处理器运行时，使所述处理器执行如权利要求1-7中任一项所述的方法。
  10. 一种计算设备，包括：
    处理器；
    存储器，存储有可由所述处理器执行的指令；
    其中所述处理器被配置为执行如权利要求1-7中任一项所述的方法。
PCT/CN2019/122406 2018-12-05 2019-12-02 计算设备的防误操作处理 Ceased WO2020114352A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811482942.2A CN109558032B (zh) 2018-12-05 2018-12-05 操作处理方法、装置以及计算机设备
CN201811482942.2 2018-12-05

Publications (1)

Publication Number Publication Date
WO2020114352A1 true WO2020114352A1 (zh) 2020-06-11

Family

ID=65868988

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/122406 Ceased WO2020114352A1 (zh) 2018-12-05 2019-12-02 计算设备的防误操作处理

Country Status (2)

Country Link
CN (1) CN109558032B (zh)
WO (1) WO2020114352A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116469420A (zh) * 2023-05-05 2023-07-21 中国工商银行股份有限公司 语音情绪识别方法、装置、设备和介质

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109558032B (zh) * 2018-12-05 2020-09-04 北京三快在线科技有限公司 操作处理方法、装置以及计算机设备
CN111913618B (zh) * 2020-07-08 2022-04-19 核动力运行研究所 检测方法及装置
CN112395200B (zh) * 2020-11-25 2024-04-02 电信科学技术第十研究所有限公司 一种软件可靠性操作剖面概率的确定方法及装置
CN113176847A (zh) * 2021-04-07 2021-07-27 上海墨案智能科技有限公司 一种墨水屏防误触的方法及设备
CN114386688B (zh) * 2022-01-07 2022-12-06 逑美科技(上海)有限公司 一种基于多数据融合的用户意图预测方法及系统
CN116028934B (zh) * 2023-02-07 2025-11-21 深圳海云安网络安全技术有限公司 安卓app的多源信息安全处理方法及系统
CN117686757B (zh) * 2023-12-07 2024-06-11 浙江万昌电力设备有限公司 一种户外电力计量箱的智能预警方法及系统

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101645711A (zh) * 2009-08-31 2010-02-10 深圳华为通信技术有限公司 一种键盘纠错方法及装置
CN102314436A (zh) * 2010-06-30 2012-01-11 国际商业机器公司 网页自动调整方法和系统
CN105183538A (zh) * 2014-06-03 2015-12-23 联想(北京)有限公司 一种信息处理方法及电子设备
CN106095295A (zh) * 2016-06-15 2016-11-09 维沃移动通信有限公司 一种基于指纹识别的处理方法和移动终端
CN106203380A (zh) * 2016-07-20 2016-12-07 中国科学院计算技术研究所 超声波手势识别方法及系统
CN108089897A (zh) * 2017-12-21 2018-05-29 广东欧珀移动通信有限公司 应用程序的启动管理方法、装置、存储介质及移动终端
CN109558032A (zh) * 2018-12-05 2019-04-02 北京三快在线科技有限公司 操作处理方法、装置以及计算机设备

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6954744B2 (en) * 2001-08-29 2005-10-11 Honeywell International, Inc. Combinatorial approach for supervised neural network learning
TWI557722B (zh) * 2012-11-15 2016-11-11 緯創資通股份有限公司 語音干擾的濾除方法、系統，與電腦可讀記錄媒體
US9959517B2 (en) * 2014-12-22 2018-05-01 International Business Machines Corporation Self-organizing neural network approach to the automatic layout of business process diagrams
KR102182583B1 (ko) * 2016-05-17 2020-11-24 실리콘 스토리지 테크놀로지 인크 비휘발성 메모리 어레이를 사용하는 딥러닝 신경망 분류기
US11494654B2 (en) * 2016-11-03 2022-11-08 Avanseus Holdings Pte. Ltd. Method for machine failure prediction using memory depth values
CN107358169A (zh) * 2017-06-21 2017-11-17 厦门中控智慧信息技术有限公司 一种人脸表情识别方法及人脸表情识别装置
CN107519641B (zh) * 2017-08-04 2020-12-18 网易（杭州）网络有限公司 控制游戏技能释放的方法、装置、存储介质及移动终端
CN108021292B (zh) * 2017-10-27 2021-01-26 努比亚技术有限公司 光感控件触发控制方法、设备及计算机可读存储介质
CN108021235B (zh) * 2017-12-27 2021-03-09 Oppo广东移动通信有限公司 触发操作的方法、装置、终端及存储介质
CN108710847B (zh) * 2018-05-15 2020-11-27 北京旷视科技有限公司 场景识别方法、装置及电子设备

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101645711A (zh) * 2009-08-31 2010-02-10 深圳华为通信技术有限公司 一种键盘纠错方法及装置
CN102314436A (zh) * 2010-06-30 2012-01-11 国际商业机器公司 网页自动调整方法和系统
CN105183538A (zh) * 2014-06-03 2015-12-23 联想(北京)有限公司 一种信息处理方法及电子设备
CN106095295A (zh) * 2016-06-15 2016-11-09 维沃移动通信有限公司 一种基于指纹识别的处理方法和移动终端
CN106203380A (zh) * 2016-07-20 2016-12-07 中国科学院计算技术研究所 超声波手势识别方法及系统
CN108089897A (zh) * 2017-12-21 2018-05-29 广东欧珀移动通信有限公司 应用程序的启动管理方法、装置、存储介质及移动终端
CN109558032A (zh) * 2018-12-05 2019-04-02 北京三快在线科技有限公司 操作处理方法、装置以及计算机设备

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116469420A (zh) * 2023-05-05 2023-07-21 中国工商银行股份有限公司 语音情绪识别方法、装置、设备和介质

Also Published As

Publication number Publication date
CN109558032B (zh) 2020-09-04
CN109558032A (zh) 2019-04-02

Similar Documents

Publication Publication Date Title
WO2020114352A1 (zh) 计算设备的防误操作处理
US12585989B2 (en) Feature effectiveness assessment method and apparatus, electronic device, and storage medium
US11551239B2 (en) Characterizing and modifying user experience of computing environments based on behavior logs
CN107613022B (zh) 内容推送方法、装置及计算机设备
CN110163647B (zh) 一种数据处理方法及装置
US20140279739A1 (en) Resolving and merging duplicate records using machine learning
US20160357790A1 (en) Resolving and merging duplicate records using machine learning
WO2019153604A1 (zh) 人机识别模型的建立装置、方法及计算机可读存储介质
CN114969561B (zh) 一种基于自监督图学习的会话推荐方法、系统及设备
WO2017190610A1 (zh) 目标用户定向方法、装置和计算机存储介质
US10878451B2 (en) Change point detection in a multi-armed bandit recommendation system
CN110490625A (zh) 用户偏好确定方法及装置、电子设备、存储介质
CN112487283A (zh) 训练模型的方法、装置、电子设备及可读存储介质
CN112950218A (zh) 一种业务风险评估方法、装置、计算机设备和存储介质
CN117194779B (zh) 基于人工智能的营销系统优化方法、装置及设备
CN111382346B (zh) 用于推荐内容的方法及系统
CN109978624A (zh) 信息处理方法、电子设备及计算机可读存储介质
CN114791982B (zh) 对象推荐方法和装置
CN113378067A (zh) 基于用户挖掘的消息推荐方法、设备、介质及程序产品
CN112099889A (zh) 一种信息显示方法、设备、装置及存储介质
CN111737472A (zh) 文本分类模型的更新方法及系统、电子设备及存储介质
CN118927231A (zh) 机器人控制方法、装置和机器人
CN108109052A (zh) 物品推送方法、装置及服务器
CN111159558B (zh) 一种推荐列表的生成方法、装置和电子设备
CN111651643B (zh) 候选内容的处理方法及相关设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19893180

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19893180

Country of ref document: EP

Kind code of ref document: A1