Method, apparatus, and computer-readable storage medium for labeling adjacent key mis-touch information

Abstract

This application belongs to the field of intelligent input technology, specifically relating to a method, apparatus, and computer-readable storage medium for labeling adjacent key mis-touch information. The method includes: acquiring log information generated by a user inputting information via a keyboard; parsing the log information to obtain target information corresponding to the same input cycle, wherein the target information includes typed information and on-screen information; acquiring the information type corresponding to the typed information and determining target rules based on the information type; performing adjacent key mis-touch analysis on the target information according to the target rules, and labeling adjacent key mis-touch information based on the analysis results, so as to train an adjacent key mis-touch recognition model based on the labeled adjacent key mis-touch information. This application can achieve automated labeling of adjacent key mis-touch information, shortening the information labeling time, reducing mis-labeling caused by manual labeling, accelerating the iteration speed of the adjacent key mis-touch recognition model, and improving the model's accuracy and input accuracy.

Description

Technical Field

[0001] This application belongs to the field of intelligent input technology, specifically relating to a method for marking adjacent key mis-touch information, a device for marking adjacent key mis-touch information, a computer-readable storage medium, and an electronic device. Background Technology

[0002] After years of iteration and accumulation, input methods have increasingly higher requirements for users' online experience. From the initial functional requirements to the later user experience, with the resurgence of deep learning, some model algorithms are gradually being used to improve the user's functional experience.

[0003] In actual user input, many people accidentally press adjacent keys, for example, intending to type 'a' but pressing 's' instead. Therefore, to improve the error correction ability of the model for adjacent key mis-pressing, a large number of adjacent key mis-pressing samples are needed for model training. Previously, collecting labeled samples required sampling a portion of the logs, with manual labeling of adjacent key mis-pressing logs based on experience. While this method is feasible for observing model results with a small number of samples, manual labeling has significant shortcomings when facing the large number of samples required for machine learning. Firstly, manual labeling consumes a lot of manpower and time; secondly, it carries the risk of mislabeling, both of which are extremely detrimental to model training. Summary of the Invention

[0004] The purpose of this application is to provide a method, device, computer-readable storage medium, and electronic device for labeling adjacent key mis-touch information. This method overcomes the limitations of related technologies, which require a large amount of manpower and time to manually label adjacent key mis-touch information, hindering the rapid iteration of adjacent key mis-touch recognition models. Furthermore, it carries the risk of mis-labeling due to human negligence, resulting in poor accuracy of the trained adjacent key mis-touch recognition model.

[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0006] According to one aspect of the embodiments of this application, a method for annotating adjacent key mis-touch information is provided. The method includes: acquiring log information generated by a user inputting information via a keyboard; parsing the log information to obtain target information corresponding to the same input cycle, wherein the target information includes typing information and screen display information; acquiring the information type corresponding to the typing information and determining a target rule based on the information type; performing adjacent key mis-touch analysis on the target information according to the target rule, and annotating adjacent key mis-touch information based on the analysis results, so as to train an adjacent key mis-touch recognition model based on the annotated adjacent key mis-touch information.

[0007] According to one aspect of the embodiments of this application, a device for labeling adjacent key mis-touch information is provided. The device includes: a log parsing module configured to acquire log information generated by a user inputting information via a keyboard, and to parse the log information to obtain target information corresponding to the same input cycle, wherein the target information includes input information and on-screen information; a rule determination module configured to acquire the information type corresponding to the input information and determine a target rule based on the information type; and a mis-touch analysis module configured to perform adjacent key mis-touch analysis on the target information according to the target rule, and to label adjacent key mis-touch information based on the analysis results, so as to train an adjacent key mis-touch recognition model based on the labeled adjacent key mis-touch information.

[0008] In some embodiments of this application, based on the above technical solutions, the rule determination module includes: a first rule determination unit, configured to use a first rule as the target rule when the input information is intelligent error correction information; and a second rule determination unit, configured to use a second rule as the target rule when the input information is manual error correction information; wherein the first rule is different from the second rule.

[0009] In some embodiments of this application, the target rule is the first rule; based on the above technical solution, the accidental touch analysis module includes: a character length judgment unit, configured to judge whether the string length of the on-screen information is the same as the string length of the selected information in the intelligent error correction information; and a character information judgment unit, configured to, when the string length of the on-screen information is the same as the string length of the selected information, judge whether the original input information is adjacent key accidental touch information based on the characters contained in the on-screen information and the characters contained in the original input information in the intelligent error correction information.

[0010] In some embodiments of this application, based on the above technical solutions, the character information judgment unit is configured to: compare the characters of each character position in the original input information with the characters of the corresponding character positions in the displayed information; when the i-th character of the original input information is different from the i-th character of the displayed information, determine whether the i-th character of the original input information and the i-th character of the displayed information are adjacent keys according to the adjacent key dictionary table; when it is determined that the i-th character of the original input information and the i-th character of the displayed information are adjacent keys, mark the original input information as adjacent key mis-touch information.

[0011] In some embodiments of this application, the target rule is the second rule; based on the above technical solution, the accidental touch analysis module includes: a string extraction unit configured to extract an initial string, an intermediate string, and a termination string based on the manual error correction information; a judgment unit configured to judge whether the initial string, the intermediate string, and the termination string meet preset conditions; a character matching unit configured to compare the characters at each position in the initial string with the characters at the corresponding positions in the termination string when the initial string, the intermediate string, and the termination string meet the preset conditions; a temporary string construction unit configured to construct a temporary string based on the initial string if the characters at the corresponding positions in the initial string and the termination string are different; and an accidental touch analysis unit configured to perform adjacent key accidental touch analysis based on the temporary string and the termination string, and to mark adjacent key accidental touch information based on the analysis results.

[0012] In some embodiments of this application, based on the above technical solutions, the judging unit is configured to: judge whether the initial string, the intermediate string, and the termination string are within the normal character range; when it is determined that the initial string, the intermediate string, and the termination string are all within the normal character range, judge whether the initial string, the intermediate string, and the termination string satisfy a preset relationship; wherein, the preset relationship includes: the termination string is the same as the screen information, the initial string is different from the termination string, the initial string contains the intermediate string, the termination string contains the intermediate string, and the termination string does not contain the initial string.

[0013] In some embodiments of this application, based on the above technical solutions, the temporary string construction unit is configured to: when the j-th character in the initial string is different from the j-th character in the termination string, remove the j-th character in the initial string to obtain the temporary string; where j is a positive integer.

[0014] In some embodiments of this application, based on the above technical solutions, the accidental touch analysis unit is configured to: compare the temporary string with the termination string; if the termination string starts with the temporary string, then determine that the manual error correction information is not adjacent key accidental touch information.

[0015] In some embodiments of this application, based on the above technical solutions, the accidental touch analysis unit is further configured to: if the termination string does not start with the temporary string, insert the j-th character of the termination string between the (j-1)-th and j-th positions of the initial string to update the temporary string; perform adjacent key accidental touch analysis based on the updated temporary string and the termination string, and mark adjacent key accidental touch information based on the analysis results.

[0016] In some embodiments of this application, based on the above technical solutions, mis-touch analysis is performed according to the updated temporary string and the termination string, and adjacent key mis-touch information is marked according to the analysis results, including: comparing the updated temporary string with the termination string; if the termination string does not start with the updated temporary string, then determining whether the j-th character in the termination string and the j-th character in the initial string are adjacent keys based on the adjacent key dictionary; when it is determined that the j-th character in the termination string and the j-th character in the initial string are adjacent keys, the manual error correction information is marked as adjacent key mis-touch information.

[0017] In some embodiments of this application, based on the above technical solutions, the adjacent key mis-touch information labeling device further includes: an adjacent key dictionary table construction module, configured to, before acquiring the log information generated by the user inputting information through the keyboard, take each key in the keyboard as a target key, and construct an adjacent key dictionary table according to the correspondence between the target key and the key adjacent to the target key.

[0018] In some embodiments of this application, based on the above technical solutions, the log parsing module is configured to: collect information generated during the user's character editing process, character display process, and error correction process through set tracking points, and generate the log information based on the information generated during the character editing process, character display process, and error correction process.

[0019] According to one aspect of the embodiments of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the adjacent key mis-touch information annotation method as described in the above technical solutions.

[0020] According to one aspect of the embodiments of this application, an electronic device is provided, the electronic device comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the adjacent key mis-touch information method as described above by executing the executable instructions.

[0021] According to one aspect of the embodiments of this application, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the adjacent key mis-touch information annotation method as described in the above technical solutions.

[0022] In the technical solution provided in this application embodiment, target information corresponding to the same input cycle is obtained by parsing the log information generated by the user using the keyboard. This target information includes input information and screen display information. Target rules are determined based on the information type of the input information, and mis-touch analysis is performed on the target information according to the target rules. Then, the target information is labeled based on the analysis results. This application can realize automated labeling of adjacent key mis-touch information, shorten the sample labeling time, reduce the mis-labeling caused by manual labeling, thereby accelerating the iteration speed of the adjacent key mis-touch failure model and improving the accuracy of the model.

[0023] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0024] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0025] Figure 1 An exemplary system architecture block diagram illustrating the application of the technical solution of this application is shown schematically.

[0026] Figure 2 The flowchart illustrating the steps of the adjacent key mis-touch information annotation method in this application is shown in the schematic diagram.

[0027] Figure 3 The schematic diagram illustrates the method flow for analyzing target information based on the first rule in this application.

[0028] Figure 4 This illustration schematically shows the process of performing adjacent key mis-touch analysis based on the characters contained in the on-screen information and the characters contained in the original input information in this application.

[0029] Figure 5 The schematic diagram illustrates the method flow for analyzing adjacent key mis-touches of target information according to the second rule in this application.

[0030] Figure 6 The flowchart illustrating the process of determining whether the initial string, intermediate string, and terminating string meet the preset conditions in this application is shown in the illustration.

[0031] Figure 7 The schematic diagram illustrates the process of performing adjacent key mis-touch analysis based on the updated temporary string and the termination string in this application.

[0032] Figure 8 The diagram illustrates a performance comparison between enabling and disabling the adjacent key mis-touch recognition model during input.

[0033] Figure 9 A schematic block diagram of the adjacent key mis-touch information labeling device of this application is shown.

[0034] Figure 10 A schematic diagram of a computer system architecture suitable for implementing the embodiments of this application is shown. Detailed Implementation

[0035] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art.

[0036] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0037] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0038] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0039] Figure 1 An exemplary system architecture block diagram illustrating the application of the technical solution of this application is shown schematically.

[0040] like Figure 1 As shown, the system architecture 100 may include terminal device 110, network 120, and server 130. Terminal device 110 includes a display screen and may include various electronic devices such as smartphones, tablets, laptops, smart TVs, and smart vehicle terminals. Server 130 may be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. Network 120 may be a communication medium of various connection types capable of providing a communication link between terminal device 110 and server 130, such as a wired communication link or a wireless communication link.

[0041] Depending on the implementation requirements, the system architecture in this application embodiment can have any number of terminal devices, networks, and servers. For example, server 130 can be a server group composed of multiple server devices. In addition, the technical solutions provided in this application embodiment can be applied to terminal device 110, or to server 130, or can be implemented jointly by terminal device 110 and server 130. This application does not impose any special limitations on this.

[0042] For example, the user inputs information through the virtual keyboard built in the terminal device 110 or an external keyboard connected to the terminal device 110. Through the set data points, the information generated during the editing character process, the character input process, and the error correction process in the user input process can be collected. Log information can be formed based on the information collected by the data points; by parsing the log, the target information corresponding to the same input cycle can be extracted therefrom. The target information includes keystroke information and input information, where the keystroke information includes intelligent error correction information and manual error correction information. The intelligent error correction information is that when the user makes a mistake during the input process, the adjacent key mis-touch recognition model can perform intelligent error correction and provide a set of error correction candidates for the user to click on the information in the error correction candidate set. The manual error correction information is the information generated when the user actively realizes an input error during the input process and corrects it by means of backspace. The input information is the string corresponding to the information finally displayed on the display screen. For example, if "at home" is finally displayed on the display screen, then the input information is "zaijia"; according to the information type of the keystroke information, the corresponding target rule can be determined, and the target information can be analyzed for adjacent key mis-touch according to the target rule; further, adjacent key mis-touch information annotation can be performed according to the result of the adjacent key mis-touch analysis to obtain an adjacent key mis-touch sample. It should be noted that different information types of the keystroke information correspond to different target rules.

[0043] After obtaining a large amount of adjacent key mis-touch information, it can be used as a training sample, and the adjacent key mis-touch recognition model can be trained with the training sample to obtain a stable adjacent key mis-touch recognition model and apply it to subsequent adjacent key mis-touch recognition tasks. When the user inputs information through the keyboard, it can be timely detected whether there is an adjacent key mis-touch during the input process. If there is an adjacent key mis-touch, the error correction options will be displayed on the display screen according to the learned information for the user to select and use.

[0044] In some embodiments of the present application, the adjacent key mis-touch information annotation device can also be configured in the terminal device 110. After the terminal device 110 collects the log generated by the user using the keyboard to input information, it is parsed to obtain the target information corresponding to the same input cycle; the target rule is determined according to the information type of the keystroke information in the target information, and the target information is analyzed for adjacent key mis-touch according to the target rule, and at the same time, adjacent key mis-touch information annotation is performed based on the analysis result. Further, the adjacent key mis-touch recognition model configured in the terminal device 110 can also be trained according to the annotation information to obtain a stable adjacent key mis-touch recognition model.

[0045] In some embodiments of the present application, the adjacent key mis-touch recognition model installed on the terminal device 110 or the server 130 is a machine learning model for adjacent key mis-touch information annotation based on artificial intelligence technology.

[0046] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0047] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0048] Computer vision (CV) is a science that studies how to enable machines to "see." More specifically, it refers to machine vision, which uses cameras and computers to replace human eyes in recognizing and measuring targets, and then performs image processing to create images more suitable for human observation or transmission to instruments. As a scientific discipline, computer vision studies related theories and technologies, attempting to build artificial intelligence systems capable of extracting information from images or multidimensional data. Computer vision technologies typically include image processing, image recognition, image semantic understanding, image annotation, OCR, video processing, video semantic understanding, video content / behavior recognition, 3D object reconstruction, 3D technology, virtual reality, augmented reality, and simultaneous localization and mapping (SLAM).

[0049] Machine Learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0050] In the related technologies of this application, the training samples for the adjacent key mis-touch recognition model are mainly obtained by sampling logs and manually selecting and labeling adjacent key mis-touch samples. However, this method requires a large investment of manpower and time to meet the data volume requirements for model training, which is not conducive to the rapid iterative testing of the model, and there is a certain risk of mislabeling due to human negligence. If adjacent key mis-touch samples with mislabeling are used for model training, the accuracy of the model will be reduced, thereby reducing the input accuracy, user experience, and user stickiness of the product.

[0051] The following describes in detail the technical solutions provided in this application, including the method for annotating adjacent key accidental touch information, the device for annotating adjacent key accidental touch information, the computer-readable storage medium, and the electronic device, with reference to specific embodiments.

[0052] Figure 2 The flowchart illustrating the steps of a method for labeling accidental adjacent key touches according to one embodiment of this application is shown schematically. This method can be executed by a terminal device or a server, or by both a terminal device and a server. Figure 2 As shown, the method for labeling adjacent key mis-touch information in this application embodiment can mainly include the following steps S210 to S230.

[0053] Step S210: Obtain log information generated by the user inputting information via the keyboard, parse the log information to obtain target information corresponding to the same input cycle, wherein the target information includes typing information and screen display information;

[0054] Step S220: Obtain the information type corresponding to the entered information, and determine the target rule based on the information type;

[0055] Step S230: Perform adjacent key mis-touch analysis on the target information according to the target rules, and label the adjacent key mis-touch information according to the analysis results, so as to train the adjacent key mis-touch recognition model according to the labeled adjacent key mis-touch information.

[0056] In the adjacent key mis-touch information annotation method provided in this application embodiment, target information of the same input cycle is obtained by parsing log information. Then, the corresponding target rule is determined according to the information type of the input information in the target information. Finally, adjacent key mis-touch analysis is performed on the target information based on the target rule, and adjacent key mis-touch information is annotated according to the analysis results. This application can achieve automated annotation of adjacent key mis-touch information, avoiding manual annotation, reducing manpower and time costs, reducing the risk of mis-labeling, and thus shortening the model iteration cycle. On the other hand, it can improve the automatic error correction capability of adjacent key mis-touch during user input, improve input accuracy, and enhance user experience and product user stickiness.

[0057] The following section uses the scenario of a user inputting information using the keyboard on a touchscreen phone as an example to explain in detail the specific implementation steps of the method for marking up accidental touches of adjacent keys.

[0058] In step S210, log information generated by the user inputting information via the keyboard is obtained, and the log information is parsed to obtain target information corresponding to the same input cycle, wherein the target information includes typing information and screen display information.

[0059] In one embodiment of this application, the keyboard is a 26-key keyboard built into a touchscreen phone. When users input information using the keyboard, accidental key presses often occur due to factors such as fast hand speed, small keyboard keys, and small spacing between adjacent keys. This results in the actual typed string differing from the intended string, affecting input efficiency and user experience. In this embodiment, an adjacent key accidental key press recognition model corrects errors during user input, providing candidate correction strings and candidate words for the user to choose from. This directly provides the user with candidate correction strings and words, eliminating the need for backspace and re-entry, significantly improving input efficiency and user experience. To obtain a stable adjacent key accidental key press recognition model, a large number of training samples are necessary. The following section details how adjacent key accidental key press information is labeled in this embodiment.

[0060] In one embodiment of this application, it is first necessary to extract log information generated by the user's keyboard input, and then parse the log information to obtain target information. When extracting log information, information can be collected through one or more tracking points set in the background program. These tracking points can collect information generated during the editing, input, and error correction processes of the input process. Log information can be formed based on the information generated during these processes. After obtaining the log information, it can be parsed to obtain target information corresponding to the same input cycle. This target information includes input information and input information, where input information is the edited string entered and deleted by the user via the keyboard, and input information is the input string.

[0061] In one embodiment of the present application, the obtained log contains multiple input information corresponding to different input cycles. When obtaining the target information corresponding to the same input cycle, the log information is first divided into multiple small logs according to a preset flag. Since the adjacent input information in the small logs is not necessarily the input information with context semantic relevance, it is also necessary to determine the target information corresponding to the same round of input - on - screen in each small log according to the context semantic relationship. The target information is the input information corresponding to the same input cycle. Among them, the preset flag can specifically be the ts field, and the log information between any two ts fields is a small log. When determining the target information corresponding to the same input cycle according to the context semantic relationship, semantic recognition can be performed through an LSTM model, an NLP model, etc., and the information with context semantic relationship is determined as the target information corresponding to the same input cycle. For example, when the user inputs "lyx", candidate words such as "pull it, practice heroes" appear on the display screen, and the user selects "practice heroes", then the corresponding on - screen information is "lianyingxiong". Correspondingly, from the user's input of "lyx" to the on - screen "lianyingxiong" is the same input cycle.

[0062] In step S220, obtain the information type corresponding to the keyed - in information, and determine the target rule according to the information type.

[0063] In one embodiment of the present application, after obtaining the target information, adjacent key mis-touch analysis can be performed on the target information according to the target rules. The key-information in the target information has different classifications according to different input methods. In the embodiments of the present application, the key-information is divided into two categories. The first category is intelligent error-correction information, which can be the information corresponding to the sa field in the log information. The sa field is used to record the user's selection of the correction string for the error-correction candidate. Specifically, the sa field contains the original input information entered by the user, and at the same time records the correction string of the error-correction candidate and the displayed candidate entries, which can be separated by "#". The correction string of the error-correction candidate is the string corresponding to the candidate entry displayed on the display screen. For example, when the user enters "fa sh", the displayed candidate entries are "法师", "发誓", "发射", "发生", etc., then the corresponding correction strings of the error-correction candidates are "fa shi", "fa shi", "fa she", "fa sheng", etc., and the user can select the required entry and the corresponding correction string from the displayed candidate entries. The second category is manual error-correction information, which can be the information corresponding to the ss field in the log information and is used to represent the record of the backspace usage in the pinyin area. When the user realizes that the input is incorrect, the user will correct it by re-entering after backspacing. In the log information, when entering after backspacing, the remaining input string after the backspace operation can be surrounded by two "#", such as womem#wome#women; when directly sending to the screen without entering after backspacing, the strings before and after backspacing are only separated by one "#", such as womenn#women; if there is an operation of clearing the input string, the input string between the two "#" is empty, such as women##nihao. The backspace can be repeatedly used during the input process until it is sent to the screen, and the content recorded by the ss string can include multiple "#" pairs.

[0064] In one embodiment of the present application, the information sent to the screen is the information corresponding to the sc field. The structural examples of the sc field, sa field, and ss field are as follows:

[0065] Example of sc field:

[0066] [0]sc:zaijia在家

[0067] Among them, zaijia represents the string when the user sends it to the screen, and "在家" represents the finally selected text.

[0068] Example of sa field

[0069]

[92] sa:xi'g习惯………xi’h

[0070]

[93] sc:xi'g习惯

[0071] The user typed xi’h, but it was corrected by the adjacent key mis-touch recognition model to xi'g, and finally the text "习惯" and its corresponding pinyin string were displayed on the screen.

[0072] Example of ss field

[0073]

[65] sc:xialai down

[0074]

[66] ss:xiak#xia#xialai……

[0075] The user entered xiak, deleted k by backspace, then continued to enter lai. After editing, the final string was xialai. The text displayed on the screen in sc was "下来", and the corresponding string was xialai, indicating that the user actually wanted to enter l but accidentally entered k.

[0076] In an embodiment of the present application, for different types of typed information, the rules for adjacent key mis-touch recognition are different. When the typed information is intelligent error correction information, the first rule can be used as the target rule. When the typed information is manual error correction information, the second rule can be used as the target rule, and the first rule is different from the second rule.

[0077] In step S230, adjacent key mis-touch analysis is performed on the target information according to the target rule, and adjacent key mis-touch information annotation is performed according to the analysis result, so as to train the adjacent key mis-touch recognition model according to the annotated adjacent key mis-touch information.

[0078] In an embodiment of the present application, after determining the target rule, adjacent key mis-touch analysis can be performed on the target information according to the target rule to determine whether the user has made a mis-touch during the input process. If a mis-touch occurs, the corresponding typed information is marked as adjacent key mis-touch information. The adjacent key mis-touch sample can be composed of the typed information and the corresponding text displayed on the screen. The typed information is the input information of the model, and the text displayed on the screen is the label information, which is used to verify the output result of the adjacent key mis-touch recognition model and adjust the parameters of the adjacent key mis-touch recognition model.

[0079] Next, the methods for adjacent key mis-touch analysis using the first rule and the second rule will be described in detail.

[0080] When the target rule is the first rule, Figure 3 The method flow diagram for mis-touch analysis of the target information according to the first rule is shown, as Figure 3 shown:

[0081] In step S301, it is judged whether the string length of the text displayed on the screen is the same as the string length of the selected information in the intelligent error correction information.

[0082] In one embodiment of this application, if the string length of the displayed information is the same as the string length of the selected information in the intelligent error correction information, it indicates that the error correction candidate field given by the adjacent key mis-touch recognition model has been selected by the user, and the candidate word displayed on the screen given by the adjacent key mis-touch recognition model is what the user actually wants. Furthermore, based on the characters contained in the displayed information and the characters contained in the original input information in the intelligent error correction information, it is possible to further identify whether the user has committed an adjacent key mis-touch during the input process.

[0083] In step S302, when the string length of the displayed information is the same as the string length of the selected information, it is determined whether the typed information is a mis-touched adjacent key information based on the characters contained in the displayed information and the characters contained in the original input information in the intelligent error correction information.

[0084] In one embodiment of this application, when performing adjacent key mis-touch analysis based on the characters contained in the screen information and the characters contained in the original input information, it is necessary to compare the character of each character position in the screen information with the corresponding character position in the original input information.

[0085] Figure 4 The diagram illustrates a process for analyzing adjacent key mis-touches based on characters contained in the on-screen information and characters contained in the original input information. Figure 4 As shown, the method specifically includes the following steps:

[0086] In step S401, the characters in each position of the original input information are compared with the corresponding characters in the previous screen information.

[0087] In one embodiment of this application, if there is a mis-touch of an adjacent key during the input process, there must be a character in the original input information that is different from the character in the previous screen information. Therefore, the character in the i-th position of the original input information can be compared with the character in the i-th position of the previous screen information to determine that there is a different character, where i is a positive integer.

[0088] In step S402, when the i-th character of the original input information is different from the i-th character of the previous screen information, it is determined whether the i-th character of the original input information and the i-th character of the previous screen information are adjacent keys according to the adjacent key dictionary table.

[0089] In one embodiment of this application, when it is determined that the i-th character of the original input information is different from the i-th character of the displayed information, it is determined whether they are adjacent keys according to the adjacent key dictionary. The adjacent key dictionary is a dictionary formed based on the adjacency relationships between keys on a keyboard. Specifically, each key on the keyboard can be used as a target key, and then an adjacent key dictionary is constructed based on the correspondence between the target key and its adjacent keys. The adjacent key dictionary contains not only the correspondence between character keys, but also the correspondence between character keys and function keys, and between function keys. For example, when the target key is "h", its adjacent keys are "y, g, b, j, u"; when the target key is "a", its adjacent keys are "q, w, s, z, ↑"; and when the target key is the "backspace key", its adjacent keys are "l, m, and newline key". After determining that the i-th character of the original input information is different from the i-th character of the displayed information, the i-th character of the original input information can be compared with the adjacent key dictionary. The system matches the characters to be matched in the dictionary. When a target character that matches the i-th character of the original input information is found, the adjacent key information corresponding to that target character in the adjacent key dictionary is obtained. Then, the i-th information of the displayed information is matched with the obtained adjacent key information. If there is adjacent key information that matches the i-th information of the displayed information, it means that the i-th character of the original input information and the i-th character of the displayed information are adjacent keys. If there is no adjacent key information that matches the i-th information of the displayed information, it means that the i-th character of the original input information and the i-th character of the displayed information are not adjacent keys. It is worth noting that the mapping relationship of adjacent keys in the adjacent key dictionary in this application includes, but is not limited to, two immediately adjacent keys. It can also include keys separated from the target key by one or two keys, as well as other numbers of keys, as adjacent keys of the target key, and construct an adjacent key dictionary.

[0090] In step S403, when it is determined that the i-th character of the original input information and the i-th character of the previous screen information are adjacent keys, the original input information is marked as adjacent key mis-touch information.

[0091] In one embodiment of this application, when it is determined that the i-th character of the original input information and the i-th character of the displayed information are adjacent keys, it indicates that the user has accidentally touched the information. Therefore, the original input information can be marked as adjacent key accidental touch information. Based on the original input information and the corresponding displayed information, adjacent key accidental touch samples can be constructed and used for training the adjacent key accidental touch recognition model.

[0092] When the target rule is the second rule. Figure 5 This diagram illustrates a method for analyzing adjacent key mis-touches of target information based on the second rule. Figure 5 As shown:

[0093] In step S501, the initial string, intermediate string, and termination string are extracted based on the manual error correction information.

[0094] In one embodiment of this application, when a user actively realizes an input error, they will correct it by backspace and re-entering. Based on the manual correction information corresponding to the `ss` field extracted from the log information, the initial string during the user's first edit, the intermediate string during the backspace modification, and the final termination string used for input can be obtained. Taking the `ss` field example shown in the above embodiment, the initial string edited by the user is `xiaw`. During the process, the user backspaces and deletes `w`, so the intermediate string is `xia`. Finally, the user adds `lai` for input, so the termination string is `xialai`. By analyzing the initial string, intermediate string, and termination string, it can be determined whether accidental key presses occurred during the input process.

[0095] In step S502, it is determined whether the initial string, the intermediate string, and the terminating string meet the preset conditions.

[0096] In one embodiment of this application, when determining whether the initial string, intermediate string and termination string meet the preset conditions, it is mainly carried out from two aspects: one is whether the composition of each string meets the conditions, and the other is whether the relationship between each string meets the conditions. This ensures that the correct adjacent key mis-touch information is labeled.

[0097] Figure 6 The flowchart illustrates the process of determining whether the initial string, intermediate string, and terminating string meet preset conditions. Figure 6 As shown, in step S601, it is determined whether the initial string, the intermediate string, and the termination string are within the normal character range; in step S602, when it is determined that the initial string, the intermediate string, and the termination string are all within the normal character range, it is determined whether the initial string, the intermediate string, and the termination string satisfy a preset relationship; wherein, the preset relationship includes: the termination string is the same as the screen information, the initial string is different from the termination string, the initial string contains the intermediate string, the termination string contains the intermediate string, and the termination string does not contain the initial string.

[0098] In step S601, the normal character range refers to characters between 'az' and 'a', excluding functional characters corresponding to keyboard function keys, such as uppercase and lowercase letters, separators, etc. In step S602, backspace errors caused by user thinking or re-editing (not accidental touches) are excluded based on the relationship between the initial string, intermediate strings, and termination string. Specifically, when the manually corrected information has situations where the termination string is the same as the initial string, the termination string does not contain intermediate strings, or the termination string contains the initial string, the analysis of adjacent key mis-touches for the target information stops. This is because when the termination string is the same as the initial string, there is no difference before and after editing, indicating that the user intended to input the same character as the initial input, and there is no accidental touch. When the termination string does not contain intermediate strings, such as wo##nimen, the user cleared the entire initial string and re-edited it instead of appending input after the previous intermediate string, indicating that the final information displayed on the screen is different from the initial input, not an accidental touch. When the termination string contains the initial string, such as wo#w#women, it indicates that the user was thinking during input and used backspace due to uncertainty, not an accidental touch.

[0099] In step S503, when the initial string, intermediate string, and termination string meet the preset conditions, the characters at each position in the initial string are compared with the corresponding characters at the termination string.

[0100] In one embodiment of this application, after obtaining the initial string, intermediate string, and termination string that meet preset conditions, further analysis can be performed to determine whether there are any accidental key presses during the user's input process. Specifically, the characters in the initial string can be compared with the characters in the termination string, and adjacent key accidental key press analysis can be performed based on the comparison results.

[0101] In step S504, if the characters corresponding to the same character position in the initial string and the terminating string are different, a temporary string is constructed based on the initial string.

[0102] In one embodiment of this application, when comparing the characters at each position in the initial string with the corresponding characters at each position in the terminating string, if there is a case where the j-th character in the initial string is different from the j-th character in the terminating string, then the j-th character in the initial string is removed to obtain a temporary string, and adjacent key mis-touch analysis is performed based on the temporary string. For example, if the initial string is "won" and the terminating string is "women", and the third character in the initial string is different from the third character in the terminating string, then "n" can be removed from the initial string, and the remaining string "wo" in the initial string can be used as a temporary string for subsequent analysis.

[0103] In step S505, adjacent key mis-touch analysis is performed based on the temporary string and the termination string, and adjacent key mis-touch information is marked based on the analysis results.

[0104] In one embodiment of this application, after determining the temporary string, adjacent key mis-touch analysis can be performed based on the temporary string and the terminating string. In this embodiment, by comparing the temporary string with the terminating string, it is determined whether the terminating string begins with the temporary string. If the terminating string begins with the temporary string, it is determined that the manual correction information is not adjacent key mis-touch information, but rather a backspace caused by the user typing an extra character. For example, the manual correction information is: nity#ni#niyuongyuan. The third character position 't' in the initial string 'nity' is different from the third character position 'y' in the terminating string. After removing the third character position 't', the temporary string 'niy' is obtained. By comparison, the terminating string begins with the temporary string 'niy', indicating that the initial string was caused by the user typing an extra character, resulting in a backspace. This situation does not belong to adjacent key mis-touch, and therefore can be excluded.

[0105] Furthermore, although the above method eliminates backspace caused by extra characters, it still does not eliminate backspace caused by missing characters. Therefore, when the terminating string does not begin with the temporary string, further judgment is needed to eliminate backspace caused by missing characters. In the embodiments of this application, when the character at the j-th position of the initial string is different from the character at the j-th position of the terminating string, the character at the j-th position of the terminating string is inserted between the (j-1)-th and j-th positions of the initial string to update the temporary string. Further adjacent key mis-touch analysis is then performed based on the updated temporary string and the terminating string, and adjacent key mis-touch information is marked based on the analysis results.

[0106] Figure 7 The flowchart illustrates the process of performing adjacent key mis-touch analysis based on the updated string and the terminating string, as shown below. Figure 7 As shown, in step S701, the updated temporary string is compared with the termination string to determine whether the termination string starts with the updated temporary string; in step S702, if the termination string does not start with the updated temporary string, the j-th character in the termination string and the j-th character in the initial string are determined to be adjacent keys based on the adjacent key dictionary; in step S703, when it is determined that the j-th character in the termination string and the j-th character in the initial string are adjacent keys, the manual error correction information is marked as adjacent key mis-touch information.

[0107] pass Figure 7The illustrated adjacent key mis-typing analysis process can eliminate backspace errors caused by users omitting characters, ensuring the accuracy of the final acquired adjacent key mis-typing information. For example, if the manual correction information is nitxs#ni#niytxsgyuan, and the third character 't' of the initial string differs from the third character 'y' of the terminating string, then inserting 'y' between the second and third characters of the initial string yields the updated temporary string niytxs. By comparison, the terminating string begins with the updated temporary string, indicating that the manual correction information was generated due to the user omitting a character and re-entering it via backspace, not due to the user accidentally typing an adjacent key. Furthermore, if the terminating string does not begin with the updated temporary string, it indicates that the manual correction information may have been generated by the user accidentally touching an adjacent key. To determine whether the manual correction information was generated by the user accidentally touching an adjacent key, it is also necessary to determine whether the character at the j-th position of the initial string and the character at the j-th position of the terminating string are adjacent keys based on the adjacent key dictionary. If they are determined to be adjacent keys, it indicates that the manual correction information was generated by the user accidentally touching an adjacent key during input. Therefore, the manual correction information can be labeled as adjacent key accidental touch information, and the manual correction information and the corresponding on-screen information can be used as training samples to train the model.

[0108] The adjacent key mis-touch analysis process described in the above embodiments can eliminate the situation where users backspace due to missed or extra key presses, thus ensuring the accuracy of the final labeled adjacent key mis-touch samples.

[0109] In one embodiment of this application, one method is to remove characters from the initial string that are different from the terminating string to determine whether the backspace is caused by extra characters. Another method is to insert characters from the terminating string that are different from the initial string into the initial string to determine whether the backspace is caused by missing characters. These two methods exist in parallel. In the above embodiment, characters can be removed first to form a temporary string for judging extra characters, and then characters can be inserted to form an updated temporary string for judging missing characters. Alternatively, characters can be inserted first to form a temporary string for judging missing characters, and then characters can be removed to form an updated temporary string for judging extra characters. This embodiment of the application does not specifically limit this method.

[0110] In one embodiment of this application, the adjacent key mis-touch information annotation method described in the above embodiments can identify log information of adjacent key mis-touches from a large number of input logs and annotate it to form adjacent key mis-touch samples. The adjacent key mis-touch identification model can be trained based on these samples. The original input information or manual correction information from the intelligent error correction information containing adjacent key mis-touches is used as input information to obtain the predicted information output by the adjacent key mis-touch identification model. Then, the on-screen information corresponding to the intelligent error correction information or manual error correction information is used as label information. Finally, the loss of the adjacent key mis-touch identification model can be determined based on the predicted information, label information, and a preset loss function. The model parameters are then adjusted inversely based on the loss. Through multiple iterations of training, the optimal parameters of the model can be determined, thereby obtaining a stable adjacent key mis-touch identification model.

[0111] After obtaining a stable adjacent key mis-touch detection model, it can be used to identify adjacent key mis-touches in user input. When an adjacent key mis-touch is detected, a correction string and candidate words are provided for the user to choose from. This improves input efficiency and user experience, and increases user stickiness for the input method product. Compared to input methods that do not use an adjacent key mis-touch detection model, using one can reduce the mis-touch rate. Figure 8 The following graph shows a performance comparison between enabling and disabling the adjacent key mis-touch detection model during input. Figure 8 As shown, when the adjacent key mis-touch recognition model is not enabled, the mis-touch rate (ss mis-touch / sc) in manual error correction mode is 0.037, while when the adjacent key mis-touch recognition model is enabled, the mis-touch rate (ss mis-touch / sc) in manual error correction mode is 0.0357. Meanwhile, when the adjacent key mis-touch recognition model is not enabled, the mis-touch rate (sa mis-touch / sc) in intelligent error correction mode is 0.0343, while when the adjacent key mis-touch recognition model is enabled, the mis-touch rate (sa mis-touch / sc) in intelligent error correction mode is 0.0178. Figure 8 The data shown indicates that the adjacent key mis-touch recognition model trained using adjacent key mis-touch information from the embodiments of this application can effectively reduce the user's mis-touch rate and improve the user's input accuracy and input experience.

[0112] The adjacent key mis-touch information annotation method in this application involves parsing the log information generated by the user's keyboard input to obtain target information corresponding to the same input cycle. This target information includes input information and screen display information. Target rules are determined based on the type of input information, and adjacent key mis-touch analysis is performed on the target information according to these rules. Finally, adjacent key mis-touch information is annotated based on the analysis results. This application achieves automated annotation of adjacent key mis-touch information, shortening sample annotation time from one person's monthly annotation workload to minutes. It also reduces mis-annotation caused by manual annotation, covering 85% of all mis-touches. Furthermore, it improves the quality of training data, accelerates model iteration, and increases model accuracy. After the model goes live, the mis-touch rate is significantly reduced by 38%, thereby improving the user's input accuracy.

[0113] It should be noted that although the steps of the method in this application are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0114] The following describes an embodiment of the apparatus of this application, which can be used to execute the adjacent key mis-touch information annotation method in the above embodiments of this application. Figure 9 A schematic block diagram of the adjacent key mis-touch information labeling device provided in an embodiment of this application is shown. Figure 9 As shown, the adjacent key accidental touch information labeling device 900 includes: a log parsing module 910, a rule determination module 920, and an accidental touch analysis module 930, specifically:

[0115] The log parsing module 910 is configured to acquire log information generated by the user inputting information via the keyboard, parse the log information to obtain target information corresponding to the same input cycle, wherein the target information includes input information and on-screen information; the rule determination module 920 is configured to acquire the information type corresponding to the input information, and determine the target rule according to the information type; the accidental touch analysis module 930 is configured to perform adjacent key accidental touch analysis on the target information according to the target rule, and label adjacent key accidental touch information according to the analysis results, so as to train the adjacent key accidental touch recognition model based on the labeled adjacent key accidental touch information.

[0116] In some embodiments of this application, based on the above technical solutions, the rule determination module 920 includes: a first rule determination unit, configured to use a first rule as the target rule when the input information is intelligent error correction information; and a second rule determination unit, configured to use a second rule as the target rule when the input information is manual error correction information; wherein the first rule is different from the second rule.

[0117] In some embodiments of this application, the target rule is the first rule; based on the above technical solution, the accidental touch analysis module 930 includes: a character length judgment unit, configured to judge whether the string length of the on-screen information is the same as the string length of the selected information in the intelligent error correction information; and a character information judgment unit, configured to, when the string length of the on-screen information is the same as the string length of the error correction candidate selected information, judge whether the original input information is adjacent key accidental touch information based on the characters contained in the on-screen information and the characters contained in the original input information in the intelligent error correction information.

[0118] In some embodiments of this application, based on the above technical solutions, the character information judgment unit is configured to: compare the characters of each character position in the original input information with the characters of the corresponding character positions in the displayed information; when the i-th character of the original input information is different from the i-th character of the displayed information, determine whether the i-th character of the original input information and the i-th character of the displayed information are adjacent keys according to the adjacent key dictionary table; when it is determined that the i-th character of the original input information and the i-th character of the displayed information are adjacent keys, mark the original input information as adjacent key mis-touch information.

[0119] In some embodiments of this application, the target rule is the second rule; based on the above technical solution, the accidental touch analysis module 930 includes: a string extraction unit configured to extract an initial string, an intermediate string, and a termination string based on the manual error correction information; a judgment unit configured to judge whether the initial string, the intermediate string, and the termination string meet preset conditions; a character matching unit configured to match the characters at each character position in the initial string with the characters at the corresponding character positions in the termination string when the initial string, the intermediate string, and the termination string meet the preset conditions; a temporary string construction unit configured to construct a temporary string based on the initial string if the characters at the corresponding character positions in the initial string and the termination string are different; and an accidental touch analysis unit configured to perform adjacent key accidental touch analysis based on the temporary string and the termination string, and to mark adjacent key accidental touch information based on the analysis results.

[0120] In some embodiments of this application, based on the above technical solutions, the judging unit is configured to: judge whether the initial string, the intermediate string, and the termination string are within the normal character range; when it is determined that the initial string, the intermediate string, and the termination string are all within the normal character range, judge whether the initial string, the intermediate string, and the termination string satisfy a preset relationship; wherein, the preset relationship includes: the termination string is the same as the screen information, the initial string is different from the termination string, the initial string contains the intermediate string, the termination string contains the intermediate string, and the termination string does not contain the initial string.

[0121] In some embodiments of this application, based on the above technical solutions, the temporary string construction unit is configured to: when the j-th character in the initial string is different from the j-th character in the termination string, remove the j-th character in the initial string to obtain the temporary string; where j is a positive integer.

[0122] In some embodiments of this application, based on the above technical solutions, the accidental touch analysis unit is configured to: compare the temporary string with the termination string; if the termination string starts with the temporary string, then determine that the manual error correction information is not adjacent key accidental touch information.

[0123] In some embodiments of this application, based on the above technical solutions, the accidental touch analysis unit is further configured to: if the termination string does not start with the temporary string, insert the j-th character of the termination string between the (j-1)-th and j-th positions of the initial string to update the temporary string; perform adjacent key accidental touch analysis based on the updated temporary string and the termination string, and mark adjacent key accidental touch information based on the analysis results.

[0124] In some embodiments of this application, based on the above technical solutions, mis-touch analysis is performed according to the updated temporary string and the termination string, including: comparing the updated temporary string with the termination string; if the termination string does not start with the updated temporary string, then determining whether the j-th character in the termination string and the j-th character in the initial string are adjacent keys based on the adjacent key dictionary; when it is determined that the j-th character in the termination string and the j-th character in the initial string are adjacent keys, the manual error correction information is marked as adjacent key mis-touch information.

[0125] In some embodiments of this application, based on the above technical solutions, the adjacent key mis-touch information labeling device 900 further includes: an adjacent key dictionary table construction module, configured to, before acquiring the log information generated by the user inputting information through the keyboard, take each key in the keyboard as a target key, and construct an adjacent key dictionary table according to the correspondence between the target key and the key adjacent to the target key.

[0126] In some embodiments of this application, based on the above technical solutions, the log parsing module 910 is configured to: collect information generated during the user's character editing process, character display process, and error correction process through set tracking points, and generate the log information based on the information generated during the character editing process, character display process, and error correction process.

[0127] The specific details of the adjacent key mis-touch information marking device provided in the various embodiments of this application have been described in detail in the corresponding method embodiments, and will not be repeated here.

[0128] Figure 10 This schematically illustrates a computer system architecture block diagram for an electronic device used to implement embodiments of the present application. The electronic device may be, for example... Figure 1 The terminal device 110 or server 130 shown.

[0129] It should be noted that, Figure 10 The computer system 1000 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0130] like Figure 10 As shown, the computer system 1000 includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 1002 or programs loaded from storage section 1008 into random access memory (RAM). The RAM 1003 also stores various programs and data required for system operation. The CPU 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An input / output interface 1005 (I / O interface) is also connected to the bus 1004.

[0131] In some embodiments, the following components are connected to the input / output interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a local area network card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the input / output interface 1005 as needed. A removable medium 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 1010 as needed so that computer programs read from it can be installed into the storage section 1008 as needed.

[0132] Specifically, according to embodiments of this application, the processes described in the various method flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1009, and / or installed from removable medium 1011. When the computer program is executed by central processing unit 1001, it performs various functions defined in the system of this application.

[0133] It should be noted that the computer-readable storage medium shown in the embodiments of this application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable storage medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0134] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0135] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0136] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, and includes several instructions to cause an electronic device to execute the method according to the embodiments of this application.

[0137] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0138] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for annotating adjacent key mis-touch information, characterized in that, include: Obtain log information generated by the user inputting information via the keyboard, parse the log information to obtain target information corresponding to the same input cycle, wherein the target information includes typing information and screen display information, and the typing information is intelligent error correction information or manual error correction information; Obtain the information type corresponding to the typed information, and determine the target rule based on the information type; The target information is analyzed for adjacent key mis-touches according to the target rules, and the adjacent key mis-touch information is labeled according to the analysis results, so as to train the adjacent key mis-touch recognition model based on the labeled adjacent key mis-touch information.

2. The method according to claim 1, characterized in that, The step of obtaining the information type corresponding to the typed information and determining the target rule based on the information type includes: When the input information is intelligent error correction information, the first rule is used as the target rule; When the typed information is manual error correction information, the second rule is used as the target rule; The first rule is different from the second rule.

3. The method according to claim 2, characterized in that, The target rule is the first rule; The step of performing adjacent key mis-touch analysis on the target information according to the target rules, and labeling adjacent key mis-touch information based on the analysis results, includes: Determine whether the string length of the displayed information is the same as the string length of the selected information in the intelligent error correction information; When the string length of the on-screen information is the same as the string length of the selected information, the original input information is determined to be a mis-touch information of adjacent keys based on the characters contained in the on-screen information and the characters contained in the original input information in the intelligent error correction information.

4. The method according to claim 3, characterized in that, The step of determining whether the original input information is a mis-touch of adjacent keys based on the characters contained in the on-screen information and the characters contained in the original input information in the intelligent error correction information includes: Compare the characters at each position in the original input information with the characters at the corresponding positions in the displayed information; When the i-th character of the original input information is different from the i-th character of the displayed information, determine whether the i-th character of the original input information and the i-th character of the displayed information are adjacent keys according to the adjacent key dictionary table; When it is determined that the i-th character of the original input information and the i-th character of the displayed information are adjacent keys, the original input information is marked as adjacent key mis-touch information.

5. The method according to claim 2, characterized in that, The target rule is the second rule; The step of performing adjacent key mis-touch analysis on the target information according to the target rules, and labeling adjacent key mis-touch information based on the analysis results, includes: Extract the initial string, intermediate string, and termination string based on the manual error correction information; Determine whether the initial string, the intermediate string, and the termination string satisfy a preset condition; When the initial string, the intermediate string, and the termination string meet the preset conditions, the characters at each position in the initial string are compared with the characters at the corresponding positions in the termination string. If the characters at the same position in the initial string and the terminating string are different, then a temporary string is constructed based on the initial string; The adjacent key mis-touch analysis is performed based on the temporary string and the termination string, and the adjacent key mis-touch information is marked according to the analysis results.

6. The method according to claim 5, characterized in that, The step of determining whether the initial string, the intermediate string, and the terminating string satisfy preset conditions includes: Determine whether the initial string, the intermediate string, and the terminating string are within the normal character range; When it is determined that the initial string, the intermediate string, and the terminating string are all within the normal character range, it is determined whether the initial string, the intermediate string, and the terminating string satisfy a preset relationship; The preset relationships include: the termination string is the same as the on-screen information, the initial string is different from the termination string, the initial string contains the intermediate string, the termination string contains the intermediate string, and the termination string does not contain the initial string.

7. The method according to claim 5 or 6, characterized in that, If the characters at the same position in the initial string and the terminating string are different, then a temporary string is constructed based on the initial string, including: When the j-th character in the initial string is different from the j-th character in the terminating string, the j-th character in the initial string is removed to obtain the temporary string; Where j is a positive integer.

8. The method according to claim 7, characterized in that, The step of performing adjacent key mis-touch analysis based on the temporary string and the termination string, and annotating adjacent key mis-touch information based on the analysis results, includes: Compare the temporary string with the terminating string; If the termination string begins with the temporary string, then the manual error correction information is determined not to be adjacent key mis-touch information.

9. The method according to claim 8, characterized in that, The step of performing adjacent key mis-touch analysis based on the temporary string and the termination string, and annotating adjacent key mis-touch information based on the analysis results, further includes: If the terminating string does not begin with the temporary string, then the j-th character in the terminating string is inserted between the (j-1)-th and j-th characters of the initial string to update the temporary string; The analysis of accidental adjacent key touches is performed based on the updated temporary string and the termination string, and the accidental adjacent key touch information is marked based on the analysis results.

10. The method according to claim 9, characterized in that, The step of performing accidental touch analysis based on the updated temporary string and the termination string, and annotating adjacent key accidental touch information based on the analysis results, includes: The updated temporary string is compared with the terminating string to determine whether the terminating string begins with the updated temporary string; If the terminating string does not begin with the updated temporary string, then the neighbor key dictionary is used to determine whether the j-th character in the terminating string and the j-th character in the initial string are adjacent keys. When it is determined that the j-th character in the terminating string is an adjacent key to the j-th character in the initial string, the manual error correction information is marked as adjacent key mis-touch information.

11. The method according to claim 1, characterized in that, Before acquiring the log information generated by the user's keyboard input, the method further includes: Each key on the keyboard is used as a target key, and an adjacent key dictionary table is constructed based on the correspondence between the target keys and the keys adjacent to the target keys.

12. A device for marking information on accidental touches of adjacent keys, characterized in that, include: The log parsing module is configured to acquire log information generated by the user through keyboard input, parse the log information to obtain target information corresponding to the same input cycle, wherein the target information includes typing information and screen display information, and the typing information is intelligent error correction information or manual error correction information. The rule determination module is configured to obtain the information type corresponding to the input information and determine the target rule based on the information type; The accidental touch analysis module is configured to perform adjacent key accidental touch analysis on the target information according to the target rules, and to label adjacent key accidental touch information according to the analysis results, so as to train the adjacent key accidental touch recognition model based on the labeled adjacent key accidental touch information.

13. The adjacent key mis-touch information labeling device as described in claim 12, characterized in that, The rule determination module includes: The first rule determination unit is configured to use the first rule as the target rule when the input information is intelligent error correction information; The second rule determination unit is configured to use the second rule as the target rule when the input information is manual error correction information; wherein the first rule is different from the second rule.

14. The adjacent key mis-touch information labeling device as described in claim 13, characterized in that, The target rule is the first rule; the accidental touch analysis module includes: The character length determination unit is configured to determine whether the string length of the on-screen information is the same as the string length of the selected information in the intelligent error correction information; The character information judgment unit is configured to determine whether the original input information is adjacent key mis-touch information based on the characters contained in the screen information and the characters contained in the original input information in the intelligent error correction information when the string length of the screen information is the same as the string length of the selected information.

15. The adjacent key mis-touch information labeling device as described in claim 14, characterized in that, The character information determination unit is configured as follows: Compare the characters at each position in the original input information with the characters at the corresponding positions in the displayed information; When the i-th character of the original input information is different from the i-th character of the displayed information, the i-th character of the original input information and the i-th character of the displayed information are determined according to the adjacent key dictionary table to determine whether they are adjacent keys; when it is determined that the i-th character of the original input information and the i-th character of the displayed information are adjacent keys, the original input information is marked as adjacent key mis-touch information.

16. The adjacent key mis-touch information labeling device as described in claim 13, characterized in that, The target rule is the second rule; the accidental touch analysis module includes: The string extraction unit is configured to extract the initial string, intermediate string, and termination string based on the manual error correction information; The judgment unit is configured to determine whether the initial string, the intermediate string, and the termination string meet preset conditions; The character matching unit is configured to compare the characters at each position in the initial string with the characters at the corresponding positions in the termination string when the initial string, the intermediate string, and the termination string meet the preset conditions. A temporary string construction unit is configured to construct a temporary string based on the initial string if the characters at the same character position in the initial string and the terminating string are different. The accidental touch analysis unit is configured to perform adjacent key accidental touch analysis based on the temporary string and the termination string, and to annotate adjacent key accidental touch information based on the analysis results.

17. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the adjacent key mis-touch information annotation method according to any one of claims 1 to 11.

18. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the adjacent key mis-touch information labeling method according to any one of claims 1 to 11 by executing the executable instructions.

19. A computer program product, characterized in that, The computer program product includes a computer program adapted to be loaded by a processor and execute the adjacent key mis-touch information annotation method as described in any one of claims 1 to 11.

Images