CN110858099B - Candidate word generation method and device - Google Patents

Abstract

The invention discloses a candidate word generation method and a candidate word generation device, wherein the method comprises the following steps: acquiring the above words in real time and obtaining candidate words; determining a user emotion category according to the above; and screening the candidate words according to the emotion categories of the users to obtain candidate words to be output. By using the method and the device, the accuracy of the candidate words can be improved, and the user input experience is improved.

Description

Candidate word generation method and device

Technical Field

The present invention relates to the field of input methods, and in particular to a candidate word generation method and device.

Background technique

Input method is a coding method for inputting various symbols into electronic devices, and is an indispensable tool for humans to interact with electronic devices. In order to speed up the input speed, most existing input methods have an associative memory function, that is, after inputting a Chinese character or a word, the input method will automatically recommend candidate characters or words.

Some existing input methods predict words during the user input process, predicting the next word to be input by the user and providing it to the user to facilitate the user input. The current method for word prediction in input methods mainly uses large-scale corpus data, traditional ngram statistical models, or deep learning models to learn language models, and then uses these models to find the next word with the highest probability based on information such as the previous context and input environment.

Although this method facilitates user input to a certain extent, the current model has difficulty perceiving the overly long preceding context, and therefore often produces out-of-context prediction results. For example, when a user inputs "the dishes you make are delicious and colorful", most current input methods will only give candidate words such as "not delicious" and "delicious" based on the preceding context "the dishes you make". Obviously, "not delicious" is not a reasonable candidate word here. Such candidate words not only cover up other candidate words that express appreciation, such as "very good" and "great", but also bring a very poor experience to users.

Summary of the invention

The embodiments of the present invention provide a candidate word generation method and device to improve the accuracy of candidate words and enhance the user input experience.

To this end, the present invention provides the following technical solutions:

A candidate word generation method, the method comprising:

Get the previous text and candidate words in real time;

Determine the user's emotion category according to the above context;

The candidate words are screened according to the user emotion category to obtain candidate words to be output.

Preferably, the above context is any one of the following: text, voice, or picture.

Preferably, the method further comprises: pre-building an emotion recognition model;

Determining the user emotion category according to the above text includes:

Extracting the text information of the above text;

The user emotion category is obtained by using the text information and the emotion recognition model.

Preferably, the emotion recognition model is a deep learning model; the extracting the text information of the above text comprises: obtaining a word sequence corresponding to the above text, and determining a word vector of each word in the word sequence; or

The emotion recognition model is SVM or decision tree; the extracting the text information of the above text includes: obtaining the word sequence corresponding to the above text, and determining the ID of each word in the word sequence.

Preferably, the determining the user emotion category according to the above text further includes:

Acquire auxiliary information, the auxiliary information including any one or more of the following: current environment information, location information, user body information;

The user emotion category is obtained by using the text information, the auxiliary information and the emotion recognition model.

Preferably, the method further comprises:

Obtain the candidate score of each candidate word;

The candidate words are screened according to the user emotion category to obtain the candidate words to be output, including:

Adjusting the candidate score of the candidate word according to the user emotion category to obtain a final score of the candidate word;

The candidate words to be output are determined according to the final scores.

Preferably, adjusting the candidate score of the candidate word according to the user emotion category to obtain the final score of the candidate word includes:

Determining a sentiment score for each candidate word according to the user sentiment category;

The candidate score of the candidate word and the sentiment score are weightedly summed to obtain the final score of the candidate word.

Preferably, adjusting the candidate score of the candidate word according to the user emotion category to obtain the final score of the candidate word includes:

Determining the weight of the candidate score of each candidate word according to the user emotion category;

The final score of the candidate word is calculated based on the candidate score of the candidate word and its weight.

Preferably, determining the candidate words to be output according to the final scores comprises:

Selecting a set number of candidate words as candidate words to be output according to the final scores in descending order; or

The candidate words whose final scores are greater than the set threshold are selected as the candidate words to be output.

Preferably, the screening of the candidate words according to the user emotion category to obtain the candidate words to be output includes:

A candidate word corresponding to the emotion category among the candidate words is selected as a candidate word to be output.

Preferably, the method further comprises:

Pre-establish a list of candidate words corresponding to different emotion categories;

The selecting of the candidate words corresponding to the emotion category from the candidate words comprises:

A candidate word corresponding to the emotion category among the candidate words is selected according to the list.

Preferably, the method further comprises:

The emotion recognition model is personalized trained according to historical input information to update the emotion recognition model.

A candidate word generating device, the device comprising:

The previous text acquisition module is used to acquire the previous text in real time and obtain candidate words;

A candidate word acquisition module, used to obtain candidate words;

An emotion recognition module, used to determine the user's emotion category according to the above text;

The screening module is used to screen the candidate words according to the user emotion category to obtain the candidate words to be output.

Preferably, the above context is any one of the following: text, voice, or picture.

Preferably, the device further comprises: a model building module, for pre-building an emotion recognition model;

The emotion recognition module comprises:

A text processing unit, used to extract the text information of the above text;

The recognition unit is used to obtain the user emotion category by using the text information and the emotion recognition model.

Preferably, the emotion recognition model is a deep learning model, and the text processing unit is specifically used to obtain the word sequence corresponding to the above text and determine the word vector of each word in the word sequence; or

The emotion recognition model is a classification model; the text processing unit is specifically used to obtain the word sequence corresponding to the above text and determine the ID of each word in the word sequence.

Preferably, the emotion recognition module further includes:

An auxiliary information acquisition unit, used to acquire auxiliary information, wherein the auxiliary information includes: current environment information and/or location information;

The recognition unit is specifically used to obtain the user emotion category by using the text information, the auxiliary information and the emotion recognition model.

Preferably, the candidate word acquisition module is further used to obtain a candidate score of each candidate word;

The screening module comprises:

A score adjustment module, used to adjust the candidate score of the candidate word according to the user emotion category to obtain a final score of the candidate word;

A candidate word output module is used to determine the candidate word to be output according to the final score.

Preferably, the score adjustment module includes:

A sentiment score determination unit, used to determine the sentiment score of each candidate word according to the user's sentiment category;

The first calculation unit is used to perform weighted summation of the candidate score of the candidate word and the sentiment score to obtain a final score of the candidate word.

Preferably, the score adjustment module includes:

A weight determination unit, used to determine the weight of the candidate score of each candidate word according to the user emotion category;

The second calculation unit is used to calculate the final score of the candidate word according to the candidate score of the candidate word and its weight.

Preferably, the candidate word output module is specifically used to select a set number of candidate words as candidate words to be output in descending order of final scores; or select candidate words whose final scores are greater than a set threshold as candidate words to be output.

Preferably, the screening module is specifically used to select candidate words corresponding to the emotion category from the candidate words as candidate words to be output.

Preferably, the device further comprises:

A candidate word list building module, used to pre-build candidate word lists corresponding to different emotion categories;

The screening module selects a candidate word corresponding to the emotion category from among the candidate words according to the list.

Preferably, the device further comprises:

Information recording module, used to record historical input information;

The model updating module is used to perform personalized training on the emotion recognition model using the historical input information to update the emotion recognition model.

A computer device, comprising: one or more processors, a memory;

The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the aforementioned method.

A readable storage medium stores instructions thereon, wherein the instructions are executed to implement the above-mentioned method.

The candidate word generation method and device provided in the embodiments of the present invention identify user emotions based on the above information, screen each candidate word according to the identified user emotion category, and provide the candidate words that match the user's current mood to the user first, thereby making the candidate words provided to the user more accurate, thereby improving the user input efficiency and enhancing the user input experience.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

FIG1 is a flow chart of a candidate word generating method according to an embodiment of the present invention;

FIG2 is a schematic diagram of the structure of a candidate word generating device according to an embodiment of the present invention;

FIG3 is a schematic diagram of a specific application structure of a candidate word generating device according to an embodiment of the present invention;

FIG4 is a block diagram of a device for generating a candidate word according to an exemplary embodiment;

FIG. 5 is a schematic diagram of the structure of a server in an embodiment of the present invention.

Detailed ways

In order to enable persons skilled in the art to better understand the solutions of the embodiments of the present invention, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and implementation modes.

In order to solve the problem that the existing input method generates associative candidates that are irrelevant to the actual content of the previous text when generating candidate words, an embodiment of the present invention provides a candidate word generation method and device, which analyzes the previous text to obtain the user emotion category, and uses the emotion category to screen each candidate word to obtain the candidate word to be output.

As shown in FIG. 1 , it is a flow chart of a candidate word generation method according to an embodiment of the present invention, which includes the following steps:

Step 101, obtain the above text in real time and obtain candidate words.

For different input devices and input methods, the form of the above context may also be different, for example, the above context may be in the form of text, voice, or picture, etc. In addition, according to different application scenarios, the above context may be user input, existing context, context of peer interaction, etc.

The candidate words may be generated by an existing input method. Different input methods may have different methods and rules for generating candidate words, which are not limited in the embodiments of the present invention. Regardless of the candidate words generated by the input method, further screening of the candidate words by the subsequent solution of the present invention can obtain candidate words that better match the above.

Step 102: Determine the user emotion category according to the above text.

Specifically, a model-based approach can be used to pre-build an emotion recognition model, and the emotion recognition model can use a deep learning model, such as DNN (Deep Neural Networks), CNN (Convolutional Neural Network), etc., or other traditional classification models, such as SVM (Support Vector Machine), decision tree, etc. The training process of the model can use conventional techniques and will not be described in detail here.

When using the emotion recognition model to recognize the user emotions based on the above text, it is necessary to preprocess the above text first, extract the text information in the above text, and then input the text information into the emotion recognition model, and obtain the user emotion category according to the output of the model. For example, when the user emotion category adopts a deep learning model, it is necessary to first obtain the word sequence in the above text, and then determine the word vector of each word in the word sequence, and input the word vector into the deep learning model, and the user emotion category can be obtained according to the output of the emotion recognition model. For another example, when the emotion recognition model adopts other classification models, it is also necessary to first obtain the word sequence in the above text, and then determine the ID of each word in the word sequence to obtain an ID sequence, and input the ID sequence into the classification model to obtain the user emotion category.

The user emotions can be divided into three categories: positive, negative and others. Of course, they can be further subdivided. For example, positive emotions include happiness and preference; negative emotions include sadness, anxiety, anger, etc.

In practical applications, the word sequence in the above text can be obtained by segmenting the above text, such as a segmentation method based on string matching, a segmentation method based on understanding, a segmentation method based on statistics, etc.; or by recording each word input by the user. Of course, for other non-textual forms of the above text, the text information in the above text can be obtained first, and then segmented to obtain the corresponding word sequence. For example, for the above text in the form of an image, the corresponding text can be obtained by using image recognition technology; for the above text in the form of a voice, the corresponding text can be obtained by using voice recognition technology.

Step 103: Screen the candidate words according to the user emotion category to obtain candidate words to be output.

When screening the candidate words, there may be multiple screening methods.

For example, according to the user emotion category, the candidate words corresponding to the emotion category among the candidate words can be selected as the candidate words to be output. Specifically, a candidate word list corresponding to different emotion categories can be pre-established, and the candidate word corresponding to the emotion category among the candidate words can be selected according to the list.

For another example, the candidate scores of each candidate word based on the current input may be obtained, and the candidate scores of the candidate words may be adjusted according to the user emotion category to obtain the final scores of the candidate words; and then the candidate words to be output may be determined according to the final scores.

When adjusting the candidate score of the candidate word, there may be multiple adjustment methods, such as:

One method: first determine the sentiment score of each candidate word according to the user sentiment category. For example, the part-of-speech score of the candidate word can be determined according to the strength of the correspondence between each candidate word and the identified user sentiment category. The stronger the correspondence, the higher the part-of-speech score; then perform a weighted summation of the candidate score of the candidate word and the sentiment score to obtain the final score of the candidate word.

Another way: first determine the weight of the candidate score of each candidate word according to the user emotion category; then calculate the final score of the candidate word according to the candidate score of the candidate word and its weight. For example, for the recognized user emotion category, the weight of the candidate word related to the emotion category is set to 1; the weight of other candidate words is set to 0.5.

After calculating the final score of each candidate word, a set number of candidate words can be selected as candidate words to be output in descending order of the final score; or candidate words with final scores greater than a set threshold can be selected as candidate words to be output.

The candidate word generation method provided in the embodiment of the present invention identifies user emotions based on the above information, and screens each candidate word according to the identified user emotion category, thereby providing the user with more accurate candidate words, thereby improving user input efficiency and enhancing user input experience.

For example, when a user inputs "He always makes mistakes, I am very", the existing input method will generate candidate words such as "happy", "like", and "sad". Using the method of the embodiment of the present invention, firstly, the obtained context is used to determine that the user expresses more "sad" emotions rather than "happy" emotions. Therefore, after screening, candidate words such as "sad" and "sad" will be ranked higher, and association candidates such as "happy" and "like" will be filtered or ranked lower.

For another example, in a scenario where user A and user B are talking through a chat tool, when user A inputs "Hello, cute, I'm good" by voice, candidate words such as "scared", "sad", and "like" will be generated according to the existing input method. Using the method of the embodiment of the present invention, the voice is firstly recognized to obtain the above text, and the above text is used to judge that the user expresses more "positive" emotions rather than "negative" emotions. Therefore, after screening, candidates such as "like" that express positive emotions should be ranked higher, and candidates such as "scared" and "sad" will be filtered out or ranked lower.

When the user inputs, sometimes factors such as the input environment will also have a certain impact on the user input. Therefore, in another embodiment of the method of the present invention, when constructing an emotion recognition model and using the model to perform user emotion recognition, some other factors can also be considered as auxiliary information. For example, the auxiliary information may include any one or more of the following: current environmental information, location information, user body information. The environmental information may include, for example, temperature, climate, etc. The user body information may include, for example, body temperature, movement status, current input speed, etc. Specifically, the auxiliary information is obtained, and the user emotion category is obtained using the text information, the auxiliary information and the emotion recognition model. The auxiliary information can be obtained by calling the application program interface provided by the input application, or by obtaining it through a third-party APP.

Since different users may have different input habits, in another embodiment of the method of the present invention, historical input information may also be recorded, such as the candidate words selected by the user each time, and the emotion recognition model may be personalized trained using the recorded historical input information to update the emotion recognition model. Specifically, the historical input information recorded each time may be used as a training sample. After the number of training samples reaches a certain number, the new samples may be used to retrain based on the original emotion recognition model parameters, thereby obtaining a personalized emotion recognition model that better matches the user. In subsequent inputs, the updated emotion recognition model may be used to identify the user's emotions, which may make the recognition results more accurate, and thus may make the candidate words provided to the user have a higher degree of match with the above text, further improving the accuracy of the candidate words and improving the user's input efficiency.

Accordingly, an embodiment of the present invention further provides a candidate word generating device, which can be integrated in a user device, which can be a notebook, a computer, a PAD, a mobile phone, etc. When a user performs input operations, he can use a physical keyboard or a virtual keyboard on the touch screen of the user device.

As shown in FIG. 2 , it is a schematic diagram of the structure of a candidate word generating device according to an embodiment of the present invention.

In this embodiment, the apparatus comprises:

A text acquisition module 201 is used to acquire text in real time;

A candidate word acquisition module 202 is used to obtain candidate words;

An emotion recognition module 203, used to determine the user's emotion category according to the above text;

The screening module 204 is used to screen the candidate words according to the user emotion category to obtain the candidate words to be output.

For different input devices and input methods, the form of the above context may also be different, for example, the above context may be in the form of text, voice, or picture, etc. In addition, according to different application scenarios, the above context may be user input, existing context, context of peer interaction, etc.

The candidate words may be generated by an existing input method. Different input methods may have different methods and rules for generating candidate words, which are not limited in the embodiments of the present invention. Regardless of the candidate words generated by the input method, further screening of the candidate words by the subsequent solution of the present invention can obtain candidate words that better match the above.

The emotion recognition module 203 may specifically recognize the user emotion category in a model-based manner.

The model can be pre-constructed by a model construction module, and the model construction module can be used as an independent module or integrated into the device as a part of the device of the present invention.

In practical applications, the emotion recognition model may adopt a deep learning model, such as DNN, CNN, etc., or a traditional classification model, such as SVM, decision tree, etc. The training process of the model may adopt conventional techniques and will not be described in detail here.

Based on the pre-built emotion recognition model, when the emotion recognition module 203 determines the user emotion category based on the above text, it is necessary to pre-process the above text, extract the text information of the above text, and then input the text information into the emotion recognition model to obtain the user emotion category according to the output of the model. Accordingly, a specific structure of the emotion recognition module 203 may include: a text processing unit and a recognition unit. Among them:

The text processing unit is used to extract the text information of the above text;

The recognition unit is used to obtain the user emotion category by using the text information and the emotion recognition model.

For different models, the text processing unit needs to obtain different text information. For example, when the emotion recognition model is a deep learning model, the text processing unit needs to obtain the word sequence corresponding to the above text and determine the word vector of each word in the word sequence; when the emotion recognition model is a traditional classification model, the text processing unit needs to obtain the word sequence corresponding to the above text and determine the ID of each word in the word sequence. Accordingly, the recognition unit needs to input the word vector of each word in the word sequence into the deep learning model to obtain the user emotion category, or input the ID of each word in the word sequence into the traditional classification model to obtain the user emotion category.

For the textual context, the text processing unit can obtain the word sequence corresponding to the textual context by performing word segmentation on the context; for other forms of context, the corresponding text can be obtained by first using corresponding recognition technology, and then word segmentation can be performed to obtain the word sequence. If the context is user input, the word sequence corresponding to the context can also be obtained by recording each word input by the user.

Furthermore, the emotion recognition module 203 may also include: an auxiliary information acquisition unit for acquiring auxiliary information, wherein the auxiliary information includes any one or more of the following: current environment information, location information, user body information, etc. Accordingly, the recognition unit can use the text information, the auxiliary information and the emotion recognition model to obtain the user emotion category. It should be noted that in this case, the auxiliary information also needs to be considered when constructing the emotion recognition model.

In actual applications, the screening module 204 may use a variety of screening methods when screening the candidate words.

For example, in a specific embodiment, a candidate word among the candidate words whose part of speech is the same as the part of speech information of the next word may be selected as the candidate word to be output.

For another example, as shown in FIG. 3 , in another specific embodiment, the candidate word acquisition module 202 is further configured to acquire a candidate score for each candidate word.

Accordingly, in this embodiment, the screening module 204 includes: a score adjustment module 241 and a candidate word output module 242. Among them:

The score adjustment module 241 is used to adjust the candidate score of the candidate word according to the user emotion category to obtain the final score of the candidate word;

The candidate word output module 242 is used to determine the candidate word to be output according to the final score.

In practical applications, the score adjustment module 241 may also use a variety of methods to adjust the candidate scores of the candidate words.

For example, a specific implementation of the score adjustment module 241 may include: a sentiment score determination unit and a first calculation unit. The sentiment score determination unit is used to determine the sentiment score of each candidate word according to the user emotion category; the first calculation unit is used to perform a weighted summation of the candidate score of the candidate word and the sentiment score to obtain the final score of the candidate word.

For another example, another specific implementation of the score adjustment module 241 may include: a weight determination unit and a second calculation unit. The weight determination unit is used to determine the weight of the candidate score of each candidate word according to the user emotion category; the second calculation unit is used to calculate the final score of the candidate word according to the candidate score of the candidate word and its weight.

The candidate word output module 242 may specifically select a set number of candidate words as candidate words to be output in descending order of final scores; or select candidate words whose final scores are greater than a set threshold as candidate words to be output.

The candidate word generation device provided in the embodiment of the present invention determines the user emotion category based on the acquired context information, and screens each candidate word according to the user emotion category, thereby providing the user with more accurate candidate words, thereby improving the user input efficiency and enhancing the user input experience.

Since different users may have different input habits, in another embodiment of the device of the present invention, it may further include: an information recording module and a model updating module (not shown). The information recording module is used to record historical input information; the model updating module is used to use the historical input information recorded by the information recording module to perform personalized training on the emotion recognition model and update the emotion recognition model. Specifically, each recorded historical input information can be used as a training sample. After the number of training samples reaches a certain number, the new samples can be used to retrain on the basis of the original emotion recognition model parameters, so as to obtain a personalized emotion recognition model that better matches the user.

Accordingly, in the subsequent user input process, using the updated emotion recognition model to identify user emotions can make the recognition results more accurate, and thus can make the candidate words provided to the user have a higher match with the previous text, further improving the accuracy of the candidate words and improving user input efficiency.

It should be noted that, in practical applications, the method and device of the present invention can be applied to various input methods, and can be applicable to input methods such as pinyin input, Wubi input, voice input or other input methods.

Fig. 4 is a block diagram of an apparatus 800 for generating a candidate word according to an exemplary embodiment. For example, the apparatus 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

4 , the device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input/output (I/O) interface 812 , a sensor component 814 , and a communication component 816 .

The processing component 802 generally controls the overall operation of the device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the device 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

The power component 806 provides power to the various components of the device 800. The power component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 800.

The multimedia component 808 includes a screen that provides an output interface between the device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and/or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and/or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

The audio component 810 is configured to output and/or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the device 800 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

I/O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: home button, volume button, start button, and lock button.

The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the device 800. For example, the sensor assembly 814 can detect the open/closed state of the device 800, the relative positioning of components, such as the display and keypad of the device 800, and the sensor assembly 814 can also detect the position change of the device 800 or a component of the device 800, the presence or absence of user contact with the device 800, the orientation or acceleration/deceleration of the device 800, and the temperature change of the device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

The communication component 816 is configured to facilitate wired or wireless communication between the device 800 and other devices. The device 800 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

In an exemplary embodiment, the apparatus 800 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above method.

In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by the processor 820 of the device 800 to complete the above-mentioned key mis-touch error correction method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

The present invention also provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by a processor of a mobile terminal, the mobile terminal is enabled to execute all or part of the steps in the above-mentioned method embodiment of the present invention.

FIG5 is a schematic diagram of the structure of a server in an embodiment of the present invention. The server 1900 may have relatively large differences due to different configurations or performances, and may include one or more central processing units (CPU) 1922 (for example, one or more processors) and a memory 1932, and one or more storage media 1930 (for example, one or more mass storage devices) storing application programs 1942 or data 1944. Among them, the memory 1932 and the storage medium 1930 may be short-term storage or permanent storage. The program stored in the storage medium 1930 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the server. Furthermore, the central processing unit 1922 may be configured to communicate with the storage medium 1930 and execute a series of instruction operations in the storage medium 1930 on the server 1900.

The server 1900 may also include one or more power supplies 1926, one or more wired or wireless network interfaces 1950, one or more input and output interfaces 1958, one or more keyboards 1956, and/or, one or more operating systems 1941, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

It should be understood that the present invention is not limited to the exact construction that has been described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A method of generating a candidate word, the method comprising:

acquiring the above words in real time and obtaining candidate words;

extracting the text information; acquiring auxiliary information, wherein the auxiliary information comprises any one or more of the following: current environmental information, location information, user body information;

pre-constructing an emotion recognition model, recording historical input information, wherein the historical input information comprises candidate words selected by a user each time, taking the recorded historical input information each time as a training sample, and re-training by using the training sample on the basis of pre-constructed emotion recognition model parameters after the training sample reaches a certain number to obtain a personalized emotion recognition model matched with the user;

obtaining emotion categories of the user by using the text information, the auxiliary information and the personalized emotion recognition model corresponding to the user;

obtaining candidate scores of candidate words;

determining part-of-speech scores of the candidate words according to the strength of the corresponding relation between each candidate word and the identified emotion category of the user, wherein the part-of-speech score is higher when the corresponding relation is stronger;

weighting and summing the candidate score of the candidate word and the part-of-speech score to obtain a final score of the candidate word;

and determining candidate words to be output according to the final score.

2. The method according to claim 1, wherein the above is any one of the following: text, speech, pictures.

3. The method of claim 1, wherein the step of determining the position of the substrate comprises,

the emotion recognition model is a deep learning model; the extracting the text information includes: acquiring the word sequence corresponding to the above, and determining word vectors of words in the word sequence; or alternatively

The emotion recognition model is an SVM or a decision tree; the extracting the text information includes: and acquiring the word sequence corresponding to the word sequence, and determining the ID of each word in the word sequence.

4. The method of claim 1, wherein the determining candidate words to be output according to the final score comprises:

selecting a set number of candidate words as candidate words to be output according to the sequence from high to low of the final score; or alternatively

And selecting the candidate words with the final score larger than the set threshold value as candidate words to be output.

5. A candidate word generation device, the device comprising:

the above acquisition module is used for acquiring the above in real time and obtaining candidate words;

the candidate word acquisition module is used for acquiring candidate words;

the emotion recognition module is used for extracting the text information; acquiring auxiliary information, wherein the auxiliary information comprises any one or more of the following: current environmental information, location information, user body information; pre-constructing an emotion recognition model, recording historical input information, wherein the historical input information comprises candidate words selected by a user each time, taking the recorded historical input information each time as a training sample, and re-training by using the training sample on the basis of pre-constructed emotion recognition model parameters after the training sample reaches a certain number to obtain a personalized emotion recognition model matched with the user; obtaining emotion categories of the user by using the text information, the auxiliary information and the personalized emotion recognition model corresponding to the user;

the screening module is used for determining part-of-speech scores of the candidate words according to the strength of the corresponding relation between each candidate word and the identified emotion category of the user, wherein the part-of-speech score is higher when the corresponding relation is stronger; weighting and summing the candidate score of the candidate word and the part-of-speech score to obtain a final score of the candidate word; and determining candidate words to be output according to the final score.

6. The apparatus of claim 5, wherein the foregoing is any one of: text, speech, pictures.

7. The apparatus of claim 5, wherein the device comprises a plurality of sensors,

the emotion recognition model is a deep learning model, and the text processing unit is specifically used for acquiring the word sequence corresponding to the text and determining word vectors of words in the word sequence; or alternatively

The emotion recognition model is a classification model; the text processing unit is specifically configured to obtain the word sequence corresponding to the above, and determine an ID of each word in the word sequence.

8. The apparatus of claim 5, wherein the device comprises a plurality of sensors,

the candidate word output module is specifically configured to select a set number of candidate words as candidate words to be output according to a sequence from high to low of the final score; or selecting the candidate words with the final scores larger than the set threshold value as candidate words to be output.

9. A computer device, comprising: one or more processors, memory;

the memory is for storing computer executable instructions and the processor is for executing the computer executable instructions to implement the method of any one of claims 1 to 4.

10. A readable storage medium having stored thereon instructions that are executed to implement the method of any of claims 1 to 4.