CN117251061B

CN117251061B - AIGC-based on-screen intelligent input method and device

Info

Publication number: CN117251061B
Application number: CN202311532682.6A
Authority: CN
Inventors: 严益强; 赵颖
Original assignee: Guangdong Planning and Designing Institute of Telecommunications Co Ltd
Current assignee: Guangdong Planning and Designing Institute of Telecommunications Co Ltd
Priority date: 2023-11-17
Filing date: 2023-11-17
Publication date: 2024-03-15
Anticipated expiration: 2043-11-17
Also published as: CN117251061A

Abstract

The invention relates to the technical field of text analysis, and discloses an AIGC-based on-screen intelligent input method and device, wherein the method is applied to the on-screen input method, an input interface of the on-screen input method integrates at least two input areas, and the method comprises the following steps: acquiring an input character string input by a user based on an input interface; dividing the input character string according to the character types corresponding to all the input areas to obtain word segmentation results corresponding to the input character string; based on a pre-trained language processing model, generating a candidate word set corresponding to an input character string according to word segmentation results corresponding to the input character string; and determining a target input word corresponding to the input character string according to the obtained target candidate word, wherein the target candidate word is a candidate word selected by a user from the candidate word set. Therefore, the method and the device can input multiple types of characters on the same screen, improve the character input efficiency and are beneficial to improving the character input experience of users.

Description

AIGC-based on-screen intelligent input method and device

Technical Field

The invention relates to the technical field of text analysis, in particular to an AIGC-based on-screen intelligent input method and device.

Background

At present, intelligent terminals such as computers and mobile phones become an indispensable tool in daily work and life of people, and users need to complete text input at the intelligent terminals by means of an input method in daily work and life.

Current chinese input methods generally include pinyin input, handwriting input, stroke input, voice input, and other input methods. However, it is found in practice that when a user needs to input a rare word or a strange Chinese character, the user needs to input a correct word by switching the input mode, or the user needs to perform multiple filtering among candidate words provided by the input method to correctly select a required word, and both the above two modes need the user to consume additional operation time, which increases the operation complexity of the user. Therefore, it is important to provide a technical scheme capable of improving the text input efficiency so as to improve the text input experience of the user.

Disclosure of Invention

The technical problem to be solved by the invention is to provide the AIGC-based on-screen intelligent input method and the AIGC-based on-screen intelligent input device, so that the text input efficiency can be improved, and the text input experience of a user can be improved.

In order to solve the technical problem, a first aspect of the present invention discloses an AIGC-based on-screen intelligent input method, the method is applied to an on-screen input method, an input interface of the on-screen input method integrates at least two input areas, each input area corresponds to a corresponding character type, the character types include a language type and/or a coding type, the method includes:

Acquiring an input character string input by a user based on the input interface, wherein the input character string comprises at least one input character;

according to the character types corresponding to all the input areas, the input character strings are segmented to obtain word segmentation results corresponding to the input character strings, and the word segmentation results corresponding to the input character strings comprise at least one character sub-string corresponding to the input character strings;

generating a candidate word set corresponding to the input character string based on a pre-trained language processing model according to a word segmentation result corresponding to the input character string, wherein the candidate word set comprises at least one candidate word;

and determining a target input word corresponding to the input character string according to the obtained target candidate word, wherein the target candidate word is a candidate word selected by the user from the candidate word set.

In a first aspect of the present invention, the input interface includes a first input area and a second input area, where a character type corresponding to the first input area is a first character type, and a character type corresponding to the second input area is a second character type;

the step of segmenting the input character string according to the character types corresponding to all the input areas to obtain a word segmentation result corresponding to the input character string, comprises the following steps:

Classifying all the input characters according to the character types corresponding to all the input areas to obtain classification results corresponding to the input character strings; the classification result comprises a first type character set and a second type character set, wherein the first type character set is null or the character types of input characters included in the first type character set are all the first character types, the second type character set is null or the character types of input characters included in the second type character set are all the second character types, and the first type character set and the second type character set are not null at the same time;

determining character type characteristics corresponding to the input character strings according to the classification results; when the first type character set included in the classification result is not empty, the character type features include first type character features; when the second type character set included in the classification result is not empty, the character type features include second type character features;

analyzing the input character strings based on the character type characteristics to obtain spelling association degrees between each input character and adjacent input characters;

And according to all the spelling relevancy, the input character string is segmented, and a word segmentation result corresponding to the input character string is obtained.

In an optional implementation manner, in a first aspect of the present invention, the generating, based on a pre-trained language processing model, a candidate word set corresponding to the input character string according to a word segmentation result corresponding to the input character string includes:

for each character sub-string, determining a sub-string coding mode corresponding to the character sub-string according to the character type corresponding to each input character in the character sub-string;

for each character sub-string, generating a sub-string candidate word corresponding to the character sub-string according to a sub-string coding mode corresponding to the character sub-string based on a pre-trained language processing model;

for each character sub-string, analyzing the semantic association degree between the sub-string candidate word corresponding to the character sub-string and the sub-string candidate word corresponding to the adjacent character sub-string;

based on the language processing model, performing clustering operation on sub-string candidate words corresponding to all the character sub-strings according to all the semantic relevancy, so as to obtain a clustering result, wherein the clustering result comprises a plurality of clustering sets, each clustering set comprises at least one sub-string candidate word corresponding to all the character sub-strings, and the semantic relevancy among all the sub-string candidate words in each clustering set is in the same preset relevancy range;

Screening at least one target cluster set meeting preset semantic association conditions from all the cluster sets according to the cluster result;

and determining a candidate word set corresponding to the input character string according to all the sub-string candidate words corresponding to the target cluster set.

In an optional implementation manner, in a first aspect of the present invention, for each of the character sub-strings, determining, according to a character type corresponding to each of the input characters in the character sub-string, a sub-string coding manner corresponding to the character sub-string includes:

comparing the character types corresponding to each input character in each character sub-string to obtain a type comparison result of the character sub-string;

for each character sub-string, when the type comparison result of the character sub-string is used for indicating that the character sub-string only contains one character type, determining a sub-string coding mode corresponding to the character sub-string as a coding mode corresponding to the character type contained in the character sub-string;

for each character sub-string, when the type comparison result of the character sub-string is used for indicating that the character sub-string contains at least two character types, determining the sub-string coding mode corresponding to the character sub-string as the mixed coding mode corresponding to all the character types contained in the character sub-string.

As an alternative embodiment, in the first aspect of the present invention, the method further includes:

determining the input preference corresponding to the user according to the acquired historical input record corresponding to the user;

wherein, according to the clustering result, at least one target clustering set meeting the preset semantic association condition is screened out from all the clustering sets, which comprises the following steps:

and screening at least one target cluster set meeting preset semantic association conditions or matching with the input preference from all the cluster sets according to the cluster result and the input preference corresponding to the user.

In a first aspect of the present invention, the determining, according to the obtained history input record corresponding to the user, the input preference corresponding to the user includes:

determining input statistical information corresponding to the history input records according to the acquired history input records corresponding to the users, wherein the input statistical information comprises input frequency of each history input word, vocabulary type corresponding to each history input word and sentence pattern structure corresponding to each history input sentence, and the history input sentence comprises at least one history input word;

According to the input statistical information, determining the historical input characteristics corresponding to the user;

analyzing the style matching degree between the history input features and each style corpus according to the history input features and the style corpuses corresponding to the acquired multiple language styles;

screening at least one target style corpus with the style matching degree higher than or equal to the preset style matching degree from all the style corpuses according to all the style matching degrees;

determining the user language style corresponding to the user according to the language styles corresponding to all the target language style corpus;

and determining the input preference corresponding to the user according to the language style of the user and the history input record.

generating a predicted word set corresponding to the target input word according to the target input word and the user language style, wherein the predicted word set comprises at least one predicted word;

wherein the generating, according to the target input word and the user language style, a predicted word set corresponding to the target input word includes:

Acquiring an input position corresponding to the target input word and an input background text corresponding to the target input word, wherein the input background text is the text where the input position is located;

determining adjacent texts of the target input words according to the input positions and the input position texts;

analyzing an input background text corresponding to the target input word to obtain a text analysis result corresponding to the input background text, wherein the text analysis result comprises one or more of a text genre type, a text language style, a text content type and a text word number;

predicting the input content type after the target input word according to the target input word, the adjacent text and the text analysis result;

and generating a predicted word set corresponding to the target input word according to the target input word, the user language style and the input content type based on the language processing model.

The invention discloses an AIGC-based on-screen intelligent input device, which is applied to an on-screen input method, wherein an input interface of the on-screen input method integrates at least two input areas, each input area corresponds to a corresponding character type, the character types comprise language types and/or coding types, and the device comprises:

The acquisition module is used for acquiring an input character string input by a user based on the input interface, wherein the input character string comprises at least one input character;

the segmentation module is used for segmenting the input character string according to the character types corresponding to all the input areas to obtain a word segmentation result corresponding to the input character string, wherein the word segmentation result corresponding to the input character string comprises at least one character sub-string corresponding to the input character string;

the generation module is used for generating a candidate word set corresponding to the input character string according to a word segmentation result corresponding to the input character string based on a pre-trained language processing model, wherein the candidate word set comprises at least one candidate word;

and the determining module is used for determining a target input word corresponding to the input character string according to the obtained target candidate word, wherein the target candidate word is a candidate word selected by the user from the candidate word set.

In a second aspect of the present invention, the input interface includes a first input area and a second input area, where a character type corresponding to the first input area is a first character type, and a character type corresponding to the second input area is a second character type;

The specific way for the segmentation module to obtain the word segmentation result corresponding to the input character string comprises the following steps of:

In a second aspect of the present invention, the specific manner of generating, by the generating module, the candidate word set corresponding to the input character string according to the word segmentation result corresponding to the input character string based on a pre-trained language processing model includes:

In a second aspect of the present invention, the generating module determines, for each of the character sub-strings, a specific manner of a sub-string encoding manner corresponding to each of the input characters in the character sub-string according to a character type corresponding to the character sub-string, where the specific manner includes:

As an optional implementation manner, in the second aspect of the present invention, the determining module is further configured to determine, according to the obtained history input record corresponding to the user, an input preference corresponding to the user;

the specific way for the generating module to screen at least one target cluster set meeting the preset semantic association condition from all the cluster sets according to the cluster result comprises the following steps:

In a second aspect of the present invention, as an optional implementation manner, the determining module determines, according to the obtained history input record corresponding to the user, a specific manner of the input preference corresponding to the user includes:

In a second aspect of the present invention, as an optional implementation manner, the generating module is further configured to generate, according to the target input word and the user language style, a set of predicted words corresponding to the target input word, where the set of predicted words includes at least one predicted word;

the specific mode of generating the predicted word set corresponding to the target input word by the generating module according to the target input word and the user language style comprises the following steps:

The third aspect of the invention discloses another AIGC-based on-screen intelligent input device, which comprises:

a memory storing executable program code;

a processor coupled to the memory;

the processor calls the executable program codes stored in the memory to execute the AIGC-based on-screen intelligent input method disclosed in the first aspect of the invention.

A fourth aspect of the present invention discloses a computer storage medium storing computer instructions for executing the AIGC-based on-screen intelligent input method disclosed in the first aspect of the present invention when the computer instructions are called.

Compared with the prior art, the embodiment of the invention has the following beneficial effects:

in the embodiment of the invention, the method can be applied to a same screen input method, an input interface of the same screen input method integrates at least two input areas, each input area corresponds to a corresponding character type, the character types comprise language types and/or coding types, and the method can acquire an input character string input by a user based on the input interface, wherein the input character string comprises at least one input character; dividing an input character string according to the character types corresponding to all the input areas to obtain a word segmentation result corresponding to the input character string, wherein the word segmentation result corresponding to the input character string comprises at least one character sub-string corresponding to the input character string; based on a pre-trained language processing model, generating a candidate word set corresponding to an input character string according to a word segmentation result corresponding to the input character string, wherein the candidate word set comprises at least one candidate word; and determining a target input word corresponding to the input character string according to the obtained target candidate word, wherein the target candidate word is a candidate word selected by a user from the candidate word set. Therefore, the method and the device can be applied to the same screen input method, input character strings input by a user can be obtained based on an input interface, the input character strings are segmented according to character types corresponding to all input areas of the same screen input method, word segmentation results corresponding to the input character strings are obtained, then a candidate word set corresponding to the input character strings is generated according to the word segmentation results based on a trained language processing model, and then target input words corresponding to the input character strings are determined according to target candidate words selected by the user, so that the processing efficiency of multi-type characters can be improved, the generation efficiency of the candidate words is improved, intelligent input of various types of characters based on the multi-type characters input by the user is realized, the frequency of switching of input modes of the user among various input modes is further reduced, the character input efficiency is further improved, and the character input experience of the user is facilitated to be improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic flow chart of an AIGC-based on-screen intelligent input method disclosed in an embodiment of the invention;

FIG. 2 is a schematic flow chart of another AIGC-based on-screen intelligent input method disclosed in an embodiment of the invention;

FIG. 3 is a schematic diagram of an input interface of a same screen input method of the same screen intelligent input method based on AIGC disclosed in the embodiment of the invention;

fig. 4 is a schematic diagram of a system architecture of a same-screen input method of the same-screen intelligent input method based on an AIGC according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of a structure of an AIGC-based on-screen intelligent input device according to an embodiment of the present invention;

fig. 6 is a schematic structural diagram of another intelligent input device based on an AIGC according to an embodiment of the present invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The terms first, second and the like in the description and in the claims and in the above-described figures are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, apparatus, article, or article that comprises a list of steps or elements is not limited to only those listed but may optionally include other steps or elements not listed or inherent to such process, method, article, or article.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the invention. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

The invention discloses an AIGC-based on-screen intelligent input method and device, which can be applied to an on-screen input method, can acquire input character strings input by a user based on an input interface, segments the input character strings according to character types corresponding to all input areas of the on-screen input method to obtain word segmentation results corresponding to the input character strings, then generates a candidate word set corresponding to the input character strings according to the word segmentation results based on a trained language processing model, and then determines target input words corresponding to the input character strings according to target candidate words selected by the user, so that the processing efficiency of multi-type characters can be improved, the generation efficiency of the candidate words is improved, the intelligent input of various types of characters based on the multi-type characters input by the user is realized, the frequency of switching the input modes among the various input modes of the user is further reduced, the character input efficiency is further improved, and the character input experience of the user is facilitated. The following will describe in detail.

Example 1

Referring to fig. 1, fig. 1 is a schematic flow chart of an intelligent input method based on an AIGC according to an embodiment of the present invention. The on-screen intelligent input method based on the AIGC described in fig. 1 may be applied to an on-screen input method, where an input interface of the on-screen input method integrates at least two input areas, where each input area corresponds to a corresponding character type, the character type includes a language type and/or an encoding type, and the language type corresponding to each input area may include one of a chinese type, an english type, a japanese type, a korean type, a japanese type and other language types, which is not limited in the embodiment of the present invention; the coding type corresponding to each input area can be determined based on the input mode corresponding to the input area, when the language type corresponding to the input area is a Chinese type, the coding type corresponding to each input area can comprise one of a pinyin coding type, a stroke coding type, a shape-sound coding type and a mixed coding type, and the embodiment of the invention is not limited; for example, when the input interface includes two input areas, a schematic diagram of the input interface of the on-screen input method may be shown in fig. 3; further alternatively, a system architecture schematic diagram of the on-screen input method may be shown in fig. 4, which is not limited by the embodiment of the present invention; the method can also be applied to an AIGC-based on-screen intelligent input device, and the device can comprise one of an input device, an input terminal, an input system and a server, wherein the server comprises a local server or a cloud server, and the embodiment of the invention is not limited. As shown in fig. 1, the on-screen intelligent input method based on the AIGC may include the following operations:

101. And acquiring an input character string input by a user based on the input interface.

In the embodiment of the invention, the input character string comprises at least one input character; optionally, the input character string may include one or more input characters of a character type, which is not limited by the embodiment of the present invention.

102. And according to the character types corresponding to all the input areas, segmenting the input character strings to obtain word segmentation results corresponding to the input character strings.

In the embodiment of the invention, the word segmentation result corresponding to the input character string comprises at least one character sub-string corresponding to the input character string, and each character sub-string comprises at least one input character; optionally, the candidate word corresponding to the character sub-string may be generated based on each character sub-string, or the candidate word corresponding to the character sub-string combination may be generated based on the character sub-string combination, where the character sub-string combination includes a plurality of continuous character sub-strings, and the embodiment of the present invention is not limited.

103. Based on a pre-trained language processing model, generating a candidate word set corresponding to the input character string according to word segmentation results corresponding to the input character string.

In the embodiment of the invention, the candidate word set comprises at least one candidate word; wherein, alternatively, the language processing model can be a natural language processing model constructed based on AIGC (Artificial Intelligence Generated Content, generative artificial intelligence) technology; further alternatively, in the training process of the language processing model, various model optimization algorithms may be used to improve accuracy and efficiency of the model, and exemplary model optimization algorithms may include gradient descent algorithms, back propagation algorithms, and the like, which are not limited by the embodiment of the present invention. Optionally, when the input interface of the on-screen input method further includes a candidate word region, the candidate word set may be displayed to the candidate word region.

104. And determining a target input word corresponding to the input character string according to the obtained target candidate word.

In the embodiment of the invention, the target candidate word is a candidate word selected by a user from a candidate word set; optionally, when the input interface of the on-screen input method further includes a candidate word region, the target candidate word may be obtained based on the candidate word region, which is not limited in the embodiment of the present invention.

Therefore, the method described by implementing the embodiment of the invention can be applied to the same-screen input method, input character strings input by a user can be obtained based on an input interface, the input character strings are segmented according to character types corresponding to all input areas of the same-screen input method, word segmentation results corresponding to the input character strings are obtained, then a candidate word set corresponding to the input character strings is generated according to the word segmentation results based on a trained language processing model, and then target input words corresponding to the input character strings are determined according to target candidate words selected by the user, so that the processing efficiency of multi-type characters can be improved, the generation efficiency of the candidate words is improved, intelligent input of various types of characters based on the multi-type characters input by the user is realized, the frequency of switching input modes among the various input modes of the user is further reduced, the character input efficiency is further improved, and the character input experience of the user is facilitated.

In an alternative embodiment, based on a pre-trained language processing model, generating a candidate word set corresponding to an input character string according to word segmentation results corresponding to the input character string may include the following operations:

based on a language processing model, performing clustering operation on sub-string candidate words corresponding to all character sub-strings according to all semantic relevance, so as to obtain a clustering result, wherein the clustering result comprises a plurality of clustering sets, each clustering set comprises at least one sub-string candidate word corresponding to all character sub-strings, and the semantic relevance among all the sub-string candidate words in each clustering set is in the same preset relevance range;

Screening at least one target cluster set meeting preset semantic association conditions from all cluster sets according to the cluster result;

and determining a candidate word set corresponding to the input character string according to the sub-string candidate words corresponding to all the target cluster sets.

Wherein, the adjacent character sub-strings of each character sub-string may include a previous adjacent character sub-string and/or a next adjacent character sub-string, which is not limited in the embodiment of the present invention; the at least one sub-string candidate word corresponding to all the character sub-strings in each cluster set may include at least one sub-string candidate word corresponding to each character sub-string, or may include at least one sub-string candidate word corresponding to a sub-string combination formed by combining a plurality of character sub-strings. For example, when the character input string includes three character sub-strings of "wo", "sh" and "r", the sub-string candidate words in each cluster set may include sub-string candidate words corresponding to "wo", "sh" and "r", may also include sub-string candidate words corresponding to "wo" and "sh" together, and may also include sub-string candidate words corresponding to "r" together, and may also include sub-string candidate words corresponding to "wo", "sh" and "r" together.

Optionally, screening at least one target cluster set meeting the preset semantic association condition from all cluster sets according to the cluster result may include the following operations:

and screening at least one target cluster set with a preset association range in which the semantic association between all substring candidate words in the cluster set is located as a target association range from all the cluster sets according to the cluster result.

It can be seen that, in this alternative embodiment, for each character sub-string, a sub-string encoding manner corresponding to the character sub-string is determined according to a character type corresponding to each input character in the character sub-string, a sub-string candidate word corresponding to the character sub-string is generated based on a language processing model, then a semantic association degree between the sub-string candidate word corresponding to the character sub-string and a sub-string candidate word corresponding to an adjacent character sub-string is analyzed, then all sub-string candidate words are clustered based on the language processing model and all semantic association degrees to obtain a clustering result, then a target clustering set meeting a preset semantic association condition is screened out according to the clustering result, and a candidate word set corresponding to the input character string is determined according to the sub-string candidate included in the target clustering set, so that the processing accuracy of the input character string can be improved, and further the generation accuracy and screening accuracy of the candidate words are improved, and the association of the candidate word obtained by final screening and the input character string is improved.

In this optional embodiment, optionally, for each character sub-string, determining, according to a character type corresponding to each input character in the character sub-string, a sub-string encoding manner corresponding to the character sub-string may include the following operations:

When the type comparison result of a certain character sub-string is used for indicating that the character sub-string only comprises a pinyin coding type, the sub-string coding mode corresponding to the character sub-string is a pinyin coding mode; when the type comparison result of a certain character sub-string is used for indicating that the character sub-string contains at least two character types, wherein the two character types comprise a pinyin coding type and a stroke coding type, the sub-string coding mode corresponding to the character sub-string is a mixed coding mode determined based on pinyin coding and stroke coding.

It can be seen that, in this alternative embodiment, for each character sub-string, the character types corresponding to each input character in the character sub-string can be compared, when the character types corresponding to all input characters in the character sub-string are the same, the sub-string codes corresponding to the character sub-string are determined to be the coding modes corresponding to the same character types, when the character types corresponding to all input characters in the character sub-string are different, the sub-string codes corresponding to the character sub-string are determined to be the hybrid coding modes corresponding to all the character types included in the character sub-string, so that the determination accuracy of the coding modes of the character sub-string can be improved, thereby improving the processing accuracy and the processing efficiency of multi-type characters, and further being beneficial to improving the generation accuracy of candidate words.

In this alternative embodiment, optionally, the method may further comprise the operations of:

wherein, according to the clustering result, at least one target clustering set meeting the preset semantic association condition is selected from all the clustering sets, and the method comprises the following operations:

and screening at least one target cluster set meeting preset semantic association conditions or matched with the input preference from all cluster sets according to the cluster result and the input preference corresponding to the user.

The input preference corresponding to the user may include one or more of a preferred input mode corresponding to the user, a preferred language style corresponding to the user, a preferred vocabulary corresponding to the user, and a preferred operation corresponding to the user, which is not limited in the embodiment of the present invention.

It can be seen that the optional embodiment can also determine the input preference of the user according to the historical input record of the user, and screen out the target cluster set meeting the preset semantic association condition or matched with the input preference according to the clustering result and the input preference of the user, so that personalized generation of candidate words is realized, the determination accuracy of the input preference of the user can be improved, the screening accuracy of the cluster set is improved, the matching degree of the target cluster set and the user preference is improved, and further the screening accuracy of the candidate words is improved, so that the text input accuracy is improved.

In this alternative embodiment, when the input interface of the on-screen input method further includes a candidate word region, the method may further include:

according to the input preference corresponding to the user, determining a candidate word sequence corresponding to the candidate word set, wherein the candidate word sequence comprises an arrangement sequence corresponding to each candidate word;

And displaying the candidate word set to the candidate word area according to the candidate word sequence corresponding to the candidate word set.

It can be seen that the optional embodiment can also determine the candidate word sequence according to the input preference of the user, so that the candidate word set is displayed to the candidate word area based on the candidate word sequence, the determination accuracy of the candidate word sequence can be improved, the matching degree of the candidate word sequence and the user preference is improved, and further the text input experience of the user is improved.

In this optional embodiment, further optionally, determining, according to the obtained historical input record corresponding to the user, the input preference corresponding to the user may include the following operations:

according to the obtained historical input records corresponding to the users, determining input statistical information corresponding to the historical input records, wherein the input statistical information comprises input frequency of each historical input word, vocabulary type corresponding to each historical input word and sentence pattern structure corresponding to each historical input sentence, and the historical input sentence comprises at least one historical input word;

according to the input statistical information, determining a history input characteristic corresponding to the user;

according to the historical input characteristics and the obtained style corpuses corresponding to the multiple language styles, analyzing the style matching degree between the historical input characteristics and each style corpus;

determining the user language style corresponding to the user according to the language styles corresponding to all the target language databases;

and determining the input preference corresponding to the user according to the language style and the historical input record of the user.

Wherein, by way of example, all language styles may include one or more combinations of popular language styles, web language styles, poetry language styles, written language styles, dialect language styles corresponding to various dialects, and spoken language styles.

It can be seen that the optional embodiment can also determine corresponding input statistics information according to the historical input records, determine the historical input features corresponding to the user according to the input statistics information, analyze the style matching degree between the historical input features and each style corpus according to the historical input features and the style corpuses corresponding to multiple language styles, then screen out at least one target style corpus with the style matching degree higher than or equal to the preset style matching degree, determine the language style of the user according to the language style corresponding to the target style corpus, and determine the input preference corresponding to the user according to the language style of the user and the historical input records, so that the analysis accuracy and analysis comprehensiveness of the historical input records of the user can be improved, the determination accuracy of the language style of the user can be improved, the determination accuracy of the user input preference can be improved, and the screening accuracy of candidate words and the personalized arrangement determination accuracy of an input method can be improved, so that the text input experience of the user is improved.

In this alternative embodiment, further optionally, the method may further comprise the operations of:

the generating a predicted word set corresponding to the target input word according to the target input word and the user language style may include the following operations:

acquiring an input position corresponding to a target input word and an input background text corresponding to the target input word, wherein the input background text is the text where the input position is located;

analyzing an input background text corresponding to a target input word to obtain a text analysis result corresponding to the input background text, wherein the text analysis result comprises one or more of a text genre type, a text language style, a text content type and a text word number;

based on the language processing model, a predicted word set corresponding to the target input word is generated according to the target input word, the user language style and the input content type.

The text type of the input background text is different if the input positions are different, and the text type of the input background text can be one of a text in a text file, a news text, an article text, a question text corresponding to the input position and a chat record text corresponding to the input position, or can be other text types; the adjacent text of the target input word may include at least one adjacent word or at least one adjacent sentence before and after the target input word, which is not limited in the embodiment of the present invention.

It can be seen that, in this optional embodiment, the adjacent text of the target input word can be determined according to the obtained input position and the input background text, the input background text is analyzed to obtain a corresponding text analysis result, the input content type after the target input word is predicted according to the target input word, the adjacent text and the text analysis result, and the predicted word set is generated according to the target input word, the user language style and the input content type based on the language processing model, so that the analysis accuracy of the input background can be improved, the determination accuracy of the predicted content can be improved, the generation accuracy of the predicted word can be improved, the text input efficiency can be improved, and the text input experience of the user can be improved.

after the target input word is input to the input position, displaying a predicted word set corresponding to the target input word to a candidate word area;

or after inputting the target input word to the input position, evaluating the input probability corresponding to each predicted word in the predicted word set;

screening one target predicted word meeting a preset probability condition from the predicted word set according to the input probabilities corresponding to all the predicted words;

and inputting the target predicted word to the next position of the input position corresponding to the target input word.

Therefore, in the alternative embodiment, after the target input word is input to the input position, the predicted word is displayed in the candidate word area, or the target predicted word is screened out according to the estimated input probability corresponding to each predicted word, and the target predicted word is directly input to the next position, so that the display efficiency or the input efficiency of the predicted word can be improved, the analysis accuracy of the predicted word is improved, and further the text input efficiency is improved.

Example two

Referring to fig. 2, fig. 2 is a schematic flow chart of an intelligent input method based on an AIGC according to an embodiment of the present invention. The on-screen intelligent input method based on the AIGC described in fig. 2 may be applied to an on-screen input method, where an input interface of the on-screen input method integrates at least two input areas, where each input area corresponds to a corresponding character type, the character type includes a language type and/or an encoding type, and the language type corresponding to each input area may include one of a chinese type, an english type, a japanese type, a korean type, a japanese type and other language types, which is not limited in the embodiment of the present invention; the coding type corresponding to each input area can be determined based on the input mode corresponding to the input area, when the language type corresponding to the input area is a Chinese type, the coding type corresponding to each input area can comprise one of a pinyin coding type, a stroke coding type, a shape-sound coding type and a mixed coding type, and the embodiment of the invention is not limited; for example, when the input interface includes two input areas, a schematic diagram of the input interface of the on-screen input method may be shown in fig. 3; further alternatively, a system architecture schematic diagram of the on-screen input method may be shown in fig. 4, which is not limited by the embodiment of the present invention; the method can also be applied to an AIGC-based on-screen intelligent input device, and the device can comprise one of an input device, an input terminal, an input system and a server, wherein the server comprises a local server or a cloud server, and the embodiment of the invention is not limited. As shown in fig. 2, the on-screen intelligent input method based on the AIGC may include the following operations:

201. And acquiring an input character string input by a user based on the input interface.

In the embodiment of the invention, the input interface comprises a first input area and a second input area, wherein the character type corresponding to the first input area is a first character type, and the character type corresponding to the second input area is a second character type; for example, when the input interface includes a first input area and a second input area, the schematic diagram of the input interface of the on-screen input method may be shown in fig. 3, in which in fig. 3, columns 1-4 of the input interface are divided into the first input area, columns 5-8 of the input interface are divided into the second input area, where a character type corresponding to the first input area is a pinyin coding type, and a character type corresponding to the second input area is a stroke coding type, and embodiments of the present invention are not limited.

In the embodiment of the present invention, optionally, the input interface of the on-screen input method may further include one or more combinations of a voice input area, a handwriting input area, an expression input area and a picture input area.

202. And classifying all input characters according to the character types corresponding to all input areas to obtain classification results corresponding to the input character strings.

In the embodiment of the invention, the classification result comprises a first type character set and a second type character set, wherein the first type character set is null or the character types of input characters included in the first type character set are all of the first character type, the second type character set is null or the character types of input characters included in the second type character set are all of the second character type, and the first type character set and the second type character set are null when different.

In the embodiment of the invention, for example, when the character type corresponding to the first input area is the pinyin coding type and the character type corresponding to the second input area is the stroke coding type, if the first type character set is not null, the character types of the input characters included in the first type character set are both pinyin coding types, and if the second type character set is not null, the character types of the input characters included in the second type character set are both stroke coding types.

203. And determining character type characteristics corresponding to the input character strings according to the classification result.

In the embodiment of the invention, when a first type character set included in a classification result is not empty, the character type features include first type character features; when the second type character set included in the classification result is not empty, the character type features include second type character features.

204. Based on character type characteristics, the input character strings are analyzed, and the spelling association degree between each input character and the adjacent input characters is obtained.

In the embodiment of the invention, the spelling association degree between each input character and the adjacent input characters is used for representing the spelling relation between the input character and the adjacent input characters.

205. And according to all spelling association degrees, the input character string is segmented, and word segmentation results corresponding to the input character string are obtained.

206. Based on a pre-trained language processing model, generating a candidate word set corresponding to the input character string according to word segmentation results corresponding to the input character string.

207. And determining a target input word corresponding to the input character string according to the obtained target candidate word.

In the embodiment of the present invention, for other detailed descriptions of step 201 and step 206-step 207, please refer to the detailed descriptions of step 101 and step 103-step 104 in the first embodiment, and the detailed descriptions of the embodiment of the present invention are omitted.

Therefore, the method described by implementing the embodiment of the invention can be applied to the same-screen input method, input character strings input by a user can be obtained based on an input interface, the input character strings are segmented according to character types corresponding to all input areas of the same-screen input method, word segmentation results corresponding to the input character strings are obtained, then a candidate word set corresponding to the input character strings is generated according to the word segmentation results based on a trained language processing model, and then target input words corresponding to the input character strings are determined according to target candidate words selected by the user, so that the processing efficiency of multi-type characters can be improved, the generation efficiency of the candidate words is improved, intelligent input of various types of characters based on the multi-type characters input by the user is realized, the frequency of switching input modes among the various input modes of the user is further reduced, the character input efficiency is further improved, and the character input experience of the user is facilitated. In addition, the input characters can be classified according to the character types corresponding to all the input areas to obtain a classification result corresponding to the input character strings, character type features corresponding to the input character strings are determined according to the classification result, the input character strings are analyzed based on the character type features to obtain spelling association degrees between each input character and adjacent input characters, the input character strings are segmented according to the spelling association degrees to obtain corresponding word segmentation results, the classification efficiency and classification accuracy of the input characters can be improved, the determination accuracy and determination comprehensiveness of the character features are improved, the analysis accuracy of association relations among the characters is improved, the word segmentation accuracy of the character strings is improved, the generation efficiency and the generation accuracy of candidate words are improved, and the word input accuracy is improved.

In an alternative embodiment, the input character string is segmented according to all spelling association degrees to obtain a word segmentation result corresponding to the input character string, which may include the following operations:

screening at least one target relevancy lower than or equal to a preset spelling relevancy from all spelling relevancy;

for each target association degree, determining a segmentation point of the input character string between two input characters corresponding to the target association degree;

based on all segmentation points of the input character string, segmenting the input character string to obtain a word segmentation result corresponding to the input character string.

Therefore, the optional embodiment can screen at least one target association degree meeting the condition, determine the segmentation point of the input character string based on the target association degree, and segment the character string based on the segmentation point, so that the analysis accuracy of the spelling association degree can be improved, the determination accuracy of the segmentation point is improved, and the word segmentation accuracy of the character string is improved.

acquiring an error correction record corresponding to the same screen input method;

based on the language processing model and the error correction record, analyzing two input characters corresponding to each target association degree to obtain a character misinput analysis result corresponding to each target association degree;

For each target association, when the character misinput analysis result corresponding to the target association is used for indicating that the misinput probability of two input characters corresponding to the target association is higher than the preset probability, correcting errors of the two input characters corresponding to the target association is carried out based on the language processing model and the input character string, the corrected input character string is obtained, the input character string is analyzed again based on character type characteristics, and the spelling association between each input character and adjacent input characters is obtained.

It can be seen that, according to the obtained error correction record and language processing model, the alternative embodiment can also analyze two input characters corresponding to each target association degree to obtain a corresponding character error input analysis result, if the error input possibility obtained by analysis is higher than the preset possibility, error correction is performed on the two input characters corresponding to the target association degree to obtain an input character string after error correction, and the spelling association degree is re-analyzed, so that the error input analysis accuracy of the input character can be improved, and the error correction accuracy of the character string is improved, and the text input accuracy is further improved.

In another alternative embodiment, the method may further comprise the operations of:

Determining a user corresponding input requirement, wherein the input requirement can comprise one or more of an input mode requirement, an input language requirement, an input efficiency requirement and an input preference requirement;

determining the number of input areas integrated by an input interface of the same-screen input method and the character type corresponding to each input area according to the input requirement corresponding to the user;

determining the layout characteristics of the input areas corresponding to each input area according to the character types corresponding to each input area;

and determining interface layout corresponding to the same-screen input method according to the input requirements, the number of the input areas and the layout characteristics of the input areas corresponding to each input area.

Therefore, according to the alternative embodiment, the number of the input areas and the character types corresponding to each input area can be determined according to the input requirements of the user, the layout characteristics of the corresponding input areas are determined according to the character types corresponding to each input area, the interface layout corresponding to the same screen input method is determined according to the input requirements, the number of the input areas and the layout characteristics of each input area, the analysis accuracy of the user requirements can be improved, the analysis accuracy of the layout characteristics is improved, the determination flexibility and the determination accuracy of the interface layout corresponding to the same screen input method are improved, the modular design of the same screen input method is facilitated, the expansion maintenance efficiency and the expansion maintenance flexibility of the same screen input method are improved, and the text input efficiency of the user is further improved.

In the embodiment of the present invention, when the on-screen intelligent input method based on the AIGC is applied to the on-screen input method, and the on-screen input method includes an input area corresponding to pinyin codes and an input area corresponding to stroke codes, a system architecture diagram of the on-screen input method may be as shown in fig. 4, and a system function flow corresponding to the system architecture may include the following flows:

the data collection and processing module can collect a large amount of language data and user data and preprocess the collected data; the model training and optimizing module can perform model training and model optimization on the language processing model based on the preprocessed data; the system deployment and maintenance module can deploy the optimized language processing model to the intelligent terminal. The pinyin and stroke coding module can realize pinyin input by adopting a pinyin coding mode (such as a 9-key coding mode) and stroke input by adopting a stroke coding mode (such as a five-stroke coding mode); the personalized recommendation module can assist in generating candidate words and determining the sequence of the candidate words according to the input history record of the user, so that a personalized recommendation function is realized; the intelligent error correction and prediction module can realize the intelligent error correction function of the input character string and the prediction function of generating candidate words and predicted words; the multi-task processing module can support various input modes, such as a keyboard input mode, a voice input mode, a handwriting input mode and the like, so as to meet different input requirements of users; and an interface design module: the design is simple and clear, and the operation is easy, so that the user can conveniently use the one-screen input of pinyin and strokes; and a user feedback mechanism module: and collecting the use condition and feedback opinion of the user on the input method so as to perfect the function of the input method. The operating system is applied on the physical layer, and the multitasking module and the interface design module of the input method can be adjusted according to the difference of the operating systems.

Example III

Referring to fig. 5, fig. 5 is a schematic structural diagram of an intelligent input device with a screen based on an AIGC according to an embodiment of the present invention. The on-screen intelligent input device based on the AIGC described in fig. 5 may be applied to an on-screen input method, where an input interface of the on-screen input method integrates at least two input areas, where each input area corresponds to a corresponding character type, the character type includes a language type and/or an encoding type, and the language type corresponding to each input area may include one of a chinese type, an english type, a japanese type, a korean type, a japanese type, and other language types, where embodiments of the present invention are not limited; the coding type corresponding to each input area can be determined based on the input mode corresponding to the input area, when the language type corresponding to the input area is a Chinese type, the coding type corresponding to each input area can comprise one of a pinyin coding type, a stroke coding type, a shape-sound coding type and a mixed coding type, and the embodiment of the invention is not limited; for example, when the input interface includes two input areas, a schematic diagram of the input interface of the on-screen input method may be shown in fig. 3; further alternatively, a system architecture schematic diagram of the on-screen input method may be shown in fig. 4, which is not limited by the embodiment of the present invention; the apparatus may include one of an input device, an input terminal, an input system, and a server, where the server includes a local server or a cloud server, and embodiments of the present invention are not limited. As shown in fig. 5, the AIGC-based on-screen intelligent input device may include:

An obtaining module 301, configured to obtain an input string input by a user based on an input interface, where the input string includes at least one input character;

the segmentation module 302 is configured to segment an input string according to the character types corresponding to all the input areas, so as to obtain a word segmentation result corresponding to the input string, where the word segmentation result corresponding to the input string includes at least one character sub-string corresponding to the input string;

the generating module 303 is configured to generate, based on a pre-trained language processing model, a candidate word set corresponding to an input character string according to a word segmentation result corresponding to the input character string, where the candidate word set includes at least one candidate word;

the determining module 304 is configured to determine, according to the obtained target candidate word, a target input word corresponding to the input character string, where the target candidate word is a candidate word selected by the user from the candidate word set.

Therefore, the device described by the embodiment of the invention can be applied to the same screen input method, can acquire the input character string input by the user based on the input interface, segments the input character string according to the character types corresponding to all input areas of the same screen input method, obtains the word segmentation result corresponding to the input character string, then generates the candidate word set corresponding to the input character string according to the word segmentation result based on the trained language processing model, and then determines the target input word corresponding to the input character string according to the target candidate word selected by the user, thereby improving the processing efficiency of the multi-type character, further improving the generation efficiency of the candidate word, realizing the intelligent input of various types of characters based on the multi-type character input by the user, further reducing the frequency of switching the input modes among the various input modes by the user, further improving the character input efficiency and being beneficial to improving the character input experience of the user.

In an alternative embodiment, the input interface includes a first input area and a second input area, the first input area corresponds to a first character type, and the second input area corresponds to a second character type;

the specific manner of the segmentation module 302 for segmenting the input character string according to the character types corresponding to all the input areas to obtain the word segmentation result corresponding to the input character string may include:

classifying all input characters according to the character types corresponding to all input areas to obtain classification results corresponding to the input character strings; the classification result comprises a first type character set and a second type character set, wherein the first type character set is null or the character types of input characters included in the first type character set are all of a first character type, the second type character set is null or the character types of input characters included in the second type character set are all of a second character type, and the first type character set and the second type character set are not null at the same time;

Based on character type characteristics, analyzing input character strings to obtain spelling association degrees between each input character and adjacent input characters;

and according to all spelling association degrees, the input character string is segmented, and word segmentation results corresponding to the input character string are obtained.

Therefore, the device described by implementing the alternative embodiment can classify all input characters according to the character types corresponding to all input areas to obtain the classification result corresponding to the input character strings, determine the character type characteristics corresponding to the input character strings according to the classification result, analyze the input character strings based on the character type characteristics to obtain the spelling association degree between each input character and the adjacent input characters, and then segment the input character strings according to the spelling association degree to obtain the corresponding word segmentation result, so that the classification efficiency and classification accuracy of the input characters can be improved, the determination accuracy and the determination comprehensiveness of the character characteristics are improved, the analysis accuracy of the association relation among the characters is improved, the word segmentation accuracy of the character strings is improved, the generation efficiency and the generation accuracy of candidate words are improved, and the word input accuracy is improved.

In another alternative embodiment, the specific manner of generating, by the generating module 303, the candidate word set corresponding to the input character string according to the word segmentation result corresponding to the input character string based on the pre-trained language processing model may include:

It can be seen that, the device described in this optional embodiment is capable of determining, for each character sub-string, a sub-string encoding manner corresponding to the character sub-string according to a character type corresponding to each input character in the character sub-string, generating a sub-string candidate word corresponding to the character sub-string based on a language processing model, analyzing a semantic association degree between the sub-string candidate word corresponding to the character sub-string and a sub-string candidate word corresponding to an adjacent character sub-string, clustering all the sub-string candidate words based on the language processing model and all the semantic association degrees to obtain a clustering result, screening a target cluster set meeting a preset semantic association condition according to the clustering result, and determining a candidate word set corresponding to the input character string according to the sub-string candidate included in the target cluster set, so that the processing accuracy of the input character string can be improved, further improving the generation accuracy and screening accuracy of the candidate word, and further being beneficial to improving the association of the finally screened candidate word and the input character string.

In this optional embodiment, optionally, for each character sub-string, the specific manner of determining, by the generating module 303, the sub-string encoding manner corresponding to the character sub-string according to the character type corresponding to each input character in the character sub-string may include:

It can be seen that, the device described in this optional embodiment may also be configured to compare, for each character sub-string, a character type corresponding to each input character in the character sub-string, determine, when the character types corresponding to all input characters in the character sub-string are the same, a sub-string code corresponding to the character sub-string as a coding manner corresponding to the same character type, and determine, when the character types corresponding to all input characters in the character sub-string are different, a sub-string code corresponding to the character sub-string as a hybrid coding manner corresponding to all character types included in the character sub-string, so that accuracy in determining a coding manner of the character sub-string can be improved, thereby improving accuracy and efficiency in processing multi-type characters, and further being beneficial to improving accuracy in generating candidate words.

In this optional embodiment, optionally, the determining module 304 is further configured to determine, according to the obtained historical input record corresponding to the user, an input preference corresponding to the user;

the specific manner of the generating module 303 selecting at least one target cluster set that meets the preset semantic association condition from all cluster sets according to the cluster result may include:

It can be seen that the device described by implementing the alternative embodiment can also determine the input preference of the user according to the historical input record of the user, and screen out the target cluster set meeting the preset semantic association condition or matched with the input preference according to the clustering result and the input preference of the user, so as to implement personalized generation of the candidate word, improve the determination accuracy of the input preference of the user, and improve the screening accuracy of the cluster set, thereby improving the matching degree of the target cluster set and the user preference, further being beneficial to further improving the screening accuracy of the candidate word, and improving the text input accuracy.

In this optional embodiment, further optionally, the specific manner of determining, by the determining module 304, the input preference corresponding to the user according to the obtained history input record corresponding to the user may include:

It can be seen that the device described in this optional embodiment may further determine corresponding input statistics according to the history input record, determine the history input feature corresponding to the user according to the input statistics, analyze the style matching degree between the history input feature and each style corpus according to the history input feature and the style corpuses corresponding to multiple language styles, then screen out at least one target style corpus with the style matching degree higher than or equal to the preset style matching degree, determine the language style of the user according to the language style corresponding to the target style corpus, and determine the input preference corresponding to the user according to the language style of the user and the history input record, thereby improving the analysis accuracy and analysis comprehensiveness of the history input record of the user, further improving the determination accuracy of the language style of the user, and further improving the screening accuracy of the candidate words and the determination accuracy of the personalized arrangement of the input method, so as to improve the text input experience of the user.

In this optional embodiment, further optionally, the generating module 303 is further configured to generate, according to the target input word and the language style of the user, a set of predicted words corresponding to the target input word, where the set of predicted words includes at least one predicted word;

The specific manner of generating the predicted word set corresponding to the target input word by the generating module 303 according to the target input word and the user language style may include:

It can be seen that the device described in this optional embodiment may further determine, according to the obtained input position and the input background text, the adjacent text of the target input word, analyze the input background text to obtain a corresponding text analysis result, predict the input content type after the target input word according to the target input word, the adjacent text and the text analysis result, and generate the predicted word set according to the target input word, the user language style and the input content type based on the language processing model, so as to improve the analysis accuracy of the input background, thereby improving the determination accuracy of the predicted content, further improving the generation accuracy of the predicted word, and being beneficial to improving the text input efficiency so as to improve the text input experience of the user.

Example IV

Referring to fig. 6, fig. 6 is a schematic structural diagram of another intelligent input device based on an AIGC according to an embodiment of the present invention. As shown in fig. 6, the AIGC-based on-screen intelligent input device may include:

a memory 401 storing executable program codes;

a processor 402 coupled with the memory 401;

the processor 402 invokes executable program codes stored in the memory 401 to perform the steps in the on-screen intelligent input method based on the AIGC described in the first embodiment of the present invention or the second embodiment of the present invention.

Example five

The embodiment of the invention discloses a computer storage medium which stores computer instructions for executing the steps in the AIGC-based on-screen intelligent input method described in the first or second embodiment of the invention when the computer instructions are called.

Example six

An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the on-screen intelligent input method based on AIGC described in the first embodiment or the second embodiment.

The apparatus embodiments described above are merely illustrative, wherein the modules illustrated as separate components may or may not be physically separate, and the components shown as modules may or may not be physical, i.e., may be located in one place, or may be distributed over a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

From the above detailed description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course by means of hardware. Based on such understanding, the foregoing technical solutions may be embodied essentially or in part in the form of a software product that may be stored in a computer-readable storage medium including Read-Only Memory (ROM), random-access Memory (Random Access Memory, RAM), programmable Read-Only Memory (Programmable Read-Only Memory, PROM), erasable programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM), one-time programmable Read-Only Memory (OTPROM), electrically erasable programmable Read-Only Memory (EEPROM), compact disc Read-Only Memory (Compact Disc Read-Only Memory, CD-ROM) or other optical disc Memory, magnetic disc Memory, tape Memory, or any other medium that can be used for computer-readable carrying or storing data.

Finally, it should be noted that: the embodiment of the invention discloses an AIGC-based on-screen intelligent input method and device, which are disclosed by the embodiment of the invention and are only used for illustrating the technical scheme of the invention, but not limiting the technical scheme; although the invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art will understand that; the technical scheme recorded in the various embodiments can be modified or part of technical features in the technical scheme can be replaced equivalently; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims

1. The AIGC-based on-screen intelligent input method is characterized in that the method is applied to an on-screen input method, an input interface of the on-screen input method integrates at least two input areas, each input area corresponds to a corresponding character type, the character types comprise language types and/or coding types, and the method comprises the following steps:

determining a target input word corresponding to the input character string according to the obtained target candidate word, wherein the target candidate word is a candidate word selected by the user from the candidate word set;

the generating, based on a pre-trained language processing model, a candidate word set corresponding to the input character string according to a word segmentation result corresponding to the input character string includes:

2. The AIGC-based on-screen intelligent input method according to claim 1, wherein the input interface comprises a first input area and a second input area, wherein the character type corresponding to the first input area is a first character type, and the character type corresponding to the second input area is a second character type;

3. The AIGC-based on-screen intelligent input method of claim 1 or 2, wherein for each of the character sub-strings, determining a sub-string encoding mode corresponding to the character sub-string according to a character type corresponding to each of the input characters in the character sub-string, includes:

4. The AIGC-based on-screen intelligent input method of claim 1 or 2, further comprising:

5. The AIGC-based on-screen intelligent input method of claim 4, wherein the determining the input preference corresponding to the user according to the obtained history input record corresponding to the user comprises:

6. The AIGC-based on-screen intelligent input method of claim 5, further comprising:

7. An AIGC-based on-screen intelligent input device, wherein the device is applied to an on-screen input method, an input interface of the on-screen input method integrates at least two input areas, each input area corresponds to a corresponding character type, the character types include language types and/or coding types, and the device includes:

the determining module is used for determining a target input word corresponding to the input character string according to the obtained target candidate word, wherein the target candidate word is a candidate word selected by the user from the candidate word set;

the specific mode of generating the candidate word set corresponding to the input character string based on the pre-trained language processing model according to the word segmentation result corresponding to the input character string comprises the following steps:

8. An AIGC-based on-screen intelligent input device, the device comprising:

a memory storing executable program code;

a processor coupled to the memory;

the processor invokes the executable program code stored in the memory to perform the AIGC-based on-screen intelligent input method of any of claims 1-6.

9. A computer storage medium storing computer instructions which, when invoked, are operable to perform an AIGC-based on-screen intelligent input method according to any one of claims 1-6.