CN113313001B - Semantic model-based handwriting input optimization method, system and medium - Google Patents
Semantic model-based handwriting input optimization method, system and medium Download PDFInfo
- Publication number
- CN113313001B CN113313001B CN202110558066.2A CN202110558066A CN113313001B CN 113313001 B CN113313001 B CN 113313001B CN 202110558066 A CN202110558066 A CN 202110558066A CN 113313001 B CN113313001 B CN 113313001B
- Authority
- CN
- China
- Prior art keywords
- vector
- sentence
- semantic model
- individual
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/32—Digital ink
- G06V30/36—Matching; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Machine Translation (AREA)
- User Interface Of Digital Computer (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method, a system and a medium for optimizing handwriting input based on a semantic model, wherein the method comprises the following steps: acquiring data and application scenes of individual voice input, and creating a semantic model according to the data of the individual voice input; establishing a mapping relation between an application scene of individual voice input and a semantic model; performing classification operation on the semantic model according to the mapping relation, and extracting a plurality of sentence vectors according to the classification operation result; configuring a weight relation, splitting sentence vectors according to the weight relation, and generating a plurality of word vectors; the invention can generate and train a model through personalized voice input, extract sentence vectors, distinguish the occurrence frequency of the sentence vectors, and recommend sentence vectors or word vectors for individuals according to time domain and space domain when the individuals input by handwriting, thereby improving the accuracy of handwriting input and greatly improving the efficiency of handwriting input.
Description
Technical Field
The invention relates to the technical field of handwriting input, in particular to a semantic model-based handwriting input optimization method, a semantic model-based handwriting input optimization system and a semantic model-based handwriting input medium.
Background
The prior art scheme is calculated by calculating the edit distance of the text, the similarity coefficient or the conventional weighting technology of TF-IDF (term frequency-inverse document frequency) information retrieval data mining, and the like.
The Edit Distance, english called Edit Distance, also called Levenshtein Distance, refers to the minimum number of Edit operations between two strings required to change from one to the other, and if their distances are larger, indicating that they are different, the allowable Edit operations include replacing one character with another, inserting a character, and deleting a character.
And the similarity coefficient is used for comparing the similarity and the difference between the limited sample sets, and the larger the similarity coefficient value is, the higher the sample similarity is. The calculation mode is very simple, the value obtained by dividing the intersection of two samples by the union is 1 when the two samples are completely consistent, and the value is 0 when the two samples are completely different.
The third scheme is to directly calculate the similarity of two vectors in the TF matrix, and in fact, solve the cosine value of the included angle of the two vectors, namely the dot product divided by the modular length of the two vectors.
The existing text similarity calculation method is based on word similarity calculation methods, semantic understanding for reference to working conditions, personal preference or family factors for text input is not carried out, sentences similar in semantic or syntactic structure cannot be distinguished only based on keyword information on the surface layers of sentences, and in intelligent screens of various offices or universities, the optimization method of handwriting input based on a semantic model has important influence on accelerating text input speed and associated word recommendation.
Disclosure of Invention
The invention mainly solves the problem that the existing handwriting input does not refer to the use condition, working factors and personal preference of a user to recommend related words.
In order to solve the technical problems, the invention adopts a technical scheme that: the method for optimizing handwriting input based on the semantic model comprises the following steps:
acquiring data and application scenes of individual voice input, and creating a semantic model according to the data of the individual voice input;
establishing a mapping relation between the application scene of the individual voice input and the semantic model;
performing classification operation on the semantic model according to the mapping relation, and extracting a plurality of sentence vectors according to the classification operation result;
configuring a weight relation, splitting the sentence vectors according to the weight relation, and generating a plurality of word vectors;
and acquiring individual handwriting operation, and executing associated recommendation on the individual handwriting operation according to the weight relation between the word vector and the sentence vector.
Preferably, the step of obtaining the data and the application scene of the individual voice input and creating the semantic model according to the data of the individual voice input further comprises:
configuring a first threshold and a threshold interval, performing data characteristic analysis operation on the data input by the individual voice, and taking the data meeting the first threshold as a characteristic vector;
extracting the characteristic vector to create a semantic model.
Preferably, the step of establishing a mapping relationship between the application scenario of the handwriting input of the individual and the semantic model further includes:
and acquiring the characteristic vector of the application scene, and establishing a mapping relation between the characteristic vector and the corresponding application scene and semantic model.
Preferably, the step of performing a classification operation on the semantic model according to the mapping relationship, and extracting a plurality of sentence vectors according to the classification operation result further includes:
calculating the use times of the sentence vectors in a time domain and a space domain, and executing classification operation according to a threshold value interval where the use times of the sentence vectors are located;
the threshold interval comprises a first interval, a second interval and a third interval, and sentence vector recommendation strategies, word vector recommendation strategies or association recommendation strategy operations are classified and executed according to the interval where the threshold interval of the sentence vector is located.
Preferably, the step of executing the sentence vector recommendation policy further includes: if the threshold interval of the sentence vector is the first interval, the sentence vector is used as a high-frequency sentence vector, the sentence head of the high-frequency sentence vector is identified as a high-frequency word vector, and when the high-frequency word vector is triggered by the handwriting input of the individual, the high-frequency sentence vector corresponding to the sentence head is recommended to the individual.
Preferably, the step of executing the word vector recommendation policy further includes: if the threshold interval of the sentence vector is the second interval, the sentence vector is used as an intermediate frequency sentence vector, the intermediate frequency sentence vector is split into a plurality of word vectors and used as an intermediate frequency word vector, and when the intermediate frequency word vector is triggered by the handwriting input of the individual, the intermediate frequency word vector corresponding to the intermediate frequency sentence vector is recommended to the individual.
Preferably, the step of associatively recommending a policy further includes: and if the threshold interval of the sentence vector is a third interval, taking the sentence vector as a low-frequency sentence vector, splitting the low-frequency sentence vector into a plurality of word vectors and taking the word vectors as low-frequency word vectors, and recommending the medium-frequency word vectors associated with the low-frequency word vectors for the individual when the low-frequency word vectors are triggered by the handwriting input of the individual.
Preferably, the step of executing the sentence vector recommendation policy, the word vector recommendation policy, or the association recommendation policy, respectively, further includes: and performing unsupervised self-adaptive training on the semantic model according to the times of executing the sentence vector recommendation strategy, the word vector recommendation strategy or the association recommendation strategy.
The invention also provides a handwriting input optimization system based on the semantic model, which comprises: the system comprises an acquisition module, a semantic module, a voice recognition module, a mapping module, a classification module, a splitting module, a handwriting recognition module and a recommendation module;
the acquisition module is used for acquiring data input by individual voice and sending the data to the semantic module;
the semantic module is used for creating a semantic model according to the data;
the voice recognition module is used for recognizing voice input of the individual in an application scene;
the mapping module is used for establishing a mapping relation between the semantic model and the application scene;
the classification module is used for performing classification on the application scene according to the mapping relation and extracting a plurality of sentence vectors;
the splitting module is used for splitting the sentence vectors according to the weight relation to generate a plurality of word vectors;
the handwriting recognition module is used for recognizing the handwriting input of the individual and sending the handwriting input recognition result to the recommendation module;
and the recommendation module is used for executing association recommendation on the individual according to the weight relation between the word vector and the sentence vector.
The present invention also provides a computer readable storage medium storing a computer program which, when executed by a processor, implements any of the semantic model based handwriting input optimization method steps.
The beneficial effects of the invention are as follows:
1. the method for optimizing the handwriting input based on the semantic model can realize generation and training of the model through personalized voice input, extract sentence vectors, distinguish the occurrence frequency of the sentence vectors, recommend the sentence vectors or the word vectors for individuals according to a time domain and a space domain when the individuals input the handwriting, improve the accuracy of the handwriting input and greatly improve the efficiency of the handwriting input.
2. The semantic model-based handwriting input optimizing system can realize the establishment of a semantic model, acquire voice input and establish a mapping relation between an application scene and the voice input after the acquisition.
3. The optimized medium for handwriting input based on the semantic model can realize the division of threshold intervals and execute the recommendation strategy of the corresponding interval after the division.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of a method for optimizing handwriting input based on a semantic model according to embodiment 1 of the present invention;
FIG. 2 is a schematic diagram of a semantic model based handwriting input optimization system according to embodiment 2 of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made apparent and fully in view of the accompanying drawings, in which some, but not all embodiments of the invention are shown. 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.
In the description of the present invention, "individual" is used; "semantic model"; "Voice input"; "application scenario"; "sort operations"; "weight relationship"; "sentence vector"; "word vector"; "associated recommendations"; "data property resolution"; "feature vector"; "time domain"; "spatial domain"; "threshold interval"; "first interval"; "second interval"; "third interval"; "sentence vector recommendation strategy"; "word vector recommendation strategy"; "associative recommendation policy"; "high frequency sentence vector"; "high frequency word vector"; "intermediate frequency sentence vector"; "intermediate frequency word vector"; "Low-frequency sentence vector"; an "acquisition module"; "semantic Module"; a "speech recognition module"; "mapping Module"; a "classification module"; "split Module"; a handwriting recognition module; the term "recommendation module" should not be limited to a literal meaning, but should be construed broadly in accordance with the technical means available to those skilled in the art.
Example 1
The embodiment of the invention provides a method for optimizing handwriting input based on a semantic model, referring to fig. 1, comprising the following steps:
s100, acquiring data and an application scene of individual voice input, and creating a semantic model according to the data of the individual voice input;
the step S100 specifically includes:
for detailed description, the embodiment provides an example of "when a train of lights slowly passes over a longcourt, where a henry of 32 years old is there, a deep look is a shadow of 22 years old of the user himself, or that train is sent away in a hurry with time, if the user uses the data satisfying the threshold of the number of times of use as a feature vector, for example, the user uses the data as an individual and the user refers to the same section of speech input for a plurality of times. Still open to another continent, 33 years of henry is there, the deep eye is looking to lean and emerge the model of the age of 25 years of oneself, the distant is expected to be naturally beautiful, and the short stay of the train is more beautiful looking back like years, at that time, the light is looking back to home, 35 years of henry is there, the deep eye is looking to outline the moving picture of the age of 29 years of oneself, the evolutionary rut silently takes away the noise belonging to four cities, but directs the people who follow all the way to seek the time that the section is elapsed. When the years have the tear and the rest of the body, the henry of 37 years is in the place, the deep eye looks and tries to return to the original point in the past, and the station from which the station starts remembers the bag of the back, the seventeen years is the situation, for example, in a high school, a teacher emphasizes the section for a plurality of times, the data characteristics of the section can be analyzed, the henry of two words appears for a plurality of times in the section, and the henry can be used as a feature vector, and a semantic model can be created or supplemented around the feature vector.
S200, establishing a mapping relation between an application scene of individual voice input and a semantic model;
the step S200 specifically includes:
because in the above step S100, we propose an application scenario of a semantic model, it can be known that the application scenario is very important for the related operation of the semantic model, after extracting the feature vector, we need to distinguish the application scenario of the feature vector, i.e. "henry" and "a certain class of a college" establish a mapping relationship, and in the semantic model, the mapping relationship is provided with a spatial domain and a temporal domain, i.e. in a certain time period, the mapping relationship is applicable.
S300, performing classification operation on the semantic model according to the mapping relation, and extracting a plurality of sentence vectors according to the classification operation result;
s301, calculating the use times of the sentence vector in the time domain, for example, a teacher often emphasizes a sentence in a large section as a central idea or as a summary, so that the use times of the sentence vector are calculated in the time domain, and statistics is performed on the use frequency, for example, when the time domain is 16:00-16:30, the teacher emphasizes three times of ' distant hopes of being naturally good ', and when the train stays briefly more like the beautiful look-back of years ', the threshold interval of the use times of the sentence vector can be identified, and it is required to say that in the half hour time interval in the time domain, if the sentence vector appears three times, the threshold interval of the sentence vector can be identified as the first interval, if the sentence vector appears twice, the threshold interval of the sentence vector can be identified as the second interval, and if the sentence vector appears once, the threshold interval of the sentence vector can be identified as the third interval.
S302, if the sentence vector is in the first interval, executing a sentence vector recommendation strategy; if the sentence vector is in the second interval, executing a word vector recommendation strategy; and if the sentence vector is in the third interval, executing the associative recommendation strategy.
S400, configuring a weight relation, splitting sentence vectors according to the weight relation, and generating a plurality of word vectors;
the step S400 specifically includes: in this embodiment, since a single chinese character is input at the time of handwriting input of an individual, in semantic recognition, sentence vectors need to be decomposed into a plurality of word vectors, that is, "henry," "time," "court," "train," etc. in step S100 may be used as word vectors, for example, in one segment in step S100, "henry" may be used as the word vector with the highest occurrence frequency, that is, as a representative of the segment.
S500, acquiring individual handwriting operation, and executing associated recommendation on the individual handwriting operation according to the weight relation between the word vector and the sentence vector;
the step S500 specifically includes:
s501, executing sentence vector recommendation strategies;
if the threshold of the sentence vector is the first section, the sentence vector is regarded as a high-frequency sentence vector, for example, "the train in time slowly passes through the headpitch, the henry of 32 years old is there, the deep looking at the train is the shadow of 22 years old, or the train is sent away in a hurry to pass the passenger with the time". The sentence head in the sentence is 'time-out', namely when a teacher in a college triggers the sentence head by handwriting, the sentence head is automatically recommended to be supplemented by a later section of sentence according to a sentence vector recommendation strategy.
S502, executing a word vector recommendation strategy;
if the sentence vector is the second interval, the sentence vector is used as an intermediate frequency sentence vector, for example, a 'calm open to another continent, 33 years old henry is there, deep looking is expected to be thin and emerge in a 25 years old model, a distant place is expected to be naturally good, a short stay of a train is more beautiful looking back as years' this section is an intermediate frequency sentence vector, the sentence vector is split into a plurality of word vectors with parallel weights, when a teacher of a college uses handwriting to trigger one of the word vectors, the parallel word vectors are recommended for the teacher of the college, and through the method, the associative writing of the associated intermediate frequency word vectors in the teaching process is realized, and the writing time is reduced.
S503, executing an association recommendation strategy;
if the sentence vector is the third interval, the sentence vector is used as a low-frequency sentence vector, for example, "when the years of age have the tear-containing messages turned around, the henry of 37 years old is there, the deep eye looks to try to return to the origin in the past, when the station from which the station starts remembers to back up the bag, the seventeen years old splits the low-frequency sentence vector into a plurality of low-frequency word vectors, and when a teacher in a college triggers a low-frequency word by handwriting, for example," turn around "," far-end "," bag "," henry "and the like, the medium-frequency word vector associated with" turn around "," far-end "," bag "," henry "and the like is recommended.
S504, perfecting the semantic model by continuously executing sentence vector recommendation strategies, word vector recommendation strategies or associative recommendation strategies, and performing unsupervised self-adaptive learning on the semantic model.
Example 2
An embodiment of the present invention provides a semantic model-based handwriting input optimization system, referring to fig. 2, including: the system comprises an acquisition module, a semantic module, a voice recognition module, a mapping module, a classification module, a splitting module, a handwriting recognition module and a recommendation module.
The acquisition module is used for acquiring data input by individual voice and sending the data to the semantic module.
The semantic module is used for creating a semantic model according to the data.
The acquisition module performs a data characteristic analysis operation on the data input by the individual, and the acquisition module uses the data meeting the threshold of the number of times of use as a characteristic vector, for example, the user is used as the individual, and the same section of speech is mentioned in the speech recognition module for a plurality of times, so for the sake of detailed description, the embodiment provides an example of "the train with time gradually passes through the longcourt, the henry of 32 years old is there, the looking deep in the past is the shadow of 22 years old of the user, or the train is sent away with time to hurry to get away. Still open to another continent, 33 years of henry is there, the deep eye is looking to lean and emerge the model of the age of 25 years of oneself, the distant is expected to be naturally beautiful, and the short stay of the train is more beautiful looking back like years, at that time, the light is looking back to home, 35 years of henry is there, the deep eye is looking to outline the moving picture of the age of 29 years of oneself, the evolutionary rut silently takes away the noise belonging to four cities, but directs the people who follow all the way to seek the time that the section is elapsed. When the years have the tear and the rest of the body, the henry of 37 years is in the place, the deep eye looks and tries to return to the original point in the past, and the station from which the station starts remembers the bag of the back, the seventeen years is the same as in a high school, for example, a teacher can analyze the data characteristics of the section through the voice recognition module to emphasize the section for multiple times, and the henry can be used as a feature vector when the henry is presented for multiple times in the section, and the semantic module can create or supplement a semantic model around the feature vector.
The mapping module establishes a mapping relation between the application scene of the voice input acquired by the individual voice recognition module and the semantic model.
Because we propose an application scenario of a semantic model, it can be known that the application scenario is very important for the related operation of the semantic model, after the feature vector is extracted, the application scenario of the feature vector needs to be distinguished, and the mapping module establishes a mapping relationship between "henry" and "a class of a college", and in the semantic model, the mapping relationship of the mapping module is provided with a spatial domain and a temporal domain, that is, the mapping relationship in the mapping module is applicable only in a certain time period.
The classification module performs classification operation on the semantic model according to the mapping relation, and extracts a plurality of sentence vectors;
the classification module calculates the number of times of use of the sentence vector in the time domain, for example, a teacher often emphasizes a sentence in a large section as a central idea or as a summary, so the classification module calculates the number of times of use of the sentence vector in the time domain, for example, when the time domain is 16:00-16:30, the teacher emphasizes three times of "distant expectations are naturally good", and when the train stays briefly more like a beautiful look-back of years ", the classification module can identify the threshold interval of the number of times of use of the sentence vector, it is to be noted that in the half-hour interval in the time domain, if the sentence vector appears three times, the classification module can identify the threshold interval of the sentence vector as the first interval, if the sentence vector appears twice, the classification module can identify the threshold interval of the sentence vector as the second interval, and if the sentence vector appears once, the classification module can identify the threshold interval of the sentence vector as the third interval.
If the sentence vector is in the first interval, the recommendation module executes a sentence vector recommendation strategy; if the sentence vector is in the second interval, the recommendation module executes a word vector recommendation strategy; if the sentence vector is in the third interval, the recommendation module executes the associative recommendation strategy.
The splitting module splits the sentence vectors according to the weight relation to generate a plurality of word vectors;
in this embodiment, when an individual performs handwriting input to the handwriting recognition module, the handwriting recognition module needs to decompose the sentence vector into a plurality of word vectors, that is, "henry," "time," "court," "train," etc. as word vectors, for example, "henry" as the word vector with the highest occurrence frequency, that is, as a representative of the section of speech in semantic recognition.
When an individual performs handwriting input to the handwriting recognition module, the recommendation module performs association recommendation on the individual according to the weight relation between the word vector and the sentence vector;
the recommendation module executes sentence vector recommendation strategies;
if the threshold of the sentence vector is the first section, the sentence vector is regarded as a high-frequency sentence vector, for example, "the train in time slowly passes through the headpitch, the henry of 32 years old is there, the deep looking at the train is the shadow of 22 years old, or the train is sent away in a hurry to pass the passenger with the time". The sentence head in the sentence is 'time-out', namely when a teacher in a college triggers the sentence head by using the handwriting recognition module, the recommendation module automatically recommends that the later sentence is full of the sentence head according to the sentence vector recommendation strategy.
The recommendation module executes a word vector recommendation strategy;
if the sentence vector is the second interval, the sentence vector is used as an intermediate frequency sentence vector, for example, a still open is conducted to another continent, 33 years old henry is located there, deep looking is expected to be thin and emerge in a 25 years old model, a distant place is expected to be naturally good, the short stay of a train is more beautiful looking back as years is the intermediate frequency sentence vector, the sentence vector is split into a plurality of word vectors with parallel weights, when a teacher handwriting recognition module of a college triggers one word vector, a recommendation module recommends the parallel word vector for the teacher of the college, and through the method, the associative writing of the associated intermediate frequency word vector in the teaching process is realized, and the writing time of a blackboard is reduced.
The recommendation module executes an associative recommendation strategy;
if the sentence vector is the third interval, the sentence vector is used as a low-frequency sentence vector, for example, "when the years of age have the tear-containing messages turned around, the henry of 37 years old is there, the deep eye is expected to return to the origin in the past, when the station from which the station starts remembers the back of the station, the seventeen years old splits the low-frequency sentence vector into a plurality of low-frequency word vectors, and when a teacher in a college triggers a low-frequency word using the handwriting recognition module, for example," turn around "," far-end "," bag "," henry "and the like, the medium-frequency word vector associated with" turn around "," far-end "," bag "," henry "and the like is recommended.
The semantic module perfects the semantic model by continuously executing sentence vector recommendation strategies, word vector recommendation strategies or association recommendation strategies, and performs unsupervised self-adaptive learning on the semantic model.
Example 3
The invention also provides a computer readable storage medium storing a computer program which when executed by a processor performs the method as above.
Finally, it should be noted that, as will be understood by those skilled in the art, implementing all or part of the above-described methods in the embodiments may be implemented by a computer program to instruct related hardware, and the program of the method for monitoring software may be stored in a computer readable storage medium, where the program may include the flow of the embodiments of the methods described above when executed. The storage medium of the program may be a magnetic disk, an optical disk, a read-only memory (ROM), a random-access memory (RAM), or the like. The computer program embodiments described above may achieve the same or similar effects as any of the method embodiments described above.
Furthermore, the method disclosed according to the embodiment of the present invention may also be implemented as a computer program executed by a processor, which may be stored in a computer-readable storage medium. The above-described functions defined in the methods disclosed in the embodiments of the present invention are performed when the computer program is executed by a processor.
Furthermore, the above-described method steps and system units may also be implemented using a controller and a computer-readable storage medium storing a computer program for causing the controller to implement the above-described steps or unit functions.
Further, it should be appreciated that the computer-readable storage medium (e.g., memory) herein can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory.
Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the disclosure herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as software or hardware depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.
The foregoing embodiment of the present invention has been disclosed with reference to the number of embodiments for the purpose of description only, and does not represent the advantages or disadvantages of the embodiments.
It will be appreciated by those of ordinary skill in the art that all or part of the steps of implementing the above embodiments may be implemented by hardware, or a program implemented by a program to instruct related hardware may be stored in a computer readable storage medium, where the storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The foregoing description is only illustrative of the present invention and is not intended to limit the scope of the invention, and all equivalent structures or equivalent processes or direct or indirect application in other related technical fields are included in the scope of the present invention.
Claims (4)
1. The method for optimizing handwriting input based on the semantic model is characterized by comprising the following steps of:
acquiring data and application scenes of individual voice input, and creating a semantic model according to the data of the individual voice input;
establishing a mapping relation between the application scene of the individual voice input and the semantic model;
performing classification operation on the semantic model according to the mapping relation, and extracting a plurality of sentence vectors according to the classification operation result;
configuring a weight relation, splitting the sentence vector according to the weight relation, and generating a plurality of word vectors;
acquiring individual handwriting operation, and executing associated recommendation on the individual handwriting operation according to the weight relation between the word vector and the sentence vector;
the step of obtaining the data and the application scene of the individual voice input and creating the semantic model according to the data of the individual voice input further comprises the following steps:
configuring a first threshold and a threshold interval, performing data characteristic analysis operation on the data input by the individual voice, and taking the data meeting the first threshold as a characteristic vector;
extracting the characteristic vector to create a semantic model;
the step of establishing a mapping relation between the application scene of the individual handwriting input and the semantic model further comprises the following steps:
acquiring the characteristic vector of the application scene, and establishing a mapping relation between the characteristic vector and the corresponding application scene and semantic model;
the step of executing classification operation on the semantic model according to the mapping relation and extracting a plurality of sentence vectors according to the classification operation result further comprises the following steps:
calculating the use times of the sentence vectors in a time domain and a space domain, and executing classification operation according to a threshold value interval where the use times of the sentence vectors are located;
the threshold interval comprises a first interval, a second interval and a third interval, and if the threshold interval of the sentence vector is in the first interval, a sentence vector recommendation strategy is executed; if the threshold interval of the sentence vector is in the second interval, executing a word vector recommendation strategy; if the threshold interval of the sentence vector is in the third interval, executing an association recommendation strategy;
the step of executing the sentence vector recommendation policy further includes: if the threshold interval of the sentence vector is the first interval, the sentence vector is used as a high-frequency sentence vector, the sentence head of the high-frequency sentence vector is marked as a high-frequency word vector, and when the high-frequency word vector is triggered by the handwriting input of the individual, the high-frequency sentence vector corresponding to the sentence head is recommended to the individual;
the step of executing the word vector recommendation policy further comprises: if the threshold interval of the sentence vector is the second interval, the sentence vector is used as an intermediate frequency sentence vector, the intermediate frequency sentence vector is split into a plurality of intermediate frequency word vectors with parallel weights, and when one intermediate frequency word vector is triggered by the handwriting input of the individual, the intermediate frequency word vector with the parallel weight with the triggered intermediate frequency word vector is recommended to the individual;
the step of associatively recommending a policy further comprises: and if the threshold interval of the sentence vector is a third interval, taking the sentence vector as a low-frequency sentence vector, splitting the low-frequency sentence vector into a plurality of word vectors and taking the word vectors as low-frequency word vectors, and recommending the medium-frequency word vectors associated with the low-frequency word vectors for the individual when the low-frequency word vectors are triggered by the handwriting input of the individual.
2. The method for optimizing handwriting input based on semantic model according to claim 1, wherein: the optimization method further comprises the following steps: the semantic model is perfected through continuously executing sentence vector recommendation strategies, word vector recommendation strategies or association recommendation strategies, and unsupervised self-adaptive learning is conducted on the semantic model.
3. A semantic model-based handwriting input optimization system employing the semantic model-based handwriting input optimization method of claim 1, the optimization system comprising: the system comprises an acquisition module, a semantic module, a voice recognition module, a mapping module, a classification module, a splitting module, a handwriting recognition module and a recommendation module;
the acquisition module is used for acquiring data input by individual voice and sending the data to the semantic module;
the semantic module is used for creating a semantic model according to the data;
the voice recognition module is used for recognizing voice input of the individual in an application scene;
the mapping module is used for establishing a mapping relation between the semantic model and the application scene;
the classification module is used for performing classification on the application scene according to the mapping relation and extracting a plurality of sentence vectors;
the splitting module is used for splitting the sentence vectors according to the weight relation to generate a plurality of word vectors;
the handwriting recognition module is used for recognizing the handwriting input of the individual and sending the handwriting input recognition result to the recommendation module;
and the recommendation module is used for executing association recommendation on the individual according to the weight relation between the word vector and the sentence vector.
4. A computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the method steps of optimizing handwriting input based on a semantic model according to claim 1 or 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110558066.2A CN113313001B (en) | 2021-05-21 | 2021-05-21 | Semantic model-based handwriting input optimization method, system and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110558066.2A CN113313001B (en) | 2021-05-21 | 2021-05-21 | Semantic model-based handwriting input optimization method, system and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113313001A CN113313001A (en) | 2021-08-27 |
| CN113313001B true CN113313001B (en) | 2023-10-10 |
Family
ID=77374139
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110558066.2A Active CN113313001B (en) | 2021-05-21 | 2021-05-21 | Semantic model-based handwriting input optimization method, system and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113313001B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20180129166A (en) * | 2017-05-25 | 2018-12-05 | 주식회사 스터디맥스 | Learning system and method using voice input of the learner |
| CN110263160A (en) * | 2019-05-29 | 2019-09-20 | 中国电子科技集团公司第二十八研究所 | A kind of Question Classification method in computer question answering system |
| CN112164391A (en) * | 2020-10-16 | 2021-01-01 | 腾讯科技（深圳）有限公司 | Statement processing method and device, electronic equipment and storage medium |
-
2021
- 2021-05-21 CN CN202110558066.2A patent/CN113313001B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20180129166A (en) * | 2017-05-25 | 2018-12-05 | 주식회사 스터디맥스 | Learning system and method using voice input of the learner |
| CN110263160A (en) * | 2019-05-29 | 2019-09-20 | 中国电子科技集团公司第二十八研究所 | A kind of Question Classification method in computer question answering system |
| CN112164391A (en) * | 2020-10-16 | 2021-01-01 | 腾讯科技（深圳）有限公司 | Statement processing method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113313001A (en) | 2021-08-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108984745B (en) | Neural network text classification method fusing multiple knowledge maps | |
| CN108763284B (en) | Question-answering system implementation method based on deep learning and topic model | |
| CN110866117B (en) | Short text classification method based on semantic enhancement and multi-level label embedding | |
| WO2021057424A1 (en) | Virtual image behavior control method and device based on text, and medium | |
| Ghosh et al. | Fracking sarcasm using neural network | |
| CN110148318B (en) | Digital teaching assistant system, information interaction method and information processing method | |
| CN109597891A (en) | Text emotion analysis method based on two-way length Memory Neural Networks in short-term | |
| CN110309503A (en) | A kind of subjective item Rating Model and methods of marking based on deep learning BERT--CNN | |
| CN114064918A (en) | Multi-modal event knowledge graph construction method | |
| KR102186326B1 (en) | A system and method for improving the performance of deep learning-based relation extraction by pre-training on structural information | |
| Wu et al. | Hierarchical attention-based multimodal fusion for video captioning | |
| CN112966525B (en) | Law field event extraction method based on pre-training model and convolutional neural network algorithm | |
| US20230395075A1 (en) | Human-machine dialogue system and method | |
| CN106682089A (en) | RNNs-based method for automatic safety checking of short message | |
| CN110047502A (en) | The recognition methods of hierarchical voice de-noising and system under noise circumstance | |
| CN112131876A (en) | Method and system for determining standard problem based on similarity | |
| Harwath et al. | Zero resource spoken audio corpus analysis | |
| CN112861524A (en) | Deep learning-based multilevel Chinese fine-grained emotion analysis method | |
| CN110851601A (en) | Cross-domain emotion classification system and method based on layered attention mechanism | |
| CN113823272A (en) | Voice processing method, device, electronic equipment and storage medium | |
| CN113360610A (en) | Dialog generation method and system based on Transformer model | |
| Chen et al. | Improving the prediction of therapist behaviors in addiction counseling by exploiting class confusions | |
| Yang | Natural language processing based on convolutional neural network and semi supervised algorithm in deep learning | |
| CN113313001B (en) | Semantic model-based handwriting input optimization method, system and medium | |
| CN110633468A (en) | Information processing method and device for object feature extraction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |