CN111783427A - Method, device, equipment and storage medium for training model and outputting information - Google Patents
Method, device, equipment and storage medium for training model and outputting information Download PDFInfo
- Publication number
- CN111783427A CN111783427A CN202010615558.6A CN202010615558A CN111783427A CN 111783427 A CN111783427 A CN 111783427A CN 202010615558 A CN202010615558 A CN 202010615558A CN 111783427 A CN111783427 A CN 111783427A
- Authority
- CN
- China
- Prior art keywords
- text
- test
- recognition model
- training
- negative
- 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.)
- Granted
Links
- 238000012549 training Methods 0.000 title claims abstract description 179
- 238000000034 method Methods 0.000 title claims abstract description 72
- 238000012360 testing method Methods 0.000 claims abstract description 208
- 238000012216 screening Methods 0.000 claims abstract description 69
- 230000006870 function Effects 0.000 claims description 41
- 230000015654 memory Effects 0.000 claims description 23
- 230000007704 transition Effects 0.000 claims description 19
- 238000004458 analytical method Methods 0.000 claims description 18
- 230000004044 response Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 239000000126 substance Substances 0.000 claims 2
- 238000013135 deep learning Methods 0.000 abstract description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 11
- 238000004891 communication Methods 0.000 description 8
- 239000000284 extract Substances 0.000 description 6
- 238000004590 computer program Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000001737 promoting effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Abstract
The application discloses a method, a device, equipment and a storage medium for training a model and outputting information, and relates to the fields of big data, artificial intelligence and deep learning. The specific implementation scheme is as follows: acquiring a first training sample set, wherein the first training sample set comprises a training set and a test set, and training samples comprise positive texts and negative texts; training the initial text recognition model by using a training set to obtain an intermediate text recognition model; testing the intermediate text recognition model by using the test set to obtain a test result; screening a second training sample set from the test set according to the test result and preset screening conditions; and training the intermediate text recognition model by using the second training sample set to obtain a target text recognition model. The realization mode can improve the accuracy of negative text recognition.
Description
Technical Field
The application relates to the field of artificial intelligence, in particular to the technical field of big data and deep learning. In particular, the application provides a method, a device, equipment and a storage medium for training a model and outputting information.
Background
With the continuous development of internet technology, people rely on internet applications more and more for communication, information acquisition and dissemination. In recent years, the rapid development of mobile internet has brought forth a new internet application mode for mobile terminals. The migration of PCs to the mobile end allows users to become interactive and socializing among the large-traffic apps. The large-flow app data are complicated, texts such as dynamic texts, news comments and posts need to be analyzed, and it is difficult to accurately identify whether marketing advertisements, pornography and other violation information exist in the texts. Therefore, a model capable of accurately identifying complex data needs to be trained on the basis of big data by means of a deep learning technology, but a deep learning algorithm needs to train more neural network parameters, such as weight values, threshold values and the like, so that a bottleneck exists in improving the accuracy of model identification data.
Disclosure of Invention
The present disclosure provides a method, apparatus, device, and storage medium for training a model and outputting information.
According to an aspect of the present disclosure, there is provided a method for training a model, comprising: acquiring a first training sample set, wherein the first training sample set comprises a training set and a test set; training the initial text recognition model by using a training set to obtain an intermediate text recognition model; testing the intermediate text recognition model by using the test set to obtain a test result; screening a second training sample set from the test set according to the test result and preset screening conditions; and training the intermediate text recognition model by using the second training sample set to obtain a target text recognition model.
According to another aspect of the present disclosure, there is provided a method for outputting information, including: acquiring a target text; performing dependency syntax analysis on the target text, and determining a main word set included in the target text; determining whether the target text is a negative text according to the set of main words and the target text recognition model of claim 1; generating output information in response to determining that the target text is a negative text; and outputting the output information.
According to another aspect of the present disclosure, there is provided an apparatus for training a model, comprising: a training sample set acquisition unit configured to acquire a first training sample set, the first training sample set including a training set and a test set; an initial text recognition model training unit configured to train an initial text recognition model using a training set to obtain an intermediate text recognition model; the intermediate text recognition model testing unit is configured to test the intermediate text recognition model by using the test set to obtain a test result; the second training sample set determining unit is configured to screen out a second training sample set from the test set according to the test result and a preset screening condition; and the target text recognition model determining unit is configured to train the intermediate text recognition model by using the second training sample set to obtain the target text recognition model.
According to another aspect of the present disclosure, there is provided an apparatus for outputting information, including: a target text acquisition unit configured to acquire a target text; the main word set determining unit is configured to perform dependency syntax analysis on the target text and determine a main word set included in the target text; a negative text determination unit configured to determine whether the target text is a negative text based on the set of main words and the target text recognition model of claim 1; an output information generating unit configured to generate output information in response to determining that the target text is a negative text; an information output unit configured to output the output information.
According to another aspect of the present disclosure, there is provided an electronic device for training a model, comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform the method for training a model as described above.
According to another aspect of the present disclosure, there is provided an electronic device for outputting information, including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform the method for outputting information as described above.
According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method for training a model as described above.
According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method for outputting information as described above.
According to the technology of the application, the problem that whether the marketing advertisement, the pornography and other violation information exist in the text is difficult to accurately identify is solved, and the accuracy of negative text identification is improved.
It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.
Drawings
The drawings are included to provide a better understanding of the present solution and are not intended to limit the present application. Wherein:
FIG. 1 is an exemplary system architecture diagram in which one embodiment of the present application may be applied;
FIG. 2 is a flow diagram of one embodiment of a method for training a model according to the present application;
FIG. 3 is a schematic diagram of an application scenario of a method for training a model according to the present application;
FIG. 4 is a flow diagram of another embodiment of a method for training a model according to the present application;
FIG. 5 is a schematic diagram of an embodiment of an apparatus for training a model according to the present application;
FIG. 6 is a flow diagram for one embodiment of a method for outputting information, in accordance with the present application;
FIG. 7 is a schematic diagram of an application scenario of a method for outputting information according to the present application;
FIG. 8 is a schematic block diagram illustrating one embodiment of an apparatus for outputting information according to the present application;
FIG. 9 is a block diagram of an electronic device for implementing a method for training a model and outputting information according to an embodiment of the present application.
Detailed Description
The following description of the exemplary embodiments of the present application, taken in conjunction with the accompanying drawings, includes various details of the embodiments of the application for the understanding of the same, which are to be considered exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
Fig. 1 illustrates an exemplary system architecture 100 to which embodiments of the present method for training a model or apparatus for training a model, method for outputting information, or apparatus for outputting information may be applied.
As shown in fig. 1, the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 serves as a medium for providing communication links between the terminal devices 101, 102, 103 and the server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.
The user may use the terminal devices 101, 102, 103 to interact with the server 105 via the network 104 to receive or send messages or the like. Various communication client applications, such as a web browsing application, a search application, etc., may be installed on the terminal devices 101, 102, 103.
The terminal apparatuses 101, 102, and 103 may be hardware or software. When the terminal devices 101, 102, 103 are hardware, they may be various electronic devices including, but not limited to, smart phones, tablet computers, e-book readers, car computers, laptop portable computers, desktop computers, and the like. When the terminal apparatuses 101, 102, 103 are software, they can be installed in the electronic apparatuses listed above. It may be implemented as multiple pieces of software or software modules (e.g., to provide distributed services) or as a single piece of software or software module. And is not particularly limited herein.
The server 105 may be a server providing various services, such as a background server analyzing data provided on the terminal devices 101, 102, 103. The background server can receive the data, train the target text recognition model by using the data, recognize the text to be recognized by using the trained target text recognition model, and output the text recognition result to the terminal devices 101, 102, and 103.
It should be noted that after the terminal devices 101, 102, and 103 acquire data, the terminal devices 101, 102, and 103 may train a target text recognition model, recognize a text to be recognized by using the trained target text recognition model, and output a recognition result.
The server 105 may be hardware or software. When the server 105 is hardware, it may be implemented as a distributed server cluster composed of a plurality of servers, or may be implemented as a single server. When the server 105 is software, it may be implemented as multiple pieces of software or software modules (e.g., to provide distributed services), or as a single piece of software or software module. And is not particularly limited herein.
It should be noted that the method for training the model and the method for outputting the information provided in the embodiment of the present application may be performed by the terminal devices 101, 102, 103 or the server 105. Accordingly, the means for training the model and the means for outputting the information are typically provided in the terminal device 101, 102, 103 or in the server 105.
It should be understood that the number of terminal devices, networks, and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
With continued reference to FIG. 2, a flow 200 of one embodiment of a method for training a model according to the present application is shown. The method for training the model of the embodiment comprises the following steps:
In this embodiment, an executing entity (e.g., the server 105 shown in fig. 1) of the method for training the model may obtain the first training sample set from the network through a wired connection or a wireless connection. The first training sample set may be sent to the execution subject after the user labels the text through the terminal device. The first set of training samples includes a training set and a test set, the training samples including positive text and negative text. The forward text may be text whose content conforms to the network platform specification, and may be, for example, educational text or the like. The network platform specification requires that characters with pornographic, advertising, abusive, and the like meanings are not likely to appear in text. The negative text may be text that does not conform to the network platform specification, and may be pornographic text, advertising promotional text, and the like, for example. Both the training set and the test set contain a certain amount of positive text and negative text.
After the executing subject obtains the first training sample set, the executing subject may train the initial text recognition model by using the training set to obtain an intermediate text recognition model. Specifically, an initial text recognition model may be understood as a text recognition model constructed before training the model, which has a poor ability to recognize text and needs to be further trained and iteratively optimized, so that the text recognition model has an enhanced ability to recognize text. After the execution subject trains the initial text recognition model by using the training set, the capability of the obtained text recognition model for text recognition can be further improved based on training, and the model is called as an intermediate text recognition model. The executive agent trains the initial text recognition model with the training set so that negative text in the training set can be recognized. The recognition capability of the obtained intermediate text recognition model to the negative text in the text is not high enough, and further training is needed to improve the recognition capability of the intermediate text recognition model to the negative text.
And step 203, testing the intermediate text recognition model by using the test set to obtain a test result.
After the execution main body obtains the intermediate text recognition model, the test set can be used for testing the intermediate text recognition model to obtain a test result. Specifically, the test set contains a certain amount of positive text and negative text, but the positive text and the negative text in the test set are different from the positive text and the negative text in the training set. And inputting the test set into an intermediate text recognition model, and recognizing the negative text in the test set by using the intermediate text recognition model. And judging the recognition capability of the intermediate text recognition model on the negative text according to the ratio of the number of the recognized negative texts to the number of all the negative texts in the test set. The execution subject may use the ratio of the number of identified negative text to the number of all negative text in the test set and the identified negative text as the test result.
And 204, screening a second training sample set from the test set according to the test result and preset screening conditions.
After the execution main body obtains the test result, the second training sample set can be screened out from the test set according to the test result and the preset screening condition. Specifically, the preset screening condition includes that a screening function value corresponding to the test text is greater than a preset value. Because the positive text and the negative text in the test set are different from those in the training set, the executive subject can take the negative text recognized by the intermediate text recognition model as a second training sample set to train the intermediate text recognition model again. The preset screening condition may be that all negative texts identified by the intermediate text identification model may be added to the second training sample set.
After the execution subject obtains the second training sample set, the intermediate text recognition model can be trained by using the second training sample set to obtain the target text recognition model. Specifically, the intermediate text recognition model is trained by using the second training sample set, and each parameter in the intermediate text recognition model is optimized. For example, the second training sample set is used to adjust the character spacing of the negative text recognizable by the intermediate text recognition model, so as to improve the sensitivity of the intermediate text recognition model to the negative text recognition. And finally, successfully training by the second training sample set, and accurately recognizing the intermediate text recognition model of the negative text, namely the target text recognition model.
With continued reference to FIG. 3, a schematic diagram of one application scenario of a method for training a model according to the present application is shown. In the application scenario of fig. 3, the desktop 301 collects a first training sample set 302, where the first training sample set 302 includes a training set 3021 and a test set 3022, and the training set samples include positive text and negative text. A server (not shown) may obtain the training set 3021 from the desktop 301 to train the initial text recognition model 303, resulting in the intermediate text recognition model 304. After training is completed, the server obtains a test result by using the intermediate text recognition model 304 obtained by testing in the test set 3022. And according to the test result and the preset screening condition, screening a second training sample set 306 from the test set 3022, and using the second training sample set 306 to train the intermediate text recognition model 304 again, so as to obtain the target text recognition model 305.
The method and the device can improve the accuracy of negative text recognition.
With continued reference to FIG. 4, a flow 400 of another embodiment of a method for training a model according to the present application is shown. As shown in fig. 4, the method for training a model of the present embodiment may include the following steps:
The principle of step 401 and step 402 is similar to that of step 201 and step 202, and is not described here again.
Specifically, step 402 can be determined by the following steps 4021-4022:
The execution subject may perform dependency parsing on at least one positive text and at least one negative text, respectively, to determine a set of stem words corresponding to each positive text and a set of stem words corresponding to each negative text. Specifically, the dependency syntax analysis is to analyze the text into a dependency syntax tree, which describes the dependency relationship between words in the text, i.e. indicates the syntactic collocation relationship between words in the text, and this collocation relationship is associated with semantics. The stems may be subjects, predicates, objects in the text. And the execution subject performs dependency syntactic analysis on the target text, extracts the subject, the predicate and the object in the target text, and performs permutation and combination on the extracted subject, the predicate and the object, so that the result of the permutation and combination is determined as a main word set included in the target text. For example, the executing entity obtains a forward text "father likes running very much", analyzes a set of stem words included in the text, and first extracts the stem words in the text as: father, liking, running. Then the set of stem words may include: father, liking, running. For another example, the executing entity obtains the negative text "he is a dog", analyzes the set of main words included in the text, and first extracts the main words in the text, including: he, Ye, dog. Then the set of stems corresponding to the negative text may be: he, Ye, dog.
After the main execution body obtains the main word set corresponding to the positive text and the main word set corresponding to the negative text, the main word set corresponding to the positive text and the main word set corresponding to the negative text can be used for training an initial text recognition model to obtain an intermediate text recognition model. Specifically, when the execution subject trains the initial text recognition model, a main word set corresponding to a negative text is input into the model, and the distance between characters in the main word set is adjusted, so that the model recognizes all or most of, for example, more than 90% of the input main word set as main words corresponding to the negative text, and recognizes main words corresponding to positive texts, which are not recognized or whose number ratio is less than 10%, of the input main word set corresponding to the positive text as the negative text, to obtain the intermediate text recognition model.
In the embodiment, the initial text recognition model is trained through the stem word sets corresponding to the positive and negative texts, so that the accuracy of model training is improved, and the accuracy of the model for recognizing the negative text is improved.
And 403, testing the intermediate text recognition model by using the test set to obtain a test result.
The principle of step 403 is similar to that of step 203, and is not described in detail here.
Specifically, step 403 can be determined through the following steps 4031-4034:
After the execution subject trains the initial text recognition model by using the training set to obtain the intermediate text recognition model, the test set can be used to test and optimize the obtained intermediate text recognition model. Specifically, the intermediate text recognition model includes different states, and when a state transition condition is satisfied, the state of the text transitions, where the different states include a start state, an intermediate state, at least one end state, and recognition results corresponding to the end states, and the state transition condition includes that the text includes target words, and a distance between the target words is smaller than a preset distance range. The execution body may determine a set of stem words corresponding to each test text by performing dependency parsing on each test text in the test set. The execution subject performs dependency parsing on each test text, and extracts a stem of each test text, where the stem may be a subject, a predicate, and an object of each test text, or a subject and an object of each test text, and the content of the stem is not specifically limited in the present application. And arranging and combining the extracted main words of each test text to form a main word set.
After the execution main body obtains the main word sets corresponding to the test texts, the distance between the main words in each main word set can be determined according to each test text. Specifically, the "distance" here may refer to a distance between extracted main words for each test text in which characters after each main word is put in place. For example, for a test text "he is a handsome fool dog", the main words are "he, dog", and the distance between "he" and "dog" in the test text "he is a handsome fool dog" is 8. After the execution main body obtains the main word sets corresponding to the test texts, the distance between the main words in each main word set can be determined according to the character spacing of the main words in each main word set in the corresponding test texts.
After determining the distance between the main words in the test text, the execution main body may determine the termination state of the test text according to the main word set, the distance between the main words, and the state transition condition. Specifically, a state may be interpreted as a permutation of the stem words, with each permutation forming a state. Determining the termination state of the test text according to the stem word set, the distances between the stem words and the state transition condition, for example, when the stem word set of the test text in the input model includes any character a in a negative text stem word set, no judgment is made on the test text first, i.e., the state is not transitioned first, then it is continuously judged whether another character b in the negative text stem word set exists in the stem word set of the test text, and if so, it is continuously judged whether the character spacing of a and b in the test text is within a preset threshold range, if so, the test text is recognized as a negative text, e.g., the preset threshold is [2, 4], that is, when the character spacing of a and b in the test text is 3, the state of the test text is transitioned to a state that can determine the test text as a negative text, determining that the recognition of the test text reaches a termination state, and outputting the recognition result of the test text as a negative text.
After determining the termination state of the test text, the execution main body may determine the test result of the test text according to the recognition result corresponding to the termination state. Specifically, if the recognition result corresponding to the termination state is that the test text is a negative text, the test result for the test text is a negative text.
According to the embodiment, the termination state of the text is determined according to the stem word set, the distance between the stem words and the state transition condition, the text can be accurately identified by combining a plurality of conditions, and the accuracy of the model for identifying the text is improved.
And step 404, screening out a second training sample set from the test set according to the test result and a preset screening condition.
The principle of step 404 is similar to that of step 204, and is not described here again.
Specifically, step 404 can be determined by steps 4041-4042 as follows:
After the execution main body obtains the test result of the test text, the parameters of the intermediate text recognition model can be adjusted according to the test result and the screening condition. The test results may be the number of recognized negative text and the number of recognized positive text. Specifically, the parameters of the intermediate text recognition model include a distance range value, and the screening condition includes that a screening function value corresponding to the test text is greater than a preset value. The execution main body combines the screening function according to the number of the identified negative texts and the number of the identified positive texts, and adjusts the distance of the main word corresponding to the negative texts which can be identified by the intermediate text identification model in the negative texts so as to enable the number of the negative texts screened by the screening function to be maximum. That is, the adjusted parameter of the intermediate text recognition model may be a character spacing of a main word of the negative text that can be recognized in the negative text thereof.
Specifically, step 4041 can be determined by steps 40411 to 40414 as follows:
Specifically, for each test text in the test set, the intermediate text recognition model may output a test result of the test text, where the test result may include recognizing the test text as a positive text or recognizing the test text as a negative text. By counting the test results of all test texts in the test set, a first number of positive texts and a second number of negative texts can be determined.
At step 40412, a third number of positive text and a fourth number of negative text in the test set are determined.
Specifically, each test text in the test set has been labeled, that is, the test set includes a positive test text and a negative test text. By counting the labels of the test texts in the test set, a third number of positive texts and a fourth number of negative texts in the test set can be determined.
Specifically, the executing body sets the distance between the main words of each test text to a, and the distance range value may be set to [0, a ], [0, 0], [0, 2a/5], [0, 3a/5], [0, 4a/5], for example, according to setting the distance between the main words of each test text to a preset value. Then, on the basis of the preset distance between the main words and the distance range value of each test text, obtaining a ratio X from the first quantity/the third quantity and obtaining a ratio Y from the second quantity/the fourth quantity, and determining whether the screening function value f (X, Y) corresponding to each test text is greater than 0 or less than 0. And if the screening function value f (X, Y) is greater than 0, identifying the test text as a negative text.
After the execution main body obtains the screening function value corresponding to each test text, the character spacing which enables the screening function value of each test text to be larger than 0 and has the largest number is obtained by adjusting the character spacing between the main words in each test text, namely, the distance range value is adjusted, the character spacing serves as the adjusted distance value, the distance value enables the middle text recognition model to be more sensitive to recognition of the negative text, and accuracy of recognition of the negative text can be improved.
The embodiment can enable the text recognition model to accurately recognize the negative text.
Wherein step 40414 further comprises the steps of:
and 404141, adjusting the distance range value according to the number of the screening function values larger than the preset value.
Specifically, the number of the filtering function values larger than the preset value is the number of the identified negative texts. The execution main body determines the number of the identified negative texts according to the number of the screening functions larger than the preset value, and when the number of the screening function values larger than the preset value is too small, the distance range value needs to be adjusted, so that the screening function values corresponding to the negative texts in each test set are all larger than the preset value, and the text recognition model can accurately recognize the negative texts.
Specifically, by adjusting the character spacing of the main words corresponding to the negative text in the negative text, the value of the screening function corresponding to the negative text is greater than the preset value, and the negative text with the value of the screening function greater than the preset value can be used as the second training sample set screened from the test set.
The method and the device can improve the accuracy of the model for recognizing the negative text.
Specifically, step 4042 may be implemented by:
Specifically, by adjusting the character spacing of the main word corresponding to the negative text in the negative text, the value of the screening function corresponding to the negative text is greater than a preset value, for example, the preset value is 0, and the negative text with the value of the corresponding screening function greater than 0 can be used as the second training sample set screened from the test set.
The method and the device can improve the accuracy of the model for recognizing the negative text.
And 405, training the intermediate text recognition model by using the second training sample set to obtain a target text recognition model.
The principle of step 405 is similar to that of step 205, and is not described here again.
With further reference to fig. 5, as an implementation of the method shown in the above figures, the present application provides an embodiment of an apparatus for training a model, which corresponds to the embodiment of the method shown in fig. 2, and which can be applied in various electronic devices.
As shown in fig. 5, the apparatus 500 for training a model of the present embodiment includes: a training sample set obtaining unit 501, an initial text recognition model training unit 502, an intermediate text recognition model testing unit 503, a second training sample set determining unit 504 and a target text recognition model determining unit 505.
A training sample set obtaining unit 501 configured to obtain a first training sample set, where the first training sample set includes a training set and a test set.
An initial text recognition model training unit 502 configured to train the initial text recognition model with a training set, resulting in an intermediate text recognition model.
The intermediate text recognition model testing unit 503 is configured to test the intermediate text recognition model by using the test set, and obtain a test result.
The second training sample set determining unit 504 is configured to filter out the second training sample set from the test set according to the test result and a preset filtering condition.
And a target text recognition model determining unit 505 configured to train the intermediate text recognition model by using the second training sample set, resulting in a target text recognition model.
In some optional implementations of this embodiment, the training samples include positive text and negative text; and an initial text recognition model training unit 502, further configured to: performing dependency syntax analysis on at least one positive text and at least one negative text respectively, and determining a main word set corresponding to each positive text and a main word set corresponding to each negative text; and training an initial text recognition model by using the main word set corresponding to the positive text and the main word set corresponding to the negative text to obtain an intermediate text recognition model.
In some optional implementation manners of this embodiment, the intermediate text recognition model includes different states, and when a state transition condition is satisfied, the state of the text transitions, where the different states include a start state, an intermediate state, at least one end state, and recognition results corresponding to the end states, and the state transition condition includes that the text includes target words, and distances between the target words are smaller than a preset distance range; and an intermediate text recognition model test unit 503, further configured to: performing dependency syntax analysis on each test text, and determining a main word set corresponding to each test text; determining the distance between the main words in each main word set according to each test text; determining the termination state of the test text according to the stem word set, the distance between the stem words and the state transition condition; and determining the test result of the test text according to the identification result corresponding to the termination state.
In some optional implementations of this embodiment, the second training sample set determining unit 504 is further configured to: adjusting parameters of the intermediate text recognition model according to the test result and preset screening conditions; and screening out a second training sample set from the test set according to the adjusted parameters and preset screening conditions.
In some optional implementation manners of this embodiment, the parameter of the intermediate text recognition model includes a distance range value, and the screening condition includes that a screening function value corresponding to the test text is greater than a preset value; and a second training sample set determining unit 504, further configured to: determining a first quantity of positive texts and a second quantity of negative texts in the test result; determining a third quantity of positive texts and a fourth quantity of negative texts in the test set; determining a screening function value corresponding to each test text according to the first number, the second number, the third number and the fourth number, and the distance and distance range value between the main words of each test text; and adjusting the distance range value according to each screening function value.
In some optional implementations of this embodiment, the second training sample set determining unit 504 is further configured to: and adjusting the distance range value according to the number of the screening function values larger than the preset value.
In some optional implementations of this embodiment, the second training sample set determining unit 504 is further configured to: and adding the test text with the corresponding screening function value larger than the preset value into a second training sample set according to the adjusted parameters.
It should be understood that units 501 to 505, respectively, recited in the apparatus 500 for training a model correspond to the respective steps in the method described with reference to fig. 2. Thus, the operations and features described above for the method for training a model are equally applicable to the apparatus 500 and the units included therein and will not be described in detail here.
With continued reference to FIG. 6, a flow 600 of one embodiment of a method for outputting information in accordance with the present application is shown. The method for outputting information of the embodiment comprises the following steps:
In this embodiment, an execution subject of the method for outputting information (e.g., the server 105 shown in fig. 1) may acquire the target text by wired connection or wireless connection. The target text can be input by the user through the terminal device and then sent to the execution subject. The target text may include positive text and negative text. The forward text may be text whose content conforms to a network platform specification. The network platform specification requires that characters with pornographic, advertising, abusive, and the like meanings are not likely to appear in text. The negative text may be text that does not conform to the network platform specification, and may be pornographic text, advertising promotional text, and the like, for example.
After the execution main body obtains the target text, dependency syntax analysis can be performed on the target text, and a main word set included in the target text is determined. Specifically, the dependency syntax analysis is to analyze the text into a dependency syntax tree, which describes the dependency relationship between words in the text, i.e. indicates the syntactic collocation relationship between words in the text, and this collocation relationship is associated with semantics. The execution main body determines a main word set included in the target text according to dependency parsing on the target text, for example, the execution main body acquires a text "father likes running very much", analyzes the main word set included in the text, and first extracts main words in the text as follows: father, liking, running. Then the set of stem words may include: father, liking, running. For another example, the executing entity obtains the negative text "he is a dog", analyzes the set of main words included in the text, and first extracts the main words in the text, including: he, Ye, dog. Then the set of stems corresponding to the negative text may be: he, Ye, dog.
After determining the main word set included in the target text, the execution main body can determine whether the target text is a negative text according to the main word set and the target text recognition model. In particular, the set of stem words may be a set of sequentially combined stem words. The target text recognition model may be a Finite state Automaton (DFA) determined by a trained DFA. The trained determined finite state automaton is used for analyzing an input main word set, specifically, judging whether the input main word contains preset characters and whether the character spacing between the preset characters is within a preset range, and when the input main word contains the preset characters and the character spacing between the preset characters is within the preset range, judging that a target text is a negative text, otherwise, judging that the target text is a positive text. For example, if the trained target text recognition model identifies that the main word of the input model contains "a" and "b" characters, and the distance between the "a" and "b" characters in the sequentially combined main word is between 2 characters and 4 characters, the target text recognition model outputs the determination result of the sequentially combined main word as a negative text. In this embodiment, the target text recognition model may be a certain finite state automaton, which may be composed of the following 5 parts: m ═ Si，Σ，f，S0Z) in which SiIs a negative text collection. S0Is a unique starting state (S)0∈Si) F is the state transfer function, Σ is the set of input stem words, and Z is the set of termination states. The term "determine" in "determined finite state automata" means that only one output result is provided for the text of an input DFA, that is, the output result for the text of the input DFA can only be one of negative text and positive text.
After determining that the target text is a negative text, the execution main body can generate output information according to the target text. Specifically, after determining that the target text is negative text, the executing entity continues to generate a classification corresponding to the target text according to the negative text, for example, pornographic text, advertising promotional text, or abusive text.
The execution subject may output the output information after generating the output information. So as to carry out corresponding treatment according to production practice. Specifically, the output information includes, in addition to the classification corresponding to the target text, negative text in a training set used to derive the target text recognition model. The method is convenient to implement the recognition of the target text recognition model on the negative text in combination with the actual rapid landing of production, and optimizes the network environment.
With continued reference to fig. 7, a schematic illustration of one application scenario of a method for outputting information according to the present application is shown. In the application scenario of fig. 7, the target text is obtained through the desktop computer 701, and after the server obtains the target text 702 from the desktop computer 701, the server performs dependency parsing 703 on the target text 702 to determine a set of main words included in the target text. The server determines whether the target text is negative text according to the set of main words and the target text recognition model 704. When the server determines that the target text is the negative text 705, output information 706 is generated and output information 706 is output.
The method and the device can improve the accuracy of negative text recognition.
With further reference to fig. 8, as an implementation of the methods shown in the above-mentioned figures, the present application provides an embodiment of an apparatus for outputting information, which corresponds to the method embodiment shown in fig. 6, and which is particularly applicable to various electronic devices.
As shown in fig. 8, the apparatus 800 for outputting information of the present embodiment includes: a target text acquisition unit 801, a main word set determination unit 802, a negative text determination unit 803, an output information generation unit 804, and an information output unit 805.
A target text acquisition unit 801 configured to acquire a target text.
A stem word set determining unit 802, configured to perform dependency syntax analysis on the target text, and determine a stem word set included in the target text.
A negative text determination unit 803 configured to determine whether the target text is a negative text according to the set of main words and the target text recognition model.
An output information generating unit 804 configured to generate output information in response to determining that the target text is a negative text.
An information output unit 805 configured to output the output information.
It should be understood that units 801 to 805 described in the apparatus 800 for outputting information correspond to respective steps in the method described with reference to fig. 6. Thus, the operations and features described above for the method for outputting information are equally applicable to the apparatus 600 and the units included therein and will not be described in detail here.
According to an embodiment of the present application, an electronic device and a readable storage medium are also provided.
FIG. 9 is a block diagram of an electronic device for training a model and a method for outputting information according to an embodiment of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present application that are described and/or claimed herein.
As shown in fig. 9, the electronic apparatus includes: one or more processors 901, memory 902, and interfaces for connecting the various components, including a high-speed interface and a low-speed interface. The various components are interconnected using different buses 905 and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions for execution within the electronic device, including instructions stored in or on the memory to display graphical information of a GUI on an external input/output apparatus (such as a display device coupled to the interface). In other embodiments, multiple processors and/or multiple buses 905 may be used, along with multiple memories and multiple memories, as desired. Also, multiple electronic devices may be connected, with each device providing portions of the necessary operations (e.g., as a server array, a group of blade servers, or a multi-processor system). Fig. 9 illustrates an example of a processor 901.
The memory 902, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs, non-transitory computer executable programs, and units, such as program instructions/units corresponding to the method for training a model in the embodiment of the present application (for example, the training sample set acquisition unit 501, the initial text recognition model training 502, the intermediate text recognition model test unit 503, the second training sample set determination unit 504, and the target text recognition model determination unit 505 shown in fig. 5). Also, for example, the program instructions/units corresponding to the method for outputting information in the embodiment of the present application (for example, the target text acquisition unit 801, the main word set determination unit 802, the negative text determination unit 803, the output information generation unit 804, and the information output unit 805 shown in fig. 8). The processor 901 executes various functional applications of the server and data processing, i.e., implements the methods for training the model and for outputting information in the above-described method embodiments, by executing non-transitory software programs, instructions, and modules stored in the memory 902.
The memory 902 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of the electronic device for training the model and the method for outputting information, and the like. Further, the memory 902 may include high speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory 902 optionally includes memory located remotely from the processor 901, which may be connected over a network to an electronic device for use in a method of training a model and for outputting information. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The electronic device for training the model and the method for outputting information may further include: an input device 903 and an output device 904. The processor 901, the memory 902, the input device 903, and the output device 904 may be connected by a bus 905 or in other ways, and are exemplified by the bus 905 in fig. 9.
The input device 903 may receive input numeric or character information and generate key signal inputs related to user settings and function controls of the electronic device for training the model and for methods of outputting information, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, a pointer, one or more mouse buttons, a track ball, a joystick, and the like. The output devices 904 may include a display device, auxiliary lighting devices (e.g., LEDs), tactile feedback devices (e.g., vibrating motors), and the like. The display device may include, but is not limited to, a Liquid Crystal Display (LCD), a Light Emitting Diode (LED) display, and a plasma display. In some implementations, the display device can be a touch screen.
Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, integrated circuitry, application specific ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.
These computer programs (also known as programs, software applications, or code) include machine instructions for a programmable processor, and may be implemented using high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.
The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
According to the technical scheme of the embodiment of the application, the accuracy rate of negative text recognition can be improved.
It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present application may be executed in parallel, sequentially, or in different orders, and the present invention is not limited thereto as long as the desired results of the technical solutions disclosed in the present application can be achieved.
The above-described embodiments should not be construed as limiting the scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (20)
1. A method for training a model, comprising:
acquiring a first training sample set, wherein the first training sample set comprises a training set and a test set;
training an initial text recognition model by using the training set to obtain an intermediate text recognition model;
testing the intermediate text recognition model by using the test set to obtain a test result;
screening a second training sample set from the test set according to the test result and a preset screening condition;
and training the intermediate text recognition model by using the second training sample set to obtain a target text recognition model.
2. The method of claim 1, wherein the training samples comprise positive text and negative text; and
the training of the initial text recognition model by using the training set to obtain an intermediate text recognition model comprises the following steps:
performing dependency syntax analysis on at least one positive text and at least one negative text respectively, and determining a main word set corresponding to each positive text and a main word set corresponding to each negative text;
and training the initial text recognition model by using the stem word set corresponding to the positive text and the stem word set corresponding to the negative text to obtain an intermediate text recognition model.
3. The method of claim 2, wherein the intermediate text recognition model comprises different states, the states of the text transition when a state transition condition is met, the different states comprise a start state, an intermediate state, at least one end state and a recognition result corresponding to each end state, the state transition condition comprises that the text comprises target words, and the distance between each target word is smaller than a preset distance range; and
the testing the intermediate text recognition model by using the test set to obtain a test result comprises the following steps:
performing dependency syntax analysis on each test text, and determining a main word set corresponding to each test text;
determining the distance between the main words in each main word set according to each test text;
determining the termination state of the test text according to the main word set, the distance between the main words and the state transition condition;
and determining the test result of the test text according to the identification result corresponding to the termination state.
4. The method of claim 3, wherein the screening out a second set of training samples from the test set according to the test result and a preset screening condition comprises:
adjusting parameters of the intermediate text recognition model according to the test result and the preset screening condition;
and screening out a second training sample set from the test set according to the adjusted parameters and the preset screening conditions.
5. The method of claim 4, wherein the parameters of the intermediate text recognition model comprise distance range values, and the preset screening condition comprises that the screening function value corresponding to the test text is greater than a preset value; and
adjusting parameters of the intermediate text recognition model according to the test result and the preset screening condition, wherein the parameters comprise:
determining a first number of positive texts and a second number of negative texts in the test result;
determining a third number of positive text and a fourth number of negative text in the test set;
determining a screening function value corresponding to each test text according to the first number, the second number, the third number, the fourth number, the distance between the main words of each test text and the distance range value;
and adjusting the distance range value according to each screening function value.
6. The method of claim 5, wherein said adjusting the distance range value according to each filter function value comprises:
and adjusting the distance range value according to the number of the screening function values larger than the preset value.
7. The method of claim 5, wherein the screening out a second set of training samples from the test set according to the adjusted parameters and the screening conditions comprises:
and adding the test text with the corresponding screening function value larger than a preset value into the second training sample set according to the adjusted parameters.
8. A method for outputting information, comprising:
acquiring a target text;
performing dependency syntax analysis on the target text, and determining a main word set included in the target text;
determining whether the target text is a negative text according to the set of main words and the target text recognition model of claim 1;
generating output information in response to determining that the target text is a negative text;
and outputting the output information.
9. An apparatus for training a model, comprising:
a training sample set acquisition unit configured to acquire a first training sample set, the first training sample set including a training set and a test set;
an initial text recognition model training unit configured to train an initial text recognition model using the training set, resulting in an intermediate text recognition model;
the intermediate text recognition model testing unit is configured to test the intermediate text recognition model by using the test set to obtain a test result;
a second training sample set determining unit configured to screen out a second training sample set from the test set according to the test result and a preset screening condition;
and the target text recognition model determining unit is configured to train the intermediate text recognition model by using the second training sample set to obtain a target text recognition model.
10. The apparatus of claim 9, wherein the training samples comprise positive text and negative text; and
the initial text recognition model training unit is further configured to:
performing dependency syntax analysis on at least one positive text and at least one negative text respectively, and determining a main word set corresponding to each positive text and a main word set corresponding to each negative text;
and training the initial text recognition model by using the stem word set corresponding to the positive text and the stem word set corresponding to the negative text to obtain an intermediate text recognition model.
11. The apparatus according to claim 9, wherein the intermediate text recognition model includes different states, the states of the text transition when a state transition condition is satisfied, the different states include a start state, an intermediate state, at least one end state, and a recognition result corresponding to each end state, the state transition condition includes that the text includes target words and a distance between each target word is smaller than a preset distance range; and
the intermediate text recognition model testing unit is further configured to:
performing dependency syntax analysis on each test text, and determining a main word set corresponding to each test text;
determining the distance between the main words in each main word set according to each test text;
determining the termination state of the test text according to the main word set, the distance between the main words and the state transition condition;
and determining the test result of the test text according to the identification result corresponding to the termination state.
12. The apparatus of claim 11, wherein the second training sample set determination unit is further configured to:
adjusting parameters of the intermediate text recognition model according to the test result and the preset screening condition;
and screening out a second training sample set from the test set according to the adjusted parameters and the preset screening conditions.
13. The apparatus of claim 10, wherein the parameters of the intermediate text recognition model include distance range values, and the filtering condition includes that a filtering function value corresponding to the test text is greater than a preset value; and
the second training sample set determination unit is further configured to:
determining a first number of positive texts and a second number of negative texts in the test result;
determining a third number of positive text and a fourth number of negative text in the test set;
determining a screening function value corresponding to each test text according to the first number, the second number, the third number, the fourth number, the distance between the main words of each test text and the distance range value;
and adjusting the distance range value according to each screening function value.
14. The apparatus of claim 12, wherein the second training sample set determination unit is further configured to:
and adjusting the distance range value according to the number of the screening function values larger than the preset value.
15. The apparatus of claim 12, wherein the second training sample set determination unit is further configured to:
and adding the test text with the corresponding screening function value larger than a preset value into the second training sample set according to the adjusted parameters.
16. An apparatus for outputting information, comprising:
a target text acquisition unit configured to acquire a target text;
a main word set determining unit, configured to perform dependency syntax analysis on the target text, and determine a main word set included in the target text;
a negative text determination unit configured to determine whether the target text is a negative text according to the set of main words and the target text recognition model of claim 1;
an output information generation unit configured to generate output information in response to determining that the target text is a negative text;
an information output unit configured to output the output information.
17. An electronic device for training a model, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.
18. An electronic device for outputting information, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of claim 8.
19. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-7.
20. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010615558.6A CN111783427B (en) | 2020-06-30 | 2020-06-30 | Method, device, equipment and storage medium for training model and outputting information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010615558.6A CN111783427B (en) | 2020-06-30 | 2020-06-30 | Method, device, equipment and storage medium for training model and outputting information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111783427A true CN111783427A (en) | 2020-10-16 |
| CN111783427B CN111783427B (en) | 2024-04-02 |
Family
ID=72761299
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010615558.6A Active CN111783427B (en) | 2020-06-30 | 2020-06-30 | Method, device, equipment and storage medium for training model and outputting information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111783427B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112464661A (en) * | 2020-11-25 | 2021-03-09 | 马上消费金融股份有限公司 | Model training method, voice conversation detection method and related equipment |
| CN112989789A (en) * | 2021-03-15 | 2021-06-18 | 京东数科海益信息科技有限公司 | Test method and device of text audit model, computer equipment and storage medium |
| CN113987569A (en) * | 2021-10-14 | 2022-01-28 | 武汉联影医疗科技有限公司 | Anti-crawler method and device, computer equipment and storage medium |
| CN112989789B (en) * | 2021-03-15 | 2024-05-17 | 京东科技信息技术有限公司 | Test method and device of text auditing model, computer equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477544A (en) * | 2009-01-12 | 2009-07-08 | 腾讯科技（深圳）有限公司 | Rubbish text recognition method and system |
| US20120254333A1 (en) * | 2010-01-07 | 2012-10-04 | Rajarathnam Chandramouli | Automated detection of deception in short and multilingual electronic messages |
| CN107229610A (en) * | 2017-03-17 | 2017-10-03 | 咪咕数字传媒有限公司 | The analysis method and device of a kind of affection data |
| WO2019192397A1 (en) * | 2018-04-04 | 2019-10-10 | 华中科技大学 | End-to-end recognition method for scene text in any shape |
| WO2019232869A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技（深圳）有限公司 | Handwriting model training method, text recognition method and apparatus, device, and medium |
-
2020
- 2020-06-30 CN CN202010615558.6A patent/CN111783427B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477544A (en) * | 2009-01-12 | 2009-07-08 | 腾讯科技（深圳）有限公司 | Rubbish text recognition method and system |
| US20120254333A1 (en) * | 2010-01-07 | 2012-10-04 | Rajarathnam Chandramouli | Automated detection of deception in short and multilingual electronic messages |
| CN107229610A (en) * | 2017-03-17 | 2017-10-03 | 咪咕数字传媒有限公司 | The analysis method and device of a kind of affection data |
| WO2019192397A1 (en) * | 2018-04-04 | 2019-10-10 | 华中科技大学 | End-to-end recognition method for scene text in any shape |
| WO2019232869A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技（深圳）有限公司 | Handwriting model training method, text recognition method and apparatus, device, and medium |
Non-Patent Citations (1)
| Title |
|---|
| 阳庆玲;郑志伟;邱佳玲;郭山清;贾忠伟;宋超;顾菁;李菁华;郝元涛;郝春;: "基于表情符号的文本情感分析研究", 现代预防医学, no. 09 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112464661A (en) * | 2020-11-25 | 2021-03-09 | 马上消费金融股份有限公司 | Model training method, voice conversation detection method and related equipment |
| CN112989789A (en) * | 2021-03-15 | 2021-06-18 | 京东数科海益信息科技有限公司 | Test method and device of text audit model, computer equipment and storage medium |
| CN112989789B (en) * | 2021-03-15 | 2024-05-17 | 京东科技信息技术有限公司 | Test method and device of text auditing model, computer equipment and storage medium |
| CN113987569A (en) * | 2021-10-14 | 2022-01-28 | 武汉联影医疗科技有限公司 | Anti-crawler method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111783427B (en) | 2024-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111428008B (en) | Method, apparatus, device and storage medium for training a model | |
| CN112560912B (en) | Classification model training method and device, electronic equipment and storage medium | |
| CN111221984A (en) | Multimodal content processing method, device, equipment and storage medium | |
| CN110991427B (en) | Emotion recognition method and device for video and computer equipment | |
| CN111680145A (en) | Knowledge representation learning method, device, equipment and storage medium | |
| CN111709247A (en) | Data set processing method and device, electronic equipment and storage medium | |
| CN111325020A (en) | Event argument extraction method and device and electronic equipment | |
| CN112148881B (en) | Method and device for outputting information | |
| CN111598164A (en) | Method and device for identifying attribute of target object, electronic equipment and storage medium | |
| CN111104514A (en) | Method and device for training document label model | |
| CN111859982A (en) | Language model training method and device, electronic equipment and readable storage medium | |
| CN111611990A (en) | Method and device for identifying table in image | |
| CN111078878A (en) | Text processing method, device and equipment and computer readable storage medium | |
| CN111522944A (en) | Method, apparatus, device and storage medium for outputting information | |
| CN111680517A (en) | Method, apparatus, device and storage medium for training a model | |
| CN112507090A (en) | Method, apparatus, device and storage medium for outputting information | |
| CN111782785B (en) | Automatic question and answer method, device, equipment and storage medium | |
| CN111177462B (en) | Video distribution timeliness determination method and device | |
| CN111127191A (en) | Risk assessment method and device | |
| CN111783427B (en) | Method, device, equipment and storage medium for training model and outputting information | |
| CN112380104A (en) | User attribute identification method and device, electronic equipment and storage medium | |
| CN112330455A (en) | Method, device, equipment and storage medium for pushing information | |
| CN111582477A (en) | Training method and device of neural network model | |
| CN111563198A (en) | Material recall method, device, equipment and storage medium | |
| CN114357951A (en) | Method, device, equipment and storage medium for generating standard report |
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 |