CN113505786A

CN113505786A - Test question photographing and judging method and device and electronic equipment

Info

Publication number: CN113505786A
Application number: CN202110707922.6A
Authority: CN
Inventors: 崔寅生; 刘培娜; 王辰成; 李雨桐; 潘东; 王伟戌; 韩均雷
Original assignee: Zuoyebang Education Technology Beijing Co Ltd
Current assignee: Beijing Baige Feichi Technology Co ltd
Priority date: 2021-06-24
Filing date: 2021-06-24
Publication date: 2021-10-15

Abstract

The invention belongs to the technical field of data information processing, and provides a test question photographing judgment method, a test question photographing judgment device, electronic equipment and a recording medium, wherein the method comprises the following steps: acquiring an image file submitted by a user, identifying first text data corresponding to a title in the image file, and identifying second text data corresponding to a user answer in the image file; cutting and splitting the first text data based on the spatial position, and determining the question stem content to be retrieved; matching answer text data corresponding to the question stem content from an information database by using a matching model based on semantic retrieval; the second text data is evaluated based on the answer text data. According to the method, the corresponding answer text data is retrieved based on the matching model of semantic retrieval, so that the problems with the same meaning but different contents exist in the problem base but the problems and the answers cannot be found in actual retrieval are avoided, the flexibility of photographing and searching the problems is improved, the accuracy is also improved, and the user experience is better.

Description

Test question photographing and judging method and device and electronic equipment

Technical Field

The invention belongs to the technical field of data information processing, is particularly suitable for data information processing in online education service, and more particularly relates to a test question photographing judgment method and device and electronic equipment.

Background

In the traditional teaching mode, the students finish homework and examination papers and then are manually judged by teachers, the efficiency is relatively low, and the students cannot know whether answer questions are correct or not in the first time.

With the development of internet technology, various applications of photographing and searching questions gradually appear at present, so that students can obtain answers and a question solving process of test questions by photographing without waiting for judgment of teachers when independently learning to do the questions, and whether the answers are correct or not is judged in time.

However, the photo-taking question searching application commonly used at present uses an answer library to give answers, if no answer is available in the answer library, the answers are not given, and the question searching mode is inflexible.

Disclosure of Invention

Technical problem to be solved

The invention at least aims to solve the problems that the search answers are inflexible and inaccurate in the use process of the conventional photographing question searching application.

(II) technical scheme

In order to solve the above technical problem, an aspect of the present invention provides a method for photographing and evaluating test questions, including:

acquiring an image file submitted by a user, identifying first text data corresponding to a title in the image file, and identifying second text data corresponding to a response of the user in the image file;

cutting and splitting the first text data based on the spatial position, and determining the question stem content to be retrieved;

matching answer text data corresponding to the question stem content from an information database by using a matching model based on semantic retrieval;

and judging the second text data based on the answer text data.

In an exemplary embodiment of the present invention, a neural network-based recognition model identifies first text data corresponding to a topic in the image file;

optionally, the first text data corresponds to print text in the image file, and the second text data corresponds to handwriting in the image file;

optionally, the neural network-based recognition model employs CPTN in combination with a CRNN neural network model.

In an exemplary embodiment of the present invention, the cutting and splitting the first text data based on the spatial position to determine the stem content to be retrieved includes:

cutting the first text data based on the spatial position, and splitting the first text data into a plurality of layers of titles;

and determining the test question type and the question stem content to be retrieved based on the questions of the multiple layers.

In an exemplary embodiment of the present invention, the matching answer text data corresponding to the stem content from an information database using a matching model based on semantic retrieval includes:

performing text structuring processing on the question stem content;

inputting the subject stem content after text structuralization processing into the matching model based on semantic retrieval, and matching answer text data corresponding to the subject stem content from an information database;

optionally, the matching model based on semantic retrieval is a Bi-LSTM neural network model.

In an exemplary embodiment of the present invention, the evaluating the second text data based on the answer text data includes: judging whether the answer text data is matched with the second text data or not, and if so, judging;

optionally, if the answer text data does not match the second text data, matching corresponding reverse-thrust topic text data in an information database based on the second text data; and calculating the similarity between the reverse-thrust topic text data and the first text data, if the similarity exceeds a similarity threshold value, judging the similarity, and otherwise, judging the similarity by mistake.

In an exemplary embodiment of the present invention, it is determined whether the answer text data and the second text data match through a semantic-based matching model.

In an exemplary embodiment of the invention, a prompt information database is preset, and after the judgment operation is completed, the prompt information matched with the first text data is retrieved from the prompt information database and recommended to a user.

The second aspect of the present invention provides a device for photographing and evaluating test questions, comprising:

the identification module is used for acquiring an image file submitted by a user, and identifying first text data corresponding to a title in the image file and second text data corresponding to a response of the user in the image file;

the determining module is used for cutting and splitting the first text data based on the spatial position and determining the question stem content to be retrieved;

the matching module is used for matching answer text data corresponding to the question stem content from an information database by using a matching model based on semantic retrieval;

and the judging module is used for judging the second text data based on the answer text data.

A third aspect of the present invention provides an electronic device, comprising:

one or more processors;

storage means for storing one or more programs;

when executed by the one or more processors, cause the one or more processors to implement the method.

A fourth aspect of the invention proposes a computer-readable medium, on which a computer program is stored which, when being executed by a processor, carries out any of the methods.

(III) advantageous effects

According to the method and the device, the question data and the user response data are identified through the identification model, the corresponding answer text data are retrieved based on the matching model of semantic retrieval, and the user response data are judged through the answer text data, so that the condition that answers cannot be found for questions with the same question meaning but different contents is avoided, the flexibility of photographing and searching the questions is improved, the accuracy is also improved, and the user experience is better.

Drawings

FIG. 1 is a schematic view of an application scene structure of examination question photographing evaluation according to the present invention;

FIG. 2 is a flowchart illustrating a method for evaluating a test question by photographing according to an embodiment of the present invention;

FIG. 3 is a schematic illustration of a photograph taken in accordance with an embodiment of the present invention;

FIG. 4 is a flow chart illustrating text data of matching answers according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a photographing evaluation device for test questions based on a neural network according to an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of an electronic device of one embodiment of the invention;

fig. 7 is a schematic diagram of a computer-readable recording medium of an embodiment of the present invention.

Detailed Description

In describing particular embodiments, specific details of structures, properties, effects, or other features are set forth in order to provide a thorough understanding of the embodiments by one skilled in the art. However, it is not excluded that a person skilled in the art may implement the invention in a specific case without the above-described structures, performances, effects or other features.

The flow chart in the drawings is only an exemplary flow demonstration, and does not represent that all the contents, operations and steps in the flow chart are necessarily included in the scheme of the invention, nor does it represent that the execution is necessarily performed in the order shown in the drawings. For example, some operations/steps in the flowcharts may be divided, some operations/steps may be combined or partially combined, and the like, and the execution order shown in the flowcharts may be changed according to actual situations without departing from the gist of the present invention.

The block diagrams in the figures generally represent functional entities and do not necessarily correspond to physically separate entities. I.e. these functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different network and/or processing unit devices and/or microcontroller devices.

The same reference numerals denote the same or similar elements, components, or parts throughout the drawings, and thus, a repetitive description thereof may be omitted hereinafter. It will be further understood that, although the terms first, second, third, etc. may be used herein to describe various elements, components, or sections, these elements, components, or sections should not be limited by these terms. That is, these phrases are used only to distinguish one from another. For example, a first device may also be referred to as a second device without departing from the spirit of the present invention. Furthermore, the term "and/or", "and/or" is intended to include all combinations of any one or more of the listed items.

In order to solve the technical problems, the invention provides a test question photographing evaluation method and device, after a user uploads a test question photo with an answer, evaluation can be performed according to the question content and the answer content of the user, the error is directly judged, and no longer waiting for a teacher to manually evaluate is needed.

FIG. 1 is a schematic view of an application scene structure of examination question photographing evaluation. As shown in fig. 1, a user downloads and installs a client in a terminal 101 or a terminal 102, and uploads a photo of a test question with an answer written thereto to an application server 103 through a network 105 using the client. The

terminals

101, 102 include, but are not limited to, smart phones, tablet computers, laptop portable computers, desktop computers, and the like. Network 105 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.

In the application server 103, the information database 104 is searched according to the contents of the test questions. The knowledge points summarized and sorted according to the collected various test questions are stored in the information database.

By way of example, the knowledge points are illustrated such that the sum of the internal angles of the gap filling question "(____) is 180 °," the sum of the internal angles of the triangle is (____) ", and the question" the sum of the internal angles of the triangle is 270 °. (×) ", choosing the sum of the internal angles of the title triangle isB. A. 90 °, B, 180 °, C, 270 °, D, 360 °. The topics of these questions are all different, but all involve a unified knowledge point, i.e. the sum of the internal angles of the triangle is 180 °. Through summarization, the collected massive topic contents are organized into individual knowledge points and stored in the information database 104.

The application server 103 determines whether the answer of the user is correct based on the retrieved knowledge point, and if correct, marks "v" and if incorrect, marks "x". And returning the marked result to the user.

The examination question photographing and judging method and the examination question photographing and judging device can be applied to different subjects, such as mathematics, Chinese, foreign languages, history and other subjects, and are mainly applied to the Chinese in the embodiment of the invention for illustration.

The method and the device for photographing and judging test questions can be applied to different question types, such as blank filling, selection, judgment, application questions and the like, and in the embodiment of the embodiment, the method and the device are mainly applied to the blank filling, and particularly, the blank filling of Chinese characters according to pinyin is taken as an example for explanation.

Fig. 2 is a schematic flow chart of a test question photographing evaluation method according to an embodiment of the present invention, as shown in fig. 2, the method includes:

s201, acquiring an image file submitted by a user, for example, identifying first text data corresponding to a topic in the image file and second text data corresponding to a user response in the image file through a recognition model based on a neural network.

In this embodiment, a user uses a mobile phone application to photograph a test question, and then uploads a photographed photo through a network, where the photo includes the question and an answer handwritten by the user. And constructing a recognition model based on a neural network in advance, and recognizing the questions and answers in the picture through the recognition model.

On the basis of the above technical solution, further, the first text data generally corresponds to a print form text in the image file, and the second text data generally corresponds to a handwriting form in the image file.

In the present embodiment, the titles are usually printed, the characters are arranged in order, the fonts are unified into a song style, a simulated song style, a black body, and the like, and the recognition result corresponds to the first text data. However, the answer is handwritten by the user, so that the writing is messy, and the recognition result corresponds to the second text data. Fig. 3 is a schematic diagram of a photographed photograph, and as shown in fig. 3, the first text data 301 corresponds to bold and song characters, and the second text data 302 corresponds to handwriting.

The specific way of identifying the question and answering by the user in this step is not limited, for example, the question is generally printed, and the answer to be answered by the user is generally handwritten, so that the question and the user can be identified and answered based on the answer. In addition, differentiation can be made based on differences in the format and location of the topics and user responses. Alternatively, different backgrounds or watermarks can be set in the topic and user response areas in advance and then distinguished based on the backgrounds or watermarks.

For example, a recognition model based on a neural network may be constructed in advance, and the topics and answers in the photos are recognized through the recognition model.

On the basis of the technical scheme, further, the neural network-based recognition model adopts a combination of CPTN and a CRNN neural network model.

In the present embodiment, the recognition model based on the Neural Network is a CPTN (connected Text forward Network, natural scene Text detection) Neural Network model combined with CRNN (cyclic Convolutional Neural Network). The recognition model can be trained by using a large number of images containing printed questions and handwritten answers, and the images used for training should mark the question areas and the answer areas in advance.

In the recognition process, the photograph is first converted to a gray scale map. Although the answers to the titles and handwriting are usually black and the picture looks black and white, the picture is actually an RGB three-channel color picture, and needs to be grayed out to become a single-channel grayscale image. And then, positioning and cutting the region range of the characters by using a CPTN model, wherein the font of the first text data is regular, and the region edge is regular. The area of the second text data is relatively messy, and the CPTN model can distinguish the areas of the first text data and the second text data through area positioning. The CPTN model mainly includes a convolutional layer, Conv1D, and a full link layer.

After the positioning and cutting of the text area range are completed, the CRNN model carries out the text conversion of the title information in the picture into first text data, and the text conversion of the handwritten answer of the user into second text data.

S202, cutting and splitting the first text data based on the spatial position, and determining the question stem content to be retrieved.

And S203, matching answer text data corresponding to the question stem content from an information database by using a matching model based on semantic retrieval.

Because of the multi-meaning and ambiguity of a language, a computer cannot directly understand a human language, especially an understanding of a sentence or a piece of text. With the continuous improvement of computer performance, currently, by constructing a semantic retrieval model and using a large number of sample training, the semantic retrieval model can gradually understand the true meaning of a sentence, for example, "weather is good today" and "weather is good today" can be identified as expressing one meaning. When semantic retrieval is performed, the semantic retrieval model expands words in the retrieved sentence, for example, the retrieved sentence uses a "mobile phone", and the semantic retrieval model expands words such as a "mobile phone", "telephone", and a "mobile terminal" when retrieval is performed.

The semantic retrieval model can be trained through supervised learning or unsupervised learning at present, and in order to make the semantic retrieval more accurate, supervised learning is generally used. A large number of language samples are collected and manually indexed to index the true intent of the sentence. Then dividing the sample into a training sample and a checking sample, firstly training the semantic retrieval model by using the training sample, and adjusting the parameters of the model until the data is converged. And then, checking the semantic retrieval model by using the checking sample, finishing the training if the semantic retrieval model passes the checking, and otherwise, re-indexing and training.

The semantic retrieval model of the embodiment is used for retrieving based on the semantics of the question, and is not based on the words of the question as in the common retrieval. The technical problem to be solved and solved by the embodiment is that if an answer corresponding to the question is not directly available in the answer library, the answer is not given in the ordinary search, and the question and the answer with different characters but the same actual meaning (the same or similar semantics) can be matched in the semantic search mode of the embodiment, so that the question search range is wider, and the answer is easier to find.

In the semantic retrieval system, besides providing keywords to realize topic retrieval, the system can also combine with natural language processing and knowledge expression language to express various structured, semi-structured and unstructured information and provide multi-path and multifunctional retrieval.

In the embodiment, a large number of knowledge points are stored in the information database, the knowledge points are summarized and summarized from collected topics, each topic corresponds to one or more knowledge points, and one knowledge point can derive one or more topics. The information database can be sorted in a manual induction and summarization mode, and can also be sorted, induced and sorted by using a deep learning-based TextCNN neural network model. And searching the knowledge points, namely answer text data, which are matched with the first text data semantics (such as knowledge research type topics) in the information database by using a matching model based on semantic search.

Through the semantic retrieval information database, the situation that in the prior art, although the titles have the same meaning, correct answers cannot be found due to different description characters is avoided. Even if the question bank does not have the same question as the uploaded question, the corresponding answer can be found according to the knowledge points.

On the basis of the above technical solution, further, matching answer text data corresponding to the first text data from an information database using a matching model specifically includes the following steps, as shown in fig. 4:

s2021, cutting the first text data based on the spatial position, and splitting the first text data into a plurality of layers of titles.

In the present embodiment, the test questions are generally divided into large and small questions, and the large questions are generally types of introduction questions, such as a blank filling question, a selection question, an application question, and the like. The small topic is usually a specific topic stem, that is, a requirement for the user, and what the user is allowed to complete. FIG. 3 shows a diagram with large title "one, write: when looking at Pinyin to write words, please write correctly and normatively in the grid of Tian character. (10 min) ", the subject, i.e., the subject-stem content, was" gu ī l ǜ "," zh im "u", "m ǐ n ji e", "p i ng h ng".

In this embodiment, when the first text data is cut, the second text data is cut according to the content of the cut of the first text data, for example, after "gu ī l ǜ" in fig. 3, the "cut" of the second text data is also cut at the same time, so that the content of the first text data and the content of the second text data always correspond to each other.

For another example, a general application question may include a main part of the question and also include a plurality of small questions, and each small question may be added with some known conditions as a part of the question; in the step, the first text data is cut based on the spatial position, split and combined into a plurality of associated questions, and targeted retrieval is performed.

S2022, determining the test question type and the question stem content based on the questions of the multiple layers.

In this embodiment, the first text data is split into a large topic and a small topic by splitting. For example, in fig. 3, the question type is determined as a blank filling question according to the large question, and the small question is the corresponding Chinese character for filling pinyin.

In this step, the type and content of the question stem to be searched currently are determined. S2023, performing text structuring processing on the question stem content.

In the embodiment, different question types have different structuring rules, after a good question type is determined according to a big question, the corresponding structuring rule is selected to structure the question stem content, namely, the small question content, and text segment information is selected from the small question content and converted into a specific format text, so that the subsequent matching model can be conveniently calculated.

The content structuring requires defining the structure. Given an unordered data, the process of defining the structure is by what logic to group. With the structure, the reuse technique realizes the structuring of the unordered data. For example, in fig. 3, data such as "gu ī l ǜ", "zh hu," m ǐ n ji "," p i ng "," regularity "," wisdom "," agility "," balance ", and the like are analyzed, and are unordered data for a computer, and the computer cannot correspond" gu ī l ǜ "and" regularity "," zh hu "and" wisdom "," m ǐ n ji "and" agility "," p i ng hu ng "and" balance "to one subject, thereby constituting an improved basic unit. When the Chinese characters are actually corrected, the pinyin and the following handwritten Chinese characters need to be known as a question, and meanwhile, the correction can be realized only according to the question types of the pinyin written Chinese characters. For example, "gu ī l ǜ" and "regular" are associated with a modified basic unit, and the process of determining the title as writing Chinese characters according to pinyin is a structured process. By structuring, a computer program or model is enabled to know what to do and how to judge the topic.

The structure of word selection and blank filling is different from the above problem. Similarly, the question type styles are various, and the same question type has different structures, so that the computer needs to determine the structures and the question types and know what the question needs to be made and how to modify the question. And S2024, inputting the subject stem content after the text structuring processing into a matching model, and matching answer text data corresponding to the subject stem content from an information database.

For example, in the present embodiment, the topic contents of the text structuring process are matched in the information database, and the knowledge point with the highest matching degree is selected as the basis for evaluating the second text data.

On the basis of the technical scheme, the matching model can be a Bi-LSTM neural network model.

In this embodiment, the Bi-LSTM (Bidirectional Long short term Memory) neural network model is a kind of cyclic neural network, and can better consider preceding and following words of a sentence, for example, "i do not feel good weather today", where "not" defines the following "good weather" to mean negation of good weather, and the Bidirectional Long and short term Memory network model can better capture the dependency of a longer distance, and can also consider the definition of the following words to the preceding words, for example, "not travel in today's cold", where "does not travel" is the modification and limitation of "cold". In the present embodiment, the bidirectional long-term and short-term memory network model is trained by using data of the question bank as training data in a supervised learning manner.

S204, judging the second text data based on the answer text data.

In this embodiment, the answer text data is knowledge point data obtained by searching through the matching model, and whether the second text data is correct is determined according to the knowledge point data.

On the basis of the above technical solution, further, the judging the second text data based on the answer text data specifically includes:

and judging whether the answer text data is matched with the second text data or not, and if so, judging.

In this embodiment, it is preferable that the answer text data is matched with the second text data by semantic meaning, and when the answer text data is judged to be matched with the second text data, it is indicated that the answer of the user matches the content of the knowledge point.

On the basis of the technical scheme, further, if the answer text data is not matched with the second text data, matching corresponding reverse topic text data in an information database based on the second text data;

setting a similarity threshold, calculating the similarity between the reverse-reasoning topic text data and the first text data, if the similarity exceeds the similarity threshold, judging the similarity, and otherwise, judging the similarity in error.

The reverse-deducing question text data mainly comprises a question stem, and when the similarity of the reverse-deducing question text data and the first text data is compared, the question stem of the reverse-deducing question is mainly compared with the question stem of the question corresponding to the first text data.

In this embodiment, there may be a problem of a recognition error, for example, the "gu ī l ǜ" in fig. 3 is recognized as "g lu" due to unclear photo shooting or other reasons, at this time, the answer text data matched by the matching model is "bone lu", and the judgment on the "rule" of the second text data by using "bone lu" may be wrong. In order to avoid the occurrence of this situation, at this time, the "rule" is used to obtain the reverse-thrust question text data "gu ī l ǜ" in the information data control in a matching manner, the "gu ī l ǜ" is used to calculate the similarity with "g lu", and the similarity obtained by calculation is 85%. The preset similarity threshold is 70%, so that it can be judged that there is a high possibility that an error occurs in the recognition of the first text data, and the first text data should be "gu ī l ǜ", at this time, the user answer "rule" is correct, and it should be judged that the error occurs.

Preferably, when the similarity between the reverse-inferred topic text and the topic stem content data exceeds the preset similarity threshold, the user is prompted while a decision is given, for example, whether the topic stem is "gu ī l ǜ"?

Further, in order to enhance management of wrong question identification, after the similarity between the reversely-deduced question text and the question stem content data exceeds the similarity threshold, a prompt message is provided to the user to ask the user to confirm whether the identified question stem content data is correct, answer data is continuously determined according to the question stem content data confirmed by the user, and evaluation is continuously performed according to the answer data. And after the user confirms, correcting the question stem content data, recalculating answer text data, and judging second text data according to the recalculated answer text data. And the related information of the wrong-identification stem content data, such as the related information of pictures, is stored in the system, so that the subsequent managers and the system can conveniently analyze and learn the identification errors, and the integral identification accuracy of the system is improved. On the basis of the technical scheme, a prompt information database is further preset, and after the judgment operation is completed, prompt information matched with the first text data is retrieved from the prompt information database and recommended to a user.

In the embodiment, a prompt information database is further provided, the prompt information is a supplement to the knowledge points, for example, the number of pronunciations of the word, "regular" stroke order is what, meaning is what, and the like, and the knowledge plane of the user can be enriched through the prompt information, so that the understanding of the knowledge points is deepened. The image can be returned to the user in a link form, and can also be directly marked on the image uploaded by the user.

In the embodiment, a wrong question database is further arranged, wrong questions of the user are collected, the wrong questions of the user are collected and classified, when the total number of the wrong questions or the number of one classification reaches a certain numerical value, for example, the number of the wrong questions of the pinyin blank filling questions reaches 20, corresponding knowledge points are selected to generate exercise test questions, the exercise test questions are pushed to the user to strengthen exercise, understanding of the user on the knowledge points is strengthened, and the learning achievement of the user can be well improved.

Fig. 5 is a schematic structural diagram of a neural network-based examination question photographing evaluation device 500 according to an embodiment of the present invention, as shown in fig. 5, including:

the identification module 501 is configured to acquire an image file submitted by a user, and identify first text data corresponding to a topic in the image file and second text data corresponding to a user response in the image file based on an identification model of a neural network.

In this embodiment, a user uses a mobile phone application to photograph a test question, and then uploads a photographed photo through a network, where the photo includes the question and an answer handwritten by the user. And constructing a recognition model based on the neural network in advance, and recognizing the questions and answers in the piece through the recognition model.

On the basis of the above technical solution, further, the first text data corresponds to a print form text in the image file, and the second text data corresponds to a handwriting form in the image file.

In the present embodiment, the title is usually printed, characters are arranged in order, and the first text data is usually a font of song, a song-like font, a black body, or the like. However, the answer is the second text data because the user writes by hand, and the writing is messy. Fig. 3 is a diagram illustrating a photographed photograph, and as shown in fig. 3, the first text data 301 uses bold and song characters, and the second text data 302 uses handwriting.

In the present embodiment, the recognition model based on the Neural Network is a CPTN (connected Text forward Network, natural scene Text detection) Neural Network model combined with CRNN (cyclic Convolutional Neural Network). The recognition model is trained using a large amount of historical print text data and handwritten text data.

A determining module 502, configured to cut and split the first text data based on the spatial position, and determine a stem content;

and a matching module 503, for matching answer text data corresponding to the question stem content from the information database by using a matching model based on semantic retrieval.

In the embodiment, a large number of knowledge points are stored in the information database, the knowledge points are summarized and summarized from collected topics, each topic corresponds to one or more knowledge points, and one knowledge point can derive one or more topics. The information database can be sorted in a manual induction and summarization mode, and can also be sorted, induced and sorted by using a deep learning-based TextCNN neural network model.

And searching the knowledge points semantically matched with the first text data, namely answer text data in the information database by using a matching model based on semantic search.

On the basis of the technical scheme, the matching model is a Bi-LSTM neural network model.

A judging module 504, configured to judge the second text data based on the answer text data.

In this embodiment, matching the answer text data with the second text data is also performed by semantic meaning, and when the answer text data and the second text data are judged to match, it is indicated that the answer of the user matches the content of the knowledge point, and therefore, the position of the second text data in the picture uploaded by the user can be plotted as √.

Further, in order to enhance management of wrong question identification, after the similarity between the reversely-deduced question text and the question stem content data exceeds the similarity threshold, a prompt message is provided to the user to ask the user to confirm whether the identified question stem content data is correct, answer data is continuously determined according to the question stem content data confirmed by the user, and evaluation is continuously performed according to the answer data. And after the user confirms, correcting the question stem content data, recalculating answer text data, and judging second text data according to the recalculated answer text data. And the related information of the wrong-identification stem content data, such as the related information of pictures, is stored in the system, so that the subsequent managers and the system can conveniently analyze and learn the identification errors, and the integral identification accuracy of the system is improved.

On the basis of the technical scheme, a prompt information database is further preset, and after the judgment operation is completed, prompt information matched with the first text data is retrieved from the prompt information database and recommended to a user.

FIG. 7 is a block diagram illustrating an electronic device in accordance with an example embodiment.

As shown in fig. 7, the electronic device is in the form of a general purpose computing device. The processor can be one or more and can work together. The invention also does not exclude that distributed processing is performed, i.e. the processors may be distributed over different physical devices. The electronic device of the present invention is not limited to a single entity, and may be a sum of a plurality of entity devices.

The memory stores a computer executable program, typically machine readable code. The computer readable program may be executed by the processor to enable an electronic device to perform the method of the invention, or at least some of the steps of the method.

The memory may include volatile memory, such as Random Access Memory (RAM) and/or cache memory, and may also be non-volatile memory, such as read-only memory (ROM).

Optionally, in this embodiment, the electronic device further includes an I/O interface, which is used for data exchange between the electronic device and an external device. The I/O interface may be a local bus representing one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, and/or a memory storage device using any of a variety of bus architectures.

It should be understood that the electronic device shown in fig. 7 is only one example of the present invention, and elements or components not shown in the above example may be further included in the electronic device of the present invention. For example, some electronic devices further include a display unit such as a display screen, and some electronic devices further include a human-computer interaction element such as a button, a keyboard, and the like. Electronic devices are considered to be covered by the present invention as long as the electronic devices are capable of executing a computer-readable program in a memory to implement the method of the present invention or at least a part of the steps of the method.

Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, as shown in fig. 7, the technical solution according to the embodiment of the present invention may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which may be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present invention.

The software product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).

The computer readable medium carries one or more programs which, when executed by a device, cause the computer readable medium to perform the functions of: starting a virtual interaction function based on a starting instruction of a user; acquiring a real-time video of the user based on a virtual interaction function; inputting the real-time video into an action recognition model to generate an action recognition label, wherein the action recognition model is realized through a deep learning model; generating a virtual object according to the action identification label; and drawing the target virtual object in a real-time video of the user for displaying.

Those skilled in the art will appreciate that the modules described above may be distributed in the apparatus according to the description of the embodiments, or may be modified accordingly in one or more apparatuses unique from the embodiments. The modules of the above embodiments may be combined into one module, or further split into multiple sub-modules.

Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein may be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present invention.

Exemplary embodiments of the present invention are specifically illustrated and described above. It is to be understood that the invention is not limited to the precise construction, arrangements, or instrumentalities described herein; on the contrary, the invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A test question photographing and judging method is characterized by comprising the following steps:

and judging the second text data based on the answer text data.

2. The examination question photographing and judging method of claim 1, wherein: identifying first text data corresponding to the titles in the image file based on a recognition model of a neural network;

3. The examination question photographing and judging method of claim 1, wherein the cutting and splitting the first text data based on the spatial position to determine the subject stem content to be retrieved comprises:

4. The examination question photographing and judging method of any one of claims 1 to 3, wherein the matching of answer text data corresponding to the subject matter content from an information database using a matching model based on semantic retrieval comprises:

performing text structuring processing on the question stem content;

5. The examination question photographing judging method of claim 1, wherein the judging the second text data based on the answer text data comprises: judging whether the answer text data is matched with the second text data or not, and if so, judging;

6. The examination question photographing and judging method of claim 5, wherein: and judging whether the answer text data is matched with the second text data or not through a semantic-based matching model.

7. The examination question photographing and judging method of claim 1, wherein:

and presetting a prompt information database, and after finishing the evaluation operation, retrieving prompt information matched with the first text data from the prompt information database and recommending the prompt information to a user.

8. The utility model provides a device is judged to examination question shooing which characterized in that, the device includes:

9. An electronic device, comprising:

one or more processors;

storage means for storing one or more programs;

when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-6.

10. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-6.