CN115757731A - Dialogue question rewriting method, device, computer equipment and storage medium - Google Patents
Dialogue question rewriting method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN115757731A CN115757731A CN202211458362.6A CN202211458362A CN115757731A CN 115757731 A CN115757731 A CN 115757731A CN 202211458362 A CN202211458362 A CN 202211458362A CN 115757731 A CN115757731 A CN 115757731A
- Authority
- CN
- China
- Prior art keywords
- question
- text
- answer
- dialogue
- latest
- 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.)
- Pending
Links
Images
Abstract
The embodiment of the application belongs to the technical field of artificial intelligence and financial science and technology, is applied to the field of intelligent conversation application of robots, and relates to a method, a device, computer equipment and a storage medium for rewriting dialogue questions, which comprise receiving question-answer context texts sent by a dialogue question rewriting request end; inputting a dialogue question sentence rewriting model by taking the question and answer context text as an input text, and performing text rewriting on the latest question sentence text according to the historical dialogue text; the latest question text after rewriting is obtained and output to a responder under the current dialogue scene, characters in the question and answer text can be accurately classified according to intention characteristics by using the advantages of the U-net neural network, enough text semantic information can be explored from multiple angles and full scale of the historical question and answer text by using the advantages of the depllabv 3 neural network, user information omission is more accurately complemented, user problem semantics are enriched, the understanding capacity of the dialogue robot on the user intention is enhanced, and the resolution of user problems is improved.
Description
Technical Field
The present application relates to the field of artificial intelligence and robot intelligent dialogue technologies, and in particular, to a method and an apparatus for rewriting a dialogue question, a computer device, and a storage medium.
Background
The man-machine conversation is an important task, and the problems of front-back cross reference and information omission often occur in daily communication of people at present. This is a problem that it is difficult for a robot to accurately understand the intention of a speaker, and particularly, in early consultation and sale stages of insurance and later underwriting stages, it is often difficult for an electric marketing robot to automatically deal with the consultation problem of an insurance client.
The conventional rewriting model is generally classified into a pointer network and a generation model. The two methods decode and output the rewritten sentence word by word according to the input, namely, the current sentence and the above text are semantically coded by using deep learning coding, the current sentence and the above text are combined into a dictionary during decoding, and one word is selected from the dictionary at each step until the decoding is finished. This approach does not take into account that the rewritten sentence has substantially the same structure as the current sentence, and the above is simply added to the current sentence as missing auxiliary information, and takes a long time.
The RUN model is proposed by the paper Incomplex Utterance writing as Semantic Segmentation, the Rewriting task is converted into a Semantic Segmentation task based on U-net, the problem is changed into the prediction of a word level editing matrix by introducing editing operation, and the model is used for editing the current sentence, so that the approximate structure of the current sentence is reserved, and the speed is higher. Although the problems are solved, the model still has some defects, firstly, the similarity among words extracted by a coding layer from multiple angles is not comprehensive enough, and other angles are lacked; secondly, a U-net network is adopted at the semantic segmentation layer, the network lacks the capability of exploring enough information from a full scale, and therefore the position and the boundary of the segmentation cannot be definitely obtained.
Disclosure of Invention
An object of the embodiments of the present application is to provide a method, an apparatus, a computer device, and a storage medium for rewriting a dialog question, so that not only can enough text information be explored from a full scale, but also a dialog intention of a user can be accurately predicted, and a resolution of a problem of the user can be improved.
In order to solve the above technical problem, an embodiment of the present application provides a method for rewriting a dialog question, which adopts the following technical solutions:
a dialogue question rewriting method comprises the following steps:
receiving a question and answer context text sent by a dialogue question rewriting request end, wherein the question and answer context text comprises a latest question text which is provided by a user in real time under a current dialogue scene and a historical dialogue text which is generated by a question and answer party based on the current dialogue scene;
inputting a dialogue question sentence rewriting model by taking the question and answer context text as an input text, and performing text rewriting on the latest question sentence text according to the historical dialogue text;
and acquiring the rewritten latest question text, and outputting the latest question text as an output result to a responder in the current conversation scene, wherein the responder and the user have a question-answer conversation logical relationship in the current conversation scene.
Further, the dialogue question sentence rewriting model is formed by using a U-net neural network as a context analysis layer and a depeplabv 3 neural network as an encoding layer and a semantic segmentation layer, and the step of performing text rewriting on the latest question sentence text according to the historical dialogue text specifically includes:
inputting the historical dialogue text and the latest question text into the U-net neural network;
performing word segmentation and first labeling processing on the historical dialogue text and the latest question text respectively according to the U-net neural network, and constructing a two-dimensional question-answer matrix according to a first labeling processing result;
and performing second labeling prediction on the latest question text according to the two-dimensional question-answer matrix and the depeplabv 3 neural network, and rewriting the latest question text according to a second labeling prediction result.
Further, the step of performing word segmentation and first labeling processing on the historical dialog text and the latest question text respectively according to the U-net neural network, and constructing a two-dimensional question-answer matrix according to a first labeling processing result specifically includes:
introducing a BERT word segmentation model into the U-net neural network in advance, wherein the BERT word segmentation model is provided with a corresponding BERT word segmentation dictionary and character number information corresponding to different characters in the BERT word segmentation dictionary;
performing word segmentation processing on the historical dialogue text and the latest question text according to the BERT word segmentation model to obtain character number information of each character in the historical dialogue text and the latest question text in the BERT word segmentation dictionary;
and outputting the character number information in a two-dimensional matrix form according to the dialogue sequence of the historical dialogue text and the latest question text to obtain the two-dimensional question-answer matrix.
Further, the step of performing second labeling prediction on the latest question text according to the two-dimensional question-answer matrix and the depeplabv 3 neural network, and rewriting the latest question text according to a second labeling prediction result specifically includes:
inputting the two-dimensional question-answer matrix into the deplapbv 3 neural network;
according to a part-of-speech similarity analysis function preset in the deplapv 3 neural network, obtaining part-of-speech similarities between different characters corresponding to different character number information in the two-dimensional question-answer matrix;
predicting characters capable of carrying out reference resolution, omitted characters and replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question sentence text according to the part-of-speech similarity;
acquiring character number information corresponding to the omitted character and the replaced character according to the BERT word segmentation dictionary, and taking the character number information as a second labeling prediction result;
acquiring first label processing information corresponding to the latest question text after the omitted characters are added and the characters capable of carrying out reference resolution are updated to the replaced characters;
and acquiring corresponding characters according to the first labeling processing information and the BERT word segmentation dictionary to generate the latest question text after rewriting.
Further, the step of outputting the character number information in a form of a two-dimensional matrix according to the dialogue sequence of the historical dialogue text and the latest question text to obtain the two-dimensional question-answer matrix specifically includes:
acquiring the character number information corresponding to each character in each sentence of the dialog text according to the dialog sequence;
taking character number information corresponding to each character in each sentence of dialog text as matrix row information;
and adding matrix row information corresponding to each sentence of dialog text line by line according to the dialog sequence until the matrix row information corresponding to the latest question text is added finally, and finishing the construction of the two-dimensional question-answer matrix.
Further, the step of obtaining part-of-speech similarities between different characters corresponding to different character number information in the two-dimensional question-answer matrix according to a part-of-speech similarity analysis function preset in the depeplabv 3 neural network specifically includes:
according to a preset part-of-speech similarity analysis function: f (x) n ,c m )＝[element(h n ,u m )；cos(h n ,u m )；h n Wu m ；match_score(h n ,u m )]And acquiring part-of-speech similarity between different characters corresponding to different character number information in the two-dimensional question-answer matrix, wherein the element (h) is n ,u m ) Function representation acquiring the two dimensionsElement similarity, cos (h), between two target characters in the question-answer matrix n ,u m ) The function representation obtains the cosine similarity between the two target characters, h n Wu m Representing the acquisition of bilinear similarity between the two target characters, wherein W is a learnable parameter, match _ score (h) n ,u m )＝1/(1+|h n -u m L), wherein l h n -u m L is the Euclidean distance between the two target characters, match _ score (h) n ,u m ) H represents the similarity between the two target characters obtained according to the Euclidean distance between the two target characters n And u m And representing target characters corresponding to two arbitrary elements in the two-dimensional question-answering matrix, wherein n and m are positive integers and respectively represent the row number and the column number of the matrix.
Further, the step of predicting the characters capable of performing resolution, the omitted characters and the replacement characters corresponding to the characters capable of performing resolution according to the part-of-speech similarity in the latest question sentence text specifically includes:
presetting a reference resolution judgment condition and an omission judgment condition;
if the part-of-speech similarity meets the reference resolution judging condition, identifying characters capable of carrying out reference resolution and replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question text as corresponding predicted characters;
and if the part-of-speech similarity meets the omission judgment condition, identifying the omitted character in the latest question sentence text as a corresponding predicted character.
In order to solve the above technical problem, an embodiment of the present application further provides a device for rewriting a dialog question, which adopts the following technical solutions:
a dialogue question rewriting apparatus comprising:
the system comprises a question and answer text receiving module, a question and answer text receiving module and a question and answer text processing module, wherein the question and answer text receiving module is used for receiving a question and answer context text sent by a dialogue question rewriting request end, and the question and answer context text comprises a latest question text which is provided by a user in real time under a current dialogue scene and a historical dialogue text which is generated by a question and answer party at this time based on the current dialogue scene;
the model rewriting module is used for inputting a dialogue question sentence rewriting model by taking the question and answer context text as an input text, and performing text rewriting on the latest question sentence text according to the historical dialogue text;
and the model output module is used for acquiring the rewritten latest question sentence text and outputting the latest question sentence text as an output result to the responder in the current conversation scene, wherein the responder and the user have a question-answer conversation logic relationship in the current conversation scene.
In order to solve the above technical problem, an embodiment of the present application further provides a computer device, which adopts the following technical solutions:
a computer device comprising a memory and a processor, the memory having stored therein computer-readable instructions, the processor implementing the steps of the dialog question rewriting method described above when executing the computer-readable instructions.
In order to solve the above technical problem, an embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solutions:
a computer readable storage medium having computer readable instructions stored thereon which, when executed by a processor, implement the steps of a dialog question rewrite method as described above.
Compared with the prior art, the embodiment of the application mainly has the following beneficial effects:
according to the method for rewriting the dialogue question, a question and answer context text sent by a dialogue question rewriting request end is received, wherein the question and answer context text comprises a latest question text which is provided by a user in real time under a current dialogue scene and a historical dialogue text which is generated by a question and answer party at this time based on the current dialogue scene; taking the question-answer context text as an input text, inputting a dialogue question rewriting model, and rewriting the latest question text according to the historical dialogue text; the latest question sentence text after rewriting is obtained and is used as an output result to be output to a responder under the current dialogue scene, the dialogue question sentence rewriting model is formed by a U-net neural network as a context analysis layer and a deplabv 3 neural network as a coding layer and a semantic segmentation layer, characters in the question and answer text can be accurately classified according to the advantages of the U-net neural network, enough text semantic information can be searched from multiple angles and full scales of historical question and answer text according to the advantages of the deplabv 3 neural network, user information omission is more accurately complemented, user problem semantics are enriched, the understanding capacity of a dialogue robot on user intentions is enhanced, and the solution rate of user problems is improved.
Drawings
In order to more clearly illustrate the solution of the present application, the drawings needed for describing the embodiments of the present application will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present application, and that other drawings can be obtained by those skilled in the art without inventive effort.
FIG. 1 is an exemplary system architecture diagram in which the present application may be applied;
FIG. 2 is a flow diagram of one embodiment of a dialog question rewrite method according to the present application;
FIG. 3 is a flow diagram for one embodiment of step 202 shown in FIG. 2;
FIG. 4 is a flow diagram of one embodiment of step 302 of FIG. 3;
FIG. 5 is a flow diagram of one embodiment of step 403 shown in FIG. 4;
FIG. 6 is a flowchart of one embodiment of step 303 of FIG. 3;
FIG. 7 is a flowchart of one embodiment of step 603 of FIG. 6;
FIG. 8 is a schematic diagram illustrating the structure of one embodiment of a dialog question rewriting apparatus according to the present application;
FIG. 9 is a schematic diagram of one embodiment of 802 shown in FIG. 8;
FIG. 10 is a schematic block diagram of one embodiment of a computer device according to the present application.
Detailed Description
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description of the application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "including" and "having," and any variations thereof, in the description and claims of this application and the description of the above figures are intended to cover non-exclusive inclusions. The terms "first," "second," and the like in the description and claims of this application or in the above-described drawings are used for distinguishing between different objects and not for describing a particular order.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.
As shown in fig. 1, the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. Network 104 is the medium used to provide communication links between terminal devices 101, 102, 103 and server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
A user may use terminal devices 101, 102, 103 to interact with a server 105 over a network 104 to receive or send messages or the like. The terminal devices 101, 102, 103 may have various communication client applications installed thereon, such as a web browser application, a shopping application, a search application, an instant messaging tool, a mailbox client, social platform software, and the like.
The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to a smart phone, a tablet computer, an e-book reader, an MP3 player (Moving Picture experts Group Audio Layer III, motion Picture experts compression standard Audio Layer 3), an MP4 player (Moving Picture experts Group Audio Layer IV, motion Picture experts compression standard Audio Layer 4), a laptop portable computer, a desktop computer, and the like.
The server 105 may be a server providing various services, such as a background server providing support for pages displayed on the terminal devices 101, 102, 103.
It should be noted that the dialog question rewriting method provided in the embodiments of the present application is generally executed by a server/terminal device, and accordingly, the dialog question rewriting apparatus is generally installed in the server/terminal device.
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 diagram of one embodiment of a dialog question rewrite method according to the present application is shown. The method for rewriting the dialogue question comprises the following steps:
In this embodiment, the question and answer context text includes a latest question text provided by the user in real time in the current dialog scenario and a historical dialog text generated by the question and answer party based on the current dialog scenario.
In this embodiment, the dialog question rewriting model is formed by using a U-net neural network as a context analysis layer and using a depeplabv 3 neural network as a coding layer and a semantic segmentation layer.
By using the deplapv 3 neural network as the coding layer and the semantic segmentation layer of the dialogue question rewriting model, on one hand, the problem that the similarity among words cannot be comprehensively extracted from multiple angles when the U-net neural network is directly used is avoided, and on the other hand, the defect that sufficient semantic information is not explored from a full scale when the U-net neural network is directly used is overcome; on the other hand, the U-net neural network is used as a context analysis layer of the dialogue question and answer rewriting model, the defect that the depeplabv 3 neural network can only directly analyze the feature context of the picture target and cannot directly analyze the feature context of the question and answer text is overcome, finally, the U-net neural network is used as the context analysis layer, the depeplabv 3 neural network is used as a coding layer and a semantic segmentation layer to jointly form the dialogue question and answer rewriting model, the characters in the question and answer text can be accurately classified according to the advantages of the U-net neural network, and sufficient text semantic information can be searched from multiple angles and full scale of the history question and answer text according to the advantages of the depeplabv 3 neural network.
And 202, taking the question-answer context text as an input text, inputting a dialogue question rewriting model, and rewriting the latest question text according to the historical dialogue text.
In this embodiment, the step of text-rewriting the latest question text according to the historical dialog text specifically includes: inputting the historical dialog text and the latest question text into the U-net neural network; respectively carrying out word segmentation and first labeling on the historical dialogue text and the latest question text according to the U-net neural network, and constructing a two-dimensional question-answer matrix according to a first labeling processing result; and performing second labeling prediction on the latest question text according to the two-dimensional question-answer matrix and the depeplabv 3 neural network, and rewriting the latest question text according to a second labeling prediction result.
Dividing characters and carrying out first labeling processing on the historical dialog text and the latest question text through a U-net neural network, constructing a two-dimensional question-answer matrix according to a first labeling processing result, converting characters in the dialog text into different character number information, constructing the two-dimensional question-answer matrix by taking the character number information as an element, converting the dialog text into a two-dimensional matrix form, inputting the two-dimensional question-answer matrix form into a deplabv 3 neural network, and facilitating the deplabv 3 neural network to carry out semantic prediction on the question-answer context text through the two-dimensional question-answer matrix; and then, according to the semantic prediction result of the deplabv 3 neural network, performing second labeling prediction on the latest question text in the deplabv 3 neural network, namely predicting characters which can refer to resolution and omitted characters in the latest question text, and rewriting the latest question text according to the second labeling prediction result.
With continued reference to FIG. 3, FIG. 3 is a flowchart of one embodiment of step 202 of FIG. 2, comprising:
In this embodiment, the step of performing word segmentation and first labeling processing on the historical dialog text and the latest question text respectively according to the U-net neural network, and constructing a two-dimensional question-answer matrix according to a first labeling processing result specifically includes: introducing a BERT word segmentation model into the U-net neural network in advance, wherein the BERT word segmentation model is provided with a corresponding BERT word segmentation dictionary and character number information corresponding to different characters in the BERT word segmentation dictionary; performing word segmentation processing on the historical dialogue text and the latest question text according to the BERT word segmentation model to obtain character number information of each character in the historical dialogue text and the latest question text in the BERT word segmentation dictionary; and outputting the character number information in a two-dimensional matrix form according to the dialogue sequence of the historical dialogue text and the latest question text to obtain the two-dimensional question-answer matrix.
The method comprises the steps of introducing a BERT word segmentation model and a BERT word segmentation dictionary into a U-net neural network, segmenting characters in a question and answer context text and accurately obtaining character number information corresponding to each segmented character, then outputting the character number information in a two-dimensional matrix form according to a conversation sequence to obtain a two-dimensional question and answer matrix, and facilitating semantic prediction of the question and answer context text by a coding layer of a depeplabv 3 neural network through the two-dimensional question and answer matrix.
With continued reference to FIG. 4, FIG. 4 is a flowchart of one embodiment of step 302 of FIG. 3, comprising:
In this embodiment, the step of outputting the character number information in a form of a two-dimensional matrix according to the dialogue sequence of the historical dialogue text and the latest question text to obtain the two-dimensional question-answer matrix specifically includes: acquiring the character number information corresponding to each character in each sentence of the dialog text according to the dialog sequence; taking character number information corresponding to each character in each sentence of dialog text as matrix row information; and adding matrix row information corresponding to each sentence of dialog text line by line according to the dialog sequence until the matrix row information corresponding to the latest question text is added finally, and finishing the construction of the two-dimensional question-answer matrix.
And through a dialogue sequence, taking character number information corresponding to each character in each sentence of dialogue text as matrix row information, adding the matrix row information corresponding to each sentence of dialogue text line by line until the matrix row information corresponding to the latest question text is finally added, completing the construction of the two-dimensional question-answer matrix, fully considering the characteristics of the question-answer text in a dialogue task to construct the two-dimensional question-answer matrix, and ensuring that the original dialogue time sequence is not changed when the two-dimensional question-answer matrix is constructed and the two-dimensional question-answer matrix is more suitable for a dialogue service scene.
With continuing reference to FIG. 5, FIG. 5 is a flowchart of one embodiment of step 403 shown in FIG. 4, including:
In this embodiment, the step of performing second labeling prediction on the latest question text according to the two-dimensional question-answer matrix and the depeplabv 3 neural network, and rewriting the latest question text according to a second labeling prediction result specifically includes: inputting the two-dimensional question-answer matrix into the deplapbv 3 neural network; according to a part-of-speech similarity analysis function preset in the deplapv 3 neural network, obtaining part-of-speech similarities between different characters corresponding to different character number information in the two-dimensional question-answer matrix; predicting characters capable of carrying out reference resolution, omitted characters and replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question sentence text according to the part-of-speech similarity; acquiring character number information corresponding to the omitted character and the replaced character according to the BERT word segmentation dictionary, and taking the character number information as a second labeling prediction result; acquiring first labeling processing information corresponding to the latest question sentence text after the omitted characters are added and the characters capable of carrying out reference resolution are updated to the replacing characters; and acquiring corresponding characters according to the first label processing information and the BERT word segmentation dictionary to generate the rewritten latest question sentence text.
By utilizing the advantages of the deplaybv 3 neural network, enough text semantic information is explored from multiple angles and full scale of a historical question and answer text, characters capable of carrying out reference resolution, omitted characters and replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question and answer text are predicted, corresponding omitted characters and replacement characters corresponding to the characters capable of carrying out reference resolution are obtained according to the BERT word segmentation dictionary, the latest question and answer text is rewritten, the problem that the similarity among words cannot be comprehensively extracted from multiple angles when the U-net neural network is directly used is avoided, and the defect that the sufficient semantic information is lacked from full scale exploration due to the fact that the U-net neural network is directly used is overcome.
With continued reference to FIG. 6, FIG. 6 is a flowchart of one embodiment of step 303 of FIG. 3, including:
in this embodiment, the step of obtaining part-of-speech similarities between different characters corresponding to different character number information in the two-dimensional question and answer matrix according to a part-of-speech similarity analysis function preset in the depeplabv 3 neural network specifically includes: according to a preset part-of-speech similarity analysis function: f (x) n ，c m )＝[element(h n ，u m )；cos(h n ，u m )；h n Wu m ；match_score(h n ，u m )]And acquiring part-of-speech similarity between different characters corresponding to different character number information in the two-dimensional question-answer matrix, wherein the element (h) is n ,u m ) Function representation obtaining the twoElement similarity, cos (h), between two target characters in a dimensional question-answer matrix n ,u m ) The function representation obtains the cosine similarity between the two target characters, h n Wu m Representing the acquisition of bilinear similarity between the two target characters, wherein W is a learnable parameter, match _ score (h) n ,u m )＝1/(1+|h n -u m L), wherein, | h n -u m L is the Euclidean distance between the two target characters, match _ score (h) n ,u m ) Means for obtaining the similarity between the two target characters according to the Euclidean distance between the two target characters, h n And u m And representing target characters corresponding to two arbitrary elements in the two-dimensional question-answering matrix, wherein n and m are positive integers and respectively represent the row number and the column number of the matrix.
The similarity of target characters corresponding to two arbitrary elements in the two-dimensional question-answering matrix is obtained through multiple angles, the similarity comprises element similarity, cosine similarity, bilinear similarity and similarity based on Euclidean distance, the cosine similarity is the similarity of the target characters obtained based on data symmetry, the bilinear similarity enables the similarity of asymmetry analysis target characters to be introduced when the similarity is calculated through introducing learnable parameters, and meanwhile the similarity between the two target characters is obtained through using a symmetric mode and an asymmetric mode; in addition, the cosine similarity is the similarity between two target characters obtained from the angle of the direction between the text vectors, the Euclidean distance is the similarity between the two target characters obtained from the angle of the distance between the text vectors, and the part-of-speech similarity between the target characters is comprehensively obtained through a plurality of data analysis angles, so that the prediction accuracy of the part-of-speech similarity between the two target characters is ensured.
in this embodiment, the step of predicting the characters capable of performing resolution, the omitted characters, and the replacement characters corresponding to the characters capable of performing resolution according to the part-of-speech similarity in the latest question text specifically includes: presetting a reference resolution judgment condition and an omission judgment condition; if the part-of-speech similarity meets the reference resolution judging condition, identifying characters capable of carrying out reference resolution and replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question text as corresponding predicted characters; and if the part-of-speech similarity meets the omission judgment condition, identifying the omitted character in the latest question sentence text as a corresponding predicted character.
The characters capable of carrying out reference resolution, the omitted characters and the replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question text are predicted by presetting a reference resolution judgment condition and a omission judgment condition and combining the part-of-speech similarity, and the characters capable of carrying out reference resolution, the omitted characters and the replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question text are predicted under the condition of ensuring high accuracy and predictability of the part-of-speech similarity between two target characters, so that the accuracy of rewriting the latest question text is further improved.
In this embodiment, the resolution-indicating determination condition and the omission determination condition may be implemented by presetting the part-of-speech similarity result corresponding to the word similarity analysis function in step 602 and a threshold, for example, presetting the threshold corresponding to the resolution-indicating determination condition as a first similarity threshold, if the part-of-speech similarity between a character in the latest question text and a corresponding character in the historical question-answer text is greater than the first similarity threshold, the character in the latest question text satisfies the resolution-indicating determination condition, replacing the character in the latest question text with the corresponding character in the historical question-answer text as a replacement character, and similarly, presetting the threshold corresponding to the omission determination condition as a second similarity threshold, if the part-of-speech similarity between the character in the latest question text and the corresponding character in the historical question-answer text is greater than the second similarity threshold, the character in the latest question text satisfies the omission determination condition, and using the corresponding character in the historical question-answer text as a character appended to the position of the latest question text where the character is appended to the latest question.
With continuing reference to FIG. 7, FIG. 7 is a flowchart of one embodiment of step 603 of FIG. 6, including:
701, presetting a reference resolution judgment condition and an omission judgment condition;
and 703, if the part-of-speech similarity meets the omission judgment condition, identifying the omitted character in the latest question sentence text as a corresponding predicted character.
In this embodiment, after the step of obtaining the latest question text after rewriting and outputting the latest question text as an output result to the responder in the current dialog scenario, the method further includes: and the responder carries out semantic recognition on the rewritten latest question sentence text, generates a corresponding answer text according to a semantic recognition result and sends the answer text to the user.
By using the dialogue question rewriting model, the missing of user information and the contents of front and back mutual reference are supplemented, the semantics of user problems are enriched, the understanding capability of the dialogue robot on the user intentions is enhanced, and the solution rate of the user problems is improved.
The method comprises the steps that a question-answer context text sent by a dialogue question rewriting request end is received, wherein the question-answer context text comprises the latest question text which is provided by a user in real time under the current dialogue scene and historical dialogue texts which are generated by a question-answer party and a question-answer party based on the current dialogue scene; inputting a dialogue question sentence rewriting model by taking the question and answer context text as an input text, and performing text rewriting on the latest question sentence text according to the historical dialogue text; the latest question text after rewriting is obtained and is used as an output result to be output to a responder under the current dialogue scene, the dialogue question rewriting model is formed by taking a U-net neural network as a context analysis layer and taking a depllabv 3 neural network as a coding layer and a semantic segmentation layer, characters in the question and answer text can be accurately classified according to the advantages of the U-net neural network, enough text semantic information can be searched from multiple angles and full scale of the historical question and answer text according to the advantages of the depllabv 3 neural network, and the dialogue question rewriting model is used to complement the content of user information omission and front-back mutual reference, enrich the user problem semantics, enhance the comprehension ability of a dialogue robot to the user intention and improve the resolution of user problems.
The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.
The artificial intelligence base technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.
In the embodiment of the application, the dialogue question rewriting model is formed by a U-net neural network as a context analysis layer and a deplab v3 neural network as a coding layer and a semantic segmentation layer, not only can the characters in the question and answer text be accurately classified according to the advantages of the U-net neural network, but also enough text semantic information can be searched from multiple angles and full scales of the historical question and answer text according to the advantages of the deplab v3 neural network, the dialogue question rewriting model is used to complement the content of user information omission and front-back mutual reference, the user problem semantics are enriched, the understanding ability of a dialogue robot on the user intention is enhanced, the user problem solving rate is improved, and the dialogue robot is more intelligent and automatic.
With further reference to fig. 8, as an implementation of the method shown in fig. 2, the present application provides an embodiment of a dialog question rewriting apparatus, which corresponds to the embodiment of the method shown in fig. 2, and which is specifically applicable to various electronic devices.
As shown in fig. 8, a dialogue question rewriting apparatus 800 according to the present embodiment includes: a question and answer text receiving module 801, a model rewriting module 802 and a model output module 803. Wherein:
a question-answer text receiving module 801, configured to receive a question-answer context text sent by a dialog question rewriting request end, where the question-answer context text includes a latest question text that is provided by a user in real time in a current dialog scenario and a historical dialog text that is generated by both the question-answer party and the current dialog scenario based on the current dialog scenario;
a model rewriting module 802, configured to use the question-answer context text as an input text, input a dialogue question-sentence rewriting model, and rewrite the latest question-sentence text according to the historical dialogue text;
and a model output module 803, configured to obtain the latest question text after rewriting, and output the latest question text as an output result to a responder in the current dialog scenario, where a question-and-answer dialog logical relationship exists between the responder and the user in the current dialog scenario.
With continuing reference to FIG. 9, FIG. 9 is a block diagram illustrating an embodiment of 802 shown in FIG. 8, where the model rewrite module 802 includes: a text input sub-module 901, a U-net neural network processing sub-module 902, and a depeplabv 3 neural network processing sub-module 903. Wherein:
a text input submodule 901, configured to input the historical dialog text and the latest question text into the U-net neural network;
the U-net neural network processing sub-module 902 is configured to perform word segmentation and first labeling processing on the historical dialog text and the latest question text according to the U-net neural network, and construct a two-dimensional question-answer matrix according to a first labeling processing result;
and the depllabv 3 neural network processing submodule 903 is used for performing second labeling prediction on the latest question text according to the two-dimensional question-answer matrix and the depllabv 3 neural network, and rewriting the latest question text according to a second labeling prediction result.
The method comprises the steps that a question-answer context text sent by a dialogue question rewriting request end is received, wherein the question-answer context text comprises the latest question text which is provided by a user in real time under the current dialogue scene and historical dialogue texts which are generated by a question-answer party and a question-answer party based on the current dialogue scene; inputting a dialogue question sentence rewriting model by taking the question and answer context text as an input text, and performing text rewriting on the latest question sentence text according to the historical dialogue text; the latest question text after rewriting is obtained and is used as an output result to be output to a responder under the current dialogue scene, the dialogue question rewriting model is formed by taking a U-net neural network as a context analysis layer and taking a depllabv 3 neural network as a coding layer and a semantic segmentation layer, characters in the question and answer text can be accurately classified according to the advantages of the U-net neural network, enough text semantic information can be explored from multiple angles and full scale of the historical question and answer text according to the advantages of the depllabv 3 neural network, and the dialogue question rewriting model is used to complement the content of user information omission and front-back mutual reference, enrich the user question semantics, enhance the comprehension capability of a dialogue robot on the user intention, improve the user question resolution, and be more intelligent and automatic.
Those skilled in the art will appreciate that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to computer readable instructions, which can be stored in a computer readable storage medium, and when executed, the programs can include the processes of the embodiments of the methods described above. The storage medium may be a non-volatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM).
It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and may be performed in other orders unless otherwise indicated herein. Moreover, at least a portion of the steps in the flow chart of the figure may include multiple sub-steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of execution is not necessarily sequential, but may be performed alternately or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
In order to solve the technical problem, the embodiment of the application further provides computer equipment. Referring to fig. 10, fig. 10 is a block diagram of a basic structure of a computer device according to the present embodiment.
The computer device 10 includes a memory 10a, a processor 10b, and a network interface 10c, which are communicatively connected to each other via a system bus. It should be noted that only computer device 10 having components 10a-10c is shown, but it should be understood that not all of the shown components are required to be implemented, and that more or fewer components may be implemented instead. As will be understood by those skilled in the art, the computer device is a device capable of automatically performing numerical calculation and/or information processing according to a preset or stored instruction, and the hardware includes, but is not limited to, a microprocessor, an Application Specific Integrated Circuit (ASIC), a Programmable Gate Array (FPGA), a Digital Signal Processor (DSP), an embedded device, and the like.
The computer device can be a desktop computer, a notebook, a palm computer, a cloud server and other computing devices. The computer equipment can carry out man-machine interaction with a user in a keyboard mode, a mouse mode, a remote controller mode, a touch panel mode or a voice control equipment mode.
The memory 10a includes at least one type of readable storage medium including flash memory, hard disks, multimedia cards, card-type memory (e.g., SD or DX memory, etc.), random Access Memory (RAM), static Random Access Memory (SRAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), programmable Read Only Memory (PROM), magnetic memory, magnetic disks, optical disks, etc. In some embodiments, the storage 10a may be an internal storage unit of the computer device 10, such as a hard disk or a memory of the computer device 10. In other embodiments, the memory 10a may also be an external storage device of the computer device 10, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), or the like provided on the computer device 10. Of course, the memory 10a may also include both an internal storage unit and an external storage device of the computer device 10. In this embodiment, the memory 10a is generally used for storing an operating system and various types of application software installed in the computer device 10, such as computer-readable instructions of a dialog question rewriting method. Further, the memory 10a may also be used to temporarily store various types of data that have been output or are to be output.
The processor 10b may be a Central Processing Unit (CPU), a controller, a microcontroller, a microprocessor, or other data Processing chip in some embodiments. The processor 10b is typically used to control the overall operation of the computer device 10. In this embodiment, the processor 10b is configured to execute computer-readable instructions stored in the memory 10a or process data, for example, execute computer-readable instructions of the dialog question rewriting method.
The network interface 10c may comprise a wireless network interface or a wired network interface, and the network interface 10c is generally used for establishing communication connections between the computer device 10 and other electronic devices.
The embodiment provides computer equipment, and belongs to the technical field of intelligent robot dialogue application. The method comprises the steps that a question-answer context text sent by a dialogue question rewriting request end is received, wherein the question-answer context text comprises the latest question text which is provided by a user in real time under the current dialogue scene and historical dialogue texts which are generated by a question-answer party and a question-answer party based on the current dialogue scene; taking the question-answer context text as an input text, inputting a dialogue question rewriting model, and rewriting the latest question text according to the historical dialogue text; the latest rewritten question text is obtained and is used as an output result to be output to a responder under the current dialogue scene, characters in the question and answer text can be accurately classified according to intention characteristics by using the advantages of the U-net neural network, enough text semantic information can be explored from multiple angles and full scales of historical question and answer text by using the advantages of the depeplabv 3 neural network, the user information omission and the content of front and back mutual reference can be supplemented by using a dialogue question rewriting model, the user problem semantics can be enriched, the understanding ability of a dialogue robot on the user intention can be enhanced, the resolution of the user problem can be improved, and the dialogue robot is more intelligent and automatic.
The present application further provides another embodiment, which is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, which can be executed by a processor, so as to make the processor execute the steps of the dialogue question rewriting method.
The embodiment provides a computer readable storage medium, and belongs to the technical field of robot intelligent dialogue application. The method comprises the steps that a question-answer context text sent by a dialogue question rewriting request end is received, wherein the question-answer context text comprises the latest question text which is provided by a user in real time in the current dialogue scene and historical dialogue texts which are generated by a question-answer party and a question-answer party at this time based on the current dialogue scene; inputting a dialogue question sentence rewriting model by taking the question and answer context text as an input text, and performing text rewriting on the latest question sentence text according to the historical dialogue text; and acquiring the rewritten latest question text, and outputting the latest question text as an output result to a responder in the current dialogue scene, wherein the characters in the question and answer text can be accurately classified according to the intention characteristics by using the advantages of the U-net neural network, and sufficient text semantic information can be searched from multiple angles and full scales of the historical question and answer text by using the advantages of the depeplabv 3 neural network.
Through the description of the foregoing embodiments, it is clear to those skilled in the art that the method of the foregoing embodiments may be implemented by software plus a necessary general hardware platform, and certainly may also be implemented by hardware, but in many cases, the former is a better implementation. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present application.
It is to be understood that the above-described embodiments are merely illustrative of some, but not restrictive, of the broad invention, and that the appended drawings illustrate preferred embodiments of the invention and do not limit the scope of the invention. This application is capable of embodiments in many different forms and the embodiments are provided so that this disclosure will be thorough and complete. Although the present application has been described in detail with reference to the foregoing embodiments, it will be apparent to one skilled in the art that the present application may be practiced without modification or with equivalents of some of the features described in the foregoing embodiments. All equivalent structures made by using the contents of the specification and the drawings of the present application are directly or indirectly applied to other related technical fields and are within the protection scope of the present application.
Claims (10)
1. A dialogue question rewriting method is characterized by comprising the following steps:
receiving a question-answer context text sent by a dialogue question rewriting request end, wherein the question-answer context text comprises the latest question text which is provided by a user in real time under the current dialogue scene and a historical dialogue text which is generated by both the question-answer party and the current dialogue scene based on the current dialogue scene;
taking the question-answer context text as an input text, inputting a dialogue question rewriting model, and rewriting the latest question text according to the historical dialogue text;
and acquiring the rewritten latest question text, and outputting the latest question text as an output result to a responder in the current conversation scene, wherein the responder and the user have a question-answer conversation logical relationship in the current conversation scene.
2. The method for rewriting a question dialog according to claim 1, wherein the question dialog rewriting model is formed by using a U-net neural network as a context analysis layer and a deplabv 3 neural network as an encoding layer and a semantic segmentation layer, and the step of performing text rewriting on the latest question text according to the historical dialog text specifically includes:
inputting the historical dialog text and the latest question text into the U-net neural network;
respectively carrying out word segmentation and first labeling on the historical dialogue text and the latest question text according to the U-net neural network, and constructing a two-dimensional question-answer matrix according to a first labeling processing result;
and performing second labeling prediction on the latest question text according to the two-dimensional question-answer matrix and the depeplabv 3 neural network, and rewriting the latest question text according to a second labeling prediction result.
3. The method for rewriting a question and sentence in a dialog according to claim 2, wherein the step of performing word segmentation and first labeling processing on the historical dialog text and the latest question and sentence text respectively according to the U-net neural network, and constructing a two-dimensional question and answer matrix according to a first labeling processing result specifically includes:
introducing a BERT word segmentation model into the U-net neural network in advance, wherein a corresponding BERT word segmentation dictionary and character number information corresponding to different characters in the BERT word segmentation dictionary are arranged in the BERT word segmentation model;
performing word segmentation processing on the historical dialogue text and the latest question text according to the BERT word segmentation model to obtain character number information of each character in the historical dialogue text and the latest question text in the BERT word segmentation dictionary;
and outputting the character number information in a two-dimensional matrix form according to the dialogue sequence of the historical dialogue text and the latest question text to obtain the two-dimensional question-answer matrix.
4. The method for rewriting dialogue questions according to claim 3, wherein the step of performing second labeling prediction on the latest question text according to the two-dimensional question-answer matrix and the depeplabv 3 neural network, and rewriting the latest question text according to a second labeling prediction result specifically includes:
inputting the two-dimensional question-answer matrix into the depllabv 3 neural network;
according to a part-of-speech similarity analysis function preset in the deplapv 3 neural network, obtaining part-of-speech similarities between different characters corresponding to different character number information in the two-dimensional question-answer matrix;
predicting characters capable of carrying out reference resolution, omitted characters and replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question sentence text according to the part-of-speech similarity;
acquiring character number information corresponding to the omitted character and the replaced character according to the BERT word segmentation dictionary, and taking the character number information as a second labeling prediction result;
acquiring first labeling processing information corresponding to the latest question sentence text after the omitted characters are added and the characters capable of carrying out reference resolution are updated to the replacing characters;
and acquiring corresponding characters according to the first label processing information and the BERT word segmentation dictionary to generate the rewritten latest question sentence text.
5. The method for rewriting a question sentence in a dialog according to claim 3, wherein the step of outputting the character number information in a form of a two-dimensional matrix according to the dialog sequence between the history dialog text and the latest question sentence text to obtain the two-dimensional question-answer matrix specifically comprises:
acquiring the character number information corresponding to each character in each sentence of the dialog text according to the dialog sequence;
taking character number information corresponding to each character in each sentence of dialog text as matrix row information;
and adding matrix row information corresponding to each sentence of dialogue text line by line according to the dialogue sequence until the matrix row information corresponding to the latest question text is added finally, and finishing the construction of the two-dimensional question-answer matrix.
6. The method for rewriting a dialog question sentence according to claim 4, wherein the step of obtaining the part-of-speech similarity between different characters corresponding to different character number information in the two-dimensional question-answer matrix according to a part-of-speech similarity analysis function preset in the depeplabv 3 neural network specifically includes:
according to a preset part-of-speech similarity analysis function: f (x) n ，c m )＝[element(h n ，u m )；cos(h n ，u m )；h n Wu m ；match_score(h n ，u m )]And acquiring part-of-speech similarity between different characters corresponding to different character number information in the two-dimensional question-answer matrix, wherein the element (h) is n ，u m ) Function representation obtains element similarity, cos (h), between two target characters in the two-dimensional question-answer matrix n ，u m ) The function representation obtains the cosine similarity between the two target characters, h n Wu m Indicating that bilinear similarity between the two target characters is obtained, wherein W is a learnable parameter, match _ score (h) n ，u m )＝1/(1+|h n -u m L), wherein l h n -u m I is the Euclidean distance between the two target characters, match _ score (h) n ,u m ) Means for obtaining the similarity between the two target characters according to the Euclidean distance between the two target characters, h n And u m And representing target characters corresponding to two arbitrary elements in the two-dimensional question-answering matrix, wherein n and m are positive integers and respectively represent the row number and the column number of the matrix.
7. The method for rewriting a dialog question sentence according to claim 4, wherein the step of predicting characters which can be subject to resolution, omitted characters, and replacement characters corresponding to the characters which can be subject to resolution in the latest question sentence text according to the part-of-speech similarity specifically includes:
presetting a reference resolution judgment condition and an omission judgment condition;
if the part-of-speech similarity meets the reference resolution judging condition, identifying characters capable of carrying out reference resolution and replacement characters corresponding to the characters capable of carrying out reference resolution in the latest question text as corresponding predicted characters;
and if the part-of-speech similarity meets the omission judgment condition, identifying the omitted character in the latest question sentence text as a corresponding predicted character.
8. A dialogue question rewriting apparatus comprising:
the system comprises a question-answer text receiving module, a question-answer text receiving module and a question-answer text sending module, wherein the question-answer context text is used for receiving a question-answer context text sent by a dialogue question rewriting request end, and comprises the latest question text which is provided by a user in real time under a current dialogue scene and a historical dialogue text which is generated by a question-answer party and a question-answer party at this time based on the current dialogue scene;
the model rewriting module is used for inputting a dialogue question sentence rewriting model by taking the question and answer context text as an input text, and performing text rewriting on the latest question sentence text according to the historical dialogue text;
and the model output module is used for acquiring the rewritten latest question sentence text and outputting the latest question sentence text as an output result to the responder in the current conversation scene, wherein the responder and the user have a question-answer conversation logic relationship in the current conversation scene.
9. A computer device comprising a memory having computer readable instructions stored therein and a processor which when executed implements the steps of a dialog question rewriting method according to any one of claims 1 to 7.
10. A computer-readable storage medium having computer-readable instructions stored thereon which, when executed by a processor, implement the steps of the dialog question rewriting method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211458362.6A CN115757731A (en) | 2022-11-16 | 2022-11-16 | Dialogue question rewriting method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211458362.6A CN115757731A (en) | 2022-11-16 | 2022-11-16 | Dialogue question rewriting method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115757731A true CN115757731A (en) | 2023-03-07 |
Family
ID=85334036
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211458362.6A Pending CN115757731A (en) | 2022-11-16 | 2022-11-16 | Dialogue question rewriting method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115757731A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116881429A (en) * | 2023-09-07 | 2023-10-13 | 四川蜀天信息技术有限公司 | Multi-tenant-based dialogue model interaction method, device and storage medium |
| CN117057325A (en) * | 2023-10-13 | 2023-11-14 | 湖北华中电力科技开发有限责任公司 | Form filling method and system applied to power grid field and electronic equipment |
-
2022
- 2022-11-16 CN CN202211458362.6A patent/CN115757731A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116881429A (en) * | 2023-09-07 | 2023-10-13 | 四川蜀天信息技术有限公司 | Multi-tenant-based dialogue model interaction method, device and storage medium |
| CN116881429B (en) * | 2023-09-07 | 2023-12-01 | 四川蜀天信息技术有限公司 | Multi-tenant-based dialogue model interaction method, device and storage medium |
| CN117057325A (en) * | 2023-10-13 | 2023-11-14 | 湖北华中电力科技开发有限责任公司 | Form filling method and system applied to power grid field and electronic equipment |
| CN117057325B (en) * | 2023-10-13 | 2024-01-05 | 湖北华中电力科技开发有限责任公司 | Form filling method and system applied to power grid field and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112685565B (en) | Text classification method based on multi-mode information fusion and related equipment thereof | |
| CN115757731A (en) | Dialogue question rewriting method, device, computer equipment and storage medium | |
| CN112863683A (en) | Medical record quality control method and device based on artificial intelligence, computer equipment and storage medium | |
| CN112328761A (en) | Intention label setting method and device, computer equipment and storage medium | |
| CN113723105A (en) | Training method, device and equipment of semantic feature extraction model and storage medium | |
| CN112836521A (en) | Question-answer matching method and device, computer equipment and storage medium | |
| CN115438149A (en) | End-to-end model training method and device, computer equipment and storage medium | |
| CN114021582B (en) | Spoken language understanding method, device, equipment and storage medium combined with voice information | |
| CN112084779A (en) | Entity acquisition method, device, equipment and storage medium for semantic recognition | |
| CN115544560A (en) | Desensitization method and device for sensitive information, computer equipment and storage medium | |
| CN112434746B (en) | Pre-labeling method based on hierarchical migration learning and related equipment thereof | |
| CN115730603A (en) | Information extraction method, device, equipment and storage medium based on artificial intelligence | |
| CN115757725A (en) | Question and answer processing method and device, computer equipment and storage medium | |
| CN115238077A (en) | Text analysis method, device and equipment based on artificial intelligence and storage medium | |
| CN112732896B (en) | Target information display method, device, electronic equipment and medium | |
| CN114398466A (en) | Complaint analysis method and device based on semantic recognition, computer equipment and medium | |
| CN109933788B (en) | Type determining method, device, equipment and medium | |
| CN113657104A (en) | Text extraction method and device, computer equipment and storage medium | |
| CN113569741A (en) | Answer generation method and device for image test questions, electronic equipment and readable medium | |
| CN113450764A (en) | Text voice recognition method, device, equipment and storage medium | |
| CN114647733B (en) | Question and answer corpus evaluation method and device, computer equipment and storage medium | |
| CN115114407B (en) | Intention recognition method, device, computer equipment and storage medium | |
| CN116821298A (en) | Keyword automatic identification method applied to application information and related equipment | |
| CN117275466A (en) | Business intention recognition method, device, equipment and storage medium thereof | |
| CN117057362A (en) | Artificial intelligence-based intention recognition method, apparatus, device and storage medium |
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 |