CN113434650B - Question-answer pair expansion method and device, electronic equipment and readable storage medium - Google Patents
Question-answer pair expansion method and device, electronic equipment and readable storage medium Download PDFInfo
- Publication number
- CN113434650B CN113434650B CN202110729964.XA CN202110729964A CN113434650B CN 113434650 B CN113434650 B CN 113434650B CN 202110729964 A CN202110729964 A CN 202110729964A CN 113434650 B CN113434650 B CN 113434650B
- Authority
- CN
- China
- Prior art keywords
- text
- question
- label
- answer pair
- answer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000013515 script Methods 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 29
- 238000013136 deep learning model Methods 0.000 claims abstract description 20
- 238000012216 screening Methods 0.000 claims abstract description 8
- 239000013598 vector Substances 0.000 claims description 92
- 238000013507 mapping Methods 0.000 claims description 16
- 230000014509 gene expression Effects 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 12
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000004806 packaging method and process Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 14
- 238000004891 communication Methods 0.000 description 8
- 238000000605 extraction Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000004378 air conditioning Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/353—Clustering; Classification into predefined classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/90335—Query processing
- G06F16/90344—Query processing by using string matching techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention relates to the field of intelligent decision making, and discloses a question-answer pair expansion method, which comprises the following steps: constructing a fine-granularity label and a label marking script by utilizing a keyword set and one-class labels and two-class labels of the keyword set; screening keywords in each question-answer pair text in the question-answer text set by using a label marking script, and marking the keywords according to the corresponding fine granularity labels to obtain marked question-answer pair text; keyword replacement is carried out on the text by the marked questions and answers to obtain a similar text; training the text on the deep learning model by using the similarity text and the marked question-answering to obtain a text generation model; and generating an extended question-answer pair text of the to-be-extended question-answer pair text by using the text generation model. The invention also relates to a blockchain technique, wherein the labeled question-answer pair text can be stored in a blockchain node. The invention also provides a question-answer pair expanding device, electronic equipment and a storage medium. The invention can improve the expansion diversity of question-answering pairs.
Description
Technical Field
The invention relates to the field of intelligent decision making, in particular to a question-answer pair expansion method, a question-answer pair expansion device, electronic equipment and a readable storage medium.
Background
The search type question-answering system is one of the question-answering systems commonly used nowadays, and can receive questions input by users, search similar questions of the input questions in a question-answering library, and return the existing answers of the similar questions in the question-answering library as answers to the users, so that the search type question-answering system is widely applied to the industry due to the stable and controllable characteristics.
The user experience of the search question-answer system is closely related to the number of question-answer pairs in the question-answer library and the quality of the question-answer pairs. The number of question-answer pairs in the question-answer library is generally realized by expanding original question-answer pairs in the question-answer library.
Currently, the expansion of question-answer pairs relies on the generation of fixed templates that require user maintenance, are limited in number and are of a single template type, resulting in an expanded question-answer pair that is too single in number and type.
Disclosure of Invention
The invention provides a question-answer pair expansion method, a question-answer pair expansion device, electronic equipment and a computer readable storage medium, and aims to improve the diversity of question-answer pair expansion.
In order to achieve the above object, the invention provides a question-answer pair expansion method, which comprises the following steps:
obtaining a tag mapping table, wherein the tag mapping table comprises: a keyword set, a class label and a class label of each keyword in the keyword set;
Constructing a fine-granularity label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-granularity label and the keyword set;
acquiring a question-answer pair text set, screening keywords in each question-answer pair text in the question-answer text set by using the label marking script, and marking the keywords according to the corresponding fine-granularity labels to obtain marked question-answer pair texts;
replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different types of labels in the keyword set, so as to obtain a similar text;
training a pre-constructed deep learning model by using the similarity text and the marked question-answering text to obtain a text generation model;
and when receiving the to-be-expanded question-answer pair text, generating the expanded question-answer pair text by utilizing the text generation model.
Optionally, the constructing a fine-grained label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-grained label and the corresponding keyword set includes:
splicing the first class labels and the second class labels corresponding to each keyword in the keyword set by using preset characters to obtain corresponding fine-grained labels;
Constructing a regular expression according to the fine granularity label and all corresponding keywords in the keyword set;
and summarizing and packaging all the regular expressions into scripts to obtain the label marking scripts.
Optionally, training the text on the pre-constructed deep learning model by using the similarity text and the labeled question-answer to obtain a text generation model, including:
extracting characteristics of the text by the marked questions and answers and encoding the labels to obtain text sentence vectors;
extracting features of the similar texts to obtain similar text sentence vectors;
and carrying out iterative training on the deep learning model by using the text sentence vector and the similar text sentence vector to obtain the text generation model.
Optionally, the feature extraction and tag coding are performed on the text by the marking questions and answers to obtain text sentence vectors, which includes:
the text is segmented by utilizing a preset word segmentation dictionary to obtain an initial text word set;
combining each word in the initial text word set according to the sequence of the text in the marked questions and answers to obtain a text sequence;
converting each word in the text sequence into a vector, and combining all the vectors obtained by conversion according to the sequence of the corresponding words in the text sequence to obtain an initial sentence vector;
Carrying out tag coding on the text sequence according to the fine granularity tag corresponding to the text of the tag question and answer to obtain a tag vector;
and splicing the tag vector and the initial sentence vector to obtain the text sentence vector.
Optionally, the performing tag coding on the text sequence according to the fine granularity tag corresponding to the text by the tag questions and answers to obtain a tag vector includes:
acquiring a feature coding value of the fine granularity label;
determining the fine granularity label corresponding to the label question-answer pair text as a target fine granularity label;
constructing zero vectors of corresponding dimensions according to the number of words in the text sequence;
counting the positions of words marked by the target fine granularity labels in the text sequence to obtain label positions;
and replacing the element of the label position corresponding position in the zero vector with the characteristic coding value of the target fine granularity label corresponding to the label position to obtain the label vector.
Optionally, after the generating the text of the question-answer pair to be expanded by using the text generation model, the method includes:
calculating the confusion degree of the extended questions and answers to the text;
and carrying out text quality identification on the text by the question and answer to be expanded according to the confusion degree and a preset confusion degree threshold value, and obtaining an identification result of the text by the question and answer to be expanded.
Optionally, the text quality identification of the text by the question and answer to be expanded according to the confusion degree and a preset confusion degree threshold value, so as to obtain an identification result of the text by the question and answer to be expanded, which includes:
comparing the confusion degree with a preset confusion degree threshold value;
when the confusion degree is larger than or equal to the confusion degree threshold value, the authentication result is that authentication is not passed;
when the confusion degree is smaller than the confusion degree threshold value, the authentication result is authentication passing.
In order to solve the above problems, the present invention further provides a question-answer pair expansion device, which includes:
the script construction module is used for acquiring a tag mapping table, wherein the tag mapping table comprises: a keyword set, a class label and a class label of each keyword in the keyword set; constructing a fine-granularity label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-granularity label and the keyword set;
the model training module is used for acquiring a question-answer pair text set, screening keywords in each question-answer pair text in the question-answer text set by using the label marking script, marking the keywords according to the corresponding fine granularity labels, and obtaining marked question-answer pair text; replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different types of labels in the keyword set, so as to obtain a similar text; training a pre-constructed deep learning model by using the similarity text and the marked question-answering text to obtain a text generation model;
And the text expansion module is used for generating the to-be-expanded question-answer pair text by utilizing the text generation model when the to-be-expanded question-answer pair text is received.
In order to solve the above-mentioned problems, the present invention also provides an electronic apparatus including:
a memory storing at least one computer program; a kind of electronic device with high-pressure air-conditioning system
And the processor executes the computer program stored in the memory to realize the question-answer pair expansion method.
In order to solve the above-mentioned problems, the present invention also provides a computer-readable storage medium having stored therein at least one computer program that is executed by a processor in an electronic device to implement the question-answer pair expansion method described above.
According to the embodiment of the invention, the first class labels and the second class labels of the keywords are utilized to construct the fine-grained labels, the keywords in each question-answer pair text in the question-answer text set are analyzed, and the keywords are marked according to the corresponding fine-grained labels, so that marked question-answer pair texts are obtained; the keywords in the marked question-answering pair text are replaced by keywords with the same type of labels and different types of labels, so that similar texts are obtained, and the expansion types of the similar texts are more comprehensive and diversified; and training the pre-constructed deep learning model by using the similarity text and the marked question-answering text, wherein the obtained text generation model has better performance. Therefore, the question-answer pair expansion method, the question-answer pair expansion device, the electronic equipment and the readable storage medium provided by the embodiment of the invention improve the question-answer pair expansion diversity.
Drawings
FIG. 1 is a flow chart of a question-answer pair expansion method according to an embodiment of the application;
FIG. 2 is a schematic block diagram of a question-answer pair expansion device according to an embodiment of the present application;
fig. 3 is a schematic diagram of an internal structure of an electronic device for implementing a question-answer pair extension method according to an embodiment of the present application;
the achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The embodiment of the application provides a question-answer pair expansion method. The execution subject of the question-answer pair extension method includes, but is not limited to, at least one of a server, a terminal and the like capable of being configured to execute the method provided by the embodiment of the application. In other words, the question-answer pair extension method may be performed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The service end includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like.
Referring to fig. 1, a flow chart of a question-answer pair expansion method according to an embodiment of the present application is shown, where in the embodiment of the present application, the question-answer pair expansion method includes:
S1, acquiring a tag mapping table, wherein the tag mapping table comprises: a keyword set, a class label and a class label of each keyword in the keyword set;
in detail, the keyword set in the embodiment of the invention is a set of text words of different categories, the one-class label is a first-class category corresponding to each text word in the keyword set, and the two-class label is a second-class category under the first-class category, which is a further limitation to the first-class category. If the first class is "sports" and the second class is "basketball", one application example of the invention is: the keyword set comprises keywords such as father, child and the like, and then one type of label corresponding to the keyword "father" of the keyword set is an object, and the corresponding second type of label is a parent; one type of label of the keyword 'child' is an 'object', and the other type of label is a 'child'.
S2, constructing a fine-grained label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-grained label and the keyword set;
in detail, the embodiment of the invention uses preset characters to splice the first class labels and the second class labels corresponding to each keyword in the keyword set to obtain the corresponding fine-grained labels; if the first class label corresponding to the keyword set a is "object", the second class label corresponding to the keyword set a is "parent", and the preset character is "-", then the fine-grained label corresponding to the keyword set is "object-parent".
Further, in order to better label the subsequent text, a regular expression is constructed according to the fine-granularity label and all the keywords corresponding to the keyword set, wherein the regular expression is used for labeling the words in the keyword set in the text to be labeled. Further, since the keyword set includes a plurality of keywords, the embodiment of the invention can obtain a plurality of regular expressions, and further, the embodiment of the invention gathers and encapsulates all the regular expressions into a script to obtain the tag label script.
S3, acquiring a question-answer pair text set, screening keywords in each question-answer pair text in the question-answer text set by using the label marking script, and marking the keywords according to the corresponding fine-granularity labels to obtain a marked question-answer pair text;
the question-answer pair text set in the embodiment of the invention is a set of question-answer pair texts of different types, wherein the question-answer pair text is a text formed by questions and corresponding answers in a certain search type question-answer system. Optionally, in the embodiment of the present invention, the question-answer pair text set may be obtained from a background server of a search-type question-answer system.
In detail, in the embodiment of the invention, the regular expression in the label marking script is utilized to analyze and obtain the keyword in each question-answer pair text, and the keyword is marked according to the corresponding fine-granularity label to obtain the marked answer text. For example: question and answer pair text is "why buy insurance for parents? The parents are ensured to buy insurance, parents can be better guaranteed, the parents are better cared, and the 4 keywords 'parents' in the question-answer text are marked with fine-grained labels 'objects-parents', so that marked answer texts are obtained.
In another embodiment of the invention, the marking question-answer pair text can be stored in the blockchain node, and the efficiency of taking the marking question-answer pair text is improved by utilizing the characteristic of high throughput of the blockchain node.
S4, replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different type of labels in the keyword set, and obtaining a similar text;
in detail, in the embodiment of the present invention, the keywords in the labeled question-answer pair text are replaced with keywords in the keyword set, which have the same type of labels as the keywords and the type of labels are different from each other, so as to obtain similar text, for example, the labeled question-answer pair text is "why is the parents to buy insurance? And the keyword 'parent' word corresponds to the fine-grained label of 'object-parent', so that the 'parent' is replaced by the keyword 'child' corresponding to the fine-grained label of 'object-child' in the keyword set, and then the text is subjected to similar sentence conversion by the replaced marked questions and answers, so that the similar text can be 'buying insurance for children is not good'.
S5, training a pre-constructed deep learning model by using the similarity text and the marked question-answering text to obtain a text generation model;
in detail, in the embodiment of the present invention, training a text on a pre-constructed deep learning model by using the similarity text and the labeled question-answer to obtain a text generation model, including:
step A: extracting features of the text by the marked questions and answers and encoding labels to obtain text sentence vectors, and extracting features of the similar text to obtain similar text sentence vectors;
in detail, in the embodiment of the invention, the feature extraction and the tag coding are carried out on the text by the mark question and answer to obtain the text vector by the mark question and answer, which comprises the following steps: the text is segmented by using a preset word segmentation dictionary to obtain an initial text word set, and the word segmentation operation in the embodiment of the invention does not influence the fine granularity label corresponding to the text by the marked question and answer; combining each word in the initial text word set according to the sequence in the text to obtain a text sequence; converting each word in the text sequence into a vector, and combining all the vectors obtained by conversion according to the sequence of the corresponding words in the text sequence to obtain an initial sentence vector, for example: the text sequence contains two words in total, and the vector converted by the first word is that The vector of the second word conversion is +.>Then the combined sentence vector is +.>Carrying out tag coding on the text sequence according to the fine granularity tag corresponding to the text of the tag question and answer to obtain a tag vector; splicing the tag vector and the initial sentence vector to obtain the text sentence vector, for example: the initial sentence vector is +.>The label vector is [5 6 ]]Then the text sentence vector obtained by splicing the label vector and the initial sentence vector is +.>
Specifically, in the embodiment of the present invention, tag encoding is performed on the text sequence according to the fine granularity tag corresponding to the text by using the tag question and answer, so as to obtain a tag vector, including: acquiring characteristic code values of the fine-grained labels, wherein different fine-grained labels in the embodiment of the invention have corresponding characteristic code values, the characteristic code values are manually set numbers for distinguishing different fine-grained labels, for example, the characteristic code value of a fine-grained label 'object-child' is 1, and the characteristic code value of a fine-grained label 'object-parent' is 2; further, in the embodiment of the present invention, each fine-granularity tag corresponding to the tag question-answer pair text is determined as a target fine-granularity tag, for example, two fine-granularity tags "object-parent" and "object-child" corresponding to the tag question-answer pair text are determined as target fine-granularity tags; further, constructing a zero vector of a corresponding dimension according to the number of words in the text sequence; according to the embodiment of the invention, zero vectors with corresponding dimensions are constructed according to the number of words in the text sequence, for example, 5 words in the text sequence are included, and then the corresponding zero vectors are [0 0 0 0 0]; counting the positions of words marked by the target fine-granularity labels in the text sequence to obtain label positions, for example, the text sequence corresponds to two target fine-granularity labels, and the second word and the fourth word in the text sequence are marked respectively, and then the label positions are 2 and 4; and replacing the element of the position corresponding to the label position in the zero vector with the characteristic code value of the target fine-granularity label corresponding to the label position to obtain the label vector, for example, the zero vector is [0 0 0 0 0], the characteristic code value of the target fine-granularity label corresponding to the label position 2 is 1, the characteristic code value of the target fine-granularity label corresponding to the label position 2 is 2, and then the label vector is [0 1 0 2 0].
Optionally, the method for extracting features of the similar text in the embodiment of the present invention is the same as the method for converting the text of the labeled question-answer pair into the initial sentence vector, which is not described herein in detail.
And (B) step (B): calculating a loss value between the text sentence vector and the similar text sentence vector by using a preset loss function;
alternatively, the loss function in the embodiment of the present invention may be a cross entropy loss function.
Step C: and (C) when the loss value is greater than or equal to a preset loss threshold value, updating model parameters of the deep learning model, returning to the step (A) for iterative training, and stopping training until the loss value is less than the preset loss threshold value, so as to obtain the text generation model.
Optionally, in an embodiment of the present invention, the deep learning model is a Seq2Seq model.
S6, when receiving a to-be-expanded question-answer pair text, generating an expanded question-answer pair text by using the text generation model;
in detail, in the embodiment of the present invention, text generation is performed on the text by using the text generation model to the question and answer to be expanded, including: marking the fine granularity labels of the to-be-expanded question-answer pair text by using the label marking script to obtain marked to-be-expanded text; extracting features of the marked text to be expanded and encoding labels to obtain sentence vectors of the text to be expanded; and generating text of the to-be-expanded question-answer text by using the text generation model to obtain the expanded question-answer text pair.
Optionally, the feature extraction and tag encoding manners of the text to be expanded marked are the same as those used in S4, and are not described in detail herein.
Further, since the expanded question-answer generated by the generation model may be unreasonable text, in order to ensure the usability of the generated expanded question-answer on the text, the method may further include calculating the confusion degree of the expanded question-answer on the text; and carrying out text quality identification on the text by the question and answer to be expanded according to the confusion degree and a preset confusion degree threshold value, and obtaining an identification result of the text by the question and answer to be expanded.
Optionally, the embodiment of the present invention may calculate the confusion of the text by the expanded question-answer using a pre-constructed confusion model, where the confusion model may be an N-gram model.
In detail, in the embodiment of the present invention, text quality identification is performed on the text by the question and answer to be expanded according to the confusion degree and a preset confusion degree threshold value, so as to obtain an identification result of the text by the question and answer to be expanded, including: comparing the confusion degree with a preset confusion degree threshold value; when the confusion degree is smaller than the confusion degree threshold, the identification result is that identification passes, and the expanded question-answer pair text is added into a preset question-answer database; when the confusion degree is larger than or equal to the confusion degree threshold, the authentication result is that authentication is not passed, and the authentication result and the extended question-answer pair text are sent to a preset terminal device;
In the embodiment of the invention. The terminal device includes: intelligent terminal equipment such as mobile phones, computers, tablets and the like.
As shown in FIG. 2, a functional block diagram of the question-answer pair expansion device of the present invention is shown.
The question-answer pair expansion device 100 of the present invention may be installed in an electronic apparatus. Depending on the functions implemented, the question-answer pair expansion means may comprise a script construction module 101, a model training module 102, a text expansion module 103, which may also be referred to as a unit, means a series of computer program segments capable of being executed by the processor of the electronic device and of performing a fixed function, which are stored in the memory of the electronic device.
In the present embodiment, the functions concerning the respective modules/units are as follows:
the script construction module 101 is configured to obtain a tag mapping table, where the tag mapping table includes: a keyword set, a class label and a class label of each keyword in the keyword set; constructing a fine-granularity label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-granularity label and the keyword set;
in detail, the keyword set in the embodiment of the invention is a set of text words of different categories, the one-class label is a first-class category corresponding to each text word in the keyword set, and the two-class label is a second-class category under the first-class category, which is a further limitation to the first-class category. If the first class is "sports" and the second class is "basketball", one application example of the invention is: the keyword set comprises keywords such as father, child and the like, and then one type of label corresponding to the keyword "father" of the keyword set is an object, and the corresponding second type of label is a parent; one type of label of the keyword 'child' is an 'object', and the other type of label is a 'child'.
In detail, the script construction module 101 according to the embodiment of the present invention uses preset characters to splice a first class label and a second class label corresponding to each keyword in the keyword set, so as to obtain a corresponding fine-grained label; if the first class label corresponding to the keyword set a is "object", the second class label corresponding to the keyword set a is "parent", and the preset character is "-", then the fine-grained label corresponding to the keyword set is "object-parent".
Further, in the embodiment of the present invention, in order to better label the subsequent text, the script construction module 101 constructs a regular expression according to the fine-grained label and all the keywords corresponding to the keyword set, where the regular expression is used to label the fine-grained label corresponding to the word in the keyword set existing in the text to be labeled. Further, since the keyword set includes a plurality of keywords, the embodiment of the invention can obtain a plurality of regular expressions, and further, the embodiment of the invention gathers and encapsulates all the regular expressions into a script to obtain the tag label script.
The model training module 102 is configured to obtain a question-answer pair text set, screen keywords in each question-answer pair text in the question-answer text set by using the label marking script, and mark the keywords according to the corresponding fine granularity labels to obtain a marked question-answer pair text; replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different types of labels in the keyword set, so as to obtain a similar text; training a pre-constructed deep learning model by using the similarity text and the marked question-answering text to obtain a text generation model;
The question-answer pair text set in the embodiment of the invention is a set of question-answer pair texts of different types, wherein the question-answer pair text is a text formed by questions and corresponding answers in a certain search type question-answer system. Optionally, in the embodiment of the present invention, the question-answer pair text set may be obtained from a background server of a search-type question-answer system.
In detail, in the embodiment of the present invention, the model training module 102 analyzes and obtains a keyword in each question-answer pair text by using a regular expression in the label marking script, and marks the keyword according to a corresponding fine-grained label, so as to obtain a marked answer text. For example: question and answer pair text is "why buy insurance for parents? The parents are ensured to buy insurance, parents can be better guaranteed, the parents are better cared, and the 4 keywords 'parents' in the question-answer text are marked with fine-grained labels 'objects-parents', so that marked answer texts are obtained.
In another embodiment of the invention, the marking question-answer pair text can be stored in the blockchain node, and the efficiency of taking the marking question-answer pair text is improved by utilizing the characteristic of high throughput of the blockchain node.
In detail, in the embodiment of the present invention, the model training module 102 replaces the keywords in the labeled question-answer pair text with the keywords in the keyword set, where the keywords have the same type of label and the type of label is different from the type of label, so as to obtain similar text, for example, "why is the labeled question-answer pair text to buy insurance for parents? And the keyword 'parent' word corresponds to the fine-grained label of 'object-parent', so that the 'parent' is replaced by the keyword 'child' corresponding to the fine-grained label of 'object-child' in the keyword set, and then the text is subjected to similar sentence conversion by the replaced marked questions and answers, so that the similar text can be 'buying insurance for children is not good'.
In detail, in the embodiment of the present invention, the model training module 102 trains the text to the pre-constructed deep learning model by using the similarity text and the labeled question-answer, so as to obtain a text generation model, which includes:
step A: extracting features of the text by the marked questions and answers and encoding labels to obtain text sentence vectors, and extracting features of the similar text to obtain similar text sentence vectors;
In detail, in the embodiment of the present invention, the model training module 102 performs feature extraction and tag coding on the labeled question-answer pair text to obtain a labeled question-answer pair text vector, which includes: the text is segmented by using a preset word segmentation dictionary to obtain an initial text word set, and the word segmentation operation in the embodiment of the invention does not influence the fine granularity label corresponding to the text by the marked question and answer; combining each word in the initial text word set according to the sequence in the text to obtain a text sequence; converting each word in the text sequence into a vector, and combining all the vectors obtained by conversion according to the sequence of the corresponding words in the text sequence to obtain an initial sentence vector, for example: the text sequence contains two words in total, and the vector converted by the first word is thatThe vector of the second word conversion is +.>Then the combined sentence vector isCarrying out tag coding on the text sequence according to the fine granularity tag corresponding to the text of the tag question and answer to obtain a tag vector; splicing the tag vector and the initial sentence vector to obtain the text sentence vector, for example: the initial sentence vector is +. >The label vector is [5 6 ]]Then the text sentence vector obtained by splicing the label vector and the initial sentence vector is +.>
Specifically, in the embodiment of the present invention, the model training module 102 performs tag coding on the text sequence according to the fine granularity tag corresponding to the text by using the tag question and answer to obtain a tag vector, which includes: acquiring characteristic code values of the fine-grained labels, wherein different fine-grained labels in the embodiment of the invention have corresponding characteristic code values, the characteristic code values are manually set numbers for distinguishing different fine-grained labels, for example, the characteristic code value of a fine-grained label 'object-child' is 1, and the characteristic code value of a fine-grained label 'object-parent' is 2; further, in the embodiment of the present invention, each fine-granularity tag corresponding to the tag question-answer pair text is determined as a target fine-granularity tag, for example, two fine-granularity tags "object-parent" and "object-child" corresponding to the tag question-answer pair text are determined as target fine-granularity tags; further, constructing a zero vector of a corresponding dimension according to the number of words in the text sequence; according to the embodiment of the invention, zero vectors with corresponding dimensions are constructed according to the number of words in the text sequence, for example, 5 words in the text sequence are included, and then the corresponding zero vectors are [0 0 0 0 0]; counting the positions of words marked by the target fine-granularity labels in the text sequence to obtain label positions, for example, the text sequence corresponds to two target fine-granularity labels, and the second word and the fourth word in the text sequence are marked respectively, and then the label positions are 2 and 4; and replacing the element of the position corresponding to the label position in the zero vector with the characteristic code value of the target fine-granularity label corresponding to the label position to obtain the label vector, for example, the zero vector is [0 0 0 0 0], the characteristic code value of the target fine-granularity label corresponding to the label position 2 is 1, the characteristic code value of the target fine-granularity label corresponding to the label position 2 is 2, and then the label vector is [0 102 0].
Optionally, the method for extracting features of the similar text in the embodiment of the present invention is the same as the method for converting the text of the labeled question-answer pair into the initial sentence vector, which is not described herein in detail.
And (B) step (B): calculating a loss value between the text sentence vector and the similar text sentence vector by using a preset loss function;
alternatively, the loss function in the embodiment of the present invention may be a cross entropy loss function.
Step C: and (C) when the loss value is greater than or equal to a preset loss threshold value, updating model parameters of the deep learning model, returning to the step (A) for iterative training, and stopping training until the loss value is less than the preset loss threshold value, so as to obtain the text generation model.
Optionally, in an embodiment of the present invention, the deep learning model is a Seq2Seq model.
The text expansion module 103 is used for generating the to-be-expanded question-answer pair text by utilizing the text generation model when the to-be-expanded question-answer pair text is received.
In detail, in the embodiment of the present invention, the text expansion module 103 uses the text generation model to generate text for the question and answer to be expanded, including: marking the fine granularity labels of the to-be-expanded question-answer pair text by using the label marking script to obtain marked to-be-expanded text; extracting features of the marked text to be expanded and encoding labels to obtain sentence vectors of the text to be expanded; and generating text of the to-be-expanded question-answer text by using the text generation model to obtain the expanded question-answer text pair.
Optionally, the feature extraction and tag encoding manners of the text to be expanded marked are the same as those used in S4, and are not described in detail herein.
Further, since the extended question-answer generated by the generation model may be unreasonable text, in order to ensure the usability of the generated extended question-answer on the text, the text extension module 103 calculates the confusion degree of the extended question-answer on the text in the embodiment of the present invention; and carrying out text quality identification on the text by the question and answer to be expanded according to the confusion degree and a preset confusion degree threshold value, and obtaining an identification result of the text by the question and answer to be expanded.
Optionally, the embodiment of the present invention may calculate the confusion of the text by the expanded question-answer using a pre-constructed confusion model, where the confusion model may be an N-gram model.
In detail, in the embodiment of the present invention, text quality identification is performed on the text by the question and answer to be expanded according to the confusion degree and a preset confusion degree threshold value, so as to obtain an identification result of the text by the question and answer to be expanded, including: comparing the confusion degree with a preset confusion degree threshold value; when the confusion degree is smaller than the confusion degree threshold, the identification result is that identification passes, and the expanded question-answer pair text is added into a preset question-answer database; when the confusion degree is larger than or equal to the confusion degree threshold, the authentication result is that authentication is not passed, and the authentication result and the extended question-answer pair text are sent to a preset terminal device;
In the embodiment of the invention. The terminal device includes: intelligent terminal equipment such as mobile phones, computers, tablets and the like.
Fig. 3 is a schematic structural diagram of an electronic device for implementing the question-answer pair expansion method according to the present invention.
The electronic device may comprise a processor 10, a memory 11, a communication bus 12 and a communication interface 13, and may further comprise a computer program, such as a question-answer pair extension program, stored in the memory 11 and executable on the processor 10.
The memory 11 includes at least one type of readable storage medium, including flash memory, a mobile hard disk, a multimedia card, a card memory (e.g., SD or DX memory, etc.), a magnetic memory, a magnetic disk, an optical disk, etc. The memory 11 may in some embodiments be an internal storage unit of the electronic device, such as a mobile hard disk of the electronic device. The memory 11 may in other embodiments also be an external storage device of the electronic device, such as a plug-in mobile hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the electronic device. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device. The memory 11 may be used not only for storing application software installed in an electronic device and various data such as codes of question-answer pair extension programs, etc., but also for temporarily storing data that has been output or is to be output.
The processor 10 may be comprised of integrated circuits in some embodiments, for example, a single packaged integrated circuit, or may be comprised of multiple integrated circuits packaged with the same or different functions, including one or more central processing units (Central Processing unit, CPU), microprocessors, digital processing chips, graphics processors, combinations of various control chips, and the like. The processor 10 is a Control Unit (Control Unit) of the electronic device, connects various components of the entire electronic device using various interfaces and lines, and executes various functions of the electronic device and processes data by running or executing programs or modules (e.g., question-answer pair extension programs, etc.) stored in the memory 11, and calling data stored in the memory 11.
The communication bus 12 may be a peripheral component interconnect standard (perIPheral component interconnect, PCI) bus, or an extended industry standard architecture (extended industry standard architecture, EISA) bus, among others. The bus may be classified as an address bus, a data bus, a control bus, etc. The communication bus 12 is arranged to enable a connection communication between the memory 11 and at least one processor 10 etc. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus.
Fig. 3 shows only an electronic device with components, and it will be understood by those skilled in the art that the structure shown in fig. 3 is not limiting of the electronic device and may include fewer or more components than shown, or may combine certain components, or a different arrangement of components.
For example, although not shown, the electronic device may further include a power source (such as a battery) for supplying power to the respective components, and preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that functions of charge management, discharge management, power consumption management, and the like are implemented through the power management device. The power supply may also include one or more of any of a direct current or alternating current power supply, recharging device, power failure detection circuit, power converter or inverter, power status indicator, etc. The electronic device may further include various sensors, bluetooth modules, wi-Fi modules, etc., which are not described herein.
Optionally, the communication interface 13 may comprise a wired interface and/or a wireless interface (e.g., WI-FI interface, bluetooth interface, etc.), typically used to establish a communication connection between the electronic device and other electronic devices.
Optionally, the communication interface 13 may further comprise a user interface, which may be a Display, an input unit, such as a Keyboard (Keyboard), or a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch, or the like. The display may also be referred to as a display screen or display unit, as appropriate, for displaying information processed in the electronic device and for displaying a visual user interface.
It should be understood that the embodiments described are for illustrative purposes only and are not limited to this configuration in the scope of the patent application.
The question-answer pair extension program stored in the memory 11 in the electronic device is a combination of a plurality of computer programs, which, when run in the processor 10, may implement:
obtaining a tag mapping table, wherein the tag mapping table comprises: a keyword set, a class label and a class label of each keyword in the keyword set;
constructing a fine-granularity label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-granularity label and the keyword set;
Acquiring a question-answer pair text set, screening keywords in each question-answer pair text in the question-answer text set by using the label marking script, and marking the keywords according to the corresponding fine-granularity labels to obtain marked question-answer pair texts;
replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different types of labels in the keyword set, so as to obtain a similar text;
training a pre-constructed deep learning model by using the similarity text and the marked question-answering text to obtain a text generation model;
and when receiving the to-be-expanded question-answer pair text, generating the expanded question-answer pair text by utilizing the text generation model.
In particular, the specific implementation method of the processor 10 on the computer program may refer to the description of the relevant steps in the corresponding embodiment of fig. 1, which is not repeated herein.
Further, the electronic device integrated modules/units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. The computer readable medium may be non-volatile or volatile. The computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM).
Embodiments of the present invention may also provide a computer readable storage medium storing a computer program which, when executed by a processor of an electronic device, may implement:
obtaining a tag mapping table, wherein the tag mapping table comprises: a keyword set, a class label and a class label of each keyword in the keyword set;
constructing a fine-granularity label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-granularity label and the keyword set;
acquiring a question-answer pair text set, screening keywords in each question-answer pair text in the question-answer text set by using the label marking script, and marking the keywords according to the corresponding fine-granularity labels to obtain marked question-answer pair texts;
replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different types of labels in the keyword set, so as to obtain a similar text;
training a pre-constructed deep learning model by using the similarity text and the marked question-answering text to obtain a text generation model;
And when receiving the to-be-expanded question-answer pair text, generating the expanded question-answer pair text by utilizing the text generation model.
Further, the computer-usable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created from the use of blockchain nodes, and the like.
In the several embodiments provided in the present invention, it should be understood that the disclosed apparatus, device and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical function division, and there may be other manners of division when actually implemented.
The modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units can be realized in a form of hardware or a form of hardware and a form of software functional modules.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof.
The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.
The blockchain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm and the like. The Blockchain (Blockchain), which is essentially a decentralised database, is a string of data blocks that are generated by cryptographic means in association, each data block containing a batch of information of network transactions for verifying the validity of the information (anti-counterfeiting) and generating the next block. The blockchain may include a blockchain underlying platform, a platform product services layer, an application services layer, and the like.
Furthermore, it is evident that the word "comprising" does not exclude other elements or steps, and that the singular does not exclude a plurality. A plurality of units or means recited in the system claims can also be implemented by means of software or hardware by means of one unit or means. The terms second, etc. are used to denote a name, but not any particular order.
Finally, it should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention.
Claims (7)
1. A question-answer pair extension method, the method comprising:
obtaining a tag mapping table, wherein the tag mapping table comprises: a keyword set, a class label and a class label of each keyword in the keyword set;
constructing a fine-granularity label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-granularity label and the keyword set;
acquiring a question-answer pair text set, screening keywords in each question-answer pair text in the question-answer pair text set by using the label marking script, and marking the keywords according to the corresponding fine-granularity labels to obtain marked question-answer pair texts;
Replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different types of labels in the keyword set, so as to obtain a similar text;
training a pre-constructed deep learning model by using the similar texts and the marked questions and answers to obtain a text generation model;
when receiving a question-answer pair text to be expanded, generating an expanded question-answer pair text by utilizing the text generation model;
training the text on the pre-constructed deep learning model by using the similar text and the marked question-answer to obtain a text generation model, wherein the training comprises the following steps: extracting characteristics of the text by the marked questions and answers and encoding the labels to obtain text sentence vectors; extracting features of the similar texts to obtain similar text sentence vectors; performing iterative training on the deep learning model by using the text sentence vector and the similar text sentence vector to obtain the text generation model;
the step of extracting the characteristics of the text and the label coding of the text by the marked questions and answers to obtain text sentence vectors comprises the following steps: the text is segmented by utilizing a preset word segmentation dictionary to obtain an initial text word set; combining each word in the initial text word set according to the sequence of the text in the marked questions and answers to obtain a text sequence; converting each word in the text sequence into a vector, and combining all the vectors obtained by conversion according to the sequence of the corresponding words in the text sequence to obtain an initial sentence vector; carrying out tag coding on the text sequence according to the fine granularity tag corresponding to the text of the tag question and answer to obtain a tag vector; splicing the tag vector and the initial sentence vector to obtain the text sentence vector;
The method for carrying out tag coding on the text sequence according to the fine granularity tag corresponding to the text according to the tag questions and answers to obtain a tag vector comprises the following steps: acquiring a feature coding value of the fine granularity label; determining the fine granularity label corresponding to the label question-answer pair text as a target fine granularity label; constructing zero vectors of corresponding dimensions according to the number of words in the text sequence; counting the positions of words marked by the target fine granularity labels in the text sequence to obtain label positions; and replacing the element of the label position corresponding position in the zero vector with the characteristic coding value of the target fine granularity label corresponding to the label position to obtain the label vector.
2. The question-answer pair extension method of claim 1, wherein the constructing a fine-grained label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-grained label and the keyword set, comprises:
splicing the first class labels and the second class labels corresponding to each keyword in the keyword set by using preset characters to obtain corresponding fine-grained labels;
constructing a regular expression according to the fine granularity label and all corresponding keywords in the keyword set;
And summarizing and packaging all the regular expressions into scripts to obtain the label marking scripts.
3. A question-answer pair expansion method according to any one of claims 1 to 2, characterized in that after said generating an expanded question-answer pair text using said text generation model, it comprises:
calculating the confusion degree of the extended questions and answers to the text;
and carrying out text quality identification on the text by the question and answer to be expanded according to the confusion degree and a preset confusion degree threshold value, and obtaining an identification result of the text by the question and answer to be expanded.
4. The method for expanding question-answer pairs according to claim 3, wherein said performing text quality authentication on the text by the question-answer to be expanded according to the confusion degree and a preset confusion degree threshold value to obtain an authentication result of the text by the expanded question-answer comprises:
comparing the confusion degree with a preset confusion degree threshold value;
when the confusion degree is larger than or equal to the confusion degree threshold value, the authentication result is that authentication is not passed;
when the confusion degree is smaller than the confusion degree threshold value, the authentication result is authentication passing.
5. A question-answer pair expansion apparatus for implementing a question-answer pair expansion method according to any one of claims 1 to 4, comprising:
The script construction module is used for acquiring a tag mapping table, wherein the tag mapping table comprises: a keyword set, a class label and a class label of each keyword in the keyword set; constructing a fine-granularity label by using the first-class label and the second-class label, and constructing a label mark script according to the fine-granularity label and the keyword set;
the model training module is used for acquiring a question-answer pair text set, screening keywords in each question-answer pair text in the question-answer pair text set by using the label marking script, and marking the keywords according to the corresponding fine granularity labels to obtain marked question-answer pair text; replacing keywords in the marked question-answer pair text with keywords which have the same type of labels as the keywords and have different types of labels in the keyword set, so as to obtain a similar text; training a pre-constructed deep learning model by using the similar texts and the marked questions and answers to obtain a text generation model;
and the text expansion module is used for generating the to-be-expanded question-answer pair text by utilizing the text generation model when the to-be-expanded question-answer pair text is received.
6. An electronic device, the electronic device comprising:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores computer program instructions executable by the at least one processor to enable the at least one processor to perform the question-answer pair expansion method of any one of claims 1 to 4.
7. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the question-answer pair expansion method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110729964.XA CN113434650B (en) | 2021-06-29 | 2021-06-29 | Question-answer pair expansion method and device, electronic equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110729964.XA CN113434650B (en) | 2021-06-29 | 2021-06-29 | Question-answer pair expansion method and device, electronic equipment and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113434650A CN113434650A (en) | 2021-09-24 |
| CN113434650B true CN113434650B (en) | 2023-11-14 |
Family
ID=77757918
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110729964.XA Active CN113434650B (en) | 2021-06-29 | 2021-06-29 | Question-answer pair expansion method and device, electronic equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113434650B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101369265A (en) * | 2008-01-14 | 2009-02-18 | 北京百问百答网络技术有限公司 | Method and system for automatically generating semantic template of problem |
| CN102770857A (en) * | 2010-02-26 | 2012-11-07 | 独立行政法人情报通信研究机构 | Relational information expansion device, relational information expansion method and program |
| CN110196929A (en) * | 2019-05-20 | 2019-09-03 | 北京百度网讯科技有限公司 | The generation method and device of question and answer pair |
| CN110245219A (en) * | 2019-04-25 | 2019-09-17 | 义语智能科技(广州)有限公司 | A kind of answering method and equipment based on automatic extension Q & A database |
| CN111061847A (en) * | 2019-11-22 | 2020-04-24 | 中国南方电网有限责任公司 | Dialogue generation and corpus expansion method and device, computer equipment and storage medium |
| JP2020074205A (en) * | 2020-01-27 | 2020-05-14 | 日本電信電話株式会社 | Answer candidate generation device, answer candidate generation method, and program |
| CN111191442A (en) * | 2019-12-30 | 2020-05-22 | 杭州远传新业科技有限公司 | Similar problem generation method, device, equipment and medium |
| CN112749262A (en) * | 2020-07-24 | 2021-05-04 | 腾讯科技（深圳）有限公司 | Question and answer processing method and device based on artificial intelligence, electronic equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115688758A (en) * | 2022-10-31 | 2023-02-03 | 深圳市思为软件技术有限公司 | Statement intention identification method and device and storage medium |
-
2021
- 2021-06-29 CN CN202110729964.XA patent/CN113434650B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101369265A (en) * | 2008-01-14 | 2009-02-18 | 北京百问百答网络技术有限公司 | Method and system for automatically generating semantic template of problem |
| CN102770857A (en) * | 2010-02-26 | 2012-11-07 | 独立行政法人情报通信研究机构 | Relational information expansion device, relational information expansion method and program |
| CN110245219A (en) * | 2019-04-25 | 2019-09-17 | 义语智能科技(广州)有限公司 | A kind of answering method and equipment based on automatic extension Q & A database |
| CN110196929A (en) * | 2019-05-20 | 2019-09-03 | 北京百度网讯科技有限公司 | The generation method and device of question and answer pair |
| CN111061847A (en) * | 2019-11-22 | 2020-04-24 | 中国南方电网有限责任公司 | Dialogue generation and corpus expansion method and device, computer equipment and storage medium |
| CN111191442A (en) * | 2019-12-30 | 2020-05-22 | 杭州远传新业科技有限公司 | Similar problem generation method, device, equipment and medium |
| JP2020074205A (en) * | 2020-01-27 | 2020-05-14 | 日本電信電話株式会社 | Answer candidate generation device, answer candidate generation method, and program |
| CN112749262A (en) * | 2020-07-24 | 2021-05-04 | 腾讯科技（深圳）有限公司 | Question and answer processing method and device based on artificial intelligence, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| Generating Question-Answer Hierarchies;Kalpesh Krishna et al;《arXiv:1906.02622v2 [cs.CL]》;第1-15页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113434650A (en) | 2021-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113157927B (en) | Text classification method, apparatus, electronic device and readable storage medium | |
| CN112380343A (en) | Problem analysis method, problem analysis device, electronic device and storage medium | |
| CN113704429A (en) | Semi-supervised learning-based intention identification method, device, equipment and medium | |
| CN112446207A (en) | Title generation method and device, electronic equipment and storage medium | |
| CN113704614A (en) | Page generation method, device, equipment and medium based on user portrait | |
| CN112860905A (en) | Text information extraction method, device and equipment and readable storage medium | |
| CN113378970A (en) | Sentence similarity detection method and device, electronic equipment and storage medium | |
| CN113627530A (en) | Similar problem text generation method, device, equipment and medium | |
| CN113360654B (en) | Text classification method, apparatus, electronic device and readable storage medium | |
| CN113505273B (en) | Data sorting method, device, equipment and medium based on repeated data screening | |
| CN113344125B (en) | Long text matching recognition method and device, electronic equipment and storage medium | |
| CN112633988A (en) | User product recommendation method and device, electronic equipment and readable storage medium | |
| CN113434650B (en) | Question-answer pair expansion method and device, electronic equipment and readable storage medium | |
| CN113536782B (en) | Sensitive word recognition method and device, electronic equipment and storage medium | |
| CN113626605B (en) | Information classification method, device, electronic equipment and readable storage medium | |
| CN113591881B (en) | Intention recognition method and device based on model fusion, electronic equipment and medium | |
| CN115496166A (en) | Multitasking method and device, electronic equipment and storage medium | |
| CN114943306A (en) | Intention classification method, device, equipment and storage medium | |
| CN115146064A (en) | Intention recognition model optimization method, device, equipment and storage medium | |
| CN113822215A (en) | Equipment operation guide file generation method and device, electronic equipment and storage medium | |
| CN113486646A (en) | Product report issuing method and device, electronic equipment and readable storage medium | |
| CN112631589A (en) | Application program home page layout configuration method and device, electronic equipment and storage medium | |
| CN113704478B (en) | Text element extraction method, device, electronic equipment and medium | |
| CN115146627B (en) | Entity identification method, entity identification device, electronic equipment and storage medium | |
| CN116540981A (en) | Form design method, device, equipment and storage medium based on attribute panel |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |