CN114579725A - Question and answer pair generation method and device, electronic equipment and storage medium - Google Patents
Question and answer pair generation method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN114579725A CN114579725A CN202210210991.0A CN202210210991A CN114579725A CN 114579725 A CN114579725 A CN 114579725A CN 202210210991 A CN202210210991 A CN 202210210991A CN 114579725 A CN114579725 A CN 114579725A
- Authority
- CN
- China
- Prior art keywords
- answer
- question
- candidate
- sentences
- sentence
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 238000004590 computer program Methods 0.000 claims description 14
- 238000001914 filtration Methods 0.000 claims description 7
- 239000000126 substance Substances 0.000 claims description 2
- 238000003058 natural language processing Methods 0.000 abstract description 8
- 238000013473 artificial intelligence Methods 0.000 abstract description 6
- 238000013135 deep learning Methods 0.000 abstract description 3
- 238000012545 processing Methods 0.000 description 15
- 238000005516 engineering process Methods 0.000 description 13
- 238000004891 communication Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000002372 labelling Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 239000012530 fluid Substances 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000010705 motor oil Substances 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 241000282412 Homo Species 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 235000012054 meals Nutrition 0.000 description 1
- 230000003924 mental process Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- 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
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
- G06F40/35—Discourse or dialogue representation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The disclosure discloses a question-answer pair generation method, a question-answer pair generation device, electronic equipment and a storage medium, and particularly relates to the technical field of artificial intelligence such as natural language processing and deep learning. The specific scheme is as follows: preprocessing the conversation record in the log to obtain a conversation contained in the conversation record and a first core word corresponding to the conversation; determining a type label corresponding to each statement in the conversation, wherein the type label is any one of the following items: question type, answer type, and other types; generating candidate question-answer pairs according to the type labels corresponding to the sentences; and rewriting each candidate question-answer pair based on the first core word corresponding to the conversation to generate a target question-answer pair. Therefore, candidate question-answer pairs can be generated based on the type label corresponding to each sentence in the conversation, and then the candidate question-answer pairs are rewritten based on the first core word corresponding to the conversation to generate target question-answer pairs, so that the number and the quality of the target question-answer pairs are improved.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a method and an apparatus for generating question and answer pairs, an electronic device, and a storage medium.
Background
With the development of computer technology, the intelligent question-answering system is more and more widely used. For example, in a customer service system, for a common question, a resource can be solved by a question and answer set in advance. However, the amount of the question and answer resources may be limited to some extent, so that some questions may not be answered by the question and answer resources, and manual answer is required. Therefore, how to generate effective question-answer pairs to expand the resources of the question-answer pairs is very important.
Disclosure of Invention
The disclosure provides a question-answer pair generation method, a question-answer pair generation device, electronic equipment and a storage medium.
In one aspect of the present disclosure, a question-answer pair generating method is provided, including:
preprocessing a conversation record in a log to obtain a conversation contained in the conversation record and a first core word corresponding to the conversation;
determining a type label corresponding to each statement in the conversation, wherein the type label is any one of the following items: question type, answer type, and other types;
generating candidate question-answer pairs according to the type labels corresponding to the sentences;
and rewriting each candidate question-answer pair based on the first core word corresponding to the conversation to generate a target question-answer pair.
In another aspect of the present disclosure, there is provided a question-answer pair generating apparatus including:
the acquisition module is used for preprocessing the conversation record in the log so as to acquire the conversation contained in the conversation record and the first core word corresponding to the conversation;
a determining module, configured to determine a type tag corresponding to each statement in the session, where the type tag is any one of: question type, answer type, and other types;
the first generation module is used for generating candidate question-answer pairs according to the type label corresponding to each statement;
and the second generation module is used for rewriting each candidate question-answer pair based on the first core word corresponding to the conversation so as to generate a target question-answer pair.
In another aspect of the present disclosure, an electronic device is provided, including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the question-answer pair generating method of the embodiment of the above aspect.
In another aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, on which a computer program is stored, where the computer program is configured to cause a computer to perform the method for generating a question and answer pair described in the embodiment of the above aspect.
In another aspect of the present disclosure, a computer program product is provided, which includes a computer program and when executed by a processor, the computer program implements the question and answer pair generating method described in the embodiment of the above aspect.
The question-answer pair generation method, device, electronic device and storage medium provided by the disclosure can be used for preprocessing the conversation record in the log to obtain the conversation contained in the conversation record and the first core word corresponding to the conversation, then determining the type label corresponding to each statement in the conversation, generating candidate question-answer pairs according to the type label corresponding to each statement, and rewriting each candidate question-answer pair based on the first core word corresponding to the conversation to generate the target question-answer pair. Therefore, candidate question-answer pairs can be generated based on the type labels corresponding to each sentence in the conversation, and then the candidate question-answer pairs can be rewritten based on the first core words corresponding to the conversation to generate target question-answer pairs, so that the number of the target question-answer pairs is increased, the quality of the target question-answer pairs is improved, and further, the smoothness and the accuracy of the intelligent conversation are guaranteed.
It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.
Drawings
The drawings are included to provide a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:
fig. 1 is a schematic flow chart of a method for generating a question-answer pair according to an embodiment of the present disclosure;
fig. 1A is a schematic view of a session record according to an embodiment of the disclosure;
fig. 2 is a schematic flow chart of a method for generating question-answer pairs according to an embodiment of the present disclosure;
fig. 3 is a schematic flow chart of a method for generating question-answer pairs according to an embodiment of the present disclosure;
FIG. 3A is a schematic diagram of a question-answer pair generation process provided by an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of a question-answer pair generating device according to another embodiment of the present disclosure;
fig. 5 is a block diagram of an electronic device for implementing a question-answer pair generation method according to an embodiment of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of the embodiments of the disclosure are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Artificial intelligence is the subject of research that makes computers simulate some human mental processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), both at the hardware level and at the software level. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing, and the like; the artificial intelligence software technology mainly comprises a computer vision technology, a voice recognition technology, a natural language processing technology, a machine learning technology, a deep learning technology, a big data processing technology, a knowledge map technology and the like.
Natural language processing is the computer processing, understanding and use of human languages (such as chinese, english, etc.), which is a cross discipline between computer science and linguistics, also commonly referred to as computational linguistics. Since natural language is the fundamental mark that humans distinguish from other animals. Without language, human thinking has never been said, so natural language processing embodies the highest task and context of artificial intelligence, that is, only when a computer has the capability of processing natural language, the machine realizes real intelligence.
Deep learning refers to a multi-layered artificial neural network and a method of training it. One layer of neural network takes a large number of matrix numbers as input, weights are taken through a nonlinear activation method, and another data set is generated as output. Through the appropriate number of matrixes, multiple layers of tissues are linked together to form a neural network brain to carry out accurate and complex processing just like people identify object labeling pictures.
A question-answer pair generation method, apparatus, electronic device, and storage medium of the embodiments of the present disclosure are described below with reference to the accompanying drawings.
The question-answer pair generation method provided by the embodiment of the disclosure can be executed by a question-answer pair generation device provided by the embodiment of the disclosure, and the device can be configured in electronic equipment.
Fig. 1 is a schematic flow chart of a question-answer pair generation method according to an embodiment of the present disclosure.
As shown in fig. 1, the question-answer pair generating method may include the following steps:
The log may be log data in the customer service system, or may also be a log in which a session record is optionally stored, such as log data in a database, and the like, which is not limited in this disclosure.
In addition, the number of sessions may be one or more, and the present disclosure does not limit this.
It will be appreciated that there are many ways in which the dialog records in the log may be pre-processed.
For example, the dialog records in the log may be segmented into one or more sessions according to the user identifier and the customer service identifier, and the dialog records in each session may be segmented into each statement.
For example, according to the user identifier and the customer service identifier, the conversation records in the log are preprocessed, an obtained certain conversation is as shown in fig. 1A, and then each sentence in the conversation can be obtained by segmenting the content in the conversation. For example, meaningless content in the session may be filtered, such as time information may be removed, and so on. The present disclosure is not limited thereto.
Additionally, the first core word may characterize the core content of the session. For example, by performing natural language processing on a sentence in a certain conversation, core content corresponding to the conversation, that is, a first core word and the like, for example, the first core word and the like may be car a, brand XX, and the like, which is not limited by the present disclosure.
Optionally, a reference probability that each statement in the session belongs to each type tag may be determined, and then the type tag corresponding to each statement may be determined according to each reference probability and a preset rule.
The preset rules may be various. For example, question type statements may be followed by answer statements rather than question type statements or other types of statements, and so on, and this disclosure is not limited thereto.
For example, for statement 1 and statement 2 in the session, by performing semantic recognition and parsing on the statements, if it is determined that the type tag corresponding to statement 1 is: the question type, statement 2, has a reference probability of 0.3 for the question type, 0.35 for the answer type, and 0.35 for the other types. If the preset rule is as follows: if the sentence of question type is followed more by answer sentence, then it can be determined that the type tag corresponding to sentence 2 is: the answer type.
Or, the type label corresponding to each statement in the conversation can be determined through the trained sequence labeling model.
For example, the trained sequence labeling model may include an ERNIE (annie) layer and a CRF (conditional random field) layer. The preprocessed sentences can be converted into sentence vectors, and then the sentence vectors can be input into an ERNIE layer, so that the reference probability that each sentence belongs to each type of label can be output after the processing of the ERNIE layer; and then, the reference probability of each statement belonging to each type label can be used as an original parameter of a CRF layer, so that the type label corresponding to each statement is determined through the processing of the CRF layer.
It should be noted that the above examples are only illustrative, and cannot be used as a limitation on the manner of determining the type label corresponding to each statement in the embodiment of the present disclosure.
And 103, generating candidate question-answer pairs according to the type labels corresponding to the sentences.
The number of candidate question-answer pairs may be one or multiple, which is not limited in this disclosure.
In addition, the candidate question-answer pairs usually include candidate question sentences and corresponding candidate answer sentences. The present disclosure is not limited thereto.
For example, the type labels of each statement in a session are: statement 1 is question type, statement 2 is answer type, statement 3 is other type, statement 4 is question type, statement 5 is question type, statement 6 is answer type. Sentence 1 may be determined as a candidate question sentence and sentence 2 may be determined as a candidate answer sentence corresponding to sentence 1, with sentence 1 and sentence 2 constituting a candidate question-answer pair 1. The sentence 4, the sentence 5, and the sentence 6 constitute a question-answer candidate pair 2 by using the sentence 4 and the sentence 5 as one question-question candidate sentence and the sentence 6 as a candidate answer sentence corresponding to the question-question candidate sentence. Wherein, candidate question sentences and the like can be generated after the sentence 4 and the sentence 5 are processed in a fusion mode.
It should be noted that the above examples are only illustrative and should not be taken as limiting the type labels and the like of the statements in the embodiments of the present disclosure.
And 104, rewriting each candidate question-answer pair based on the first core word corresponding to the conversation to generate a target question-answer pair.
It is understood that the original candidate question statement or candidate answer statement may be relatively redundant, and at this time, the candidate question statement or candidate answer statement may be rewritten based on the first core word corresponding to the session to generate the target question-answer pair. The target question-answer pair may include a target question statement and a target answer statement, which is not limited in the present disclosure.
For example, it has been determined that the first core word corresponding to the session is: an automobile. If a candidate question sentence in a candidate question-answer pair is "o, i thinks that it is this, and if the car is inflated, it is charged or not charged", the candidate question sentence may be rewritten based on the first core word "car", for example, it may be rewritten as "if the car is inflated to charge".
Or, it has been determined that the first core word corresponding to the session is: an automobile. If the candidate answer sentence in a candidate question-answer pair is "good, i understand what we mean, i find that we have a car to be inflated at this point is usually free without charge", the candidate answer sentence may be rewritten based on the first core word "car", for example, it may be rewritten as "car to be inflated at this point is free".
It should be noted that the foregoing examples are merely illustrative, and should not be taken as limitations on the first core word, candidate question-answer pairs, and target question-answer pairs in the embodiments of the present disclosure.
Therefore, in the embodiment of the disclosure, by rewriting the candidate question-answer pairs extracted from the dialogue records, more concise and effective target question-answer pairs can be generated, the number of the question-answer pairs is increased, and the quality of the question-answer pairs is improved, so that the resource of the question-answer pairs can be effectively expanded, and the fluency and the accuracy of an intelligent question-answer system are guaranteed.
In the embodiment of the disclosure, the dialog records in the log may be preprocessed to obtain the sessions included in the dialog records and the first core words corresponding to the sessions, then the type label corresponding to each sentence in the session may be determined, the candidate question-answer pairs are generated according to the type label corresponding to each sentence, and then each candidate question-answer pair is rewritten based on the first core word corresponding to the session to generate the target question-answer pair. Therefore, candidate question-answer pairs can be generated based on the type labels corresponding to each sentence in the conversation, and then the candidate question-answer pairs can be rewritten based on the first core words corresponding to the conversation to generate target question-answer pairs, so that the number of the target question-answer pairs is increased, the quality of the target question-answer pairs is improved, and further, the smoothness and the accuracy of the intelligent conversation are guaranteed.
Fig. 2 is a schematic flow chart of a question-answer pair generating method provided in the embodiment of the present disclosure, and as shown in fig. 2, the question-answer pair generating method may include the following steps:
It should be noted that specific contents and implementation manners of step 202 and step 203 may refer to descriptions of other embodiments of the present disclosure, and are not described herein again.
There may be multiple statements in the session, and there is usually a sequence between statements, so that each statement in the session can be traversed in sequence according to the sequence of each statement in the session.
And 204, under the condition that the type label of the first sentence is the question type, determining that the first sentence is a candidate question sentence in the candidate question-answer pair.
It is understood that in an actual intelligent customer service system or an intelligent dialogue system, the user and the customer service are not completely asked, and the user may express a plurality of questions continuously to express the real demand clearly. Thus, a plurality of questions may be regarded as candidate question sentences, etc., and the present disclosure does not limit this.
For example, if each statement in the session is: the sentences 1, 2, 3, 4, and 5 may be sequentially traversed in the order of the sentences. For example, in the case where the type tag of statement 1 is a question type, this statement 1 may be determined as a candidate question statement.
Or, in the session, each statement is: in the case of statements 1, 2, 3, 4, and 5, if it is determined that the type tag of statement 1 is of another type, then the traversal of each statement following statement 1 can be continued. For example, if statement 2 is tagged as a question type, then statement 2 may be determined to be a candidate question statement.
Or, in the session, each sentence is: in the case of statements 1, 2, 3, 4, and 5, if it is determined that the type tag of statement 1 is: other types, statement 2 and statement 3, have type labels: the question type can be determined as candidate question sentences by the sentences 2 and 3.
It should be noted that the above examples are only illustrative and should not be taken as limiting the manner of determining the type tag of the sentence in the implementation of the present disclosure.
It can be understood that in an actual intelligent customer service system or an intelligent dialogue system, the user and the customer service are not completely asked or answered at a time, and the user may express a plurality of questions continuously to express the actual demand clearly; alternatively, the customer service replies with multiple sentences in succession to clarify the question. Thus, a plurality of answer type sentences may be used as candidate question sentences, etc., which is not limited by the present disclosure.
For example, if each statement in the session is: sentence 1, sentence 2, sentence 3, sentence 4, sentence 5, it has been determined that sentence 1 is: and (5) candidate question sentences, and then, traversing each sentence after the sentence 1. If statement 2 has type tags as: answer type, type label of statement 3 is: other types, then statement 2 may be determined as: candidate answer sentences, sentence 1 and sentence 2 form candidate question-answer pairs.
Or, if each statement in the session is: sentence 1, sentence 2, sentence 3, sentence 4, sentence 5, it has been determined that sentence 1 is: and (6) candidate question sentences. If the type labels of statement 2, statement 3, and statement 4 are: answer type, statement 5, type label is: other types, then statement 2, statement 3, and statement 4 may be determined as: candidate answer sentences, wherein the candidate question sentences and the candidate answer sentences form a candidate question-answer pair. It should be noted that the above examples are merely illustrative, and cannot be taken as limitations on the manner of determining candidate question sentences and candidate answer sentences, and the like in the embodiments of the present disclosure.
Therefore, in the embodiment of the disclosure, the accuracy and reliability of the generation of the candidate question-answer pairs are improved by traversing each statement in the conversation.
Wherein the candidate question sentences can be rewritten based on the first core word; or, the candidate answer sentence may be rewritten based on the first core word; alternatively, the candidate question sentence and the candidate answer sentence may be rewritten based on the first core word, and the disclosure is not limited thereto.
It can be understood that if the candidate question sentence includes a plurality of sentences, the plurality of sentences may be fused, or keywords in the plurality of sentences may be extracted, and then the candidate question sentence may be rewritten based on the first core word.
For example, the first core word is "BMW" and the candidate question statement is "kay Pair". How much money I want the brake fluid in the next week. The rewritten target question sentence may be "how much money the brake fluid is next week". Therefore, a simple target question sentence with smooth semantics can be generated through rewriting.
Alternatively, if the first core word is "bmam", and the candidate answer sentence is "bmam X3 maintained twice, and if the candidate answer sentence is two times, including the engine oil and the engine oil following the working hours cost", the rewritten target question sentence may be: several times for bme X3 maintenance?
It should be noted that the above examples are merely illustrative, and should not be taken as limitations on the first core word, candidate question sentence, candidate answer sentence, and the like in the embodiments of the present disclosure.
It is understood that the candidate question sentence is generally the content input by the user, and the candidate answer sentence is generally the content returned by the customer service, so that, in the embodiment of the present disclosure, the candidate question sentence and/or the candidate answer sentence may also be rewritten and the like according to the corresponding identifier of each sentence based on the first core word.
Therefore, in the embodiment of the disclosure, by rewriting the candidate question sentences and/or candidate answer sentences in the candidate question-answer pairs, more accurate and reliable target question sentences can be generated, so that the generation quantity and quality of the target question sentences are improved, and a basis is provided for intelligent customer service conversation.
The second core word may characterize the core point of the candidate answer sentence. The present disclosure is not limited thereto.
It is to be understood that the second core word corresponding to the candidate answer sentence may also be obtained by performing natural language processing, entity recognition, and the like on the candidate answer sentence in the candidate question-answer pair. The manner of obtaining the second core word is not limited in this disclosure.
And step 208, rewriting the candidate answer sentences based on the second core words and the candidate question sentences in the candidate question-answer pairs to generate target answer sentences in the target question-answer pairs.
For example, if the candidate answer sentence is "couple, sunday is few, saturday is particularly many, and saturday is a little less than sunday", the second core word is determined to be "sunday is few" by performing semantic recognition on the candidate answer sentence. If the candidate question sentence is "there are relatively few worship days", the candidate answer sentence may be rewritten to "there are few weekdays and there are particularly many weekdays" to generate the target answer sentence.
Optionally, the candidate answer sentence may also be rewritten based on reading understanding and abstract generating technology to generate the target answer sentence, and the disclosure does not limit this.
It should be noted that the above examples are merely illustrative, and cannot be used as limitations on the manner in which the target answer sentence is generated in the embodiments of the present disclosure.
Therefore, in the embodiment of the present disclosure, the generated target answer sentence can be more accurate and reliable by rewriting the candidate answer sentence, thereby providing a condition for generating an accurate and reliable target question-answer pair.
In the embodiment of the disclosure, a dialog record in a log may be preprocessed to obtain a session included in the dialog record and a first core word corresponding to the session, then a type tag corresponding to each sentence in the session may be determined, each sentence in the session may be traversed in sequence, when the type tag of the first sentence is a question type, the first sentence is determined as a candidate question sentence in a candidate question-answer pair, then traversal is continued for the sentence after the first sentence, when the type tag of the second sentence is another type and a sentence in which the type tag is an answer type is included between the second sentence and the first sentence, the sentence in the answer type included between the second sentence and the first sentence is determined as a candidate answer sentence in the candidate question-answer pair, then the candidate question sentence and/or the candidate answer sentence in the candidate question-answer pair may be rewritten based on the first core word, and then rewriting the candidate answer sentences based on the second core words and the candidate question sentences in the candidate question-answer pairs to generate the target answer sentences in the target question-answer pairs. Therefore, after the type label corresponding to each sentence in the conversation is determined, each sentence can be traversed, a candidate question-answer pair is generated according to the type label corresponding to each sentence, and then the candidate question-answer sentence and/or the candidate answer sentence are rewritten based on the first core word corresponding to the conversation and the second core word of the candidate answer sentence to generate the target question-answer pair, so that the number of the target question-answer pair is increased, the quality of the target question-answer pair is improved, and a foundation is provided for guaranteeing the fluency and the accuracy of the intelligent conversation.
Fig. 3 is a schematic flow chart of a question-answer pair generating method provided in the embodiment of the present disclosure, and as shown in fig. 3, the question-answer pair generating method may include the following steps:
It should be noted that specific contents and implementation manners of step 301 and step 302 may refer to descriptions of other embodiments of the present disclosure, and are not described herein again.
The scenes corresponding to the conversation can be determined by performing natural language processing or semantic recognition on the conversation. For example, by analyzing the session shown in fig. 1A, it may be determined that the scene corresponding to the session is "XX decreasing activity", which is not limited in this disclosure.
And step 304, filtering the redundant words in each statement according to the scenes.
The redundant words may be spoken words, or may also be words assisted by voice, or words without practical meaning, and the like, which is not limited in this disclosure.
It is understood that the same term may have different effects in different contexts. For example, in a navigation scenario, neither "go" nor "come" are spoken words. In the catering scene, "go" in "go XX meal" belongs to a spoken word, and the word can be filtered. The present disclosure is not limited thereto.
Alternatively, a dictionary or wordrank (word rank) model may be used in combination to perform the filtering process of the redundant words. Wherein, the dictionary summarizes common spoken words, and the spoken words can be rapidly removed by using the dictionary; wordrank may make decisions, etc., taking context into account, which is not limited by this disclosure.
And 305, generating candidate question-answer pairs according to the type labels corresponding to the processed sentences.
The redundant words in the sentences are filtered, so that each processed sentence is more accurate, and candidate question-answer pairs can be generated according to the type label corresponding to each processed sentence, which is not described herein again.
Therefore, in the embodiment of the disclosure, by processing the redundant words in the sentence, the generated candidate question-answer pair can be more accurate and reliable, and the accuracy and reliability of the candidate question-answer pair are improved.
And step 306, rewriting each candidate question-answer pair based on the first core word corresponding to the session to generate a target question-answer pair.
Optionally, the question-answer pair matching degree between the target question sentences and the target answer sentences in each target question-answer pair may be determined through the question-answer pair matching model. For example, the target question-answer pair may be input into the question-answer pair matching model, and the question-answer pair matching degree between the target question sentences and the target answer sentences in the target question-answer pair may be determined through processing the question-answer pair matching model.
Alternatively, the target question sentences and the target answer sentences in each target question-answer pair may be processed by semantic recognition, natural language processing, or the like to determine the degree of association between the target question sentences and the target answer sentences. It can be understood that the higher the degree of association is, the higher the matching degree between the target question sentence and the target answer sentence is; the lower the degree of association, the lower the degree of matching between the target question sentence and the target answer sentence, and so on.
It should be noted that the above examples are merely illustrative, and cannot be taken as a limitation on the way of determining the question-answer pair matching degree between the target question sentence and the target answer sentence in the embodiments of the present disclosure.
And 308, adding the target question-answer pairs into the target question-answer library under the condition that the question-answer pair matching degree between the target question sentences and the corresponding target answer sentences is greater than a threshold value. The threshold may be a preset value, such as 0.85, 0.9, and the like, which is not limited in this disclosure.
In addition, the target question-answer library can store a large number of effective target question-answer pairs, which can correspond to a dialogue system, a customer service system and the like. For example, a frequency affected questions library (FAQ) corresponding to a certain customer service system may be used. The present disclosure is not limited thereto.
It can be understood that if the matching degree of the question-answer pair between the target question sentence and the target answer sentence in any target question-answer pair is greater than the threshold, it can be considered that a question-answer relationship exists between the target question sentence and the target answer sentence, and an effective question-answer pair can be formed, so that the target question-answer pair can be added to the target question-answer library. If the matching degree of the question-answer pairs between the target question sentences and the target answer sentences in any target question-answer pair is smaller than or equal to the threshold value, the question-answer relationship between the target question sentences and the target answer sentences can be considered to be absent, and the target question-answer pair is rejected. Therefore, in the embodiment of the disclosure, the target question-answer pairs can be filtered according to the question-answer pair matching degree, the quality of the target question-answer pairs in the target question-answer library is effectively ensured, and conditions are provided for ensuring the fluency of the intelligent dialogue system and the customer service system.
It can be understood that the question and answer pair generating method provided by the present disclosure may be applied to any dialog scenario and any intelligent customer service system, and the present disclosure does not limit this. The process of question and answer pair generation provided by the present disclosure is described below in conjunction with fig. 3A.
As shown in fig. 3A, the dialog log may be first session-sliced, and data cleansing is performed, such as removing useless data, special symbols, and so on. Then, manual marking can be carried out, and data enhancement is carried out, for example, the manually marked question sentences and answer sentences are randomly exchanged; or the colloquial prefix in the sentence, and nonsense sentences or words such as "kaihe", "feed", etc. can be removed. And then training the sequence labeling model to obtain the trained sequence labeling model. And then, predicting by using the trained sequence labeling model.
For example, based on the trained sequence tagging model, a type tag corresponding to each statement in the session is determined, and a candidate question-answer pair is generated according to the type tag corresponding to each statement. And then, redundant word filtering, namely spoken language removal processing, can be carried out on the candidate question-answer pairs. Then, based on the first core word corresponding to the conversation, rewriting of the candidate question sentences is carried out to generate target question sentences; and rewriting the candidate answer sentences based on reading understanding and abstract generation technologies to generate target answer sentences. And then, determining question-answer pair matching between the target question sentences and the target answer sentences in the target question-answer pairs, and adding the target question-answer pairs with the question-answer pair matching degree larger than a threshold value into an FAQ library.
It should be noted that the above examples are only illustrative, and should not be taken as a limitation on the generation process of the questions and answers in the embodiments of the present disclosure.
In the embodiment of the disclosure, the dialog records in the log may be preprocessed to obtain the sessions and the first core words corresponding to the sessions included in the dialog records, then the type tag corresponding to each statement in the sessions may be determined, and the sessions may be analyzed to determine the scenes corresponding to the sessions, and then, according to the scenes, filtering the redundant words in each statement, generating candidate question-answer pairs according to the type label corresponding to each processed statement, then generating a first core word corresponding to the conversation, rewriting each candidate question-answer pair to generate a target question-answer pair, then determining the question-answer pair matching degree between the target question sentences and the target answer sentences in each target question-answer pair, and under the condition that the question-answer pair matching degree between the target question sentences and the corresponding target answer sentences is greater than a threshold value, adding the target question-answer pairs into a target question-answer library. Therefore, after the target question-answer pairs are generated, the target question-answer sentences with the question-answer pair matching degree larger than the threshold value can be added into the target question-answer library, so that the number of the target question-answer pairs in the target question-answer library is increased, the quality of the target question-answer pairs in the target question-answer library is improved, and a foundation is further provided for the fluency and the accuracy of intelligent conversation.
In order to implement the above embodiments, the present disclosure also provides a question-answer pair generating device.
Fig. 4 is a schematic structural diagram of a question-answer pair generating device according to an embodiment of the present disclosure.
As shown in fig. 4, the question-answer pair generation apparatus 400 includes: the device comprises an acquisition module 410, a determination module 420, a first generation module 430 and a second generation module 440.
The obtaining module 410 is configured to pre-process a session record in a log to obtain a session included in the session record and a first core word corresponding to the session.
A determining module 420, configured to determine a type tag corresponding to each statement in the session, where the type tag is any one of: question type, answer type, and other types.
And a first generating module 430, configured to generate candidate question-answer pairs according to the type tag corresponding to each statement.
The second generating module 440 is configured to rewrite each candidate question-answer pair based on the first core word corresponding to the session, so as to generate a target question-answer pair.
Optionally, the first generating module 430 is specifically configured to:
traversing each statement in the conversation in sequence;
determining a first sentence as a candidate question sentence in a candidate question-answer pair under the condition that the type tag of the first sentence is a question type;
and continuously traversing the sentences after the first sentence, and determining the sentences of answer types contained between the second sentence and the first sentence as candidate answer sentences in the candidate question-answer pairs under the condition that the type labels of the second sentence are other types and the sentences of which the type labels are answer types are contained between the second sentence and the first sentence.
Optionally, the second generating module 440 is specifically configured to:
and rewriting the candidate question sentences and/or candidate answer sentences in the candidate question-answer pairs based on the first core words to generate the target question sentences in the target question-answer pairs.
Optionally, the second generating module 440 is specifically configured to:
performing semantic recognition on candidate answer sentences in the candidate question-answer pairs to determine second core words corresponding to the candidate answer sentences;
and rewriting the candidate answer sentences based on the second core words and the candidate question sentences in the candidate question-answer pairs to generate target answer sentences in the target question-answer pairs.
Optionally, the first generating module 430 is specifically configured to:
analyzing the conversation to determine a scene corresponding to the conversation;
according to the scene, filtering redundant words in each statement;
and generating candidate question-answer pairs according to the type label corresponding to each processed statement.
Optionally, the determining module 420 is further configured to:
determining question-answer pair matching degree between the target question sentences and the target answer sentences in each target question-answer pair;
and under the condition that the question-answer pair matching degree between the target question sentences and the corresponding target answer sentences is greater than a threshold value, adding the target question-answer pairs into a target question-answer library.
The functions and specific implementation principles of the modules in the embodiments of the present disclosure may refer to the embodiments of the methods, and are not described herein again.
The question-answer pair generation device in the embodiment of the disclosure may first preprocess a dialog record in a log to obtain a session included in the dialog record and a first core word corresponding to the session, then may determine a type tag corresponding to each sentence in the session, generate a candidate question-answer pair according to the type tag corresponding to each sentence, and then rewrite each candidate question-answer pair based on the first core word corresponding to the session to generate a target question-answer pair. Therefore, candidate question-answer pairs can be generated based on the type labels corresponding to each sentence in the conversation, and then the candidate question-answer pairs can be rewritten based on the first core words corresponding to the conversation to generate target question-answer pairs, so that the number of the target question-answer pairs is increased, the quality of the target question-answer pairs is improved, and further, the smoothness and the accuracy of the intelligent conversation are guaranteed.
The present disclosure also provides an electronic device, a readable storage medium, and a computer program product according to embodiments of the present disclosure.
FIG. 5 illustrates a schematic block diagram of an example electronic device 500 that can be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 5, the apparatus 500 comprises a computing unit 501 which may perform various appropriate actions and processes in accordance with a computer program stored in a Read Only Memory (ROM)502 or a computer program loaded from a storage unit 508 into a Random Access Memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored. The calculation unit 501, the ROM 502, and the RAM 503 are connected to each other by a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.
A number of components in the device 500 are connected to the I/O interface 505, including: an input unit 506 such as a keyboard, a mouse, or the like; an output unit 507 such as various types of displays, speakers, and the like; a storage unit 508, such as a magnetic disk, optical disk, or the like; and a communication unit 509 such as a network card, modem, wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange information/data with other devices through a computer network such as the internet and/or various telecommunication networks.
The computing unit 501 may be a variety of general-purpose and/or special-purpose processing components having processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The calculation unit 501 executes the respective methods and processes described above, such as the question-answer pair generation method. For example, in some embodiments, the question-answer pair generation method may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 500 via the ROM 502 and/or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the question-answer pair generation method described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform the challenge-pair generation method by any other suitable means (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), the internet, and blockchain networks.
The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The Server can be a cloud Server, also called a cloud computing Server or a cloud host, and is a host product in a cloud computing service system, so as to solve the defects of high management difficulty and weak service expansibility in the traditional physical host and VPS service ("Virtual Private Server", or simply "VPS"). The server may also be a server of a distributed system, or a server incorporating a blockchain.
According to the technical scheme, the conversation records in the log can be preprocessed to obtain the conversation contained in the conversation records and the first core words corresponding to the conversation, then the type label corresponding to each statement in the conversation can be determined, the candidate question-answer pairs are generated according to the type label corresponding to each statement, and then each candidate question-answer pair is rewritten based on the first core words corresponding to the conversation to generate the target question-answer pair. Therefore, candidate question-answer pairs can be generated based on the type labels corresponding to each sentence in the conversation, and then the candidate question-answer pairs can be rewritten based on the first core words corresponding to the conversation to generate target question-answer pairs, so that the number of the target question-answer pairs is increased, the quality of the target question-answer pairs is improved, and further, the smoothness and the accuracy of the intelligent conversation are guaranteed.
It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present disclosure may be executed in parallel or sequentially or in different orders, and are not limited herein as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved.
The above detailed description should not be construed as limiting the scope of the disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present disclosure should be included in the scope of protection of the present disclosure.
Claims (15)
1. A question-answer pair generating method, wherein the method comprises:
preprocessing a conversation record in a log to obtain a conversation contained in the conversation record and a first core word corresponding to the conversation;
determining a type label corresponding to each statement in the conversation, wherein the type label is any one of the following items: question type, answer type, and other types;
generating candidate question-answer pairs according to the type labels corresponding to the sentences;
and rewriting each candidate question-answer pair based on the first core word corresponding to the conversation to generate a target question-answer pair.
2. The method of claim 1, wherein the generating candidate question-answer pairs according to the type tag corresponding to each statement comprises:
traversing each statement in the conversation in sequence;
determining a first sentence as a candidate question sentence in a candidate question-answer pair under the condition that the type tag of the first sentence is a question type;
and continuously traversing the sentences after the first sentence, and determining the sentences of answer types contained between the second sentence and the first sentence as candidate answer sentences in the candidate question-answer pairs under the condition that the type labels of the second sentence are other types and the sentences of which the type labels are answer types are contained between the second sentence and the first sentence.
3. The method of claim 1, wherein the overwriting each of the candidate question-answer pairs based on the first core word corresponding to the session to generate a target question-answer pair comprises:
and rewriting the candidate question sentences and/or candidate answer sentences in the candidate question-answer pairs based on the first core words to generate target question sentences in the target question-answer pairs.
4. The method of claim 1, wherein said overwriting each of said candidate question-answer pairs to generate a target question-answer pair comprises:
performing semantic recognition on candidate answer sentences in the candidate question-answer pairs to determine second core words corresponding to the candidate answer sentences;
and rewriting the candidate answer sentences based on the second core words and the candidate question sentences in the candidate question-answer pairs to generate target answer sentences in the target question-answer pairs.
5. The method of claim 1, wherein the generating candidate question-answer pairs according to the type tag corresponding to each statement comprises:
analyzing the conversation to determine a scene corresponding to the conversation;
according to the scene, filtering redundant words in each statement;
and generating candidate question-answer pairs according to the type label corresponding to each processed statement.
6. The method of any one of claims 1-5, wherein, after said adapting each of said candidate question-answer pairs to generate a target question-answer pair, further comprises:
determining question-answer pair matching degree between the target question sentences and the target answer sentences in each target question-answer pair;
and under the condition that the question-answer pair matching degree between the target question sentences and the corresponding target answer sentences is greater than a threshold value, adding the target question-answer pairs into a target question-answer library.
7. A question-answer pair generating device, wherein the method comprises:
the acquisition module is used for preprocessing the conversation record in the log so as to acquire the conversation contained in the conversation record and the first core word corresponding to the conversation;
a determining module, configured to determine a type tag corresponding to each statement in the session, where the type tag is any one of: question type, answer type, and other types;
the first generation module is used for generating candidate question-answer pairs according to the type label corresponding to each statement;
and the second generation module is used for rewriting each candidate question-answer pair based on the first core word corresponding to the conversation so as to generate a target question-answer pair.
8. The apparatus of claim 7, wherein the first generating module is specifically configured to:
traversing each statement in the conversation in sequence;
determining a first sentence as a candidate question sentence in a candidate question-answer pair under the condition that the type tag of the first sentence is a question type;
and continuously traversing the sentences after the first sentence, and determining the sentences of answer types contained between the second sentence and the first sentence as candidate answer sentences in the candidate question-answer pairs under the condition that the type labels of the second sentence are other types and the sentences of which the type labels are answer types are contained between the second sentence and the first sentence.
9. The apparatus of claim 7, wherein the second generating module is specifically configured to:
and rewriting the candidate question sentences and/or candidate answer sentences in the candidate question-answer pairs based on the first core words to generate the target question sentences in the target question-answer pairs.
10. The apparatus of claim 7, wherein the second generating module is specifically configured to:
performing semantic recognition on candidate answer sentences in the candidate question-answer pairs to determine second core words corresponding to the candidate answer sentences;
and rewriting the candidate answer sentences based on the second core words and the candidate question sentences in the candidate question-answer pairs to generate target answer sentences in the target question-answer pairs.
11. The apparatus of claim 7, wherein the first generating means is specifically configured to
Analyzing the conversation to determine a scene corresponding to the conversation;
according to the scene, filtering redundant words in each statement;
and generating candidate question-answer pairs according to the type label corresponding to each processed statement.
12. The apparatus of any of claims 7-11, wherein the means for determining is further configured to:
determining question-answer pair matching degree between the target question sentences and the target answer sentences in each target question-answer pair;
and under the condition that the question-answer pair matching degree between the target question sentences and the corresponding target answer sentences is greater than a threshold value, adding the target question-answer pairs into a target question-answer library.
13. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-6.
14. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-6.
15. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210210991.0A CN114579725A (en) | 2022-03-04 | 2022-03-04 | Question and answer pair generation method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210210991.0A CN114579725A (en) | 2022-03-04 | 2022-03-04 | Question and answer pair generation method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114579725A true CN114579725A (en) | 2022-06-03 |
Family
ID=81772710
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210210991.0A Withdrawn CN114579725A (en) | 2022-03-04 | 2022-03-04 | Question and answer pair generation method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114579725A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118052222A (en) * | 2024-04-15 | 2024-05-17 | 北京晴数智慧科技有限公司 | Method and device for generating multi-round dialogue data |
-
2022
- 2022-03-04 CN CN202210210991.0A patent/CN114579725A/en not_active Withdrawn
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118052222A (en) * | 2024-04-15 | 2024-05-17 | 北京晴数智慧科技有限公司 | Method and device for generating multi-round dialogue data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113705187B (en) | Method and device for generating pre-training language model, electronic equipment and storage medium | |
| EP4152178A1 (en) | Method for training image-text matching model, apparatus, storage medium, and computer program product | |
| CN107679039B (en) | Method and device for determining statement intention | |
| EP3910492A2 (en) | Event extraction method and apparatus, and storage medium | |
| CN114399769B (en) | Training method of text recognition model, and text recognition method and device | |
| CN112560496A (en) | Training method and device of semantic analysis model, electronic equipment and storage medium | |
| CN113553412B (en) | Question-answering processing method, question-answering processing device, electronic equipment and storage medium | |
| CN113191256A (en) | Method and device for training lane line detection model, electronic device and storage medium | |
| CN113051374B (en) | Text matching optimization method and device | |
| EP4123496A2 (en) | Method and apparatus for extracting text information, electronic device and storage medium | |
| CN113722493A (en) | Data processing method, device, storage medium and program product for text classification | |
| EP4191544A1 (en) | Method and apparatus for recognizing token, electronic device and storage medium | |
| CN113836925A (en) | Training method and device for pre-training language model, electronic equipment and storage medium | |
| CN113221565A (en) | Entity recognition model training method and device, electronic equipment and storage medium | |
| CN112818227A (en) | Content recommendation method and device, electronic equipment and storage medium | |
| CN112926308A (en) | Method, apparatus, device, storage medium and program product for matching text | |
| CN114218940A (en) | Text information processing method, text information processing device, text information model training method, text information model training device, text information model training equipment and storage medium | |
| CN114579725A (en) | Question and answer pair generation method and device, electronic equipment and storage medium | |
| CN113378921A (en) | Data screening method and device and electronic equipment | |
| CN116383382A (en) | Sensitive information identification method and device, electronic equipment and storage medium | |
| CN113344214B (en) | Training method and device of data processing model, electronic equipment and storage medium | |
| CN115565186A (en) | Method and device for training character recognition model, electronic equipment and storage medium | |
| US11113081B2 (en) | Generating a video for an interactive session on a user interface | |
| CN113641724A (en) | Knowledge tag mining method and device, electronic equipment and storage medium | |
| CN113033179A (en) | Knowledge acquisition method and device, electronic equipment and readable 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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20220603 |