CN110427467B - Question-answer processing method, device, computer equipment and storage medium - Google Patents
Question-answer processing method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110427467B CN110427467B CN201910560681.XA CN201910560681A CN110427467B CN 110427467 B CN110427467 B CN 110427467B CN 201910560681 A CN201910560681 A CN 201910560681A CN 110427467 B CN110427467 B CN 110427467B
- Authority
- CN
- China
- Prior art keywords
- question
- user
- entity
- intention
- matching
- 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
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
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Animal Behavior & Ethology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to a question and answer processing method, a question and answer processing device, computer equipment and a storage medium. The method comprises the following steps: receiving a user inquiry instruction, and acquiring a user inquiry sentence according to the user inquiry instruction; identifying a user question, and determining an entity and an intention from the user question; matching the entity and the intention by using the trained incidence relation matching model to obtain a matching result; and searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result. By adopting the method, the accuracy of obtaining the corresponding answers of the question of the user can be improved.
Description
Technical Field
The present application relates to the field of internet technologies, and in particular, to a question and answer processing method and apparatus, a computer device, and a storage medium.
Background
With the development of internet technology, artificial customer service has not met the increasing demands of enterprises, and in order to reduce enterprise cost, enterprises begin to use customer service robots to answer questions of users. The enterprise establishes the knowledge graph of the customer service robot, and then the customer service robot searches the corresponding answer in the knowledge graph according to the problem of the user to reply. Currently, when a customer service robot performs voice question answering, the customer service robot usually obtains the intention of a user to ask a sentence, matches an answer according to the intention, and then replies according to the answer. Such direct use of intention-matching answers may result in less accurate answers when the user has more intentions to ask sentences.
Disclosure of Invention
In view of the above, it is necessary to provide a question and answer processing method, device, computer device and storage medium capable of improving query efficiency.
A question-answer processing method, the method comprising:
receiving a user query instruction, and acquiring a user query sentence according to the user query instruction;
identifying a user question, and determining an entity and an intention from the user question;
matching the entity and the intention by using the trained incidence relation matching model to obtain a matching result;
and searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result.
In one embodiment, before receiving a user query instruction and acquiring a user question according to the user query instruction, the method includes:
and establishing a preset two-dimensional table knowledge base, wherein the preset two-dimensional table knowledge base comprises an entity, an intention and answers corresponding to the entity and the intention.
In one embodiment, identifying a user question, determining an entity and an intent from the user question, comprises:
identifying the question of the user by using the trained named entity identification model, and determining an entity in the question of the user;
and identifying the question of the user by using the trained intention identification model, and determining the intention in the question of the user.
In one embodiment, identifying a user question using a trained named entity recognition model, determining entities in the user question, comprises:
performing word segmentation on the question of the user to obtain word segmentation results;
obtaining a word vector according to the word segmentation result, inputting the word vector into the trained named entity recognition model for recognition, and obtaining an entity recognition result vector;
and determining the entity in the question of the user according to the entity recognition result vector.
In one embodiment, identifying a user question using a trained intent recognition model, determining an intent in the user question, comprises:
segmenting words of the question of the user to obtain a word segmentation result;
obtaining a word vector according to the word segmentation result, and inputting the word vector into the trained intention recognition model for recognition to obtain an intention recognition result vector;
and determining the intention in the question of the user according to the intention recognition result vector.
In one embodiment, matching the entity and the intent results in a matching result, comprising:
matching the entity and the intention to obtain a matching result to be determined;
obtaining a matching vector to be determined according to a matching result to be determined, and inputting the matching vector to be determined into a trained incidence relation matching model for recognition to obtain a matching recognition result vector;
and determining a target matching result according to the matching identification result vector.
In one embodiment, searching for an answer corresponding to a question of a user in a preset two-dimensional table knowledge base according to a matching result includes:
searching a target entity in the matching result in an entity field in a preset two-dimensional table knowledge base, and searching a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base;
and when the target entity and the target intention are found, obtaining answers corresponding to the target entity and the target intention from a preset two-dimensional table knowledge base.
In one embodiment, after searching for an answer corresponding to a question of a user in a preset two-dimensional table knowledge base according to a matching result, the method further includes:
and returning the answer corresponding to the question of the user to the terminal so that the terminal displays the answer.
A question-answering processing apparatus, the apparatus comprising:
the question acquisition module is used for receiving a user inquiry instruction and acquiring a user question according to the user inquiry instruction;
the identification module is used for identifying the question of the user to obtain an entity and an intention corresponding to the question of the user;
the matching module is used for matching the entity and the intention by using a trained incidence relation matching model to obtain a matching result;
and the searching module is used for matching the preset entity and the intention two-dimensional table according to the correlation result and obtaining an answer corresponding to the question of the user according to the matching result.
In one embodiment, the question answering processing device comprises:
and the two-dimensional table establishing module is used for establishing a preset two-dimensional table knowledge base, and the preset two-dimensional table knowledge base comprises an entity, an intention and answers corresponding to the entity and the intention.
In one embodiment, the identification module comprises:
the entity determining module is used for identifying the user question by using the trained named entity identification model and determining an entity in the user question;
and the intention determining module is used for identifying the user question by using the trained intention identification model and determining the intention in the user question.
In one embodiment, the entity determination module includes:
the word segmentation module is used for segmenting words of the question asked by the user to obtain word segmentation results;
the entity recognition module is used for obtaining word vectors according to the word segmentation result, inputting the word vectors into the trained named entity recognition model for recognition, and obtaining entity recognition result vectors; and determining the entity in the question of the user according to the entity recognition result vector.
In one embodiment, the intention determining module comprises:
the word segmentation module is used for segmenting words of the question asked by the user to obtain word segmentation results;
the intention recognition module is used for obtaining word vectors according to the word segmentation result, inputting the word vectors into the trained intention recognition model for recognition, and obtaining intention recognition result vectors; and determining the intention in the question of the user according to the intention recognition result vector.
In one embodiment, the matching module comprises:
the result obtaining module is used for matching the entity and the intention to obtain a matching result to be determined;
the result identification module is used for obtaining a matching vector to be determined according to the matching result to be determined, and inputting the matching vector to be determined into the trained incidence relation matching model for identification to obtain a matching identification result vector;
and the result determining module is used for determining a target matching result according to the matching identification result vector.
In one embodiment, the lookup module includes:
the target searching module is used for searching a target entity in the matching result in an entity field in a preset two-dimensional table knowledge base and searching a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base;
and the answer obtaining module is used for obtaining the answers corresponding to the target entity and the target intention from the preset two-dimensional table knowledge base when the target entity and the target intention are found.
In one embodiment, the question answering processing device further includes:
and the answer returning module is used for returning the answer corresponding to the question of the user to the terminal so that the terminal displays the answer.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
receiving a user inquiry instruction, and acquiring a user inquiry sentence according to the user inquiry instruction;
identifying a user question, and determining an entity and an intention from the user question;
matching the entity and the intention by using a trained incidence relation matching model to obtain a matching result;
and searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
receiving a user inquiry instruction, and acquiring a user inquiry sentence according to the user inquiry instruction;
identifying a user question, and determining an entity and an intention from the user question;
matching the entity and the intention by using a trained incidence relation matching model to obtain a matching result;
and searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result.
According to the question and answer processing method, the question and answer processing device, the computer equipment and the storage medium, the entity and the intention in the question of the user are identified, then the entity and the intention are matched through the trained incidence relation matching model, and the answer corresponding to the question of the user is searched in the preset two-dimensional table knowledge base according to the matching result. When a plurality of entities or a plurality of intents exist or a plurality of entities and a plurality of intents exist simultaneously, the accuracy of obtaining the matching result is improved, namely, the associated entities and intents can be accurately obtained. Therefore, the situation that when a plurality of entities and a plurality of intentions exist, the entity and the intention are wrongly associated is reduced, then answers corresponding to the question of the user are directly searched in the preset two-dimensional table knowledge base by using the matching result, and the accuracy of obtaining the answers of the question is improved.
Drawings
FIG. 1 is a diagram of an application scenario of a question and answer processing method in one embodiment;
FIG. 2 is a schematic flow chart diagram illustrating a method for question and answer processing in one embodiment;
FIG. 3 is a flow diagram illustrating the determination of a user question entity and intent in one embodiment;
FIG. 4 is a flow diagram illustrating the determination of a user question entity in one embodiment;
FIG. 5 is a schematic flow chart illustrating the determination of user question intent in one embodiment;
FIG. 6 is a flow diagram illustrating the determination of a target match result in one embodiment;
FIG. 7 is a diagram illustrating an embodiment of a process for searching answers in a database of predefined two-dimensional tables;
fig. 7a is an application scenario diagram of a question and answer processing method in a specific embodiment;
FIG. 8 is a schematic flow chart diagram of a question and answer processing method in an exemplary embodiment;
FIG. 9 is a block diagram showing an example of a device that is a method of processing a question and answer;
FIG. 10 is a diagram showing an internal structure of a computer device in one embodiment;
fig. 11 is an internal configuration diagram of a computer device in another embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The question answering processing method provided by the application can be applied to the application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The server 104 receives a user query instruction sent by the terminal 102, and acquires a user question sentence according to the user query instruction; server 104 identifies the user question, determines the entity and intent from the user question; the server 104 matches the entity with the intent to obtain a matching result; the server 104 searches the answer corresponding to the question of the user in the preset two-dimensional table knowledge base according to the matching result. The answer corresponding to the user question may then be returned to the terminal 102. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, and the server 104 may be implemented by an independent server or a server cluster formed by multiple servers.
In one embodiment, as shown in fig. 2, a question processing method is provided, which is described by taking the application of the method to the server in fig. 1 as an example, and includes the following steps:
s202, receiving a user inquiry instruction, and acquiring a user question according to the user inquiry instruction.
Specifically, the terminal obtains the question sentence of the user, and the user voice can be converted into the question sentence text of the user through the user voice obtained by the voice device. The user question text input by the user can also be acquired through the input device. And at the moment, the terminal sends a user query instruction to the server according to the user file text, and the server receives the user query instruction and analyzes the user query instruction to obtain a user question text.
And S204, identifying the question of the user, and determining the entity and the intention from the question of the user.
The entity refers to a specific type of words in the user question, such as a name of a person, a name of a place, a name of an organization, a proper noun, and the like. Intent is meant to refer to a relationship or attribute that a user wants to query against an entity. Such as: asking what the name is, asking about the attributes of the name entity.
Specifically, the server identifies the question of the user by using a machine learning algorithm, and determines an entity and an intention corresponding to the question of the user from the question of the user. A single entity and a single intent may be determined from the user question. For example, the user question is "where a small and clear home stays". From this user question, it can be determined that the entity is "Xiaoming" and the intent is "Home". A single entity and multiple intentions, or multiple entities and a single intention, or multiple entities and multiple intentions, etc., may also be determined from the user question. For example, the user question may be "what the stock codes of the Yangtze river securities and the science news fly". Two entities, "Changjiang river securities" and "science major news fly", are determined from the user question, with a single intention "stock code". The intention determined from the user question may be a word included in the user question or a word not included in the user question. For example, "all small and clear relatives have a person". The intention that the identified entity is "Xiaoming" may be "the name of the" jijiujiu "," the name of the "niu Gu", "the name of the aunt Fu", etc.
S206, matching the entity and the intention by using the trained incidence relation matching model to obtain a matching result.
And training the trained incidence relation matching model according to the existing labeling data by using a machine learning algorithm to obtain the target object. Wherein the machine learning algorithm may be a linear regression algorithm, a neural network algorithm, or the like. For example, according to the entity and the intention in the existing question sentence, the associated entity and the intention are marked as "associated", the unassociated entity and the intention are marked as "unassociated", and then the trained association relation matching model is obtained by training with the marked data and the neural network algorithm. Matching entities and intents refers to associating entities and intents identified from the user question.
Specifically, matching the intention corresponding to the entity by using the trained incidence relation matching model to obtain a matching result. The matching result may be that one entity corresponds to one intention result, or that one entity corresponds to a plurality of intentions, or that a plurality of entities correspond to one intention, or that a plurality of entities correspond to a plurality of intentions, etc. For example, two entities, "Changjiang river stock" and "science major fly," are determined from the user question, with a single intent "stock code. Matching according to preset matching rules to obtain matching results including the Changjiang river securities and the stock code and the science news and stock code, i.e. two entities correspond to a single intention. When the user asks the sentence "what the stock code of the changjiang securities is and who the president who flies major news in science" determines the entity "changjiang securities" and "major news" and intends "stock code" and "president", matching is performed using the trained incidence relation matching model, and the matching result "changjiang securities and stock code" president news and president "is obtained.
And S208, searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result.
The preset two-dimensional table knowledge base is a preset two-dimensional data table of answers of question sentences of users, and the two dimensions are entity dimensions and intention dimensions.
Specifically, according to the matching result, namely the entity and the corresponding intention, the entity is searched in the entity dimension in the preset two-dimensional table knowledge base, and then the intention corresponding to the entity is searched in the intention dimension. When the corresponding intentions of the entity and the entity are found, the in-table information of the entity and the corresponding intentions, namely the answers of the question of the user, is obtained.
In the embodiment, the entity and the intention in the question of the user are identified, then the entity and the intention are matched through the trained incidence relation matching model, and the answer corresponding to the question of the user is searched in the preset two-dimensional table knowledge base according to the matching result. When a plurality of entities or a plurality of intents exist or a plurality of entities and a plurality of intents exist simultaneously, the accuracy of obtaining the matching result is improved, namely, the associated entities and intents can be accurately obtained. Therefore, the situation that when a plurality of entities and a plurality of intentions exist, the entity and the intention are wrongly associated is reduced, then answers corresponding to the question of the user are directly searched in the preset two-dimensional table knowledge base by using the matching result, and the accuracy of obtaining the answers of the question is improved. And according to the matching result, searching an answer corresponding to the question of the user in a preset two-dimensional table knowledge base. Therefore, the answers of the question do not need to be searched by using a knowledge graph, namely, the question of the user does not need to be converted into a special search sentence to be searched in the knowledge graph, the answer corresponding to the question of the user is directly searched in the preset two-dimensional table knowledge base by using the matching result of the entity and the intention, and the query efficiency of the answers of the question is improved.
In one embodiment, before step S202, that is, before receiving a user query instruction and acquiring a user question according to the user query instruction, the method includes the steps of:
and establishing a preset two-dimensional table knowledge base, wherein the preset two-dimensional table knowledge base comprises an entity, an intention and answers corresponding to the entity and the intention.
Specifically, the terminal receives an instruction of a user for establishing a two-dimensional table knowledge base, and acquires entity dimension information, intention dimension information and user answer information according to the instruction for establishing the two-dimensional table knowledge base. And then the terminal sends an instruction for establishing a two-dimensional table knowledge base to the server, and the server receives the instruction for establishing the two-dimensional table knowledge base and acquires entity dimension information, intention dimension information and user answer information according to the instruction. And establishing a preset two-dimensional table knowledge base according to the entity dimension information, the intention dimension information and the user answer information. In the embodiment, the two-dimensional table knowledge base is established in advance, so that the subsequent use is facilitated. In a specific embodiment, part of the contents of the two-dimensional table knowledge base is set up as shown in table 1:
TABLE 1
In one embodiment, as shown in fig. 3, the step S204 of identifying a user question and determining an entity and an intention from the user question includes the steps of:
s302, the trained named entity recognition model is used for recognizing the question of the user, and the entity in the question of the user is determined.
The named entity recognition model is used for recognizing entities in the question of the user and is a model established by using a named entity recognition algorithm, and the named entity recognition algorithm can be a keyword matching algorithm, a template matching algorithm or a sequence labeling algorithm and the like.
Specifically, the server may use the user question as an input of a keyword matching algorithm, a template matching algorithm, or a sequence tagging algorithm, and perform named entity identification according to the user question to obtain an entity output by the keyword matching algorithm, the template matching algorithm, or the sequence tagging algorithm, so as to obtain an entity included in the user question. In one embodiment, a sequence annotation algorithm can be used as the named entity recognition algorithm. The method comprises the steps of training by using a sequence marking algorithm according to existing user question sentences and marked entities in advance, taking user descending question sentences as input of the sequence marking algorithm, training by taking the marked entities as labels of the sequence marking algorithm, and obtaining a trained named entity recognition model when training completion conditions are met. The training completion condition may be that the number of training iterations exceeds a preset threshold, or that the value of the loss function is smaller than a preset threshold.
S304, recognizing the question of the user by using the trained intention recognition model, and determining the intention in the question of the user.
The intention recognition model is used for recognizing the intention in the question of the user and is obtained by training through a machine learning algorithm, a search technical scheme or other algorithms according to marked training data.
Specifically, the server takes the user question as the input of the intention identification model, and the intention identification model identifies the intention according to the input answering user question to obtain the intention output by the intention identification model, namely the intention contained in the user question. In one embodiment, the intent recognition model output is trained using a deep neural network algorithm based on existing training data that includes an existing user question and an intent that the user question has annotated. And taking the question of the user as the input of the Shen network algorithm, training by taking the marked intention as a label, and finishing the training when a preset condition is reached, namely the iteration times of the Shen network algorithm exceed a preset threshold value or the loss function value of the Shen network algorithm is smaller than a preset threshold value, so as to obtain a trained intention recognition model.
In the embodiment, the server can identify the entity and the intention in the question of the user through the trained named entity identification model and the intention identification model, so that the efficiency and the accuracy of the identification of the question of the user are improved.
In one embodiment, as shown in fig. 4, step S302, namely identifying a user question by using a trained named entity identification model, to determine an entity in the user question, includes the steps of:
s402, segmenting words of the question of the user to obtain word segmentation results.
The word segmentation is a process of recombining continuous word sequences into word sequences according to a certain standard.
Specifically, the server performs word segmentation on the user question based on a word segmentation method based on character string matching, a word segmentation method based on understanding, or a word segmentation method based on statistics to obtain a word sequence after word segmentation.
S404, obtaining word vectors according to the word segmentation result, inputting the word vectors into the trained named entity recognition model for recognition, and obtaining entity recognition result vectors.
Specifically, the server maps the word sequence into a vector space according to the word sequence after word segmentation to obtain a word vector. The word bag model may be used to vectorize the segmentation results to obtain word vectors, the word2vec (a group of related models used to generate word vectors, which are shallow and double-layer neural networks used to train to reconstruct the word text of linguistics) algorithm may be used to vectorize the segmentation results to obtain word vectors, and so on. And the server inputs the obtained word vector into the trained named entity recognition model for recognition to obtain an entity recognition result vector. The entity recognition result vector is used to represent the entities included in the user question.
S406, determining the entity in the question of the user according to the entity recognition result vector.
Specifically, a correspondence between an entity recognition result vector obtained during training and an entity is obtained. For example, a non-entity is represented as "0" in the entity recognition result vector. Entities are "LOC", "PER", "ORG", "MISC", etc., which respectively represent location, person name, organization name, miscellaneous items, etc.
And obtaining the entity in the user question sentence corresponding to the entity recognition result vector obtained at the moment according to the corresponding relation. Such as: for example, the user question is "what is the little work", and the entity recognition result vector obtained from the user question is (per, 0). The entity is the Xiaoming.
In the embodiment, the word vectors are obtained by segmenting the user question to obtain the words, the words are identified in the named entity identification model by using the word vectors to obtain the identification result vectors, and the entities in the user question are obtained according to the identification result vectors, so that the accuracy of obtaining the entities is improved.
In one embodiment, as shown in fig. 5, step S304, recognizing a user question by using a trained intention recognition model, and determining an intention in the user question includes the steps of:
s502, segmenting words of the question of the user to obtain a word segmentation result.
S504, obtaining word vectors according to the word segmentation result, inputting the word vectors into the trained intention recognition model for recognition, and obtaining intention recognition result vectors.
Specifically, the server may perform word segmentation on the user question based on various word segmentation algorithms and then obtain a word segmentation result, where the word segmentation algorithm may be a word segmentation algorithm based on string matching, a word segmentation algorithm based on understanding, or a word segmentation algorithm based on statistics. And then mapping the word sequence into a vector space according to the word sequence after word segmentation to obtain a word vector. And inputting the word vector into the trained intention recognition model for recognition to obtain an output intention recognition result vector. The intention recognition result vector is used to represent the intention included in the user question. For example, the user may ask the sentence "what the small work is" to perform word segmentation, and the word segmentation results "small work", "yes" and "what" are obtained. And obtaining a word vector by using a word2vec algorithm according to the word segmentation result.
S506, determining the intention in the question of the user according to the intention recognition result vector.
Specifically, the intention corresponding to the intention recognition result vector is obtained according to the corresponding relation between the intention recognition result vector and the intention set when the trained intention recognition model is carried out, and the intention in the question of the user is obtained. The corresponding relation means that after the user asks for a sentence and divides the word, the word is indicated by '1' if the word is intended, and is indicated by '0' if the word is not intended. For example, the user asks the sentence "what is a little work", and the resultant intention recognition result vector may be (0, 1, 0). Wherein, the word in the user question corresponding to "1" is "work", and the intention in the user file is "work".
In the embodiment, the word vectors are obtained according to the word segmentation result after the user question is segmented, the word vectors are used for recognition in the trained intention recognition model, and finally the intention in the user question is obtained according to the intention recognition result vector, so that the accuracy of obtaining the intention in the user question can be improved.
In one embodiment, as shown in fig. 6, step S206, namely matching the entity and the intention by using the trained association matching model to obtain the matching result, includes the steps of:
and S602, matching the entity and the intention to obtain a matching result to be determined.
The matching result to be determined refers to a result obtained after the entity and the intention are associated randomly.
In particular, if a single entity and a single intent are obtained, a unique matching result to be determined may be obtained. And if a plurality of entities and a single intention are obtained, directly associating the plurality of entities with the single intention one by one to obtain a plurality of matching results to be determined. If a single entity and a plurality of intents are obtained, the plurality of intents are directly associated with the single entity one by one, and then a plurality of matching results to be determined can be obtained. And when the obtained multiple entities and the multiple intents are obtained, randomly associating the multiple entities and the multiple intents one by one to obtain multiple different matching results to be determined.
S604, obtaining a matching vector to be determined according to the matching result to be determined, inputting the matching vector to be determined into the trained incidence relation matching model for recognition, and obtaining a matching recognition result vector.
The trained incidence relation matching model is obtained by training according to the existing labeling data by using a machine learning algorithm. For example, according to the existing entity and intention, the association between the entity and the intention is randomly performed to obtain the matching result. The truly associated match result is then labeled as "1". And marking the unassociated matching result as '0' to obtain a marking result. And taking the matching result as the input of the machine learning algorithm, taking the labeling result as the label of the machine learning algorithm for training, and obtaining a trained incidence relation matching model when the training completion condition is reached. The training completion condition may be that the number of iterations of training exceeds a preset number, or that a loss function of the machine learning algorithm is smaller than a preset threshold. The machine learning algorithm may be a logistic regression algorithm, a neural network algorithm, or the like.
Specifically, the server vectorizes the matching result to be determined to obtain a matching vector to be determined, and inputs the matching vector to be determined into the trained incidence relation matching model for recognition to obtain an output matching recognition result vector.
And S606, determining a target matching result according to the matching identification result vector.
Specifically, the server obtains a matching result corresponding to the matching recognition result vector according to the corresponding relationship set during training, and the matching result is used as a target matching result. For example, the matching result to be determined includes "intention 1 and entity 1", "intention 2 and entity 2", and "intention 3 and entity 3". The resulting matching recognition result vector may be (0, 1, 0). And obtaining a matching result 'intention 2 and entity 2' corresponding to the vector '1' according to the corresponding relation set during training, taking the 'intention 2 and entity 2' as a target matching result, and finally confirming the associated entity and intention.
In the embodiment, when a plurality of entities and a plurality of intentions are identified, the associated relation matching model is used to obtain the matching result of real association, the associated matching result is used as the target matching result, the answer of the question of the user is obtained by using the target matching result, and the obtained answer is more accurate.
In one embodiment, as shown in fig. 7, step S208 of searching the answer corresponding to the question from the user in the preset two-dimensional table knowledge base according to the matching result includes the steps of:
s702, searching a target entity in the matching result in an entity field in a preset two-dimensional table knowledge base, and searching a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base.
S704, when the target entity and the target intention are found, obtaining answers corresponding to the target entity and the target intention from a preset two-dimensional table knowledge base.
The entity field refers to fields to which all entities arranged in the two-dimensional table knowledge base belong, and the intention field is arranged in fields to which all intents in the two-dimensional table knowledge base belong.
Specifically, when a matching result is obtained, the server searches for a target entity in the matching result in an entity field in a preset two-dimensional table knowledge base, and searches for a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base. When the target entity can be found in the entity field and the target intention can be found in the intention field, it is indicated that the preset two-dimensional table knowledge base stores the answer corresponding to the target entity and the target intention. And then obtaining an answer from a table unit corresponding to the target entity and the target intention in a preset two-dimensional table knowledge base, wherein the answer is the answer corresponding to the target entity and the target intention, namely the answer of the question of the user.
When the target entity is not found in the entity field or the target intention is not found in the intention field, it is indicated that the answer corresponding to the target entity and the target intention is not stored in the preset two-dimensional table knowledge base. At this time, the user question has no answer. The server prompts for an error.
In the above embodiment, the answer to the question of the user is found in the preset two-dimensional table knowledge base, the question of the user does not need to be converted into a special query statement, and the answer to the question of the user can be obtained only by matching the entity and the intention in the question of the user in the preset two-dimensional table knowledge base, so that the efficiency of obtaining the answer to the question of the user can be improved.
In one embodiment, the predetermined two-dimensional table knowledge base transition bit knowledge map may be stored. Specifically, entity fields in a preset two-dimensional table knowledge base and corresponding in-table answers are used as a head entity and a tail entity of the knowledge graph triple, the corresponding entity fields in the preset two-dimensional table knowledge base are used as relations in the triple, and the knowledge graph is established and stored.
In one embodiment, after step S208, that is, after searching the answer corresponding to the question from the user in the preset two-dimensional table knowledge base according to the matching result, the method further includes the steps of:
and returning the answer corresponding to the question of the user to the terminal so that the terminal displays the answer.
The terminal is used for receiving and displaying answers of the user question, and the terminal can be a terminal corresponding to the user question or other terminals. The terminal is not limited to personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The presentation is not limited to display by text or images, play by voice, play by video, and the like. For example, the terminal may display the received answer in a terminal display interface, display and play the received answer in a video playing interface, convert the received answer into voice information through a voice device, and play the voice information.
Specifically, the server may return the obtained answer corresponding to the question of the user to the terminal corresponding to the question of the user, and the terminal corresponding to the question of the user receives the answer and then displays the answer, so that the user can obtain answer information of the question, and the use of the user is facilitated. Or returning the answer corresponding to the question of the user to a terminal set by the user or the server, and displaying the answer corresponding to the question of the user on the set terminal. For example, in a specific embodiment, as shown in fig. 7A, a user receives a user question through a mobile phone 7A, a server 7B obtains the user question, finds out an answer to the user question through the question-answer processing method in any one of the embodiments, and returns the answer to a computer 7C set by the user for display.
In a specific embodiment, as shown in fig. 8, the question answering processing method includes the steps of:
s802, a two-dimensional table knowledge base is constructed and completed, and the two-dimensional table knowledge base comprises entities, intents and answers corresponding to the entities and the intents.
And S804, an intention recognition model is obtained by using a deep learning algorithm and intention training in a two-dimensional table knowledge base.
And S806, acquiring the user question, and taking the user question as the input of the intention identification model to acquire the intention I and the intention M contained in the user question.
And S808, taking the user question as the input of the keyword matching algorithm to obtain the named entity E and the named entity N contained in the user question.
And S810, matching the obtained intention with the entity by using the trained incidence relation matching model to obtain a matching result E x I.
And S812, searching a corresponding answer of the question of the user from the two-dimensional table knowledge base according to the matching result E.
And S814, returning the answer of the question of the user to the terminal of the question of the user for displaying.
It should be understood that although the various steps in the flow diagrams of fig. 2-8 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not limited to being performed in the exact order illustrated and, unless explicitly stated herein, may be performed in other orders. Moreover, at least some of the steps in fig. 2-8 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performing the sub-steps or stages is not necessarily sequential, but may be performed alternately or alternatingly with other steps or at least some of the sub-steps or stages of other steps.
In one embodiment, as shown in fig. 9, there is provided a question-answering processing apparatus 900 including: question acquisition module 900, recognition module 904, matching module 906, and lookup module 908, wherein:
the question acquiring module 900 is configured to receive a user query instruction and acquire a user question according to the user query instruction;
an identifying module 904, configured to identify a question of the user, and determine an entity and an intention from the question of the user;
a matching module 906, configured to match the entity and the intention using the trained incidence relation matching model, to obtain a matching result;
and a searching module 908, configured to search, according to the matching result, an answer corresponding to the question of the user in a preset two-dimensional table knowledge base.
In the above embodiment, the question of the user is obtained by the question obtaining module 900, the entity and the intention in the question of the user are determined by the identifying module 904, the entity and the intention are matched in the matching module 906 to obtain a matching result, and finally, the answer corresponding to the question of the user is searched in the preset two-dimensional table knowledge base according to the matching result in the searching module 908. Through the execution of each module, the accuracy of obtaining answers of the user question sentences can be improved, the answers of the user question sentences can be found quickly, and the query efficiency is improved.
In one embodiment, the question answering processing apparatus 900 includes:
and the two-dimensional table establishing module is used for establishing a preset two-dimensional table knowledge base, and the preset two-dimensional table knowledge base comprises an entity, an intention and answers corresponding to the entity and the intention.
In one embodiment, the identifying module 904 includes:
the entity determining module is used for identifying the user question by using the trained named entity identification model and determining an entity in the user question;
and the intention determining module is used for identifying the question of the user by using the trained intention identification model and determining the intention in the question of the user.
In one embodiment, an entity determination module includes:
the word segmentation module is used for segmenting words of the question of the user to obtain word segmentation results;
the entity recognition module is used for obtaining word vectors according to the word segmentation result, inputting the word vectors into the trained named entity recognition model for recognition, and obtaining entity recognition result vectors; and determining the entity in the question of the user according to the entity recognition result vector.
In one embodiment, the intent determination module includes:
the word segmentation module is used for segmenting the question of the user to obtain a word segmentation result;
the intention recognition module is used for obtaining word vectors according to the word segmentation result, inputting the word vectors into the trained intention recognition model for recognition, and obtaining intention recognition result vectors; and determining the intention in the question of the user according to the intention recognition result vector.
In one embodiment, the matching module 906 includes:
the result obtaining module is used for matching the entity and the intention to obtain a matching result to be determined;
the result identification module is used for obtaining a matching vector to be determined according to the matching result to be determined, and inputting the matching vector to be determined into the trained incidence relation matching model for identification to obtain a matching identification result vector;
and the result determining module is used for determining a target matching result according to the matching identification result vector.
In one embodiment, the lookup module 908 includes:
the target searching module is used for searching a target entity in the matching result in an entity field in a preset two-dimensional table knowledge base and searching a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base;
and the answer obtaining module is used for obtaining the answers corresponding to the target entity and the target intention from the preset two-dimensional table knowledge base when the target entity and the target intention are found.
In one embodiment, the question answering processing apparatus 900 further includes:
and the answer returning module is used for returning the answer to the terminal corresponding to the question of the user so that the terminal corresponding to the question of the user displays the answer.
For the specific limitations of the question and answer processing device, reference may be made to the limitations of the question and answer processing method above, and details are not repeated here. The modules in the above-described question and answer processing apparatus may be wholly or partially implemented by software, hardware, and a combination thereof. The modules can be embedded in a hardware form or independent of a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 10. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operating system and the computer program to run on the non-volatile storage medium. The database of the computer device is used for storing a two-dimensional table database. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a question-answering processing method.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 11. The computer device comprises a processor, a memory, a network interface, a display screen and an input device which are connected through a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a question-answer processing method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
It will be appreciated by those skilled in the art that the configurations shown in fig. 10 or 11 are only block diagrams of a part of the configurations related to the present application, and do not constitute a limitation of the computer device to which the present application is applied, and a specific computer device may include more or less components than those shown in the drawings, or may combine some components, or have different arrangements of components.
In one embodiment, there is provided a computer device comprising a memory storing a computer program and a processor implementing the following steps when the processor executes the computer program: receiving a user inquiry instruction, and acquiring a user inquiry sentence according to the user inquiry instruction; identifying a user question, and determining an entity and an intention from the user question; matching the entity and the intention by using the trained incidence relation matching model to obtain a matching result; and searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result.
In one embodiment, the processor, when executing the computer program, further performs the steps of: and establishing a preset two-dimensional table knowledge base, wherein the preset two-dimensional table knowledge base comprises an entity, an intention and answers corresponding to the entity and the intention.
In one embodiment, the processor, when executing the computer program, further performs the steps of: identifying the question of the user by using the trained named entity identification model, and determining an entity in the question of the user; and identifying the question of the user by using the trained intention identification model, and determining the intention in the question of the user.
In one embodiment, the processor, when executing the computer program, further performs the steps of: segmenting words of the question of the user to obtain word segmentation results; obtaining a word vector according to the word segmentation result, and inputting the word vector into a trained named entity recognition model for recognition to obtain an entity recognition result vector; and determining the entity in the question of the user according to the entity recognition result vector.
In one embodiment, the processor when executing the computer program further performs the steps of: segmenting words of the question of the user to obtain a word segmentation result; obtaining a word vector according to the word segmentation result, and inputting the word vector into the trained intention recognition model for recognition to obtain an intention recognition result vector; and determining the intention in the question of the user according to the intention recognition result vector.
In one embodiment, the processor when executing the computer program further performs the steps of: matching the entity and the intention to obtain a matching result to be determined; obtaining a matching vector to be determined according to a matching result to be determined, inputting the matching vector to be determined into a trained incidence relation matching model for recognition, and obtaining a matching recognition result vector; and determining a target matching result according to the matching identification result vector.
In one embodiment, the processor, when executing the computer program, further performs the steps of: searching a target entity in the matching result in an entity field in a preset two-dimensional table knowledge base, and searching a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base; and when the target entity and the target intention are found, obtaining answers corresponding to the target entity and the target intention from a preset two-dimensional table knowledge base.
In one embodiment, the processor when executing the computer program further performs the steps of: and returning the answer to the terminal corresponding to the question of the user so as to enable the terminal corresponding to the question of the user to display the answer.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of: receiving a user inquiry instruction, and acquiring a user inquiry sentence according to the user inquiry instruction; identifying a user question, determining an entity and an intention from the user question, and matching the entity and the intention by using a trained incidence relation matching model to obtain a matching result; and searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result.
In one embodiment, the computer program when executed by the processor further performs the steps of: and establishing a preset two-dimensional table knowledge base, wherein the preset two-dimensional table knowledge base comprises an entity, an intention and answers corresponding to the entity and the intention.
In one embodiment, the computer program when executed by the processor further performs the steps of: identifying the question of the user by using the trained named entity identification model, and determining an entity in the question of the user; and identifying the question of the user by using the trained intention identification model, and determining the intention in the question of the user.
In one embodiment, the computer program when executed by the processor further performs the steps of: performing word segmentation on the question of the user to obtain word segmentation results; obtaining a word vector according to the word segmentation result, inputting the word vector into the trained named entity recognition model for recognition, and obtaining an entity recognition result vector; and determining the entity in the question of the user according to the entity recognition result vector.
In one embodiment, the computer program when executed by the processor further performs the steps of: segmenting words of the question of the user to obtain a word segmentation result; obtaining a word vector according to the word segmentation result, and inputting the word vector into the trained intention recognition model for recognition to obtain an intention recognition result vector; and determining the intention in the question of the user according to the intention recognition result vector.
In one embodiment, the computer program when executed by the processor further performs the steps of: matching the entity and the intention to obtain a matching result to be determined; obtaining a matching vector to be determined according to a matching result to be determined, and inputting the matching vector to be determined into a trained incidence relation matching model for recognition to obtain a matching recognition result vector; and determining a target matching result according to the matching identification result vector.
In one embodiment, the computer program when executed by the processor further performs the steps of: searching a target entity in the matching result in an entity field in a preset two-dimensional table knowledge base, and searching a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base; and when the target entity and the target intention are found, obtaining answers corresponding to the target entity and the target intention from a preset two-dimensional table knowledge base.
In one embodiment, the computer program when executed by the processor further performs the steps of: and returning the answer to the terminal corresponding to the question of the user so that the terminal corresponding to the question of the user displays the answer.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), rambus (Rambus) direct RAM (RDRAM), direct Rambus Dynamic RAM (DRDRAM), and Rambus Dynamic RAM (RDRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (11)
1. A question-answer processing method, the method comprising:
receiving a user inquiry instruction, and acquiring a user inquiry sentence according to the user inquiry instruction;
identifying the user question, and determining an entity and at least two intentions from the user question;
establishing an incidence relation between the entity and the at least two intentions by using a trained incidence relation matching model to obtain an entity and an associated intention, wherein the trained incidence relation matching model is obtained by marking the entity in the existing question and the entity and the intention associated in the intention as association, marking the unassociated entity and the intention as unassociated and training by using marked data;
and searching answers corresponding to the question of the user in a preset two-dimensional table knowledge base according to the entity and the associated intention.
2. The method according to claim 1, before said receiving the user query instruction and obtaining the user query sentence according to the user query instruction, comprising:
and establishing the preset two-dimensional table knowledge base, wherein the preset two-dimensional table knowledge base comprises an entity, an intention and answers corresponding to the entity and the intention.
3. The method of claim 1, wherein the identifying the user question, determining an entity and an intent from the user question, comprises:
identifying the user question by using a trained named entity identification model, and determining an entity in the user question;
and identifying the user question by using the trained intention identification model, and determining the intention in the user question.
4. The method of claim 3, wherein identifying the user question using the trained named entity recognition model to determine the entities in the user question comprises:
performing word segmentation on the question of the user to obtain a word segmentation result;
obtaining a word vector according to the word segmentation result, and inputting the word vector into a trained named entity recognition model for recognition to obtain an entity recognition result vector;
and determining the entity in the question of the user according to the entity recognition result vector.
5. The method of claim 3, wherein identifying the user question using the trained intent recognition model to determine the intent in the user question comprises:
segmenting words of the user question to obtain a word segmentation result;
obtaining a word vector according to the word segmentation result, inputting the word vector into a trained intention recognition model for recognition, and obtaining an intention recognition result vector;
and determining the intention in the question of the user according to the intention recognition result vector.
6. The method according to any one of claims 1 to 5, wherein the matching the entity and the intention using the trained incidence relation matching model to obtain a matching result comprises:
matching the entity and the intention to obtain a matching result to be determined;
obtaining a matching vector to be determined according to the matching result to be determined, and inputting the matching vector to be determined into a trained incidence relation matching model for recognition to obtain a matching recognition result vector;
and determining a target matching result according to the matching identification result vector.
7. The method according to any one of claims 1 to 5, wherein searching for the answer corresponding to the question of the user in a preset two-dimensional table knowledge base according to the matching result comprises:
searching a target entity in the matching result in an entity field in the preset two-dimensional table knowledge base, and searching a target intention matched with the target entity in the matching result in an intention field in the preset two-dimensional table knowledge base;
and when the target entity and the target intention are found, obtaining answers corresponding to the target entity and the target intention from the preset two-dimensional table knowledge base.
8. The method according to any one of claims 1 to 5, further comprising, after searching for the answer corresponding to the user question in a preset two-dimensional table knowledge base according to the matching result:
and returning the answer corresponding to the question of the user to the terminal so that the terminal displays the answer.
9. A question-answering processing apparatus characterized by comprising:
the question acquiring module is used for receiving a user inquiry instruction and acquiring a user question according to the user inquiry instruction;
the identification module is used for identifying the question of the user to obtain an entity and at least two intentions corresponding to the question of the user;
the matching module is used for establishing an incidence relation between the entity and the at least two intents by using a trained incidence relation matching model to obtain the entity and the associated intents, wherein the trained incidence relation matching model is obtained by marking the entity in the existing question sentence and the entity and the intention which are associated in the intents as associated, marking the entity and the intention which are not associated as not associated and training by using marked data;
and the searching module is used for matching the preset entity and the intention two-dimensional table according to the entity and the associated intention and obtaining an answer corresponding to the question of the user according to a matching result.
10. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 8 when executing the computer program.
11. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910560681.XA CN110427467B (en) | 2019-06-26 | 2019-06-26 | Question-answer processing method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910560681.XA CN110427467B (en) | 2019-06-26 | 2019-06-26 | Question-answer processing method, device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110427467A CN110427467A (en) | 2019-11-08 |
| CN110427467B true CN110427467B (en) | 2022-10-11 |
Family
ID=68409532
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910560681.XA Active CN110427467B (en) | 2019-06-26 | 2019-06-26 | Question-answer processing method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110427467B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110851484A (en) * | 2019-11-13 | 2020-02-28 | 北京香侬慧语科技有限责任公司 | Method and device for obtaining multi-index question answers |
| CN110930989B (en) * | 2019-11-27 | 2021-04-06 | 深圳追一科技有限公司 | Speech intention recognition method and device, computer equipment and storage medium |
| CN111209384B (en) * | 2020-01-08 | 2023-08-15 | 腾讯科技（深圳）有限公司 | Question-answer data processing method and device based on artificial intelligence and electronic equipment |
| CN111309882B (en) * | 2020-02-13 | 2023-06-09 | 支付宝(杭州)信息技术有限公司 | Method and device for realizing intelligent customer service question and answer |
| CN111368044A (en) * | 2020-02-21 | 2020-07-03 | 深圳追一科技有限公司 | Intelligent question answering method and device, computer equipment and storage medium |
| CN111444313B (en) * | 2020-03-04 | 2023-10-24 | 深圳追一科技有限公司 | Knowledge graph-based question and answer method, knowledge graph-based question and answer device, computer equipment and storage medium |
| CN111325037B (en) * | 2020-03-05 | 2022-03-29 | 苏宁云计算有限公司 | Text intention recognition method and device, computer equipment and storage medium |
| CN111414467A (en) * | 2020-03-20 | 2020-07-14 | 中国建设银行股份有限公司 | Question-answer dialogue method and device, electronic equipment and computer readable storage medium |
| CN111611366B (en) * | 2020-05-20 | 2023-08-11 | 北京百度网讯科技有限公司 | Method, device, equipment and storage medium for optimizing intention recognition |
| CN111506723B (en) * | 2020-07-01 | 2020-11-13 | 平安国际智慧城市科技股份有限公司 | Question-answer response method, device, equipment and storage medium |
| CN111858966B (en) * | 2020-08-05 | 2021-12-31 | 龙马智芯（珠海横琴）科技有限公司 | Knowledge graph updating method and device, terminal equipment and readable storage medium |
| CN112101035B (en) * | 2020-09-21 | 2022-08-09 | 贝壳技术有限公司 | Named entity identification method and device, readable storage medium and electronic equipment |
| CN112183098B (en) * | 2020-09-30 | 2022-05-06 | 完美世界(北京)软件科技发展有限公司 | Session processing method and device, storage medium and electronic device |
| CN112071300B (en) * | 2020-11-12 | 2021-04-06 | 深圳追一科技有限公司 | Voice conversation method, device, computer equipment and storage medium |
| CN115982391B (en) * | 2023-03-17 | 2023-07-25 | 恒生电子股份有限公司 | Information processing method and device |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108446286B (en) * | 2017-02-16 | 2023-04-25 | 阿里巴巴集团控股有限公司 | Method, device and server for generating natural language question answers |
| CN107918634A (en) * | 2017-06-27 | 2018-04-17 | 上海壹账通金融科技有限公司 | Intelligent answer method, apparatus and computer-readable recording medium |
| CN108268580A (en) * | 2017-07-14 | 2018-07-10 | 广东神马搜索科技有限公司 | The answering method and device of knowledge based collection of illustrative plates |
| US10678816B2 (en) * | 2017-08-23 | 2020-06-09 | Rsvp Technologies Inc. | Single-entity-single-relation question answering systems, and methods |
| CN107748757B (en) * | 2017-09-21 | 2021-05-07 | 北京航空航天大学 | Question-answering method based on knowledge graph |
| CN107992543B (en) * | 2017-11-27 | 2020-11-17 | 上海智臻智能网络科技股份有限公司 | Question-answer interaction method and device, computer equipment and computer readable storage medium |
| CN108959627B (en) * | 2018-07-23 | 2021-12-17 | 北京光年无限科技有限公司 | Question-answer interaction method and system based on intelligent robot |
| CN109522393A (en) * | 2018-10-11 | 2019-03-26 | 平安科技（深圳）有限公司 | Intelligent answer method, apparatus, computer equipment and storage medium |
-
2019
- 2019-06-26 CN CN201910560681.XA patent/CN110427467B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110427467A (en) | 2019-11-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110427467B (en) | Question-answer processing method, device, computer equipment and storage medium | |
| CN109446302B (en) | Question-answer data processing method and device based on machine learning and computer equipment | |
| CN110502608B (en) | Man-machine conversation method and man-machine conversation device based on knowledge graph | |
| CN110334179B (en) | Question-answer processing method, device, computer equipment and storage medium | |
| CN110457431B (en) | Knowledge graph-based question and answer method and device, computer equipment and storage medium | |
| CN109871446B (en) | Refusing method in intention recognition, electronic device and storage medium | |
| CN109522393A (en) | Intelligent answer method, apparatus, computer equipment and storage medium | |
| CN108427707B (en) | Man-machine question and answer method, device, computer equipment and storage medium | |
| CN111325037B (en) | Text intention recognition method and device, computer equipment and storage medium | |
| CN113157863B (en) | Question-answer data processing method, device, computer equipment and storage medium | |
| CN109063108B (en) | Search ranking method and device, computer equipment and storage medium | |
| CN109800307B (en) | Product evaluation analysis method and device, computer equipment and storage medium | |
| CN113707300B (en) | Search intention recognition method, device, equipment and medium based on artificial intelligence | |
| CN108520041B (en) | Industry classification method and system of text, computer equipment and storage medium | |
| CN111192025A (en) | Occupational information matching method and device, computer equipment and storage medium | |
| CN110442697B (en) | Man-machine interaction method, system, computer equipment and storage medium | |
| CN110263133B (en) | Knowledge graph-based question and answer method, electronic device, equipment and storage medium | |
| CN111309881A (en) | Method and device for processing unknown questions in intelligent question answering, computer equipment and medium | |
| CN110910864A (en) | Training sample selection method and device, computer equipment and storage medium | |
| CN110750523A (en) | Data annotation method, system, computer equipment and storage medium | |
| CN111274822A (en) | Semantic matching method, device, equipment and storage medium | |
| CN112949297A (en) | Intention recognition method, device, computer equipment and computer readable storage medium | |
| CN112651236A (en) | Method and device for extracting text information, computer equipment and storage medium | |
| CN112632258A (en) | Text data processing method and device, computer equipment and storage medium | |
| CN113704436A (en) | User portrait label mining method and device based on session scene |
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 |