CN110263144A - A kind of answer acquisition methods and device - Google Patents
A kind of answer acquisition methods and device Download PDFInfo
- Publication number
- CN110263144A CN110263144A CN201910570444.1A CN201910570444A CN110263144A CN 110263144 A CN110263144 A CN 110263144A CN 201910570444 A CN201910570444 A CN 201910570444A CN 110263144 A CN110263144 A CN 110263144A
- Authority
- CN
- China
- Prior art keywords
- answer
- text
- default
- enquirement
- type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000012549 training Methods 0.000 claims abstract description 9
- 238000013507 mapping Methods 0.000 claims description 33
- 238000012545 processing Methods 0.000 claims description 18
- 238000003860 storage Methods 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 abstract description 9
- 238000010586 diagram Methods 0.000 description 15
- 230000008569 process Effects 0.000 description 7
- 230000003993 interaction Effects 0.000 description 6
- 235000021167 banquet Nutrition 0.000 description 5
- 238000004590 computer program Methods 0.000 description 5
- 230000015572 biosynthetic process Effects 0.000 description 4
- 238000003786 synthesis reaction Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 238000012550 audit Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000000556 factor analysis Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3343—Query execution using phonetics
-
- 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
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Abstract
The invention discloses a kind of answer acquisition methods and device, wherein method are as follows: obtain and put question to text;By the enquirement disaggregated model of pre-training, determine whether the enquirement type for puing question to text matches with the answer type in default answer type library;If the enquirement type for puing question to text is not matched with the answer type in the default answer type library, then according at least one semantic similarity of the first deep semantic vector of each answer in the default answer library and the second deep semantic vector for puing question to text, the answer for puing question to text is obtained.When the above method is applied to financial technology (Fintech), pass through the enquirement disaggregated model of pre-training, determine whether the enquirement type for puing question to text matches with the answer type in default answer type library first, when mismatching, since at least one described semantic similarity characterizes the semantic degree of agreement of each answer in enquirement text and default answer library, to improve the accuracy rate for getting answer.
Description
Technical field
The present invention relates to the financial technology field (Fintech) and intelligence communication field more particularly to a kind of answer acquisition sides
Method and device.
Background technique
With the development of computer technology, more and more technologies (big data, distribution, block chain (Blockchain),
Artificial intelligence etc.) it applies in financial field, traditional financial industry gradually changes to financial technology (Fintech).Currently, financial
During intelligence communication in sciemtifec and technical sphere, it is often necessary to according to the enquirement of user, the answer of problem is returned to user.
The mode for obtaining customer problem answer in telephone call-in system at present is, according to the keyword of enquirement text artificial
It is scanned in the answer library of collection, and by the answer feedback of hit to user.Obviously, the answer library artificially collected is limited, and
Inefficiency, and scan for according to keyword the answer got, it is difficult to guarantee the answer semanteme got for expected language
Justice, it is also not accurate enough so as to cause the user's answer got.
Summary of the invention
The embodiment of the present application provides a kind of answer acquisition methods and device, solves the user got in the prior art and answers
The not accurate enough problem of case.
In a first aspect, the embodiment of the present application provides a kind of answer acquisition methods, comprising: obtain and put question to text;By instructing in advance
Experienced enquirement disaggregated model, determine it is described put question to text enquirement type whether with the answer type in default answer type library
Match;If the enquirement type for puing question to text is not matched with the answer type in the default answer type library, according to
The first deep semantic vector of each answer and the second deep semantic vector for puing question to text be at least in default answer library
One semantic similarity obtains the answer for puing question to text.
In the above method, pass through the enquirement disaggregated model of pre-training, it is first determined put question to text enquirement type whether with
Answer type matching in default answer type library is mentioned when mismatching since at least one described semantic similarity characterizes
The semantic degree of agreement for asking each answer in text and default answer library, to improve the accuracy rate for getting answer.
In a kind of optional embodiment, if the enquirement enquirement type of text and answering in the default answer type library
Case type matching then obtains the enquirement text matched answer type corresponding at least one in the default answer type library
A key answer element;According to the enquirement text and at least one described crucial answer element, at least one described pass is determined
The sub- answer for puing question to text is wanted at least one described crucial answer by each key answer element in key answer element
Element for it is described put question to text all sub- answers combination, as it is described put question to text answer.
In the above method, each of disassembles out crucial answer element according at least one key answer element, determine every
It is a key answer element for it is described put question to text sub- answer, thus more fine granularity obtain it is described put question to text answer,
Improve the accuracy rate for getting answer.
In a kind of optional embodiment, the first deep semantic vector according to each answer in the default answer library
With at least one semantic similarity of the second deep semantic vector for puing question to text, the answer for puing question to text is determined,
Include:
It is semantic similar by described first if first semantic similarity is greater than or equal to default semantic similarity threshold value
The first deep semantic vector corresponding answer in the default answer library of degree, as the answer for puing question to text;It is described
First semantic similarity is maximum semantic similarity at least one described semantic similarity.
By the way that default semantic similarity threshold value is arranged, above or equal to the first semantic phase of default semantic similarity threshold value
Like the first deep semantic vector corresponding answer in the default answer library of degree, as the answer for puing question to text,
Under the premise of guaranteeing semantic similarity, obtains and put question to text maximum answer of semantic similarity in the default answer library.
In a kind of optional embodiment, if first semantic similarity is less than the default semantic similarity threshold value,
Determine whether the enquirement text matches with the triple in the knowledge mapping of prebuild;If so, by the enquirement text with
Matched triple in the knowledge mapping of the prebuild, as the answer for puing question to text.
In the above method, when the first semantic similarity is less than the default semantic similarity threshold value, determines and put question to text
Whether match with the triple in the knowledge mapping of prebuild, so that the knowledge mapping of prebuild plays the role of backup, is used for
When first semantic similarity is less than the default semantic similarity threshold value, by the knowledge graph for puing question to text and the prebuild
Matched triple in spectrum, as the answer for puing question to text, to further improve the accuracy rate for getting answer.
In a kind of optional embodiment, if the enquirement text not with the triple in the knowledge mapping of the prebuild
Match, it is determined that it is described put question to text whether with the chat statement matching in default chat library, if so, by the enquirements text and
Matched chat sentence in the default chat library, as the answer for puing question to text;Alternatively, if it is not, then by default default
Answer is as the answer for puing question to text.
In the above method, matched chat sentence in the enquirement text and the default chat library is mentioned as described
Ask the answer of text, or using default default answer as the answer for puing question to text, thus the enquirement text not with
When triple in the knowledge mapping of the prebuild matches, there are also the chat sentence of backup and default default answers, thus
User's impression is improved, and improves answer accuracy rate.
Second aspect, the application provide a kind of answer acquisition device, comprising: obtain module, put question to text for obtaining;Place
Manage module, for by the enquirement disaggregated model of pre-training, determine the enquirement type for puing question to text whether with default answer
Answer type matching in typelib;If it is described put question to text enquirement type not with the answer in the default answer type library
Type matching then puts question to the second of text with described according to the first deep semantic vector of each answer in the default answer library
At least one semantic similarity of deep semantic vector obtains the answer for puing question to text.
In a kind of optional embodiment, the processing module is also used to: if the enquirement type for puing question to text with it is described
Answer type matching in default answer type library, then it is matched in the default answer type library to obtain the enquirement text
At least one corresponding crucial answer element of answer type;It is wanted according to the enquirement text and at least one described crucial answer
Element determines that each crucial answer element, will for the sub- answer for puing question to text at least one described crucial answer element
Described at least one crucial answer element for all sub- answers for puing question to text combination, as the text of puing question to
Answer.
In a kind of optional embodiment, the processing module is specifically used for: if first semantic similarity is greater than or waits
In default semantic similarity threshold value, then by the first deep semantic vector of first semantic similarity in the default answer library
In corresponding answer, as it is described put question to text answer;First semantic similarity is that at least one described semanteme is similar
Maximum semantic similarity in degree.
In a kind of optional embodiment, the processing module is also used to: if first semantic similarity be less than it is described pre-
If semantic similarity threshold value, it is determined that whether the enquirement text matches with the triple in the knowledge mapping of prebuild;If so,
Then by matched triple in the knowledge mapping of the enquirement text and the prebuild, as the answer for puing question to text.
In a kind of optional embodiment, the processing module is also used to: if the enquirement text not with the prebuild
In knowledge mapping triple matching, it is determined that it is described put question to text whether with it is default chat library in chat statement matching, if
It is, then by matched chat sentence in the enquirement text and the default chat library, as the answer for puing question to text;Or
Person, if it is not, then using default default answer as the answer for puing question to text.
The beneficial effect of above-mentioned second aspect and each embodiment of second aspect can refer to above-mentioned first aspect and first
The beneficial effect of each embodiment of aspect, which is not described herein again.
The third aspect, the embodiment of the present application provide a kind of computer equipment, including program or instruction, when described program or refer to
Order is performed, the method to execute above-mentioned first aspect and each embodiment of first aspect.
Fourth aspect, the embodiment of the present application provides a kind of storage medium, including program or instruction, when described program or instruction
It is performed, the method to execute above-mentioned first aspect and each embodiment of first aspect.
Detailed description of the invention
Fig. 1 is a kind of framework of the applicable intelligent incoming telephone call system of answer acquisition methods provided by the embodiments of the present application
Schematic diagram;
Fig. 2 is a kind of timing interaction schematic diagram of answer acquisition methods provided by the embodiments of the present application;
Fig. 3 is a kind of step flow diagram of answer acquisition methods provided by the embodiments of the present application;
Fig. 4 is a kind of specific steps flow diagram of answer acquisition methods provided by the embodiments of the present application;
Fig. 5 is a kind of structural schematic diagram of answer acquisition device provided by the embodiments of the present application.
Specific embodiment
In order to better understand the above technical scheme, below in conjunction with Figure of description and specific embodiment to above-mentioned
Technical solution is described in detail, it should be understood that the specific features in the embodiment of the present application and embodiment are to the application skill
The detailed description of art scheme, rather than the restriction to technical scheme, in the absence of conflict, the embodiment of the present application
And the technical characteristic in embodiment can be combined with each other.
In the operation process of telecom operation business, user often encounters various problem, needs to answer.Phone
Call-in system is a system for user enquirement and answering to user, and telephone call-in system supports subscriber phone incoming call simultaneously
Offer problem is answerred questions the system of guide.
Obtaining the mode of customer problem answer in telephone call-in system at present, to will lead to the user's answer got not quasi- enough
Really, the embodiment of the present application provides a kind of answer acquisition methods thus.
As shown in Figure 1, being a kind of applicable intelligent incoming telephone call system of answer acquisition methods provided by the embodiments of the present application
The configuration diagram of system.
The system architecture includes following part:
Terminal device: terminal device is the equipment that client is used to make a phone call enquirement, initiates incoming call.
Telecom operators: telecom operators are used for the incoming call of relay terminal.
Corporate networks: corporate networks are used for the incoming call that telecom operators transfer, and incoming call is relayed to intelligence machine
People.
Attend a banquet call service platform: call service platform of attending a banquet includes multiple manual positions, for manually taking over incoming call.
Intelligent robot: intelligent robot is sent a telegram here for listening user, calling record is converted to text, or text is turned
Calling record is turned to, the answer that user of incoming call is putd question to is returned.
Based on above-mentioned framework, answer can be completed by following below scheme and obtained.
S101: user dials customer service by terminal device and attends a banquet phone.
S102: incoming call is transferred by telecom operators and corporate networks, and is sent a telegram here by intelligent robot auto-pickup.
S103: interaction text is converted into voice by speech synthesis engine and plays to user by intelligent robot, and guidance is used
It puts question at family.
S104: intelligent robot, which records user's communication, to be converted into puing question to text by speech recognition engine.
S105: intelligent customer service robot obtains answer, then converts Chinese idiom by speech synthesis engine by puing question to text
Sound plays to user.
S106: repeating S103-S105, the end of conversation library if user actively hangs up, if user speech input " turns people
Work " is then conversed by operator attendance trustship, until user hangs up, end of conversation.
Above system framework is discussed in detail below with reference to Fig. 2, as shown in Fig. 2, answering for one kind provided by the embodiments of the present application
The timing interaction schematic diagram of case acquisition methods.The following steps are included:
Step 201: making a phone call.
User is made a phone call by terminal device, generates incoming call, and the talking state of terminal device is exhalation state at this time.
Step 202: notice telephony access.
Telecom operators notify intelligent robot incoming call.
Step 203: sending and connect telephone order.
After intelligent robot receives incoming call, is sent to telecom operators and connect telephone order.
Step 204: receiving calls.
Telecom operators connect phone, and the talking state of terminal device is on-state at this time.
Step 205: recording call voice, obtain calling record.
The voice of telecom operators' listening user, and obtain calling record.
Step 206: sending calling record.
The calling record of conversion is sent to speech recognition engine by telecom operators.
Step 207: converting enquirement text for calling record.
Calling record is converted enquirement text by speech recognition engine.
Step 208: the answer text of returning response enquirement text.
Intelligent robot obtains the answer text for puing question to text, and is back to speech synthesis engine.
Step 209: sending answer voice.
Speech synthesis engine answer text is converted into answer voice, and answer voice is sent to telecom operators.
Step 2010: sending answer voice.
Telecom operators send answer voice to terminal device.
If intelligent robot does not receive and turns manual command and user does not hang up the telephone, recycles and execute step 205~step
Rapid 2010.
If the problem of intelligent robot is received in step 207 text is " turning artificial ", 2011~step is thened follow the steps
Rapid 2012.
Step 2011: transmission turns manual command.
Intelligent robot is sent to telecom operators turns manual command.
Step 2012: access manual service.
Terminal device is connect call with call service platform of attending a banquet by telecom operators, to realize access manual service.
Step 2013: hanging up the telephone.
It should be noted that step 2013 may be held between the adjacent step of any two in step 201~step 2012
Row.
Intelligent robot automatic call answering used herein, and user's particular task type can be putd question to and carry out more wheels
Dialogue interaction puts question to non task type and carries out deep semantic analysis and search answer, while user also being supported to turn operator attendance clothes
Business.This scheme can obtain good balance between the service of intelligent customer service robot and the service of attending a banquet, and solve above-mentioned
Defect brings huge business earnings.
Intelligent robot is automatic come what is realized using natural language processing technique, machine learning and deep learning frame etc.
Question answering system, more wheel sessions can be carried out with user and deep semantic understands that the enquirement of user is intended to, by executing task or lookup
Knowledge base obtains answer required for user.The technology used include be intended to understanding, semantic matches, search engine, recommended engine,
Task engine, knowledge mapping etc..
The knowledge base composition of the application is as shown in table 1:
Problem | Answer | Knowledge point producing method |
Put question to text | Default answer type | Machine self-learning auxiliary |
Put question to text | Default chat library | Machine self-learning auxiliary |
Put question to text | Knowledge mapping | Periodically editor, content are relatively stable |
Put question to text | Default chat library | Periodically editor, content are relatively stable |
Table 1
It should be noted that default answer type and default chat library are to assist generating by Machine self-learning, when mentioning
Ask that text when default answer type and default chat storehouse matching are less than content, is then transmitted to operation system, it is true by business personnel
Belong to existing content if recognizing, it is accurate not match, then it is a kind of the enquirement text to be classified as existing content, if being not belonging in existing
Hold, then will edit new content, and be added in corresponding knowledge base.To constantly roll up default answer type and default chat library,
To constantly expand content, and Similar Problems are constantly classified as one kind.
The problem of this programme knowledge base, part only user putd question to text to form, rather than keyword and mode, met people
Class natural dialogue habit, greatly reduces editor's difficulty of knowledge base.Answer part then includes default answer type, default answer
Library, the knowledge mapping of prebuild and chat library, the intelligence degree of stronger support dialogue is promoted, the richness of answer mentions
It rises.
As shown in figure 3, being a kind of step flow diagram of answer acquisition methods provided by the embodiments of the present application.
Step 301: obtaining and put question to text.
Wherein, puing question to text is the text of user recording conversion.
Step 302: by the enquirement disaggregated model of pre-training, determine the enquirement type for puing question to text whether with it is default
Answer type matching in answer type library.
It is to need intelligent robot operation that must provide with the matched enquirement type of answer type in default answer type library
The enquirement type of body answer does not generally directly provide specific answer in default answer library, but first passes through default answer type
Library determines answer type, further according to the specifying information putd question in text, determines the answer for puing question to text.
Step 303: if it is described put question to text enquirement type not with the answer type in the default answer type library
Match, then according to the first deep semantic vector of each answer in the default answer library and the second depth language for puing question to text
At least one semantic similarity of adopted vector obtains the answer for puing question to text.
First deep semantic vector is a vector for describing corresponding answer semanteme in default answer library, each dimension
One attribute of the answer can be described.Correspondingly, the second deep semantic vector be for describe enquirement one of text semantic to
Amount.
In the above method, pass through the enquirement disaggregated model of pre-training, it is first determined put question to text enquirement type whether with
Answer type matching in default answer type library is mentioned when mismatching since at least one described semantic similarity characterizes
The semantic degree of agreement for asking each answer in text and default answer library, to improve the accuracy rate for getting answer.
In step 303, according at least one described semantic similarity, the answer for puing question to text is determined, it can be according to
Following manner carries out:
It is semantic similar by described first if first semantic similarity is greater than or equal to default semantic similarity threshold value
The first deep semantic vector corresponding answer in the default answer library of degree, as the answer for puing question to text;It is described
First semantic similarity is maximum semantic similarity at least one described semantic similarity.
It should be noted that the describing mode of each semantic similarity is without limitation at least one semantic similarity.It lifts
It, can be with the included angle cosine value of the first deep semantic vector and the second deep semantic vector come table with the first semantic similarity for example
Show.
By the way that default semantic similarity threshold value is arranged, above or equal to the first semantic phase of default semantic similarity threshold value
Like the first deep semantic vector corresponding answer in the default answer library of degree, as the answer for puing question to text,
Under the premise of guaranteeing semantic similarity, obtains and put question to text maximum answer of semantic similarity in the default answer library.
In above-mentioned optional embodiment, if first semantic similarity is less than the default semantic similarity threshold value,
Then determine whether the enquirement text matches with the triple in the knowledge mapping of prebuild;If so, by the enquirement text
With matched triple in the knowledge mapping of the prebuild, as it is described put question to text answer.
For example, put question to includes keyword A1 in text.Wherein, A1 matched ternary in the knowledge mapping of prebuild
Group is A1A2A3, then returns to A1A2A3.
In the above method, when the first semantic similarity is less than the default semantic similarity threshold value, determines and put question to text
Whether match with the triple in the knowledge mapping of prebuild, so that the knowledge mapping of prebuild plays the role of backup, is used for
When first semantic similarity is less than the default semantic similarity threshold value, by the knowledge graph for puing question to text and the prebuild
Matched triple in spectrum, as the answer for puing question to text, to further improve the accuracy rate for getting answer.
In above-mentioned optional embodiment, if the enquirement text not with the triple in the knowledge mapping of the prebuild
Matching, it is determined that it is described put question to text whether with it is default chat library in chat statement matching, if so, by the enquirement text
With matched chat sentence in the default chat library, as it is described put question to text answer;Alternatively, if it is not, then will be default silent
Answer is recognized as the answer for puing question to text.
For example, text B is putd question to not match with the triple in the knowledge mapping of the prebuild, it is determined that whether B
With the chat statement matching in default chat library, if B is matched with the chat sentence B in default chat library, will chat sentence B
As the answer for puing question to text B.
In the above method, matched chat sentence in the enquirement text and the default chat library is mentioned as described
Ask the answer of text, or using default default answer as the answer for puing question to text, thus the enquirement text not with
When triple in the knowledge mapping of the prebuild matches, there are also the chat sentence of backup and default default answers, thus
User's impression is improved, and improves answer accuracy rate.
In step 303, if the answer type in the enquirement type for puing question to text and the default answer type library
Match, then a kind of optional embodiment are as follows:
Obtain at least one corresponding pass of the enquirement text matched answer type in the default answer type library
Key answer element;According to the enquirement text and at least one described crucial answer element, determine that at least one described key is answered
Each key answer element, will at least one described crucial answer element pair for the sub- answer for puing question to text in case element
In the combination of all sub- answers for puing question to text, as the answer for puing question to text.
For example, if text C is putd question to match with answer type " answer type 1 ".Answer type 1 includes that crucial answer is wanted
4 this 4 element 1, crucial answer element 2, crucial answer element 3, crucial answer element crucial answer elements.It puts question to text C and answers
Crucial answer element 1, crucial answer element 2, crucial answer element 3 match in case Class1, then according to put question to text C respectively with
Crucial answer element 1, crucial answer element 2, the matched sub- answer 1 of crucial answer element 3, sub- answer 2, sub- answer 3, son is answered
Case 1, sub- answer 2, sub- answer 3 combination as enquirement text C answer.Moreover, after returning to the answer for puing question to text C, intelligence
Energy robot can also ask in reply the sub- answer for whether needing to obtain crucial answer element 4.
In the above method, each of disassembles out crucial answer element according at least one key answer element, determine every
It is a key answer element for it is described put question to text sub- answer, thus more fine granularity obtain it is described put question to text answer,
Improve the accuracy rate for getting answer.
Illustrated below with specific embodiment.
Intelligent robot is held if intelligent robot uses corresponding answer type if being matched to default answer type library
Row single-wheel or more wheel dialogue obtain institute it is necessary to slot position information, then execute task using slot position information, and by task execution
As a result user is returned to;If the problem of being matched to default answer library, as answer;If found in knowledge mapping pair
Triple is answered then to return to triple, as answer;It is last to return to chat sentence if finding answer in chat library, as
Answer.
User puts question to " 10000 yuan of interests if 10 days of borrowing money are how many? ".
Intelligent robot is first using puing question to disaggregated model to carry out enquirement classification of type to current problem, if the enquirement class
Type can be matched to answer type in default answer type library, then carry out more wheel sessions using corresponding answer type, be somebody's turn to do
Then all key answer elements needed for answer type carry out result is calculated returning to user.In this instance, if it is right
It puts question to text to carry out factor analysis and has found that the amount of money is 10000 yuan, the borrowing time is 10 days, then also lacking crucial answer element
" rate per diem ", intelligent robot can ask in reply user " your rate per diem is how many? ", answer obtaining user and parse rate per diem
Value after, the task will execute it is last interest number is calculated, and return to user.
If there is no corresponding answer type in default answer type library, intelligent robot will be selected in default answer
It is searched in library, intelligent robot is asked the deep semantic vector that user puts question to is calculated with candidate answers in default answer library
Semantic similarity between the deep semantic vector of topic, and the highest answer of semantic similarity is taken to return to user as a result,
The accuracy rate and recall rate of answer are substantially improved from actual effect.For example if keyword " loan " conduct,
Answer must can just be recalled comprising " loan " two word by puing question in the prior art, and deep semantic vector does not need keyword one then
Sample, if it is semantic the same, such as " borrowing money ", " loaning bill " etc., recall rate gets a promotion;Another aspect keyword is generally in sentence
Word accounts for smaller in son, influences, may be led due to not accounting for other words between sentence when searching answer using keyword
Cause the answer deviation searched larger, and deep semantic vector is then the semantic effect for considering all words in sentence, therefore is searched
The answer accuracy rate arrived is also higher.
If not having this to put question to the answer of text in default answer library, selection is searched in knowledge mapping and is answered by robot
Case, for example, the triple in knowledge mapping is, by subject-predicate-object (subject-predication-object,
SPO) triple can also be other types triple, it is not limited here.Intelligent robot will extract crucial in text
SPO, to one of SPO (Subject, Predicate and Object) three elements that can find missing in knowledge mapping, or in only lookup, reasoning on the way
Relationship that may be present between SPO in enquirement.As the crucial answer factor combination returned to default answer type library, or it is default
The answer supplement returned in answer library.
If above three step cannot all find reasonable answer, enters lookup chat sentence in chat library and return to use
Family, as answer.Finally, returning to default default answer to user if being not matched to chat sentence in chat library.
For example default default answer is: " it is sorry, the problem of you propose can not be answered at present? I can answer following problems ... ".
By above-mentioned multi-level knowledge base, take turns interaction mechanism, deep semantic analysis model can very good solution it is existing
Technology there are the problem of, the accuracy rate and recall rate that significant increase is answered a question, significant increase dialogue interaction in user experience.
In addition to this, the knowledge base of the application possesses stronger self-learning capability.Major embodiment are as follows: 1, to searching less than answering
The enquirement of case such as is clustered, is classified, being filtered at the processing, and collection is periodically pushed to business personnel and goes association existing the problem of will obtain
Corresponding answer is added in knowledge point.2, all user in history is putd question to and is clustered, found and ask with having in knowledge base
It inscribes similar way to put questions and submits to business personnel's audit to supplement the Similar Problems of knowledge point, to promote the accuracy rate of answer
And recall rate.3, the dialogue data after turning artificial is analyzed, excavate new knowledge point and is pushed to business personnel's audit, repairs
Change and be added in knowledge base, as the complete rate of knowledge base is promoted, user, which turns artificial rate, be will further decrease.4, to existing
Knowledge point carries out regular, for example will repeat and similar knowledge point is regular at one, will including that the knowledge point of multiple themes is split
At multiple knowledge points etc..
Below with reference to Fig. 4, a kind of answer acquisition methods provided by the embodiments of the present application are discussed in detail, as shown in figure 4, for this
Apply for the specific steps flow diagram that embodiment provides.
Step 401: obtaining the enquirement text of user.
Step 402: whether the talking state for determining user is task status.
If user, which has preset, corresponding answer type in answer type library, and is being directed to answer type call, then
Determine user in task status.
If so, thening follow the steps 406;It is no to then follow the steps 403.
Step 403: carrying out slot position analysis, determine whether slot position has expired.
Slot position is to work as the position for the problem of previous round session can be putd question to continuously, and explanation is in the wheel session if slot position has expired
It cannot additionally put question to.
If so, thening follow the steps 404;It is no to then follow the steps 405.
Step 404: re-initiating a wheel session.
After step 404, return step 401.
Step 405: executing task, search at least one corresponding crucial answer element in the answer type.
After step 405, sub- answer combination of at least one crucial answer element by this, as the answer for puing question to text,
And execute step 411.
Step 406: determining whether the enquirement type for puing question to text in default answer type library has matched answer type.
If so, thening follow the steps 407;Otherwise, step 408 is executed.
Step 407: setting task status for the talking state of user.
After step 407, step 403 is executed.
Step 408: determining and put question to whether type in default answer library has matched answer.
If so, then using the answer as the answer of enquirement text, and execute step 411;Otherwise, step 409 is executed.
Step 409: determine put question to type whether chat library in have matched answer.
If so, using the answer as the answer of enquirement text, and execute step 411;Otherwise, step 410 is executed.
Step 410: using default default answer as the answer of enquirement text, and executing step 411.
Step 411: the answer of text will be putd question to return to user.
Step 412: terminating call.
It should be noted that user may hang up at any time in step 401~step 411, that is, jump directly to step
412。
There it can be seen that technology used in this application is more advanced in terms of intelligent robot answer lookup, level is richer
Richness, process are more reasonable, greatly improve the accuracy rate and recall rate of answer return, reduce the probability for turning artificial, promote user
Experience.
As shown in figure 5, being a kind of structural schematic diagram of answer acquisition device provided by the embodiments of the present application, the application is proposed
A kind of answer acquisition device, comprising: obtain module 501, put question to text for obtaining;Processing module 502, for passing through pre-training
Enquirement disaggregated model, determine it is described put question to text enquirement type whether with the answer type in default answer type library
Match;If the enquirement type for puing question to text is not matched with the answer type in the default answer type library, according to
The first deep semantic vector of each answer and the second deep semantic vector for puing question to text be at least in default answer library
One semantic similarity obtains the answer for puing question to text.
In a kind of optional embodiment, the processing module 502 is also used to: if the enquirement type for puing question to text and institute
The answer type matching in default answer type library is stated, then obtains the enquirement text and is matched in the default answer type library
At least one corresponding crucial answer element of answer type;It is wanted according to the enquirement text and at least one described crucial answer
Element determines that each crucial answer element, will for the sub- answer for puing question to text at least one described crucial answer element
Described at least one crucial answer element for all sub- answers for puing question to text combination, as the text of puing question to
Answer.
In a kind of optional embodiment, the processing module 502 is specifically used for: if first semantic similarity be greater than or
Equal to default semantic similarity threshold value, then by the first deep semantic vector of first semantic similarity in the default answer
Corresponding answer in library, as the answer for puing question to text;First semantic similarity is at least one described semantic phase
Like semantic similarity maximum in degree.
In a kind of optional embodiment, the processing module 502 is also used to: if first semantic similarity be less than it is described
Default semantic similarity threshold value, it is determined that whether the enquirement text matches with the triple in the knowledge mapping of prebuild;If
It is, then by matched triple in the knowledge mapping of the enquirement text and the prebuild, to put question to answering for text as described
Case.
In a kind of optional embodiment, the processing module 502 is also used to: if the enquirement text not with the prebuild
Knowledge mapping in triple matching, it is determined that it is described put question to text whether with it is default chat library in chat statement matching,
If so, by matched chat sentence in the enquirement text and the default chat library, as the answer for puing question to text;
Alternatively, if it is not, then using default default answer as the answer for puing question to text.
The embodiment of the present application provides a kind of computer equipment, including program or instruction, when described program or instruction are performed
When, the method to execute above-mentioned first aspect and each embodiment of first aspect.
The embodiment of the present application provides a kind of storage medium, including program or instruction, when described program or instruction be performed,
Method to execute above-mentioned first aspect and each embodiment of first aspect.
Finally, it should be noted that it should be understood by those skilled in the art that, embodiments herein can provide as method, be
System or computer program product.Therefore, the application can be used complete hardware embodiment, complete software embodiment or combine software
With the form of the embodiment of hardware aspect.Moreover, it wherein includes that computer can use journey that the application, which can be used in one or more,
The computer implemented in the computer-usable storage medium (including but not limited to magnetic disk storage, optical memory etc.) of sequence code
The form of program product.
The application be referring to according to the present processes, equipment (system) and computer program product flow chart and/or
Block diagram describes.It should be understood that each process that can be realized by computer program instructions in flowchart and/or the block diagram and/or
The combination of process and/or box in box and flowchart and/or the block diagram.It can provide these computer program instructions to arrive
General purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices processor to generate one
Machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for realizing flowing
The device for the function of being specified in journey figure one process or multiple processes and/or block diagrams one box or multiple boxes.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
Obviously, those skilled in the art can carry out various modification and variations without departing from the model of the application to the application
It encloses.In this way, if these modifications and variations of the application belong within the scope of the claim of this application and its equivalent technologies, then
The application is also intended to include these modifications and variations.
Claims (12)
1. a kind of answer acquisition methods characterized by comprising
It obtains and puts question to text;
By the enquirement disaggregated model of pre-training, determine the enquirement type for puing question to text whether in default answer type library
Answer type matching;
If the enquirement type for puing question to text is not matched with the answer type in the default answer type library, according to
The first deep semantic vector of each answer and the second deep semantic vector for puing question to text be at least in default answer library
One semantic similarity obtains the answer for puing question to text.
2. the method as described in claim 1, which is characterized in that further include:
If the enquirement type for puing question to text is matched with the answer type in the default answer type library, mentioned described in acquisition
Ask text matched answer type in the default answer type library at least one corresponding crucial answer element;
According to the enquirement text and at least one described crucial answer element, determine at least one described crucial answer element
At least one described crucial answer element is proposed the sub- answer for puing question to text by each key answer element for described
The combination for asking all sub- answers of text, as the answer for puing question to text.
3. method according to claim 1 or 2, which is characterized in that described according to each answer in the default answer library
First deep semantic vector and it is described put question to text the second deep semantic vector at least one semantic similarity, determine described in
Put question to the answer of text, comprising:
If first semantic similarity is greater than or equal to default semantic similarity threshold value, by first semantic similarity
First deep semantic vector corresponding answer in the default answer library, as the answer for puing question to text;Described first
Semantic similarity is maximum semantic similarity at least one described semantic similarity.
4. method as claimed in claim 3, which is characterized in that further include:
If first semantic similarity be less than the default semantic similarity threshold value, it is determined that the enquirement text whether in advance
Triple matching in the knowledge mapping of building;If so, by the knowledge mapping of the enquirement text and the prebuild
The triple matched, as the answer for puing question to text.
5. method as claimed in claim 4, which is characterized in that further include:
If the enquirement text is not matched with the triple in the knowledge mapping of the prebuild, it is determined that the enquirement text is
The no chat statement matching with default chat library, if so, by the enquirement text with it is matched in the default chat library
Chat sentence, as the answer for puing question to text;Alternatively,
If it is not, then using default default answer as the answer for puing question to text.
6. a kind of answer acquisition device characterized by comprising
Module is obtained, puts question to text for obtaining;
Processing module, for by the enquirement disaggregated model of pre-training, determine the enquirement type for puing question to text whether in advance
If the answer type in answer type library matches;If the enquirement type for puing question to text is not and in the default answer type library
Answer type matching, then according to the first deep semantic vector Yu the enquirement text of each answer in the default answer library
The second deep semantic vector at least one semantic similarity, obtain it is described put question to text answer.
7. device as claimed in claim 6, which is characterized in that the processing module is also used to:
If the enquirement type for puing question to text is matched with the answer type in the default answer type library, mentioned described in acquisition
Ask text matched answer type in the default answer type library at least one corresponding crucial answer element;
According to the enquirement text and at least one described crucial answer element, determine at least one described crucial answer element
At least one described crucial answer element is proposed the sub- answer for puing question to text by each key answer element for described
The combination for asking all sub- answers of text, as the answer for puing question to text.
8. device as claimed in claims 6 or 7, which is characterized in that the processing module is specifically used for:
If first semantic similarity is greater than or equal to default semantic similarity threshold value, by first semantic similarity
First deep semantic vector corresponding answer in the default answer library, as the answer for puing question to text;Described first
Semantic similarity is maximum semantic similarity at least one described semantic similarity.
9. device as claimed in claim 8, which is characterized in that the processing module is also used to:
If first semantic similarity be less than the default semantic similarity threshold value, it is determined that the enquirement text whether in advance
Triple matching in the knowledge mapping of building;If so, by the knowledge mapping of the enquirement text and the prebuild
The triple matched, as the answer for puing question to text.
10. device as claimed in claim 9, which is characterized in that the processing module is also used to:
If the enquirement text is not matched with the triple in the knowledge mapping of the prebuild, it is determined that the enquirement text is
The no chat statement matching with default chat library, if so, by the enquirement text with it is matched in the default chat library
Chat sentence, as the answer for puing question to text;Alternatively,
If it is not, then using default default answer as the answer for puing question to text.
11. a kind of computer equipment, which is characterized in that including program or instruction, when described program or instruction are performed, as weighed
Benefit require any one of 1 to 5 described in method be performed.
12. a kind of storage medium, which is characterized in that including program or instruction, when described program or instruction are performed, such as right
It is required that method described in any one of 1 to 5 is performed.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910570444.1A CN110263144A (en) | 2019-06-27 | 2019-06-27 | A kind of answer acquisition methods and device |
PCT/CN2019/117517 WO2020258654A1 (en) | 2019-06-27 | 2019-11-12 | Answer acquisition method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910570444.1A CN110263144A (en) | 2019-06-27 | 2019-06-27 | A kind of answer acquisition methods and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110263144A true CN110263144A (en) | 2019-09-20 |
Family
ID=67922619
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910570444.1A Pending CN110263144A (en) | 2019-06-27 | 2019-06-27 | A kind of answer acquisition methods and device |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN110263144A (en) |
WO (1) | WO2020258654A1 (en) |
Cited By (34)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110837549A (en) * | 2019-11-06 | 2020-02-25 | 腾讯科技(深圳)有限公司 | Information processing method, device and storage medium |
CN111444724A (en) * | 2020-03-23 | 2020-07-24 | 腾讯科技(深圳)有限公司 | Medical question-answer quality testing method and device, computer equipment and storage medium |
CN111931514A (en) * | 2020-07-28 | 2020-11-13 | 薛杨杨 | Information processing method based on deep learning and big data and block chain service platform |
CN111966840A (en) * | 2020-08-18 | 2020-11-20 | 北京猿力未来科技有限公司 | Man-machine interaction management method and management system for language teaching |
CN112084315A (en) * | 2020-09-07 | 2020-12-15 | 腾讯科技(深圳)有限公司 | Question-answer interaction method, device, storage medium and equipment |
WO2020258654A1 (en) * | 2019-06-27 | 2020-12-30 | 深圳前海微众银行股份有限公司 | Answer acquisition method and device |
CN112163083A (en) * | 2020-10-20 | 2021-01-01 | 珠海格力电器股份有限公司 | Intelligent question and answer method and device, electronic equipment and storage medium |
CN112287085A (en) * | 2020-11-06 | 2021-01-29 | 中国平安财产保险股份有限公司 | Semantic matching method, system, device and storage medium |
WO2021252827A1 (en) * | 2020-06-12 | 2021-12-16 | Apple Inc. | Providing personalized responses based on semantic context |
CN113807148A (en) * | 2020-06-16 | 2021-12-17 | 阿里巴巴集团控股有限公司 | Text recognition matching method and device and terminal equipment |
US11467802B2 (en) | 2017-05-11 | 2022-10-11 | Apple Inc. | Maintaining privacy of personal information |
US11487364B2 (en) | 2018-05-07 | 2022-11-01 | Apple Inc. | Raise to speak |
US11538469B2 (en) | 2017-05-12 | 2022-12-27 | Apple Inc. | Low-latency intelligent automated assistant |
US11557310B2 (en) | 2013-02-07 | 2023-01-17 | Apple Inc. | Voice trigger for a digital assistant |
US11610065B2 (en) | 2020-06-12 | 2023-03-21 | Apple Inc. | Providing personalized responses based on semantic context |
US11630525B2 (en) | 2018-06-01 | 2023-04-18 | Apple Inc. | Attention aware virtual assistant dismissal |
US11675491B2 (en) | 2019-05-06 | 2023-06-13 | Apple Inc. | User configurable task triggers |
US11696060B2 (en) | 2020-07-21 | 2023-07-04 | Apple Inc. | User identification using headphones |
US11699448B2 (en) | 2014-05-30 | 2023-07-11 | Apple Inc. | Intelligent assistant for home automation |
US11705130B2 (en) | 2019-05-06 | 2023-07-18 | Apple Inc. | Spoken notifications |
US11749275B2 (en) | 2016-06-11 | 2023-09-05 | Apple Inc. | Application integration with a digital assistant |
US11783815B2 (en) | 2019-03-18 | 2023-10-10 | Apple Inc. | Multimodality in digital assistant systems |
CN116886656A (en) * | 2023-09-06 | 2023-10-13 | 北京小糖科技有限责任公司 | Chat room-oriented dance knowledge pushing method and device |
US11790914B2 (en) | 2019-06-01 | 2023-10-17 | Apple Inc. | Methods and user interfaces for voice-based control of electronic devices |
US11809886B2 (en) | 2015-11-06 | 2023-11-07 | Apple Inc. | Intelligent automated assistant in a messaging environment |
US11838579B2 (en) | 2014-06-30 | 2023-12-05 | Apple Inc. | Intelligent automated assistant for TV user interactions |
US11838734B2 (en) | 2020-07-20 | 2023-12-05 | Apple Inc. | Multi-device audio adjustment coordination |
US11837237B2 (en) | 2017-05-12 | 2023-12-05 | Apple Inc. | User-specific acoustic models |
US11888791B2 (en) | 2019-05-21 | 2024-01-30 | Apple Inc. | Providing message response suggestions |
US11893992B2 (en) | 2018-09-28 | 2024-02-06 | Apple Inc. | Multi-modal inputs for voice commands |
US11900936B2 (en) | 2008-10-02 | 2024-02-13 | Apple Inc. | Electronic devices with voice command and contextual data processing capabilities |
US11900923B2 (en) | 2018-05-07 | 2024-02-13 | Apple Inc. | Intelligent automated assistant for delivering content from user experiences |
US11914848B2 (en) | 2020-05-11 | 2024-02-27 | Apple Inc. | Providing relevant data items based on context |
US11954405B2 (en) | 2022-11-07 | 2024-04-09 | Apple Inc. | Zero latency digital assistant |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112818222B (en) * | 2021-01-26 | 2024-02-23 | 吾征智能技术(北京)有限公司 | Personalized diet recommendation method and system based on knowledge graph |
CN113032540B (en) * | 2021-03-19 | 2023-06-23 | 北京百度网讯科技有限公司 | Man-machine interaction method, device, equipment and storage medium |
CN113688252B (en) * | 2021-08-09 | 2023-09-26 | 广西师范大学 | Safe cross-domain recommendation method based on multi-feature collaborative knowledge graph and blockchain |
CN113792153B (en) * | 2021-08-25 | 2023-12-12 | 北京度商软件技术有限公司 | Question and answer recommendation method and device |
TWI790069B (en) * | 2021-12-28 | 2023-01-11 | 高雄榮民總醫院 | Learning system for appraising literature |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107220380A (en) * | 2017-06-27 | 2017-09-29 | 北京百度网讯科技有限公司 | Question and answer based on artificial intelligence recommend method, device and computer equipment |
CN108197167A (en) * | 2017-12-18 | 2018-06-22 | 深圳前海微众银行股份有限公司 | Human-computer dialogue processing method, equipment and readable storage medium storing program for executing |
CN109740077A (en) * | 2018-12-29 | 2019-05-10 | 北京百度网讯科技有限公司 | Answer searching method, device and its relevant device based on semantic indexing |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107180029A (en) * | 2016-03-09 | 2017-09-19 | 阿里巴巴集团控股有限公司 | A kind of information processing method and device based on consultation service |
CN108073576A (en) * | 2016-11-09 | 2018-05-25 | 上海诺悦智能科技有限公司 | Intelligent search method, searcher and search engine system |
CN107133349B (en) * | 2017-05-24 | 2018-02-23 | 北京无忧创新科技有限公司 | One kind dialogue robot system |
CN110263144A (en) * | 2019-06-27 | 2019-09-20 | 深圳前海微众银行股份有限公司 | A kind of answer acquisition methods and device |
-
2019
- 2019-06-27 CN CN201910570444.1A patent/CN110263144A/en active Pending
- 2019-11-12 WO PCT/CN2019/117517 patent/WO2020258654A1/en active Application Filing
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107220380A (en) * | 2017-06-27 | 2017-09-29 | 北京百度网讯科技有限公司 | Question and answer based on artificial intelligence recommend method, device and computer equipment |
CN108197167A (en) * | 2017-12-18 | 2018-06-22 | 深圳前海微众银行股份有限公司 | Human-computer dialogue processing method, equipment and readable storage medium storing program for executing |
CN109740077A (en) * | 2018-12-29 | 2019-05-10 | 北京百度网讯科技有限公司 | Answer searching method, device and its relevant device based on semantic indexing |
Cited By (42)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11900936B2 (en) | 2008-10-02 | 2024-02-13 | Apple Inc. | Electronic devices with voice command and contextual data processing capabilities |
US11557310B2 (en) | 2013-02-07 | 2023-01-17 | Apple Inc. | Voice trigger for a digital assistant |
US11862186B2 (en) | 2013-02-07 | 2024-01-02 | Apple Inc. | Voice trigger for a digital assistant |
US11699448B2 (en) | 2014-05-30 | 2023-07-11 | Apple Inc. | Intelligent assistant for home automation |
US11838579B2 (en) | 2014-06-30 | 2023-12-05 | Apple Inc. | Intelligent automated assistant for TV user interactions |
US11809886B2 (en) | 2015-11-06 | 2023-11-07 | Apple Inc. | Intelligent automated assistant in a messaging environment |
US11749275B2 (en) | 2016-06-11 | 2023-09-05 | Apple Inc. | Application integration with a digital assistant |
US11467802B2 (en) | 2017-05-11 | 2022-10-11 | Apple Inc. | Maintaining privacy of personal information |
US11862151B2 (en) | 2017-05-12 | 2024-01-02 | Apple Inc. | Low-latency intelligent automated assistant |
US11837237B2 (en) | 2017-05-12 | 2023-12-05 | Apple Inc. | User-specific acoustic models |
US11538469B2 (en) | 2017-05-12 | 2022-12-27 | Apple Inc. | Low-latency intelligent automated assistant |
US11900923B2 (en) | 2018-05-07 | 2024-02-13 | Apple Inc. | Intelligent automated assistant for delivering content from user experiences |
US11487364B2 (en) | 2018-05-07 | 2022-11-01 | Apple Inc. | Raise to speak |
US11907436B2 (en) | 2018-05-07 | 2024-02-20 | Apple Inc. | Raise to speak |
US11630525B2 (en) | 2018-06-01 | 2023-04-18 | Apple Inc. | Attention aware virtual assistant dismissal |
US11893992B2 (en) | 2018-09-28 | 2024-02-06 | Apple Inc. | Multi-modal inputs for voice commands |
US11783815B2 (en) | 2019-03-18 | 2023-10-10 | Apple Inc. | Multimodality in digital assistant systems |
US11675491B2 (en) | 2019-05-06 | 2023-06-13 | Apple Inc. | User configurable task triggers |
US11705130B2 (en) | 2019-05-06 | 2023-07-18 | Apple Inc. | Spoken notifications |
US11888791B2 (en) | 2019-05-21 | 2024-01-30 | Apple Inc. | Providing message response suggestions |
US11790914B2 (en) | 2019-06-01 | 2023-10-17 | Apple Inc. | Methods and user interfaces for voice-based control of electronic devices |
WO2020258654A1 (en) * | 2019-06-27 | 2020-12-30 | 深圳前海微众银行股份有限公司 | Answer acquisition method and device |
CN110837549B (en) * | 2019-11-06 | 2023-08-11 | 腾讯科技(深圳)有限公司 | Information processing method, device and storage medium |
CN110837549A (en) * | 2019-11-06 | 2020-02-25 | 腾讯科技(深圳)有限公司 | Information processing method, device and storage medium |
CN111444724A (en) * | 2020-03-23 | 2020-07-24 | 腾讯科技(深圳)有限公司 | Medical question-answer quality testing method and device, computer equipment and storage medium |
US11914848B2 (en) | 2020-05-11 | 2024-02-27 | Apple Inc. | Providing relevant data items based on context |
WO2021252827A1 (en) * | 2020-06-12 | 2021-12-16 | Apple Inc. | Providing personalized responses based on semantic context |
US11610065B2 (en) | 2020-06-12 | 2023-03-21 | Apple Inc. | Providing personalized responses based on semantic context |
CN113807148A (en) * | 2020-06-16 | 2021-12-17 | 阿里巴巴集团控股有限公司 | Text recognition matching method and device and terminal equipment |
US11838734B2 (en) | 2020-07-20 | 2023-12-05 | Apple Inc. | Multi-device audio adjustment coordination |
US11696060B2 (en) | 2020-07-21 | 2023-07-04 | Apple Inc. | User identification using headphones |
US11750962B2 (en) | 2020-07-21 | 2023-09-05 | Apple Inc. | User identification using headphones |
CN111931514A (en) * | 2020-07-28 | 2020-11-13 | 薛杨杨 | Information processing method based on deep learning and big data and block chain service platform |
CN111966840A (en) * | 2020-08-18 | 2020-11-20 | 北京猿力未来科技有限公司 | Man-machine interaction management method and management system for language teaching |
CN112084315A (en) * | 2020-09-07 | 2020-12-15 | 腾讯科技(深圳)有限公司 | Question-answer interaction method, device, storage medium and equipment |
CN112163083A (en) * | 2020-10-20 | 2021-01-01 | 珠海格力电器股份有限公司 | Intelligent question and answer method and device, electronic equipment and storage medium |
CN112163083B (en) * | 2020-10-20 | 2023-11-03 | 珠海格力电器股份有限公司 | Intelligent question-answering method, device, electronic equipment and storage medium |
CN112287085B (en) * | 2020-11-06 | 2023-12-05 | 中国平安财产保险股份有限公司 | Semantic matching method, system, equipment and storage medium |
CN112287085A (en) * | 2020-11-06 | 2021-01-29 | 中国平安财产保险股份有限公司 | Semantic matching method, system, device and storage medium |
US11954405B2 (en) | 2022-11-07 | 2024-04-09 | Apple Inc. | Zero latency digital assistant |
CN116886656B (en) * | 2023-09-06 | 2023-12-08 | 北京小糖科技有限责任公司 | Chat room-oriented dance knowledge pushing method and device |
CN116886656A (en) * | 2023-09-06 | 2023-10-13 | 北京小糖科技有限责任公司 | Chat room-oriented dance knowledge pushing method and device |
Also Published As
Publication number | Publication date |
---|---|
WO2020258654A1 (en) | 2020-12-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110263144A (en) | A kind of answer acquisition methods and device | |
CN106202301B (en) | A kind of intelligent response system based on deep learning | |
US20210004837A1 (en) | System and method for pre-populating forms using agent assist within a cloud-based contact center | |
CN109977218B (en) | A kind of automatic answering system and method applied to session operational scenarios | |
CN105592237B (en) | A kind of method, apparatus and intelligent customer service robot of session switching | |
CN111192060B (en) | Full-channel self-service response implementation method based on power IT service | |
CN110266899A (en) | The recognition methods and customer service system that client is intended to | |
US20110228922A1 (en) | System and method for joining conference calls | |
CN109979457A (en) | A method of thousand people, thousand face applied to Intelligent dialogue robot | |
CN109753560B (en) | Information processing method and device of intelligent question-answering system | |
CN106357942A (en) | Intelligent response method and system based on context dialogue semantic recognition | |
CN110825858A (en) | Intelligent interaction robot system applied to customer service center | |
KR101901920B1 (en) | System and method for providing reverse scripting service between speaking and text for ai deep learning | |
CN110347863A (en) | Talk about art recommended method and device and storage medium | |
CN109859032A (en) | Funds on account collection method, apparatus, equipment and storage medium based on intelligent sound | |
CN111128241A (en) | Intelligent quality inspection method and system for voice call | |
CN109829729A (en) | A kind of intelligence outgoing call system and method | |
CN109509002A (en) | Automatically configure the system and method for realizing conversation process, storage medium | |
CN102665016A (en) | User-defined interactive voice question-answer implementation method based on cloud computing | |
CN109120513A (en) | Full media system and implementation method based on intelligent robot response | |
CN113779217A (en) | Intelligent voice outbound service method and system based on human-computer interaction | |
CN112150694A (en) | Intelligent voice electric charge payment urging system and method | |
CN111858884A (en) | Method and system for robot to learn real person deep dialogue content | |
CN116340482A (en) | Multi-skill customer service auxiliary product based on enterprise WeChat combined with NLP engine | |
JP2023076003A (en) | Program, information processing system, and information processing method |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190920 |