CN109299458A - Entity recognition method, device, equipment and storage medium - Google Patents
Entity recognition method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN109299458A CN109299458A CN201811061626.8A CN201811061626A CN109299458A CN 109299458 A CN109299458 A CN 109299458A CN 201811061626 A CN201811061626 A CN 201811061626A CN 109299458 A CN109299458 A CN 109299458A
- Authority
- CN
- China
- Prior art keywords
- entity recognition
- label
- lstm
- probability
- character
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention discloses a kind of entity recognition methods, including obtaining the entity recognition model based on LSTM after the completion of training, wherein the entity recognition model based on LSTM is using the training corpus training after mark;By the entity recognition model based on LSTM after the completion of training described in Entity recognition text input, the probability for adhering to mark label separately to each character in Entity recognition text is obtained;The probability is inputted into CRF model, obtains the label of each character;LSTM network is very big to the dependence of data, the size and quality of data volume also will affect model training result, combine LSTM model and CRF model, abstraction sequence feature is solved the problems, such as using LSTM model, the mark information of sentence level can be efficiently used using CRF model, the execution efficiency of conversational system is improved by LSTM+CRF model, while realizing Entity recognition and participle, improves Entity recognition accuracy rate and efficiency.
Description
Technical field
The present invention relates to information art field more particularly to a kind of entity recognition method, device, equipment and storage mediums.
Background technique
In artificial intelligence field, the trial for imitating mankind's talk ability can trace back to the early stage of artificial intelligence.?
In past several years, the application of messaging service class is grown rapidly, domestic wechat, external WhatsApp, Facebook
Messenger etc. has almost captured all chip times of user, and any active ues are hundreds of millions of, effectively becoming movement
" browser " entrance of Internet era, user only need that most information can be obtained using an application, and downloading moves
Flow bonus brought by dynamic mobile application just slowly disappears, this is that the advantage of conversational system embodies, and development cost is low, again
It can be attached on software platform.
In conversational system, generally require to identify the entity word that user inputs in sentence, i.e. Entity recognition, at the same need into
Row participle, so as to subsequent analysis.But the two tasks are separately to be handled to Entity recognition with participle in existing conversational system
's.
When inventor implements the present invention, find the Entity recognition application Shortcomings of the prior art: Entity recognition be in order to
Entity word therein is identified from sentence surface, such as: name, place name, organization name.It is similar to participle, if by the two
Business carries out in isolation, will lead to entity word identification and the accuracy rate decline of participle, such as sentence: the Nanjing Yangtze Bridge.If not yet
It identifies " Yangtze Bridge " this entity word, cutting will be likely to when participle are as follows: the Nanjing/mayor/Jiang great Qiao.Phase
Instead, if it is considered that " Yangtze Bridge " this entity word, by cutting are as follows: the Nanjing/Yangtze Bridge.
Summary of the invention
In view of this, the embodiment of the present invention provides a kind of entity recognition method, device, equipment and storage medium, pass through by
Entity recognition can improve the execution efficiency of conversational system, while improving the standard of Entity recognition and participle in conjunction with participle task
True rate.
In a first aspect, including the following steps: the embodiment of the invention provides a kind of entity recognition method
Obtain the entity recognition model based on LSTM after the completion of training, wherein the Entity recognition mould based on LSTM
Type is using the training corpus training after mark;
By the entity recognition model based on LSTM after the completion of the training described in the Entity recognition text input, obtain it is described to
Each character adheres to the probability of mark label separately in Entity recognition text;
The probability is inputted into CRF model, obtains the label of each character.
In the first possible implementation of first aspect, the entity knowledge based on LSTM obtained after the completion of training
Other model, wherein the entity recognition model based on LSTM is using the training corpus training after mark, comprising:
Training corpus after obtaining mark;
By in the training corpus after the mark word and character be converted into vector;
The vector of institute's predicate and character is inputted in the entity recognition model based on LSTM, is instructed using back propagation
Practice the parameter in the entity recognition model based on LSTM, with the Entity recognition mould based on LSTM after the completion of being trained
Type.
With reference to first aspect and the first possible mode of first aspect, second in first aspect can the side of being able to achieve
In formula, the training corpus obtained after marking includes:
The training corpus is trained using IB mode, the training corpus after being marked.
It is described to be completed to training described in Entity recognition text input in the third possible implementation of first aspect
The entity recognition model based on LSTM afterwards obtains the probability for adhering to mark label separately to each character in Entity recognition text
Include:
By the Entity recognition based on LSTM after the completion of the character of Entity recognition text sequentially inputs the training
Model obtains the probability for adhering to mark label separately to each character in Entity recognition text.
It is described that the probability is inputted into CRF model in the 4th kind of possible implementation of first aspect, obtain each word
The label of symbol includes:
By the probability input prediction formula, the maximum value of the predictor formula is solved, obtains optimal output label sequence
Column, wherein the predictor formula isWherein, y is described to Entity recognition text
Sequence label to be predicted, y=(y1, y2..., yn), X=pi,yiMark is adhered to separately to each character in Entity recognition text to be described
Infuse the probability of label, pi,yiRefer to that i-th of word is marked as yiThe probability of a label;Ayi,yi+1Refer to yiA label is transferred to
yi+1The probability of a label;
It is labeled according to the optimal output label sequence, and then obtains the label of each character.
Second aspect, the embodiment of the invention also provides a kind of entity recognition devices, comprising:
Entity recognition model obtains module, for obtaining the entity recognition model based on LSTM after the completion of training, wherein
The entity recognition model based on LSTM is using the training corpus training after mark;
Probability obtains module, for that will know after the entity based on LSTM after the completion of training described in Entity recognition text input
Other model obtains the probability for adhering to mark label separately to each character in Entity recognition text;
Label obtains module, for the probability to be inputted CRF model, obtains the label of each character.
In the first possible implementation of second aspect, the entity recognition model obtains module and includes:
Training corpus after obtaining mark;
By in the training corpus after the mark word and character be converted into vector;
The vector of institute's predicate and character is inputted in the entity recognition model based on LSTM, is instructed using back propagation
Practice the parameter in the entity recognition model based on LSTM, with the Entity recognition mould based on LSTM after the completion of being trained
Type.
In second of possible implementation of second aspect, the label obtains module and includes:
By the probability input prediction formula, the maximum value of the predictor formula is solved, obtains optimal output label sequence
Column, wherein the predictor formula isWherein, y is described to Entity recognition text
Sequence label to be predicted, y=(y1, y2..., yn), X=pi,yiMark is adhered to separately to each character in Entity recognition text to be described
Infuse the probability of label, pi,yiRefer to that i-th of word is marked as yiThe probability of a label;Ayi,yi+1Refer to yiA label is transferred to
yi+1The probability of a label;
It is labeled according to the optimal output label sequence, and then obtains the label of each character.
The third aspect, the embodiment of the invention also provides a kind of Entity recognition equipment, which is characterized in that including processor,
Memory and storage in the memory and are configured as the computer program executed by the processor, the processor
Entity recognition method as described above is realized when executing the computer program.
Fourth aspect, the embodiment of the invention also provides a kind of computer readable storage mediums, which is characterized in that the meter
Calculation machine readable storage medium storing program for executing includes the computer program of storage, wherein controls the calculating in computer program operation
Equipment executes entity recognition method as described above where machine readable storage medium storing program for executing.
Implement the embodiment of the present invention to have the following beneficial effects:
Obtain the entity recognition model based on LSTM after the completion of training, wherein the Entity recognition mould based on LSTM
Type is using the training corpus training after mark;By the reality based on LSTM after the completion of being trained described in Entity recognition text input
Body identification model obtains the probability for adhering to mark label separately to each character in Entity recognition text;The probability is inputted
CRF model obtains the label of each character;By by the entity recognition model based on LSTM and the CRF models coupling,
And then Entity recognition and participle can be carried out simultaneously, the prediction for reducing model is time-consuming, and the entity obtained using Entity recognition
The information of word is segmented, and the accuracy rate and efficiency of participle can be improved.
Detailed description of the invention
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, of the invention other
Feature, objects and advantages will become more apparent upon:
Fig. 1 is the schematic diagram for the Entity recognition equipment that the embodiment of the present invention one provides;
Fig. 2 is a kind of flow diagram of entity recognition method provided by Embodiment 2 of the present invention;
Fig. 3 is the result schematic diagram of LSTM Entity recognition provided by Embodiment 2 of the present invention;
Fig. 4 is the result schematic diagram of LSTM+CRF Entity recognition provided by Embodiment 2 of the present invention;
Fig. 5 is that Entity recognition provided by Embodiment 2 of the present invention shows result schematic diagram;
Fig. 6 is the structural schematic diagram for the entity recognition device that third embodiment of the invention provides.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Those skilled in the art will appreciate that the present invention can be implemented as equipment, method or computer program product.
Therefore, the present disclosure may be embodied in the following forms, it may be assumed that can be complete hardware, be also possible to complete software (including
Firmware, resident software, microcode etc.), it can also be the form that hardware and software combines, referred to generally herein as " circuit ", " mould
Block " or " system ".In addition, in some embodiments, the present invention is also implemented as in one or more computer-readable mediums
In computer program product form, include computer-readable program code in the computer-readable medium.
It can be using any combination of one or more computer-readable media.Computer-readable medium can be calculating
Machine readable signal medium or computer readable storage medium.Computer readable storage medium for example can be --- but it is unlimited
In system, device or the device of --- electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, or any above combination.It calculates
The more specific example (non exhaustive list) of machine readable storage medium storing program for executing includes: electrical connection with one or more conducting wires, just
Taking formula computer disk, hard disk, random access memory (RAM), read-only memory (ROM), erasable type may be programmed read-only storage
Device (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device,
Or above-mentioned any appropriate combination.In this document, computer readable storage medium can be it is any include or storage journey
The tangible medium of sequence, the program can be commanded execution system, device or device use or in connection.
Computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including --- but
It is not limited to --- electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be
Any computer-readable medium other than computer readable storage medium, which can send, propagate or
Transmission is for by the use of instruction execution system, device or device or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including --- but it is unlimited
In --- wireless, electric wire, optical cable, RF etc. or above-mentioned any appropriate combination.
The computer for executing operation of the present invention can be write with one or more programming languages or combinations thereof
Program code, described program design language include object oriented program language-such as Java, Smalltalk, C++,
Further include conventional procedural programming language-such as " C " language or similar programming language.Program code can be with
It fully executes, partly execute on the user computer on the user computer, being executed as an independent software package, portion
Divide and partially executes or executed on a remote computer or server completely on the remote computer on the user computer.?
Be related in the situation of remote computer, remote computer can pass through the network of any kind --- including local area network (LAN) or
Wide area network (WAN)-be connected to subscriber computer, or, it may be connected to outer computer (such as mentioned using Internet service
It is connected for quotient by internet).
Below with reference to the method for the embodiment of the present invention, the flow chart of equipment (system) and computer program product and/or
The block diagram description present invention.It should be appreciated that each box in each box and flowchart and or block diagram of flowchart and or block diagram
Combination, can be realized by computer program instructions.These computer program instructions can be supplied to general purpose computer, dedicated
The processor of computer or other programmable data processing units, to produce a kind of virtual machine, these computer programs refer to
It enables and being executed by computer or other programmable data processing units, produced in the box in implementation flow chart and/or block diagram
The device of defined function/operation.
These computer program instructions can also be stored in can make computer or other programmable data processing units
In computer-readable medium operate in a specific manner, in this way, the instruction of storage in computer-readable medium just produces one
The manufacture of function/operation command device specified in a box including in implementation flow chart and/or block diagram
(manufacture)。
Computer program instructions can also be loaded into computer, other programmable data processing units or other equipment
On, so that series of operation steps are executed in computer, other programmable data processing units or other equipment, in terms of generating
The process that calculation machine is realized, so that the instruction executed on a computer or other programmable device is capable of providing implementation flow chart
And/or function/operation process specified in the box in block diagram.
Embodiment one
Referring to Figure 1, Fig. 1 is the schematic diagram for the Entity recognition equipment that the embodiment of the present invention one provides, for executing this hair
The entity recognition method that bright embodiment provides, as shown in Figure 1, the Entity recognition equipment includes: at least one processor 11, such as
CPU, at least one network interface 14 or other users interface 13, memory 15, at least one communication bus 12, communication bus
12 for realizing the connection communication between these components.Wherein, user interface 13 optionally may include USB interface and other
Standard interface, wireline interface.Network interface 14 optionally may include Wi-Fi interface and other wireless interfaces.Memory 15
It may include high speed RAM memory, it is also possible to further include non-labile memory (non-volatilememory), such as extremely
A few magnetic disk storage.Memory 15 optionally may include at least one storage dress for being located remotely from aforementioned processor 11
It sets.
In some embodiments, memory 15 stores following element, executable modules or data structures, or
Their subset or their superset:
Operating system 151 includes various system programs, for realizing various basic businesses and hardware based of processing
Business;
Program 152.
Specifically, processor 11 executes reality described in following embodiments for calling the program 152 stored in memory 15
Body recognition methods.
Alleged processor can be central processing unit (Central Processing Unit, CPU), can also be it
His general processor, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng the processor is the control centre of the entity recognition method, utilizes the entire entity of various interfaces and connection
The various pieces of recognition methods.
The memory can be used for storing the computer program and/or module, and the processor is by operation or executes
Computer program in the memory and/or module are stored, and calls the data being stored in memory, realizes entity
The various functions of the electronic device of identification.The memory can mainly include storing program area and storage data area, wherein storage
It program area can application program needed for storage program area, at least one function (such as sound-playing function, text conversion function
Deng) etc.;Storage data area, which can be stored, uses created data (such as audio data, text message data etc.) according to mobile phone
Deng.It can also include nonvolatile memory in addition, memory may include high-speed random access memory, such as hard disk, interior
It deposits, plug-in type hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card,
Flash card (Flash Card), at least one disk memory, flush memory device or other volatile solid-state parts.
Wherein, if the module that the Entity recognition integrates is realized in the form of SFU software functional unit and as independent production
Product when selling or using, can store in a computer readable storage medium.Based on this understanding, the present invention realizes
All or part of the process in above-described embodiment method can also instruct relevant hardware to complete by computer program,
The computer program can be stored in a computer readable storage medium, which is being executed by processor
When, it can be achieved that the step of above-mentioned each embodiment of the method.Wherein, the computer program includes computer program code, described
Computer program code can be source code form, object identification code form, executable file or certain intermediate forms etc..The meter
Calculation machine readable medium may include: can carry the computer program code any entity or device, recording medium, USB flash disk,
Mobile hard disk, magnetic disk, CD, computer storage, read-only memory (ROM, Read-Only Memory), random access memory
Device (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium etc..It needs to illustrate
It is that the content that the computer-readable medium includes can be fitted according to the requirement made laws in jurisdiction with patent practice
When increase and decrease, such as in certain jurisdictions, according to legislation and patent practice, computer-readable medium does not include electric carrier wave letter
Number and telecommunication signal.
Below with reference to accompanying drawings come describe the embodiment of the present invention Entity recognition method.
Embodiment two
Fig. 2 is a kind of flow diagram of entity recognition method provided by Embodiment 2 of the present invention.
A kind of entity recognition method, comprising the following steps:
The entity recognition model based on LSTM after the completion of S11, acquisition training, wherein the entity based on LSTM is known
Other model is using the training corpus training after mark;
S12, by the entity recognition model based on LSTM after the completion of the training described in the Entity recognition text input, obtain institute
State the probability for adhering to mark label separately to character each in Entity recognition text;
S13, the probability is inputted into CRF model, obtains the label of each character.
In embodiments of the present invention, in order to improve the precision and efficiency of Entity recognition, LSTM model and CRF model are carried out
In conjunction with can be realized simultaneously Entity recognition and sentence Entity recognition.
Preferably, the entity recognition model based on LSTM obtained after the completion of training, wherein described based on LSTM's
Entity recognition model is using the training corpus training after mark, comprising:
Training corpus after obtaining mark;
By in the training corpus after the mark word and character be converted into vector;
The vector of institute's predicate and character is inputted in the entity recognition model based on LSTM, is instructed using back propagation
Practice the parameter in the entity recognition model based on LSTM, with the Entity recognition mould based on LSTM after the completion of being trained
Type.
Further, the training corpus obtained after marking includes:
The training corpus is trained using IB mode, the training corpus after being marked.
In embodiments of the present invention, firstly, obtaining the training corpus after mark, mark corpus is the process manually marked,
Corpus training corpus is labeled in the way of IB (Inside, Begin) (can also be labeled using other way,
Such as replaced with 0,1,2), wherein Begin: belonging to the first character label of entity word, if corresponding character is entity word,
Then plus the suffix now answered.Inside: not being first character, if it is entity word part is belonged to, then plus the suffix now answered.
Be embroidered with afterwards: the suffix of name is P, and institution term suffix is C, and place name suffix is L, if an Entity recognition unit is one
Entity starts, then is labeled as (tagB- ...);If an Entity recognition unit is vocabulary among an entity, it is labeled as
(tag I-…).By taking name (PER) most common in entity, place name (LOC) and mechanism name (ORG) as an example, for the training
Each sentence in corpus, is marked each character, such as: Ma Huateng is the CEO of Tencent.Can mark: the label of horse be
B-P;The label of change is;The label risen is;Be is labeled as B;The label risen is;The label of news is;?
Labeled as B;The label of C is;The label of E is;The label of O is.
In embodiments of the present invention, the word in the training corpus by after the mark and character are converted into vector, because
The type for being only capable of logarithm type for computer is calculated, and the word x inputted is character type, and computer cannot be calculated directly, because
This needs to carry out vector conversion, and the vector of conversion is properly termed as term vector, and word is also made to be embedded in vector.First owned according to statistics
The vocabulary for the word predicted and trained, it is assumed that vocabulary size is k, is that each word assigns unique id in vocabulary, id's
Value range is 0 to k-1, and random initializtion matrix size is [k, dim], wherein dim is preset threshold, according to each character
Corresponding id is searched, and then obtains corresponding term vector.In building term vector (WordEmbedding), at mathematical model
The first step for managing corpus of text is exactly to convert text to mathematical notation, and there are two types of method, first method can pass through one-
Hot matrix indicates that a word, one-hot matrix refer to that every a line has and only one element is 1, and other elements are all 0
Matrix.For each word in dictionary, we distribute a number, when encoding to certain words, by each list of the inside
Word be converted into this word inside dictionary number the one-hot matrix that corresponding position is 1 can.For example we will express
" I love china ", can be used matrix and is expressed asAlso WordEmbedding can be used
Matrix, the vector that WordEmbedding matrix distributes a regular length to each word indicate that this length can voluntarily be set
It is fixed, such as 300, it can actually be far smaller than dictionary length (such as 10000).And the angle value between two word vectors can
Using a measurement as relationship between them, matrix can be used and be expressed as
In embodiments of the present invention, the vector of institute's predicate and character is inputted into the entity recognition model based on LSTM
In, using the parameter in the back propagation training entity recognition model based on LSTM, with the base after the completion of being trained
In the entity recognition model of LSTM, wherein the entity recognition model calculation formula based on LSTM is as follows:
Wherein, σ is taken to each element
Sigmoid operation, ⊙ represent dot product, xtFor input, htIt is all random initial to W, h, c and b all in the formula for output
Change, corresponding vector, which is input to the formula, can be obtained by corresponding probability, for example, " I love china " is input to
In the first layer LSTM neuron elements of the entity recognition model based on LSTM, while i-th of LSTM of first layer LSTM is mono-
The output while the input as first layer LSTM i+1 LSTM unit of member, then by each neural unit output of LSTM
Each character belongs to the probability of each label.
In the present embodiment, it after having obtained each character and having belonged to the probability of each label, is instructed using back propagation
Practice the parameter in the entity recognition model based on LSTM, to obtain the trained entity recognition model based on LSTM.Institute
Stating backpropagation is that the parameter of chain type derivation policy update LSTM is used on the basis of LSTM exports result, and chain type derivation is
By " compound function to be pooled by multiple functions, derivative is equal to the derivative that the inside function substitutes into the value of outside function, multiplied by inner
The derivative of side function ", illustratively, f (x)=x2, g (x)=2x+1, then { f [g (x)] } '=2 [g (x)] × g'(x) and=2 [2x
+ 1] × 2=8x+4.Parameter in above-mentioned entity recognition model calculation formula based on LSTM is updated with this.
Preferably, after obtaining the trained entity recognition model based on LSTM, it is described will be defeated to Entity recognition text
Enter the entity recognition model based on LSTM after the completion of the training, acquisition is described to be adhered to separately to each character in Entity recognition text
Mark label probability include:
By the Entity recognition based on LSTM after the completion of the character of Entity recognition text sequentially inputs the training
Model obtains the probability for adhering to mark label separately to each character in Entity recognition text.
In the present embodiment, the entity recognition model calculation formula based on above-mentioned based on LSTM:
The Entity recognition mould based on LSTM
Type reads in a character to Entity recognition text in each step, by the entity recognition model based on LSTM
The probability that the character belongs to IOB label has can be obtained in internal calculating.Referring to Fig. 3, " Ma Huateng is Tencent to sentence
CEO " can obtain the corresponding probability for adhering to each label separately of the character after each step inputs character.For example, character
" horse ", the probability for belonging to label B is 0.5, and the probability for belonging to label B-P is 0.9, and the probability for belonging to label B-L is 0.8, is belonged to
It is 0.2 in the probability of label B-C, the probability for belonging to label I is 0.4, and the probability for belonging to label I-P is 0.5, belongs to label I-L
Probability be 0.1, belong to label I-C probability be 0.5.
Preferably, described that the probability is inputted CRF model, the label for obtaining each character includes:
By the probability input prediction formula, the maximum value of the predictor formula is solved, obtains optimal output label sequence
Column, wherein the predictor formula isWherein, y is described to Entity recognition text
Sequence label to be predicted, y=(y1, y2..., yn), X=pi,yiMark is adhered to separately to each character in Entity recognition text to be described
The probability for infusing label, refers to that i-th of word is marked as yiThe probability of a label;Ayi,yi+1Refer to yiA label is transferred to yi+1It is a
The probability of label;
It is labeled according to the optimal output label sequence, and then obtains the label of each character.
In the present embodiment, referring to fig. 4, the structural schematic diagram of LSTM+CRF, for each input X=(x1, x2 ...,
Xn), we obtain a prediction label sequences y=(y1, y2..., yn), the score for defining this prediction is
Wherein pi,yiThe probability for being yi for i-th of position softmax output,
Ayi,yi+1For the transition probability from yi to yi+1, when tag (B-person B-location ...) number is n, transfer is general
Rate matrix is (n+2) * (n+2), because adding additional a starting position and end position.This scoring function S is with regard to fine
Ground compensates for the deficiency of traditional BiLSTM, because we, when a forecasting sequence score is very high, are not that each position is all
Softmax, which exports the corresponding label of most probable value, will also meet output it is also contemplated that front transition probability is added maximum
Regular (cannot be again with B behind B), for example assume that the most possible sequence of BiLSTM output is BBIBIOOO, then because we
Transition probability matrix in B- > B probability very little even be negative, then this sequence will not obtain highest point according to s score
Number, i.e., be not just the sequence that we want.By taking " CEO that Ma Huateng is Tencent " as an example, after CRF model, obtained maximum
Scoring sequence are as follows:
S (' Ma Huateng is the CEO ' of Tencent, (B-P, I-P, I-P, B, B-C, I-C, B, B, I, I))=A (B-P, I-P)+A
(I-P,I-P)+A(I-P,B)+A(B,B-C)+A(B-C,I-C)+A(I-C,B)+A(B,B)+A(B,I)+A(I,I)+0.9+0.9+
0.9+0.8+0.8+0.9+0.8+0.9+0.9+0.9.Wherein, Ayi,yi+1For from yiTo yi+1Transition probability numerical value by mark number
It obtains according to statistics.It can thus be appreciated that word segmentation result is: Ma Huateng/be/Tencent //CEO.
It should be noted that the CRF model introduced, is modeled to output label binary group, then using dynamic
Planning is calculated, final to be labeled according to obtained optimal path.
In the present embodiment, the label of each character can be shown on the text to Entity recognition, such as joined
See Fig. 5, to the preset position of each character of Entity recognition text, such as above or below character or subscript or subscript etc.,
The label of respective symbols is shown.
Implement the present embodiment to have the following beneficial effects:
Obtain the entity recognition model based on LSTM after the completion of training, wherein the Entity recognition mould based on LSTM
Type is using the training corpus training after mark;By the reality based on LSTM after the completion of being trained described in Entity recognition text input
Body identification model obtains the probability for adhering to mark label separately to each character in Entity recognition text;The probability is inputted
CRF model obtains the label of each character;LSTM network is very big to the dependence of data, and the size and quality of data volume also can shadows
Model training is rung as a result, combining LSTM model and CRF model, is solved the problems, such as abstraction sequence feature using LSTM model, is made
The mark information that sentence level can be efficiently used with CRF model improves the execution of conversational system by LSTM+CRF model
Efficiency, while Entity recognition and participle are realized, improve Entity recognition accuracy rate and efficiency.
Embodiment three
Referring to the structural schematic diagram of Fig. 6, third embodiment of the invention entity recognition device provided;
A kind of entity recognition device, comprising:
Entity recognition model obtains module 31, for obtaining the entity recognition model based on LSTM after the completion of training,
In, the entity recognition model based on LSTM is using the training corpus training after mark;
Probability obtains module 32, for by the entity based on LSTM after the completion of the training described in the Entity recognition text input
Identification model obtains the probability for adhering to mark label separately to each character in Entity recognition text;
Label obtains module 33, for the probability to be inputted CRF model, obtains the label of each character.
Preferably, the entity recognition model acquisition module 31 includes:
Training corpus acquiring unit, for obtaining the training corpus after marking;
Vector acquiring unit, for by the training corpus after the mark word and character be converted into vector;
Parameter training unit, for the vector of institute's predicate and character to be inputted the entity recognition model based on LSTM
In, using the parameter in the back propagation training entity recognition model based on LSTM, with the base after the completion of being trained
In the entity recognition model of LSTM.
Preferably, the training corpus acquiring unit includes:
The training corpus is trained using IOB mode, the training corpus after being marked.
Preferably, the probability acquisition module 32 includes:
By the Entity recognition based on LSTM after the completion of the character of Entity recognition text sequentially inputs the training
Model obtains the probability for adhering to mark label separately to each character in Entity recognition text.
Preferably, the label acquisition module 33 includes:
By the probability input prediction formula, the maximum value of the predictor formula is solved, obtains optimal output label sequence
Column, wherein the predictor formula isWherein, y is described to Entity recognition text
Sequence label to be predicted, y=(y1, y2..., yn), X=pi,yiMark is adhered to separately to each character in Entity recognition text to be described
The probability for infusing label, refers to that i-th of word is marked as yiThe probability of a label;Ayi,yi+1Refer to yiA label is transferred to yi+1It is a
The probability of label;
It is labeled according to the optimal output label sequence, and then obtains the label of each character.
Implement the present embodiment to have the following beneficial effects:
Obtain the entity recognition model based on LSTM after the completion of training, wherein the Entity recognition mould based on LSTM
Type is using the training corpus training after mark;By the reality based on LSTM after the completion of being trained described in Entity recognition text input
Body identification model obtains the probability for adhering to mark label separately to each character in Entity recognition text;The probability is inputted
CRF model obtains the label of each character;LSTM network is very big to the dependence of data, and the size and quality of data volume also can shadows
Model training is rung as a result, combining LSTM model and CRF model, is solved the problems, such as abstraction sequence feature using LSTM model, is made
The mark information that sentence level can be efficiently used with CRF model improves the execution of conversational system by LSTM+CRF model
Efficiency, while Entity recognition and participle are realized, improve Entity recognition accuracy rate and efficiency.
It should be noted that the apparatus embodiments described above are merely exemplary, wherein described be used as separation unit
The unit of explanation may or may not be physically separated, and component shown as a unit can be or can also be with
It is not physical unit, it can it is in one place, or may be distributed over multiple network units.It can be according to actual
It needs that some or all of the modules therein is selected to achieve the purpose of the solution of this embodiment.In addition, device provided by the invention
In embodiment attached drawing, the connection relationship between module indicate between them have communication connection, specifically can be implemented as one or
A plurality of communication bus or signal wire.Those of ordinary skill in the art are without creative efforts, it can understand
And implement.
The above is a preferred embodiment of the present invention, it is noted that for those skilled in the art
For, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also considered as
Protection scope of the present invention.
It should be noted that in the above-described embodiments, all emphasizing particularly on different fields to the description of each embodiment, in some embodiment
In the part that is not described in, reference can be made to the related descriptions of other embodiments.Secondly, those skilled in the art should also know
It knows, the embodiments described in the specification are all preferred embodiments, and related movement and simulation must be that the present invention must
Must.
Claims (10)
1. a kind of entity recognition method characterized by comprising
Obtain the entity recognition model based on LSTM after the completion of training, wherein the entity recognition model based on LSTM is
Use the training corpus training after mark;
By the entity recognition model based on LSTM after the completion of being trained described in Entity recognition text input, obtain described to entity
Each character adheres to the probability of mark label separately in identification text;
The probability is inputted into CRF model, obtains the label of each character.
2. entity recognition method according to claim 1, which is characterized in that it is described obtain training after the completion of based on LSTM
Entity recognition model, wherein the entity recognition model based on LSTM be using mark after training corpus training, packet
It includes:
Training corpus after obtaining mark;
By in the training corpus after the mark word and character be converted into vector;
The vector of institute's predicate and character is inputted in the entity recognition model based on LSTM, back propagation training institute is used
The parameter in the entity recognition model based on LSTM is stated, with the entity recognition model based on LSTM after the completion of being trained.
3. entity recognition method according to claim 2, which is characterized in that the training corpus packet obtained after mark
It includes:
The training corpus is trained using IB mode, the training corpus after being marked.
4. entity recognition method according to claim 1, which is characterized in that it is described will be to described in Entity recognition text input
The entity recognition model based on LSTM after the completion of training obtains described adhere to separately to each character in Entity recognition text and marks mark
The probability of label includes:
By the entity recognition model based on LSTM after the completion of the character of Entity recognition text sequentially inputs the training,
Obtain the probability for adhering to mark label separately to each character in Entity recognition text.
5. entity recognition method according to claim 1, which is characterized in that it is described that the probability is inputted into CRF model, it obtains
Label to each character includes:
By the probability input prediction formula, the maximum value of the predictor formula is solved, obtains optimal output label sequence,
In, the predictor formula isWherein, y be it is described to Entity recognition text to pre-
The sequence label of survey, y=(y1, y2..., yn), X=pi,yiMark mark is adhered to separately to each character in Entity recognition text to be described
The probability of label, pi,yiRefer to that i-th of word is marked as yiThe probability of a label;Ayi,yi+1Refer to yiA label is transferred to yi+1It is a
The probability of label;
It is labeled according to the optimal output label sequence, and then obtains the label of each character.
6. a kind of entity recognition device characterized by comprising
Entity recognition model obtains module, for obtaining the entity recognition model based on LSTM after the completion of training, wherein described
Entity recognition model based on LSTM is using the training corpus training after mark;
Probability obtains module, for by the Entity recognition mould based on LSTM after the completion of the training described in the Entity recognition text input
Type obtains the probability for adhering to mark label separately to each character in Entity recognition text;
Label obtains module, for the probability to be inputted CRF model, obtains the label of each character.
7. entity recognition device according to claim 6, which is characterized in that the entity recognition model obtains module packet
It includes:
Training corpus after obtaining mark;
By in the training corpus after the mark word and character be converted into vector;
The vector of institute's predicate and character is inputted in the entity recognition model based on LSTM, back propagation training institute is used
The parameter in the entity recognition model based on LSTM is stated, with the entity recognition model based on LSTM after the completion of being trained.
8. entity recognition device according to claim 6, which is characterized in that the label obtains module and includes:
By the probability input prediction formula, the maximum value of the predictor formula is solved, obtains optimal output label sequence,
In, the predictor formula isWherein, y be it is described to Entity recognition text to pre-
The sequence label of survey, y=(y1, y2..., yn), X=pi,yiMark mark is adhered to separately to each character in Entity recognition text to be described
The probability of label, pi,yiRefer to that i-th of word is marked as yiThe probability of a label;Ayi,yi+1Refer to yiA label is transferred to yi+1It is a
The probability of label;
It is labeled according to the optimal output label sequence, and then obtains the label of each character.
9. a kind of Entity recognition equipment, which is characterized in that including processor, memory and storage in the memory and by
It is configured to the computer program executed by the processor, is realized when the processor executes the computer program as right is wanted
Seek entity recognition method described in 1 to 5 any one.
10. a kind of computer readable storage medium, which is characterized in that the computer readable storage medium includes the calculating of storage
Machine program, wherein equipment where controlling the computer readable storage medium in computer program operation is executed as weighed
Benefit require any one of 1 to 5 described in entity recognition method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811061626.8A CN109299458B (en) | 2018-09-12 | 2018-09-12 | Entity identification method, device, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811061626.8A CN109299458B (en) | 2018-09-12 | 2018-09-12 | Entity identification method, device, equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109299458A true CN109299458A (en) | 2019-02-01 |
CN109299458B CN109299458B (en) | 2023-03-28 |
Family
ID=65166558
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811061626.8A Active CN109299458B (en) | 2018-09-12 | 2018-09-12 | Entity identification method, device, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109299458B (en) |
Cited By (25)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109902303A (en) * | 2019-03-01 | 2019-06-18 | 腾讯科技(深圳)有限公司 | A kind of entity recognition method and relevant device |
CN110287283A (en) * | 2019-05-22 | 2019-09-27 | 中国平安财产保险股份有限公司 | Intent model training method, intension recognizing method, device, equipment and medium |
CN110516251A (en) * | 2019-08-29 | 2019-11-29 | 秒针信息技术有限公司 | A kind of construction method, construction device, equipment and the medium of electric business entity recognition model |
CN110555102A (en) * | 2019-09-16 | 2019-12-10 | 青岛聚看云科技有限公司 | media title recognition method, device and storage medium |
CN110598210A (en) * | 2019-08-29 | 2019-12-20 | 深圳市优必选科技股份有限公司 | Entity recognition model training method, entity recognition device, entity recognition equipment and medium |
CN110705211A (en) * | 2019-09-06 | 2020-01-17 | 中国平安财产保险股份有限公司 | Text key content marking method and device, computer equipment and storage medium |
CN110738055A (en) * | 2019-10-23 | 2020-01-31 | 北京字节跳动网络技术有限公司 | Text entity identification method, text entity identification equipment and storage medium |
CN110738054A (en) * | 2019-10-14 | 2020-01-31 | 携程计算机技术(上海)有限公司 | Method, system, electronic device and storage medium for identifying hotel information in mail |
CN110738319A (en) * | 2019-11-11 | 2020-01-31 | 四川隧唐科技股份有限公司 | LSTM model unit training method and device for recognizing bid-winning units based on CRF |
CN110738182A (en) * | 2019-10-21 | 2020-01-31 | 四川隧唐科技股份有限公司 | LSTM model unit training method and device for high-precision identification of bid amount |
CN110826330A (en) * | 2019-10-12 | 2020-02-21 | 上海数禾信息科技有限公司 | Name recognition method and device, computer equipment and readable storage medium |
CN110825827A (en) * | 2019-11-13 | 2020-02-21 | 北京明略软件系统有限公司 | Entity relationship recognition model training method and device and entity relationship recognition method and device |
CN111079405A (en) * | 2019-11-29 | 2020-04-28 | 微民保险代理有限公司 | Text information identification method and device, storage medium and computer equipment |
CN111209396A (en) * | 2019-12-27 | 2020-05-29 | 深圳市优必选科技股份有限公司 | Entity recognition model training method, entity recognition method and related device |
CN111476022A (en) * | 2020-05-15 | 2020-07-31 | 湖南工商大学 | Method, system and medium for recognizing STM entity by embedding and mixing L characters of entity characteristics |
CN111681670A (en) * | 2019-02-25 | 2020-09-18 | 北京嘀嘀无限科技发展有限公司 | Information identification method and device, electronic equipment and storage medium |
CN111914561A (en) * | 2020-07-31 | 2020-11-10 | 中国建设银行股份有限公司 | Entity recognition model training method, entity recognition device and terminal equipment |
CN112182157A (en) * | 2020-09-29 | 2021-01-05 | 中国平安人寿保险股份有限公司 | Training method of online sequence labeling model, online labeling method and related equipment |
CN112214987A (en) * | 2020-09-08 | 2021-01-12 | 深圳价值在线信息科技股份有限公司 | Information extraction method, extraction device, terminal equipment and readable storage medium |
CN112733869A (en) * | 2019-10-28 | 2021-04-30 | 中移信息技术有限公司 | Method, device and equipment for training text recognition model and storage medium |
CN112733911A (en) * | 2020-12-31 | 2021-04-30 | 平安科技(深圳)有限公司 | Entity recognition model training method, device, equipment and storage medium |
CN113268673A (en) * | 2021-04-23 | 2021-08-17 | 国家计算机网络与信息安全管理中心 | Method and system for analyzing clues of internet action information |
CN113486178A (en) * | 2021-07-12 | 2021-10-08 | 恒安嘉新(北京)科技股份公司 | Text recognition model training method, text recognition device and medium |
CN113821592A (en) * | 2021-06-23 | 2021-12-21 | 腾讯科技(深圳)有限公司 | Data processing method, device, equipment and storage medium |
CN116384515A (en) * | 2023-06-06 | 2023-07-04 | 之江实验室 | Model training method and device, storage medium and electronic equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106569998A (en) * | 2016-10-27 | 2017-04-19 | 浙江大学 | Text named entity recognition method based on Bi-LSTM, CNN and CRF |
CN107203511A (en) * | 2017-05-27 | 2017-09-26 | 中国矿业大学 | A kind of network text name entity recognition method based on neutral net probability disambiguation |
CN108038103A (en) * | 2017-12-18 | 2018-05-15 | 北京百分点信息科技有限公司 | A kind of method, apparatus segmented to text sequence and electronic equipment |
-
2018
- 2018-09-12 CN CN201811061626.8A patent/CN109299458B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106569998A (en) * | 2016-10-27 | 2017-04-19 | 浙江大学 | Text named entity recognition method based on Bi-LSTM, CNN and CRF |
CN107203511A (en) * | 2017-05-27 | 2017-09-26 | 中国矿业大学 | A kind of network text name entity recognition method based on neutral net probability disambiguation |
CN108038103A (en) * | 2017-12-18 | 2018-05-15 | 北京百分点信息科技有限公司 | A kind of method, apparatus segmented to text sequence and electronic equipment |
Cited By (38)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111681670A (en) * | 2019-02-25 | 2020-09-18 | 北京嘀嘀无限科技发展有限公司 | Information identification method and device, electronic equipment and storage medium |
CN111681670B (en) * | 2019-02-25 | 2023-05-12 | 北京嘀嘀无限科技发展有限公司 | Information identification method, device, electronic equipment and storage medium |
CN109902303B (en) * | 2019-03-01 | 2023-05-26 | 腾讯科技(深圳)有限公司 | Entity identification method and related equipment |
CN109902303A (en) * | 2019-03-01 | 2019-06-18 | 腾讯科技(深圳)有限公司 | A kind of entity recognition method and relevant device |
CN110287283A (en) * | 2019-05-22 | 2019-09-27 | 中国平安财产保险股份有限公司 | Intent model training method, intension recognizing method, device, equipment and medium |
CN110287283B (en) * | 2019-05-22 | 2023-08-01 | 中国平安财产保险股份有限公司 | Intention model training method, intention recognition method, device, equipment and medium |
CN110598210B (en) * | 2019-08-29 | 2023-08-04 | 深圳市优必选科技股份有限公司 | Entity recognition model training, entity recognition method, entity recognition device, entity recognition equipment and medium |
CN110516251A (en) * | 2019-08-29 | 2019-11-29 | 秒针信息技术有限公司 | A kind of construction method, construction device, equipment and the medium of electric business entity recognition model |
CN110598210A (en) * | 2019-08-29 | 2019-12-20 | 深圳市优必选科技股份有限公司 | Entity recognition model training method, entity recognition device, entity recognition equipment and medium |
CN110516251B (en) * | 2019-08-29 | 2023-11-03 | 秒针信息技术有限公司 | Method, device, equipment and medium for constructing electronic commerce entity identification model |
CN110705211A (en) * | 2019-09-06 | 2020-01-17 | 中国平安财产保险股份有限公司 | Text key content marking method and device, computer equipment and storage medium |
CN110555102A (en) * | 2019-09-16 | 2019-12-10 | 青岛聚看云科技有限公司 | media title recognition method, device and storage medium |
CN110826330A (en) * | 2019-10-12 | 2020-02-21 | 上海数禾信息科技有限公司 | Name recognition method and device, computer equipment and readable storage medium |
CN110826330B (en) * | 2019-10-12 | 2023-11-07 | 上海数禾信息科技有限公司 | Name recognition method and device, computer equipment and readable storage medium |
CN110738054A (en) * | 2019-10-14 | 2020-01-31 | 携程计算机技术(上海)有限公司 | Method, system, electronic device and storage medium for identifying hotel information in mail |
CN110738182A (en) * | 2019-10-21 | 2020-01-31 | 四川隧唐科技股份有限公司 | LSTM model unit training method and device for high-precision identification of bid amount |
CN110738055A (en) * | 2019-10-23 | 2020-01-31 | 北京字节跳动网络技术有限公司 | Text entity identification method, text entity identification equipment and storage medium |
CN112733869A (en) * | 2019-10-28 | 2021-04-30 | 中移信息技术有限公司 | Method, device and equipment for training text recognition model and storage medium |
CN110738319A (en) * | 2019-11-11 | 2020-01-31 | 四川隧唐科技股份有限公司 | LSTM model unit training method and device for recognizing bid-winning units based on CRF |
CN110825827B (en) * | 2019-11-13 | 2022-10-25 | 北京明略软件系统有限公司 | Entity relationship recognition model training method and device and entity relationship recognition method and device |
CN110825827A (en) * | 2019-11-13 | 2020-02-21 | 北京明略软件系统有限公司 | Entity relationship recognition model training method and device and entity relationship recognition method and device |
CN111079405A (en) * | 2019-11-29 | 2020-04-28 | 微民保险代理有限公司 | Text information identification method and device, storage medium and computer equipment |
CN111209396A (en) * | 2019-12-27 | 2020-05-29 | 深圳市优必选科技股份有限公司 | Entity recognition model training method, entity recognition method and related device |
CN111476022A (en) * | 2020-05-15 | 2020-07-31 | 湖南工商大学 | Method, system and medium for recognizing STM entity by embedding and mixing L characters of entity characteristics |
CN111914561A (en) * | 2020-07-31 | 2020-11-10 | 中国建设银行股份有限公司 | Entity recognition model training method, entity recognition device and terminal equipment |
CN111914561B (en) * | 2020-07-31 | 2023-06-30 | 建信金融科技有限责任公司 | Entity recognition model training method, entity recognition device and terminal equipment |
CN112214987B (en) * | 2020-09-08 | 2023-02-03 | 深圳价值在线信息科技股份有限公司 | Information extraction method, extraction device, terminal equipment and readable storage medium |
CN112214987A (en) * | 2020-09-08 | 2021-01-12 | 深圳价值在线信息科技股份有限公司 | Information extraction method, extraction device, terminal equipment and readable storage medium |
CN112182157A (en) * | 2020-09-29 | 2021-01-05 | 中国平安人寿保险股份有限公司 | Training method of online sequence labeling model, online labeling method and related equipment |
CN112182157B (en) * | 2020-09-29 | 2023-09-22 | 中国平安人寿保险股份有限公司 | Training method of online sequence labeling model, online labeling method and related equipment |
CN112733911B (en) * | 2020-12-31 | 2023-05-30 | 平安科技(深圳)有限公司 | Training method, device, equipment and storage medium of entity recognition model |
CN112733911A (en) * | 2020-12-31 | 2021-04-30 | 平安科技(深圳)有限公司 | Entity recognition model training method, device, equipment and storage medium |
CN113268673A (en) * | 2021-04-23 | 2021-08-17 | 国家计算机网络与信息安全管理中心 | Method and system for analyzing clues of internet action information |
CN113821592A (en) * | 2021-06-23 | 2021-12-21 | 腾讯科技(深圳)有限公司 | Data processing method, device, equipment and storage medium |
CN113486178A (en) * | 2021-07-12 | 2021-10-08 | 恒安嘉新(北京)科技股份公司 | Text recognition model training method, text recognition device and medium |
CN113486178B (en) * | 2021-07-12 | 2023-12-01 | 恒安嘉新(北京)科技股份公司 | Text recognition model training method, text recognition method, device and medium |
CN116384515A (en) * | 2023-06-06 | 2023-07-04 | 之江实验室 | Model training method and device, storage medium and electronic equipment |
CN116384515B (en) * | 2023-06-06 | 2023-09-01 | 之江实验室 | Model training method and device, storage medium and electronic equipment |
Also Published As
Publication number | Publication date |
---|---|
CN109299458B (en) | 2023-03-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109299458A (en) | Entity recognition method, device, equipment and storage medium | |
CN109271631A (en) | Segmenting method, device, equipment and storage medium | |
US9892414B1 (en) | Method, medium, and system for responding to customer requests with state tracking | |
US11645517B2 (en) | Information processing method and terminal, and computer storage medium | |
CN110598213A (en) | Keyword extraction method, device, equipment and storage medium | |
CN107491547A (en) | Searching method and device based on artificial intelligence | |
CN108345692A (en) | A kind of automatic question-answering method and system | |
CN109948633A (en) | User gender prediction method, apparatus, storage medium and electronic equipment | |
CN108628830A (en) | A kind of method and apparatus of semantics recognition | |
CN108897797A (en) | Update training method, device, storage medium and the electronic equipment of dialog model | |
CN110019742A (en) | Method and apparatus for handling information | |
CN110489198A (en) | A kind of method and system of worksheet | |
CN110858226A (en) | Conversation management method and device | |
Windiatmoko et al. | Developing facebook chatbot based on deep learning using rasa framework for university enquiries | |
Windiatmoko et al. | Developing FB chatbot based on deep learning using RASA framework for university enquiries | |
CN110209782A (en) | Question-Answering Model and the generation method of answer statement, device, medium and electronic equipment | |
CN110489730A (en) | Text handling method, device, terminal and storage medium | |
US20220198487A1 (en) | Method and device for processing user interaction information | |
CN110807097A (en) | Method and device for analyzing data | |
CN114119123A (en) | Information pushing method and device | |
CN115378890B (en) | Information input method, device, storage medium and computer equipment | |
CN116757270A (en) | Data processing method and server based on man-machine interaction model or large model | |
CN116401455A (en) | Link relation prediction method, entity knowledge graph generation method and device | |
CN116167382A (en) | Intention event extraction method and device, electronic equipment and storage medium | |
CN113255231B (en) | Data processing method, device, equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |