CN109657051A - Text snippet generation method, device, computer equipment and storage medium - Google Patents
Text snippet generation method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109657051A CN109657051A CN201811455100.8A CN201811455100A CN109657051A CN 109657051 A CN109657051 A CN 109657051A CN 201811455100 A CN201811455100 A CN 201811455100A CN 109657051 A CN109657051 A CN 109657051A
- Authority
- CN
- China
- Prior art keywords
- text
- sentence
- text snippet
- term vector
- model
- 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 61
- 238000003860 storage Methods 0.000 title claims abstract description 23
- 238000013528 artificial neural network Methods 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 18
- 230000008569 process Effects 0.000 claims abstract description 14
- 230000006870 function Effects 0.000 claims description 32
- 230000002457 bidirectional effect Effects 0.000 claims description 21
- 230000001351 cycling effect Effects 0.000 claims description 20
- 238000004590 computer program Methods 0.000 claims description 17
- 230000007246 mechanism Effects 0.000 claims description 15
- 230000015654 memory Effects 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 14
- 238000012545 processing Methods 0.000 description 6
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002073 mitogenetic effect Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000005086 pumping Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The embodiment of the invention discloses a kind of text snippet generation method, device, computer equipment and storage mediums.The present invention is applied to the field of neural networks in prediction model.This method comprises: pre-process and convert to obtain term vector pretreated target text by term vector tool to target text；Building using term vector as the input of text snippet model and is trained text snippet model in a manner of loop iteration using coding and decoding as the text snippet model of frame and exports sentence to be extracted；Term vector is pre-processed and be converted to text to be processed, and the term vector of text to be processed is input in the text snippet model after training and exports multiple sentences to be extracted；It is scored according to default score function model multiple sentences to be extracted, and text snippet is generated according to the scoring of sentence to be extracted.Method by implementing the embodiment of the present invention can quickly generate text snippet, effectively improve the precision of text snippet.
Description
Technical field
The present invention relates to field of artificial intelligence more particularly to a kind of text snippet generation method, device, computer to set
Standby and storage medium.
Background technique
With advances in technology with development, the artificial intelligence epoch have come quietly, started in daily life by
Gradually it is related to artificial intelligence.In the epoch of nowadays information explosion, all the time all in the information for generating flood tide, such as web page news, micro-
Rich, wechat article and mail etc..In order to realize that user can quickly and efficiently get required information, need to text
This information, which is collected, is organized into text snippet for user's fast browsing.Existing text snippet generation method mainly extracts
Formula generates text snippet by extracting sentence from target text, but this method generates taking a long time for text snippet, and
The precision of text snippet is not high, and the reading experience of user is poor.
Summary of the invention
The embodiment of the invention provides a kind of text snippet generation method, device, computer equipment and storage mediums, it is intended to
Solve the problems, such as that time-consuming, precision is low and user's reading experience is poor for text snippet generation.
In a first aspect, the embodiment of the invention provides a kind of text snippet generation methods comprising: target text is carried out
It pre-processes and passes through term vector tool and pretreated target text is converted to obtain term vector；Building is with coding and decoding
For the text snippet model of frame, using the term vector as the input of the text snippet model and in a manner of loop iteration
The text snippet model is trained and exports sentence to be extracted；Text to be processed is pre-processed and be converted to word to
The term vector of the text to be processed is input in the text snippet model after training and exports multiple sentences to be extracted by amount；
It is scored according to default score function model the multiple sentence to be extracted, and commenting according to the sentence to be extracted
It is mitogenetic at text snippet.
Second aspect, the embodiment of the invention also provides a kind of text snippet generating means comprising: converting unit is used
In to target text carry out pretreatment and by term vector tool pretreated target text is converted with obtain word to
Amount；Construction unit is plucked for constructing the text snippet model using coding and decoding as frame using the term vector as the text
It wants the input of model and is trained the text snippet model in a manner of loop iteration to export sentence to be extracted；Abstract
The term vector of the text to be processed is input to by unit for term vector to be pre-processed and be converted to text to be processed
Multiple sentences to be extracted are exported in text snippet model after training；Score unit, for according to default score function model
It scores the multiple sentence to be extracted, and text snippet is generated according to the scoring of the sentence to be extracted.
The third aspect, the embodiment of the invention also provides a kind of computer equipments comprising memory and processor, it is described
Computer program is stored on memory, the processor realizes the above method when executing the computer program.
Fourth aspect, the embodiment of the invention also provides a kind of computer readable storage medium, the storage medium storage
There is computer program, the computer program can realize the above method when being executed by a processor.
The embodiment of the invention provides a kind of text snippet generation method, device, computer equipment and storage mediums.Its
In, which comprises target text pre-process and pretreated target text is carried out by term vector tool
Conversion is to obtain term vector；Building is using coding and decoding as the text snippet model of frame, using the term vector as the text
The input of abstract model is simultaneously trained output sentence to be extracted to the text snippet model in a manner of loop iteration；It is right
Text to be processed is pre-processed and is converted to term vector, and the term vector of the text to be processed is input to the text after training
Multiple sentences to be extracted are exported in abstract model；The multiple sentence to be extracted is carried out according to default score function model
Scoring, and text snippet is generated according to the scoring of the sentence to be extracted.The embodiment of the present invention passes through building text snippet mould
Type handles text to be processed to obtain sentence to be extracted, then by default score function model to the sentence to be extracted
It scores to generate text snippet, text snippet can be quickly generated, effectively improve the precision of text snippet, improve user's
Reading experience.
Detailed description of the invention
Technical solution in order to illustrate the embodiments of the present invention more clearly, below will be to needed in embodiment description
Attached drawing is briefly described, it should be apparent that, drawings in the following description are some embodiments of the invention, general for this field
For logical technical staff, without creative efforts, it is also possible to obtain other drawings based on these drawings.
Fig. 1 is the application scenarios schematic diagram of text snippet generation method provided in an embodiment of the present invention；
Fig. 2 is the flow diagram of text snippet generation method provided in an embodiment of the present invention；
Fig. 3 is the sub-process schematic diagram of text snippet generation method provided in an embodiment of the present invention；
Fig. 4 is the sub-process schematic diagram of text snippet generation method provided in an embodiment of the present invention；
Fig. 5 is the sub-process schematic diagram of text snippet generation method provided in an embodiment of the present invention；
Fig. 6 is the sub-process schematic diagram of text snippet generation method provided in an embodiment of the present invention；
Fig. 7 is the schematic block diagram of text snippet generating means provided in an embodiment of the present invention；
Fig. 8 is the schematic block diagram of the specific unit of text snippet generating means provided in an embodiment of the present invention；
Fig. 9 is the schematic block diagram of the training unit of text snippet generating means provided in an embodiment of the present invention；And
Figure 10 is the schematic block diagram of computer equipment provided in an embodiment of the present invention.
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 some of the embodiments of the present invention, instead of all the embodiments.Based on this hair
Embodiment in bright, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall within the protection scope of the present invention.
It should be appreciated that ought use in this specification and in the appended claims, term " includes " and "comprising" instruction
Described feature, entirety, step, operation, the presence of element and/or component, but one or more of the other feature, whole is not precluded
Body, step, operation, the presence or addition of element, component and/or its set.
It is also understood that mesh of the term used in this description of the invention merely for the sake of description specific embodiment
And be not intended to limit the present invention.As description of the invention and it is used in the attached claims, unless on
Other situations are hereafter clearly indicated, otherwise " one " of singular, "one" and "the" are intended to include plural form.
It will be further appreciated that the term "and/or" used in description of the invention and the appended claims is
Refer to any combination and all possible combinations of one or more of associated item listed, and including these combinations.
Fig. 1 and Fig. 2 are please referred to, Fig. 1 is that the application scenarios of text snippet generation method provided in an embodiment of the present invention are illustrated
Figure.Fig. 2 is the schematic flow chart of text snippet generation method provided in an embodiment of the present invention.Text abstraction generating method tool
Body is applied in terminal 10, by interacting realization between terminal 10 and server 20.
Fig. 2 is the flow diagram of text snippet generation method provided in an embodiment of the present invention.As shown, this method packet
Include following steps S110-S140.
S110, target text pre-process and convert pretreated target text by term vector tool
To obtain term vector.
In one embodiment, term vector tool is a kind of natural language processing tool, and effect is exactly will be in natural language
Words switch to the term vector that computer is understood that.Traditional term vector is easy the puzzlement by dimension disaster, and any two
All be between word it is isolated, the relationship between word and word cannot be embodied, therefore the present embodiment uses this term vector of word2vec
Tool obtains term vector, can embody the similitude between word and word by calculating the distance between vector.Word2Vec
The conversion of term vector is mainly realized using two kinds of models of Skip-Gram and CBOW, the present embodiment is realized using Skip-Gram
The conversion of term vector, Skip-Gram model is mainly the word that context is predicted by centre word, for example, " today, weather was true
It is good " this sentence, predict it above " today " and hereafter " very good " by input center word " weather ".
In one embodiment, as shown in figure 3, the step S110 may include step S111-S112.
S111, acquisition target text segment and are encoded to obtain initial term vector according to the participle.
In one embodiment, target text is used for trained text, obtains according to the demand of user, for example, target is literary
Originally it can be obtained, can also be obtained from database by user from server by way of web crawlers.Obtaining target text
Afterwards, it needs first to pre-process target text, pretreatment includes: first to remove the punctuation mark in target text, is calculated
Word frequency removes low-frequency word, then segments to target text, and participle based on target text constructs vocabulary, finally root again
It is encoded to obtain initial term vector according to participle.Wherein, initial term vector refers to indicating word, example in the form of one-hot coding
Such as, obtain today/weather/very good after " today, weather was very good " participle, further according to participle carry out one-hot coding obtain initial word to
Amount, " today " is corresponding [100], " weather " corresponding [010] and " very good " corresponding [001].
S112, it carries out the initial term vector to be converted to term vector by term vector tool.
In one embodiment, Skip-Gram model trains in advance, and Skip-Gram model includes input layer, hidden layer
And output layer, the mapping from input layer to hidden layer does not use activation primitive, and hidden layer uses Huffman to output layer
Tree optimizes.Hofman tree is a binary tree, the word in the node on behalf vocabulary of leaf, and the weight of leaf node represents
The weight of word frequency, leaf node is bigger, closer apart from root node, and the path of root node to the leaf node of Hofman tree is word
Probability, calculation amount can be greatly reduced by the optimization of Hofman tree, accelerate the formation speed of text snippet.Specifically, example
Such as, sentence " today, weather was very good ", the initial term vector [010] of input word " weather " are finally obtained to Skip-Gram model
The term vector { 0.259,0.789, -0.048 } of " very good ".
S120, building are using coding and decoding as the text snippet model of frame, using the term vector as the text snippet
The input of model is simultaneously trained output sentence to be extracted to the text snippet model in a manner of loop iteration.
In one embodiment, coding and decoding frame is a kind of neural network framework end to end, for solving a kind of sequence
The problem of being converted to another sequence is made of encoder and decoder, encode to list entries by encoder
To intermediate code vector, then decoded to obtain output sequence by decoder.Due to processing text snippet the problem of be close to it is end-to-end
Sequence problem, therefore use coding and decoding framework establishment text snippet model.
In one embodiment, as shown in figure 4, the step S120 may include step S121-S122.
S121, using bidirectional valve controlled cycling element neural network as encoder and decoder and in the solution of the decoder
The code stage introduces attention mechanism construction text snippet model.
In one embodiment, using bidirectional valve controlled cycling element neural network as encoder and decoder, due to passing
To connect encoder and decoder, there are certain limitations by a fixed intermediate code vector for the coding and decoding model of system
Property, the Information Compression of list entries entire sequence after encoder encodes into the intermediate code vector of a specific length,
Cause not can completely to indicate that the information of entire list entries, the content first inputted can be override by the content of rear input, loses
Many detailed information, especially in long sequence.Therefore, attention in order to solve this problem is introduced in decoder decoding stage
Mechanism breaks conventional codec-decoder architecture by attention mechanism and all relies on internal one fixed length in encoding and decoding
Spend the limitation of vector.Attention mechanism is for being weighted variation to target data, by retaining encoder to list entries
Then centre output is as a result, carry out the study of selectivity to intermediate output result by a Matching Model, and decoding
Output sequence is associated by device therewith when exporting, and wherein Matching Model refers to calculating the model of similarity, generally speaking, defeated
Out the generating probability of each single item in sequence depend on selected in list entries which.
S122, using the term vector as the input of the text snippet model and according to the mode of loop iteration to described
Text snippet model is trained.
In one embodiment, after building text snippet model, obtained term vector is input to text snippet model
In be trained, specifically, term vector is input in encoder first, according to the bidirectional valve controlled cycling element of encoder nerve
Network obtains the coding vector of sentence, the semantic similarity between the coding vector of sentence and a upper output is then calculated, by language
Input of the coding vector of the adopted highest sentence of similarity as decoder, according to the gating cycle unit neural network of decoder
The coding vector of sentence is decoded, the inverse process of coding is equivalent to, finally obtains output, is used as next time after being exported
The calculating parameter of loop iteration continues training pattern.
In one embodiment, as shown in figure 5, the step S122 may include step S1221-S1223.
S1221, it the term vector is input to bidirectional valve controlled cycling element neural network encoder is encoded to obtain sentence
The coding vector of son.
In one embodiment, gating cycle unit (Gated Recurrent Unit) hereinafter referred to as GRU, is LSTM
The deformation of (Long Short-Term Memory), GRU is by the input gate in LSTM and forgets door merging as door is updated, therefore
It only includes two doors, i.e. resetting door and update door.Wherein, the status information quilt that door is used to control previous moment is updated
The degree being updated in current state；Resetting door is used to control the degree for the status information for ignoring previous moment.GRU model has
Parameter is few, sample requirement is few, the fast advantage of training speed, specific formula is as follows:
zt=σ (Wz[ht-1,xt])
rt=σ (Wr[ht-1,xt])
Wherein, x is the term vector of input, and h is the output of GRU model, and σ is sigmoid function, and r is resetting door, and z is more
New door, WzIt is the weight for updating door, WrIt is the weight for resetting door, by updating door and resetting door co- controlling from previous moment
Hidden state ht-1The hidden state h at current time is calculatedt.In the present embodiment, bidirectional valve controlled cycling element nerve net
Network, hereinafter referred to as BIGRU are made of two GRU, wherein two-way refer to GRU and one propagated forward backward
The GRU of propagation, can make full use of over and following information between the mapping for inputting and exporting, and realize the letter before and after text
Breath exchange, greatly improves the precision of context-prediction.It is obtained by the way that obtained term vector is input to output in BIGRU
The coding vector of implicit layer state h, that is, sentence.
S1222, it the coding vector of the sentence is input to bidirectional valve controlled cycling element neural network decoder solves
Code obtains sentence to be extracted.
In one embodiment, decoder equally uses bidirectional valve controlled cycling element neural network, and the decoding of decoder is suitable
The coding vector of sentence is carried out using the sentence coding vector of encoder output as input in the inverse process of encoder coding
Decoded output sentence to be extracted.
S1223, the sentence to be extracted is fed back to as decoding next time by the decoder by attention mechanism
Input and be trained in a manner of loop iteration.
In one embodiment, after obtaining upper output sentence i.e. to be extracted, this is waited taking out according to attention mechanism
The sentence taken is matched with current sentence coding vector, calculates similarity distribution weight between the two, selection and upper one
Output sentence sentence coding vector the most matched i.e. to be extracted is decoded to obtain currently wait take out as the input of decoder
The sentence taken matches obtained current sentence to be extracted with next sentence coding vector again, and so circulation changes
In generation, is trained text snippet model.
S130, term vector is pre-processed and is converted to text to be processed, the term vector of the text to be processed is defeated
Enter and exports multiple sentences to be extracted in the text snippet model to after training.
In one embodiment, text to be processed refers to that user wants to generate the text of abstract, after model training is good, user
Text to be processed can be selected according to demand, text to be processed is subjected to pretreatment, term vector is converted to by term vector tool, it will
The term vector of text to be processed is input in text snippet model, is carried out processing by text snippet model and is exported sentence to be extracted
Son.
S140, according to preset score function model score the multiple sentence to be extracted, and according to it is described to
The scoring of the sentence of extraction generates text snippet.
In one embodiment, using multi-layer perception (MLP) as default score function model, multi-layer perception (MLP) is that one kind connects entirely
The feed forward Artificial Network model connect comprising input layer, hidden layer and output layer, wherein the number of plies of hidden layer can have
One group of input vector can be mapped to another group of output vector by multilayer, it can indicate the mapping relations of two different spaces.
In one embodiment, as shown in fig. 6, the step S140 may include step S141-S142.
S141, the sentence to be extracted is scored to obtain score value according to default score function model.
In one embodiment, preset score function model specific formula is as follows:
Score=W1*tanh(W2*ht+W3*si)
Wherein, W is weight, and tanh is hyperbolic tangent function, and h is to hide layer state, and s is sentence to be extracted.It is default to comment
Divide function model to train in advance, is determined as W after weight is trained1、W2And W3, according to the output of text snippet model
The input of sentence to be extracted as default score function model, the calculating by presetting score function model export each to be extracted
The corresponding score value of sentence.
S142, it is ranked up according to the score value according to sequence from high to low, and preset quantity is extracted according to sequence
Sentence generation text snippet.
In one embodiment, the score value that each sentence to be extracted is exported by presetting Rating Model, to all wait take out
It takes sentence to be ranked up from high to low according to score value, chooses the highest sentence generation abstract of score value of preset quantity, preset
Quantity can be adjusted according to the reading habit or reading time of user.For example, preset quantity is 2, " today, weather was very good ",
" wanting to go window-shopping ", " you are nearest and fat ", the scoring of these three sentences are respectively 0.5,0.4 and 0.1, therefore are selected
" today, weather was very good, wanted to go window-shopping " generates text snippet.
The embodiment of the present invention illustrates a kind of text snippet generation method, by being pre-processed and being passed through to target text
Term vector tool converts to obtain term vector pretreated target text；Building is using coding and decoding as the text of frame
Abstract model, the term vector as the input of the text snippet model and is plucked the text in a manner of loop iteration
It wants model to be trained and exports sentence to be extracted；Pre-processed and be converted to term vector to text to be processed, will it is described to
The term vector of processing text, which is input in the text snippet model after training, exports multiple sentences to be extracted；According to default scoring
Function model scores to the multiple sentence to be extracted, and generates text according to the scoring of the sentence to be extracted and pluck
It wants, text snippet can be quickly generated, effectively improve the precision of text snippet, improve the reading experience of user.
Fig. 7 is a kind of schematic block diagram of text snippet generating means 200 provided in an embodiment of the present invention.As shown in fig. 7,
Corresponding to the above text snippet generation method, the present invention also provides a kind of text snippet generating means 200.Text summarization generation
Device 200 includes the unit for executing above-mentioned text snippet generation method, which can be configured in desktop computer, plate
Computer, laptop computer, etc. in terminals.Specifically, referring to Fig. 7, text summarization generation device 200 include converting unit 210,
Construction unit 220, abstract unit 230 and scoring unit 240.
Converting unit 210, for pre-process and by term vector tool to pretreated target to target text
Text is converted to obtain term vector.
In one embodiment, as shown in figure 8, the converting unit 210 includes subelement: acquiring unit 211 and conversion
Subelement 212.
Acquiring unit 211, for obtain target text carry out segment and encoded to obtain initial word according to the participle
Vector.
Conversion subunit 212, for carrying out the initial term vector to be converted to term vector by term vector tool.
Construction unit 220, for constructing the text snippet model using coding and decoding as frame, using the term vector as institute
State the input of text snippet model and in a manner of loop iteration to the text snippet model be trained output it is to be extracted
Sentence.
In one embodiment, as shown in figure 8, the construction unit 220 includes subelement: building subelement 221 and instruction
Practice unit 222.
Construct subelement 221, for using bidirectional valve controlled cycling element neural network as encoder and decoder and
The decoding stage of the decoder introduces attention mechanism construction text snippet model.
Training unit 222, for using the term vector as the input of the text snippet model and according to loop iteration
Mode the text snippet model is trained.
In one embodiment, as shown in figure 9, the training unit 222 includes subelement: coding unit 2221, decoding are single
Member 2222 and feedback unit 2223.
Coding unit 2221 is carried out for the term vector to be input to bidirectional valve controlled cycling element neural network encoder
Coding obtains the coding vector of sentence.
Decoding unit 2222, for the coding vector of the sentence to be input to bidirectional valve controlled cycling element neural network solution
Code device is decoded to obtain sentence to be extracted.
Feedback unit 2223, for by attention mechanism using the sentence to be extracted feed back to the decoder as
Decoded input next time is simultaneously trained in a manner of loop iteration.
Abstract unit 230, for term vector to be pre-processed and be converted to text to be processed, by the text to be processed
Term vector be input to training after text snippet model in export multiple sentences to be extracted.
Score unit 240, for being scored according to default score function model the multiple sentence to be extracted, and
Text snippet is generated according to the scoring of the sentence to be extracted.
In one embodiment, as shown in figure 8, the scoring unit 240 includes subelement: scoring subelement 241 and pumping
Take unit 242.
Score subelement 241, is commented for being scored according to default score function model the sentence to be extracted
Score value.
Extracting unit 242 is taken out for being ranked up according to the score value according to sequence from high to low, and according to sequence
Take the sentence generation text snippet of preset quantity.
It should be noted that it is apparent to those skilled in the art that, above-mentioned text snippet generating means
200 and each unit specific implementation process, can with reference to the corresponding description in preceding method embodiment, for convenience of description and
Succinctly, details are not described herein.
Above-mentioned text snippet generating means can be implemented as a kind of form of computer program, which can be
It is run in computer equipment as shown in Figure 10.
Referring to Fig. 10, Figure 10 is a kind of schematic block diagram of computer equipment provided by the embodiments of the present application.The calculating
Machine equipment 500 can be terminal, wherein terminal can be smart phone, tablet computer, laptop, desktop computer, individual
Digital assistants and wearable device etc. have the electronic equipment of communication function.
Refering to fig. 10, which includes processor 502, memory and the net connected by system bus 501
Network interface 505, wherein memory may include non-volatile memory medium 503 and built-in storage 504.
The non-volatile memory medium 503 can storage program area 5031 and computer program 5032.The computer program
5032 include program instruction, which is performed, and processor 502 may make to execute a kind of text snippet generation method.
The processor 502 is for providing calculating and control ability, to support the operation of entire computer equipment 500.
The built-in storage 504 provides environment for the operation of the computer program 5032 in non-volatile memory medium 503, should
When computer program 5032 is executed by processor 502, processor 502 may make to execute a kind of text snippet generation method.
The network interface 505 is used to carry out network communication with other equipment.It will be understood by those skilled in the art that in Figure 10
The structure shown, only the block diagram of part-structure relevant to application scheme, does not constitute and is applied to application scheme
The restriction of computer equipment 500 thereon, specific computer equipment 500 may include more more or fewer than as shown in the figure
Component perhaps combines certain components or with different component layouts.
Wherein, the processor 502 is for running computer program 5032 stored in memory, to realize following step
It is rapid: to target text carry out pretreatment and by term vector tool pretreated target text is converted with obtain word to
Amount；Building is using coding and decoding as the text snippet model of frame, using the term vector as the input of the text snippet model
And the text snippet model is trained in a manner of loop iteration and exports sentence to be extracted；Text to be processed is carried out
Term vector is pre-processed and be converted to, the term vector of the text to be processed is input in the text snippet model after training and is exported
Multiple sentences to be extracted；It is scored according to default score function model the multiple sentence to be extracted, and according to institute
The scoring for stating sentence to be extracted generates text snippet.
In one embodiment, processor 502 described pre-process and by term vector tool to target text realizing
When converted to pretreated target text to obtain term vector step, it is implemented as follows step: obtaining target text
Segment and is encoded to obtain initial term vector according to the participle；By term vector tool by the initial term vector into
Row is converted to term vector.
In one embodiment, processor 502 is realizing that the building, will using coding and decoding as the text snippet model of frame
The term vector as the text snippet model input and in a manner of loop iteration to the text snippet model carry out
When training exports sentence step to be extracted, it is implemented as follows step: using bidirectional valve controlled cycling element neural network as volume
Code device and decoder simultaneously introduce attention mechanism construction text snippet model in the decoding stage of the decoder；By institute's predicate
Vector as the text snippet model input and the text snippet model is trained according to the mode of loop iteration.
In one embodiment, processor 502 is described using the term vector as the defeated of the text snippet model in realization
When entering and being trained step to the text snippet model according to the mode of loop iteration, it is implemented as follows step: by institute
Predicate vector is input to bidirectional valve controlled cycling element neural network encoder and is encoded to obtain the coding vector of sentence；It will be described
The coding vector of sentence is input to bidirectional valve controlled cycling element neural network decoder and is decoded to obtain sentence to be extracted；It is logical
Attention mechanism is crossed the sentence to be extracted is fed back to the decoder as decoded input next time and is changed with circulation
The mode in generation is trained.
In one embodiment, processor 502 is realizing the default score function model of the basis to the multiple to be extracted
Sentence score, and according to the scoring of the sentence to be extracted generate text snippet step when, be implemented as follows step
It is rapid: the sentence to be extracted being scored to obtain score value according to default score function model；According to the score value according to
Sequence from high to low is ranked up, and the sentence generation text snippet of preset quantity is extracted according to sequence.
It should be appreciated that in the embodiment of the present application, processor 502 can be central processing unit (Central
Processing Unit, CPU), which can also be other general processors, 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) or other programmable logic
Device, discrete gate or transistor logic, discrete hardware components etc..Wherein, general processor can be microprocessor or
Person's processor is also possible to any conventional processor etc..
Those of ordinary skill in the art will appreciate that be realize above-described embodiment method in all or part of the process,
It is that relevant hardware can be instructed to complete by computer program.The computer program includes program instruction, computer journey
Sequence can be stored in a storage medium, which is computer readable storage medium.The program instruction is by the department of computer science
At least one processor in system executes, to realize the process step of the embodiment of the above method.
Therefore, the present invention also provides a kind of storage mediums.The storage medium can be computer readable storage medium.This is deposited
Storage media is stored with computer program, and wherein computer program includes program instruction.The program instruction makes when being executed by processor
Processor execute following steps: to target text carry out pre-process and by term vector tool to pretreated target text into
Row conversion is to obtain term vector；Building is using coding and decoding as the text snippet model of frame, using the term vector as the text
The input of this abstract model is simultaneously trained output sentence to be extracted to the text snippet model in a manner of loop iteration；
Term vector is pre-processed and be converted to text to be processed, and the term vector of the text to be processed is input to the text after training
Multiple sentences to be extracted are exported in this abstract model；According to default score function model to the multiple sentence to be extracted into
Row scoring, and text snippet is generated according to the scoring of the sentence to be extracted.
In one embodiment, the processor is realized and described is located in advance to target text executing described program instruction
When managing and passing through term vector tool and converted to pretreated target text to obtain term vector step, it is implemented as follows
Step: it obtains target text and segment and encoded to obtain initial term vector according to the participle；Pass through term vector tool
It carries out the initial term vector to be converted to term vector.
In one embodiment, the processor realizes the building using coding and decoding as frame executing described program instruction
The text snippet model of frame, to institute using the term vector as the input of the text snippet model and in a manner of loop iteration
It states text snippet model to be trained when exporting sentence step to be extracted, is implemented as follows step: bidirectional valve controlled is recycled
Unit neural network introduces attention mechanism construction text as encoder and decoder and in the decoding stage of the decoder
This abstract model；Using the term vector as the input of the text snippet model and according to the mode of loop iteration to the text
This abstract model is trained.
In one embodiment, the processor is realized described using the term vector as institute in the instruction of execution described program
It states the input of text snippet model and the text snippet model is trained according to the mode of loop iteration, specific implementation is such as
Lower step: the term vector is input to bidirectional valve controlled cycling element neural network encoder and is encoded to obtain the coding of sentence
Vector；The coding vector of the sentence is input to bidirectional valve controlled cycling element neural network decoder to be decoded to obtain wait take out
The sentence taken；The sentence to be extracted is fed back into the decoder as decoded input next time by attention mechanism
And it is trained in a manner of loop iteration.
In one embodiment, the processor realizes the basis and presets score function mould in the instruction of execution described program
Type scores to the multiple sentence to be extracted, and generates text snippet, tool according to the scoring of the sentence to be extracted
Body realizes following steps: being scored to obtain score value to the sentence to be extracted according to default score function model；According to institute
Commentary score value is ranked up according to sequence from high to low, and the sentence generation text snippet of preset quantity is extracted according to sequence.
The storage medium can be USB flash disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), magnetic disk
Or the various computer readable storage mediums that can store program code such as CD.
Those of ordinary skill in the art may be aware that list described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware
With the interchangeability of software, each exemplary composition and step are generally described according to function in the above description.This
A little functions are implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Specially
Industry technical staff can use different methods to achieve the described function each specific application, but this realization is not
It is considered as beyond the scope of this invention.
In several embodiments provided by the present invention, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary.For example, the division of each unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation.Such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.
The steps in the embodiment of the present invention can be sequentially adjusted, merged and deleted according to actual needs.This hair
Unit in bright embodiment device can be combined, divided and deleted according to actual needs.In addition, in each implementation of the present invention
Each functional unit in example can integrate in one processing unit, is also possible to each unit and physically exists alone, can also be with
It is that two or more units are integrated in one unit.
If the integrated unit is realized in the form of SFU software functional unit and when sold or used as an independent product,
It can store in one storage medium.Based on this understanding, technical solution of the present invention is substantially in other words to existing skill
The all or part of part or the technical solution that art contributes can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be a
People's computer, terminal or network equipment etc.) it performs all or part of the steps of the method described in the various embodiments of the present invention.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (10)
1. a kind of text snippet generation method characterized by comprising
Target text pre-process and converted pretreated target text to obtain word by term vector tool
Vector；
Building is using coding and decoding as the text snippet model of frame, using the term vector as the input of the text snippet model
And the text snippet model is trained in a manner of loop iteration and exports sentence to be extracted；
Term vector is pre-processed and is converted to text to be processed, after the term vector of the text to be processed is input to training
Text snippet model in export multiple sentences to be extracted；
It is scored according to default score function model the multiple sentence to be extracted, and according to the sentence to be extracted
Scoring generate text snippet.
2. text snippet generation method according to claim 1, which is characterized in that described to be pre-processed to target text
And pretreated target text is converted to obtain term vector by term vector tool, comprising:
Target text is obtained segment and encoded to obtain initial term vector according to the participle；
It carries out the initial term vector to be converted to term vector by term vector tool.
3. text snippet generation method according to claim 1, which is characterized in that the building is using coding and decoding as frame
Text snippet model, to described using the term vector as the input of the text snippet model and in a manner of loop iteration
Text snippet model, which is trained, exports sentence to be extracted, comprising:
Draw using bidirectional valve controlled cycling element neural network as encoder and decoder and in the decoding stage of the decoder
Enter attention mechanism construction text snippet model；
Using the term vector as the input of the text snippet model and according to the mode of loop iteration to the text snippet
Model is trained.
4. text snippet generation method according to claim 3, which is characterized in that it is described using the term vector as described in
The input of text snippet model is simultaneously trained the text snippet model according to the mode of loop iteration, comprising:
By the term vector be input to bidirectional valve controlled cycling element neural network encoder encoded to obtain the coding of sentence to
Amount；
The coding vector of the sentence is input to bidirectional valve controlled cycling element neural network decoder to be decoded to obtain wait take out
The sentence taken；
By attention mechanism using the sentence to be extracted feed back to the decoder as next time it is decoded input and with
The mode of loop iteration is trained.
5. text snippet generation method according to claim 1, which is characterized in that the basis presets score function model
It scores the multiple sentence to be extracted, and text snippet is generated according to the scoring of the sentence to be extracted, comprising:
The sentence to be extracted is scored to obtain score value according to default score function model；
It is ranked up according to the score value according to sequence from high to low, and extracts the sentence generation of preset quantity according to sequence
Text snippet.
6. a kind of text snippet generating means characterized by comprising
Converting unit, for pre-process to target text and be carried out by term vector tool to pretreated target text
Conversion is to obtain term vector；
Construction unit, for constructing the text snippet model using coding and decoding as frame, using the term vector as the text
The input of abstract model is simultaneously trained output sentence to be extracted to the text snippet model in a manner of loop iteration；
Make a summary unit, for term vector to be pre-processed and is converted to text to be processed, by the word of the text to be processed to
Amount is input in the text snippet model after training and exports multiple sentences to be extracted；
Score unit, for being scored according to default score function model the multiple sentence to be extracted, and according to institute
The scoring for stating sentence to be extracted generates text snippet.
7. text snippet generating means according to claim 6 characterized by comprising
Coding unit is encoded to obtain for the term vector to be input to bidirectional valve controlled cycling element neural network encoder
The coding vector of sentence；
Decoding unit is carried out for the coding vector of the sentence to be input to bidirectional valve controlled cycling element neural network decoder
Decoding obtains sentence to be extracted；
Feedback unit, for the sentence to be extracted to be fed back to the decoder as solving next time by attention mechanism
The input of code is simultaneously trained in a manner of loop iteration.
8. text snippet generating means according to claim 6 characterized by comprising
Score subelement, for being scored to obtain score value to the sentence to be extracted according to default score function model；
Extracting unit for being ranked up according to the score value according to sequence from high to low, and is extracted according to sequence and is preset
The sentence generation text snippet of quantity.
9. a kind of computer equipment, which is characterized in that the computer equipment includes memory and processor, on the memory
It is stored with computer program, the processor is realized as described in any one of claim 1-5 when executing the computer program
Method.
10. a kind of computer readable storage medium, which is characterized in that the storage medium is stored with computer program, the meter
Calculation machine program can realize method according to any one of claims 1 to 5 when being executed by a processor.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811455100.8A CN109657051A (en) | 2018-11-30 | 2018-11-30 | Text snippet generation method, device, computer equipment and storage medium |
| PCT/CN2019/092616 WO2020107878A1 (en) | 2018-11-30 | 2019-06-25 | Method and apparatus for generating text summary, computer device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811455100.8A CN109657051A (en) | 2018-11-30 | 2018-11-30 | Text snippet generation method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109657051A true CN109657051A (en) | 2019-04-19 |
Family
ID=66112132
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811455100.8A Pending CN109657051A (en) | 2018-11-30 | 2018-11-30 | Text snippet generation method, device, computer equipment and storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109657051A (en) |
| WO (1) | WO2020107878A1 (en) |
Cited By (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110209803A (en) * | 2019-06-18 | 2019-09-06 | 腾讯科技（深圳）有限公司 | Story generation method, device, computer equipment and storage medium |
| CN110377808A (en) * | 2019-06-14 | 2019-10-25 | 北京达佳互联信息技术有限公司 | Document processing method, device, electronic equipment and storage medium |
| CN110399882A (en) * | 2019-05-29 | 2019-11-01 | 广东工业大学 | A kind of character detecting method based on deformable convolutional neural networks |
| CN111026861A (en) * | 2019-12-10 | 2020-04-17 | 腾讯科技（深圳）有限公司 | Text abstract generation method, text abstract training method, text abstract generation device, text abstract training device, text abstract equipment and text abstract training medium |
| CN111180086A (en) * | 2019-12-12 | 2020-05-19 | 平安医疗健康管理股份有限公司 | Data matching method and device, computer equipment and storage medium |
| CN111199151A (en) * | 2019-12-31 | 2020-05-26 | 联想(北京)有限公司 | Data processing method and data processing device |
| WO2020107878A1 (en) * | 2018-11-30 | 2020-06-04 | 平安科技（深圳）有限公司 | Method and apparatus for generating text summary, computer device and storage medium |
| CN111241267A (en) * | 2020-01-10 | 2020-06-05 | 科大讯飞股份有限公司 | Abstract extraction and abstract extraction model training method, related device and storage medium |
| CN111291534A (en) * | 2020-02-03 | 2020-06-16 | 苏州科技大学 | Global coding method for automatic summarization of Chinese long text |
| CN111460135A (en) * | 2020-03-31 | 2020-07-28 | 北京百度网讯科技有限公司 | Method and device for generating text abstract |
| CN111581374A (en) * | 2020-05-09 | 2020-08-25 | 联想(北京)有限公司 | Text abstract obtaining method and device and electronic equipment |
| CN111597224A (en) * | 2020-04-17 | 2020-08-28 | 北京百度网讯科技有限公司 | Method and device for generating structured information, electronic equipment and storage medium |
| CN111767395A (en) * | 2020-06-30 | 2020-10-13 | 平安国际智慧城市科技股份有限公司 | Abstract generation method and system based on picture |
| CN111862977A (en) * | 2020-07-27 | 2020-10-30 | 北京嘀嘀无限科技发展有限公司 | Voice conversation processing method and system |
| CN111966820A (en) * | 2020-07-21 | 2020-11-20 | 西北工业大学 | Method and system for constructing and extracting generative abstract model |
| CN112052329A (en) * | 2020-09-02 | 2020-12-08 | 平安科技（深圳）有限公司 | Text abstract generation method and device, computer equipment and readable storage medium |
| CN112133279A (en) * | 2019-06-06 | 2020-12-25 | Tcl集团股份有限公司 | Vehicle-mounted information broadcasting method and device and terminal equipment |
| CN112464657A (en) * | 2020-12-07 | 2021-03-09 | 上海交通大学 | Hybrid text abstract generation method, system, terminal and storage medium |
| CN112685555A (en) * | 2019-10-17 | 2021-04-20 | 中国移动通信集团浙江有限公司 | Complaint work order quality detection method and device |
| CN112883711A (en) * | 2021-01-25 | 2021-06-01 | 北京金山云网络技术有限公司 | Method and device for generating abstract and electronic equipment |
| CN112989784A (en) * | 2021-03-04 | 2021-06-18 | 广州汇才创智科技有限公司 | Text automatic scoring method and device based on twin neural network and electronic equipment |
| CN113033778A (en) * | 2021-03-17 | 2021-06-25 | 平安科技（深圳）有限公司 | Fole generation method, device, equipment and storage medium |
| CN113204956A (en) * | 2021-07-06 | 2021-08-03 | 深圳市北科瑞声科技股份有限公司 | Multi-model training method, abstract segmentation method, text segmentation method and text segmentation device |
| CN113407707A (en) * | 2020-03-16 | 2021-09-17 | 北京沃东天骏信息技术有限公司 | Method and device for generating text abstract |
| WO2021189890A1 (en) * | 2020-10-22 | 2021-09-30 | 平安科技（深圳）有限公司 | Text generation model training method and apparatus based on text editing technology |
| CN113626584A (en) * | 2021-08-12 | 2021-11-09 | 中电积至(海南)信息技术有限公司 | Automatic text abstract generation method, system, computer equipment and storage medium |
| CN114386390A (en) * | 2021-11-25 | 2022-04-22 | 马上消费金融股份有限公司 | Data processing method and device, computer equipment and storage medium |
| CN114547287A (en) * | 2021-11-18 | 2022-05-27 | 电子科技大学 | Generation type text abstract method |
| CN114722836A (en) * | 2022-05-12 | 2022-07-08 | 北京中科闻歌科技股份有限公司 | Abstract generation method, apparatus, device and medium |
| CN115099359A (en) * | 2022-07-12 | 2022-09-23 | 平安科技（深圳）有限公司 | Address recognition method, device, equipment and storage medium based on artificial intelligence |
| WO2022241913A1 (en) * | 2021-05-17 | 2022-11-24 | 同济大学 | Heterogeneous graph-based text summarization method and apparatus, storage medium, and terminal |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112000777A (en) * | 2020-09-03 | 2020-11-27 | 上海然慧信息科技有限公司 | Text generation method and device, computer equipment and storage medium |
| CN112015890B (en) * | 2020-09-07 | 2024-01-23 | 广东工业大学 | Method and device for generating movie script abstract |
| CN112183057A (en) * | 2020-09-16 | 2021-01-05 | 北京思源智通科技有限责任公司 | Article generation method and device, intelligent device and storage medium |
| CN112148871B (en) * | 2020-09-21 | 2024-04-12 | 北京百度网讯科技有限公司 | Digest generation method, digest generation device, electronic equipment and storage medium |
| CN112347756B (en) * | 2020-09-29 | 2023-12-22 | 中国科学院信息工程研究所 | Inference reading understanding method and system based on serialization evidence extraction |
| CN112507111B (en) * | 2020-10-20 | 2024-02-06 | 北京中科凡语科技有限公司 | Model building method and abstract generating method for generating automatic abstract generation |
| CN112733498B (en) * | 2020-11-06 | 2024-04-16 | 北京工业大学 | Method for improving self-attention calculation of Chinese automatic text abstract |
| CN112507711B (en) * | 2020-12-04 | 2023-08-08 | 南京擎盾信息科技有限公司 | Text abstract extraction method and system |
| CN112819622B (en) * | 2021-01-26 | 2023-10-17 | 深圳价值在线信息科技股份有限公司 | Information entity relationship joint extraction method and device and terminal equipment |
| CN112800043A (en) * | 2021-02-05 | 2021-05-14 | 凯通科技股份有限公司 | Internet of things terminal information extraction method, device, equipment and storage medium |
| CN112836520A (en) * | 2021-02-19 | 2021-05-25 | 支付宝(杭州)信息技术有限公司 | Method and device for generating user description text based on user characteristics |
| CN112862727B (en) * | 2021-03-16 | 2023-06-23 | 上海壁仞智能科技有限公司 | Cross-modal image conversion method and device |
| CN113420143B (en) * | 2021-06-02 | 2023-12-22 | 平安国际智慧城市科技股份有限公司 | Method, device, equipment and storage medium for generating document abstract |
| CN113408272B (en) * | 2021-06-30 | 2023-08-18 | 北京百度网讯科技有限公司 | Training method, device, equipment and storage medium of abstract generation model |
| CN113408636B (en) | 2021-06-30 | 2023-06-06 | 北京百度网讯科技有限公司 | Pre-training model acquisition method and device, electronic equipment and storage medium |
| CN113657454B (en) * | 2021-07-23 | 2024-02-23 | 杭州安脉盛智能技术有限公司 | Nuclear power rotating machinery state monitoring method based on autoregressive BiGRU |
| CN113806514B (en) * | 2021-09-18 | 2024-03-01 | 上海交通大学 | Entity relation extraction method and system based on Transformer model |
| CN114218932B (en) * | 2021-11-26 | 2024-02-20 | 中国航空综合技术研究所 | Aviation fault text abstract generation method and device based on fault causal map |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105930314A (en) * | 2016-04-14 | 2016-09-07 | 清华大学 | Text summarization generation system and method based on coding-decoding deep neural networks |
| CN106980683A (en) * | 2017-03-30 | 2017-07-25 | 中国科学技术大学苏州研究院 | Blog text snippet generation method based on deep learning |
| CN108287922A (en) * | 2018-02-28 | 2018-07-17 | 福州大学 | A kind of text data viewpoint abstract method for digging of fusion topic attribute and emotion information |
| US20180300400A1 (en) * | 2017-04-14 | 2018-10-18 | Salesforce.Com, Inc. | Deep Reinforced Model for Abstractive Summarization |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120253792A1 (en) * | 2011-03-30 | 2012-10-04 | Nec Laboratories America, Inc. | Sentiment Classification Based on Supervised Latent N-Gram Analysis |
| CN104834735B (en) * | 2015-05-18 | 2018-01-23 | 大连理工大学 | A kind of documentation summary extraction method based on term vector |
| CN109657051A (en) * | 2018-11-30 | 2019-04-19 | 平安科技（深圳）有限公司 | Text snippet generation method, device, computer equipment and storage medium |
-
2018
- 2018-11-30 CN CN201811455100.8A patent/CN109657051A/en active Pending
-
2019
- 2019-06-25 WO PCT/CN2019/092616 patent/WO2020107878A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105930314A (en) * | 2016-04-14 | 2016-09-07 | 清华大学 | Text summarization generation system and method based on coding-decoding deep neural networks |
| CN106980683A (en) * | 2017-03-30 | 2017-07-25 | 中国科学技术大学苏州研究院 | Blog text snippet generation method based on deep learning |
| US20180300400A1 (en) * | 2017-04-14 | 2018-10-18 | Salesforce.Com, Inc. | Deep Reinforced Model for Abstractive Summarization |
| CN108287922A (en) * | 2018-02-28 | 2018-07-17 | 福州大学 | A kind of text data viewpoint abstract method for digging of fusion topic attribute and emotion information |
Cited By (46)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020107878A1 (en) * | 2018-11-30 | 2020-06-04 | 平安科技（深圳）有限公司 | Method and apparatus for generating text summary, computer device and storage medium |
| CN110399882A (en) * | 2019-05-29 | 2019-11-01 | 广东工业大学 | A kind of character detecting method based on deformable convolutional neural networks |
| CN112133279A (en) * | 2019-06-06 | 2020-12-25 | Tcl集团股份有限公司 | Vehicle-mounted information broadcasting method and device and terminal equipment |
| CN110377808A (en) * | 2019-06-14 | 2019-10-25 | 北京达佳互联信息技术有限公司 | Document processing method, device, electronic equipment and storage medium |
| CN110209803A (en) * | 2019-06-18 | 2019-09-06 | 腾讯科技（深圳）有限公司 | Story generation method, device, computer equipment and storage medium |
| CN110209803B (en) * | 2019-06-18 | 2023-11-14 | 腾讯科技（深圳）有限公司 | Story generation method, apparatus, computer device and storage medium |
| CN112685555A (en) * | 2019-10-17 | 2021-04-20 | 中国移动通信集团浙江有限公司 | Complaint work order quality detection method and device |
| CN112685555B (en) * | 2019-10-17 | 2023-04-18 | 中国移动通信集团浙江有限公司 | Complaint work order quality detection method and device |
| CN111026861A (en) * | 2019-12-10 | 2020-04-17 | 腾讯科技（深圳）有限公司 | Text abstract generation method, text abstract training method, text abstract generation device, text abstract training device, text abstract equipment and text abstract training medium |
| CN111026861B (en) * | 2019-12-10 | 2023-07-04 | 腾讯科技（深圳）有限公司 | Text abstract generation method, training device, training equipment and medium |
| CN111180086A (en) * | 2019-12-12 | 2020-05-19 | 平安医疗健康管理股份有限公司 | Data matching method and device, computer equipment and storage medium |
| CN111180086B (en) * | 2019-12-12 | 2023-04-25 | 平安医疗健康管理股份有限公司 | Data matching method, device, computer equipment and storage medium |
| CN111199151A (en) * | 2019-12-31 | 2020-05-26 | 联想(北京)有限公司 | Data processing method and data processing device |
| CN111241267B (en) * | 2020-01-10 | 2022-12-06 | 科大讯飞股份有限公司 | Abstract extraction and abstract extraction model training method, related device and storage medium |
| CN111241267A (en) * | 2020-01-10 | 2020-06-05 | 科大讯飞股份有限公司 | Abstract extraction and abstract extraction model training method, related device and storage medium |
| CN111291534A (en) * | 2020-02-03 | 2020-06-16 | 苏州科技大学 | Global coding method for automatic summarization of Chinese long text |
| CN113407707A (en) * | 2020-03-16 | 2021-09-17 | 北京沃东天骏信息技术有限公司 | Method and device for generating text abstract |
| CN111460135B (en) * | 2020-03-31 | 2023-11-07 | 北京百度网讯科技有限公司 | Method and device for generating text abstract |
| CN111460135A (en) * | 2020-03-31 | 2020-07-28 | 北京百度网讯科技有限公司 | Method and device for generating text abstract |
| CN111597224A (en) * | 2020-04-17 | 2020-08-28 | 北京百度网讯科技有限公司 | Method and device for generating structured information, electronic equipment and storage medium |
| CN111597224B (en) * | 2020-04-17 | 2023-09-15 | 北京百度网讯科技有限公司 | Method and device for generating structured information, electronic equipment and storage medium |
| CN111581374A (en) * | 2020-05-09 | 2020-08-25 | 联想(北京)有限公司 | Text abstract obtaining method and device and electronic equipment |
| CN111767395B (en) * | 2020-06-30 | 2023-12-26 | 平安国际智慧城市科技股份有限公司 | Abstract generation method and system based on pictures |
| CN111767395A (en) * | 2020-06-30 | 2020-10-13 | 平安国际智慧城市科技股份有限公司 | Abstract generation method and system based on picture |
| CN111966820B (en) * | 2020-07-21 | 2023-03-24 | 西北工业大学 | Method and system for constructing and extracting generative abstract model |
| CN111966820A (en) * | 2020-07-21 | 2020-11-20 | 西北工业大学 | Method and system for constructing and extracting generative abstract model |
| CN111862977B (en) * | 2020-07-27 | 2021-08-10 | 北京嘀嘀无限科技发展有限公司 | Voice conversation processing method and system |
| US11862143B2 (en) | 2020-07-27 | 2024-01-02 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for processing speech dialogues |
| CN111862977A (en) * | 2020-07-27 | 2020-10-30 | 北京嘀嘀无限科技发展有限公司 | Voice conversation processing method and system |
| CN112052329A (en) * | 2020-09-02 | 2020-12-08 | 平安科技（深圳）有限公司 | Text abstract generation method and device, computer equipment and readable storage medium |
| WO2021189890A1 (en) * | 2020-10-22 | 2021-09-30 | 平安科技（深圳）有限公司 | Text generation model training method and apparatus based on text editing technology |
| CN112464657A (en) * | 2020-12-07 | 2021-03-09 | 上海交通大学 | Hybrid text abstract generation method, system, terminal and storage medium |
| CN112464657B (en) * | 2020-12-07 | 2022-07-08 | 上海交通大学 | Hybrid text abstract generation method, system, terminal and storage medium |
| CN112883711A (en) * | 2021-01-25 | 2021-06-01 | 北京金山云网络技术有限公司 | Method and device for generating abstract and electronic equipment |
| CN112989784A (en) * | 2021-03-04 | 2021-06-18 | 广州汇才创智科技有限公司 | Text automatic scoring method and device based on twin neural network and electronic equipment |
| CN113033778B (en) * | 2021-03-17 | 2023-09-29 | 平安科技（深圳）有限公司 | Buddha music generation method, device, equipment and storage medium |
| CN113033778A (en) * | 2021-03-17 | 2021-06-25 | 平安科技（深圳）有限公司 | Fole generation method, device, equipment and storage medium |
| WO2022241913A1 (en) * | 2021-05-17 | 2022-11-24 | 同济大学 | Heterogeneous graph-based text summarization method and apparatus, storage medium, and terminal |
| CN113204956A (en) * | 2021-07-06 | 2021-08-03 | 深圳市北科瑞声科技股份有限公司 | Multi-model training method, abstract segmentation method, text segmentation method and text segmentation device |
| CN113626584A (en) * | 2021-08-12 | 2021-11-09 | 中电积至(海南)信息技术有限公司 | Automatic text abstract generation method, system, computer equipment and storage medium |
| CN114547287B (en) * | 2021-11-18 | 2023-04-07 | 电子科技大学 | Generation type text abstract method |
| CN114547287A (en) * | 2021-11-18 | 2022-05-27 | 电子科技大学 | Generation type text abstract method |
| CN114386390B (en) * | 2021-11-25 | 2022-12-06 | 马上消费金融股份有限公司 | Data processing method and device, computer equipment and storage medium |
| CN114386390A (en) * | 2021-11-25 | 2022-04-22 | 马上消费金融股份有限公司 | Data processing method and device, computer equipment and storage medium |
| CN114722836A (en) * | 2022-05-12 | 2022-07-08 | 北京中科闻歌科技股份有限公司 | Abstract generation method, apparatus, device and medium |
| CN115099359A (en) * | 2022-07-12 | 2022-09-23 | 平安科技（深圳）有限公司 | Address recognition method, device, equipment and storage medium based on artificial intelligence |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2020107878A1 (en) | 2020-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109657051A (en) | Text snippet generation method, device, computer equipment and storage medium | |
| Li et al. | A dual-attention hierarchical recurrent neural network for dialogue act classification | |
| CN106202010B (en) | Method and apparatus based on deep neural network building Law Text syntax tree | |
| CN110516253B (en) | Chinese spoken language semantic understanding method and system | |
| CN111241294B (en) | Relationship extraction method of graph convolution network based on dependency analysis and keywords | |
| CN108153913B (en) | Training method of reply information generation model, reply information generation method and device | |
| Peng et al. | Text generation with exemplar-based adaptive decoding | |
| CN110348016A (en) | Text snippet generation method based on sentence association attention mechanism | |
| CN109241524A (en) | Semantic analysis method and device, computer readable storage medium, electronic equipment | |
| CN106776517B (en) | Automatic poetry method, device and system | |
| CN108153864A (en) | Method based on neural network generation text snippet | |
| CN111858932A (en) | Multiple-feature Chinese and English emotion classification method and system based on Transformer | |
| CN112262368A (en) | Natural language to API conversion | |
| CN113283244B (en) | Pre-training model-based bidding data named entity identification method | |
| Núñez-Marcos et al. | A survey on Sign Language machine translation | |
| CN109933652A (en) | Intelligent answer method, apparatus, computer equipment and storage medium | |
| CN110475129A (en) | Method for processing video frequency, medium and server | |
| CN109189862A (en) | A kind of construction of knowledge base method towards scientific and technological information analysis | |
| CN111881292B (en) | Text classification method and device | |
| US11450310B2 (en) | Spoken language understanding | |
| CN108363685B (en) | Self-media data text representation method based on recursive variation self-coding model | |
| CN116628186B (en) | Text abstract generation method and system | |
| CN113360610A (en) | Dialog generation method and system based on Transformer model | |
| CN111767697B (en) | Text processing method and device, computer equipment and storage medium | |
| CN109637527A (en) | The semantic analytic method and system of conversation sentence |
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 |