CN109408812A - A method of the sequence labelling joint based on attention mechanism extracts entity relationship - Google Patents
A method of the sequence labelling joint based on attention mechanism extracts entity relationship Download PDFInfo
- Publication number
- CN109408812A CN109408812A CN201811157788.1A CN201811157788A CN109408812A CN 109408812 A CN109408812 A CN 109408812A CN 201811157788 A CN201811157788 A CN 201811157788A CN 109408812 A CN109408812 A CN 109408812A
- Authority
- CN
- China
- Prior art keywords
- word
- sentence
- indicate
- attention mechanism
- entity
- 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 33
- 230000007246 mechanism Effects 0.000 title claims abstract description 30
- 238000002372 labelling Methods 0.000 title claims abstract description 20
- 239000000284 extract Substances 0.000 title claims abstract description 16
- 230000015654 memory Effects 0.000 claims abstract description 12
- 210000004027 cell Anatomy 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000003058 natural language processing Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 230000006403 short-term memory Effects 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 239000002131 composite material Substances 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 230000007774 longterm Effects 0.000 abstract 1
- 238000000605 extraction Methods 0.000 description 14
- 238000012545 processing Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000013519 translation 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
Abstract
The method that the sequence labelling joint based on attention mechanism that the invention discloses a kind of extracts entity relationship, first carry out the Chinese sentence corpus of magnanimity the pretreatment such as to denoise, then it is segmented, being converted into vector to single word indicates, the input as two-way length memory network in short-term in this way encodes individual character.Using two-way length, memory network can not only learn long-term and short-term Dependency Specification in short-term, the data of input layer can also be calculated by forwardly and rearwardly both direction simultaneously, to learn past contextual information and following contextual information, this is very useful to the sequence labelling of sentence.Then attention mechanism is introduced in decoding layer, so that decoding generates the information vector of available front coding stage each character hidden layer when annotated sequence, the information for making full use of list entries to carry.The entity tag probability of each word is calculated finally by softmax, can effectively obtain final annotated sequence and carries out the combination of entity and its corresponding relationship.
Description
Technical field
The invention belongs to natural language processing technique field more particularly to the attentions of some sequence labellings and deep learning
Mechanism joint extracts the entity relationship in non-structural text.
Background technique
With the arriving of big data era, various information are flooded with our life, and most of is all rambling
Data or non-structural natural language text, Yao Congzhong extract useful information and are just particularly important.Information extraction one
As include two subtasks being closely connected i.e. Entity recognition and Relation extraction, basic goal be exactly from structureless network or
The semantic relation between name entity and entity is excavated in the text of field, is extracted information and is carried out structured storage, so as to
Intuitively understand in people, receive information.Entity relation extraction is also a vital task in natural language processing simultaneously, it
It is all related in multiple fields, for example, the building of domain knowledge map, information retrieval, machine translation, automatic question answering etc., all have
There is stronger supporting role, thus there is biggish researching value and research significance.
According to the degree of dependence to labeled data, entity relation extraction method can be divided into supervised learning method, semi-supervised
Learning method, unsupervised learning method and open abstracting method.There is the learning method of supervision to regard Relation extraction task to divide
Class problem designs effective feature according to training data, to learn various disaggregated models, then uses trained classifier
Projected relationship.Existing supervised learning Relation extraction method has been achieved for preferable effect, but their heavy dependence parts of speech
The natural language processings marks such as mark, syntax parsing provide characteristic of division.And natural language processing annotation tool often exists greatly
Mistake is measured, these mistakes will constantly propagate amplification, the final effect for influencing Relation extraction in Relation extraction system.Recently,
Many researchers start the technical application by deep learning into Relation extraction.Rink et al. extracts entity first, then
Identify the relationship between them, frame of this separation is easily handled two tasks all and more flexible.Socher etc.
People proposes to solve the problems, such as Relation extraction using recurrent neural network and syntactic structure.This method is first by recurrent neural network
Syntax parsing first is carried out to sentence and constructs a syntax tree, merges according to the syntactic structure iteration of sentence, finally obtains this
The vector of sentence indicates.This method can effectively consider the syntactic structure information of sentence, but can not consider two realities well
Position and semantic information of the body in sentence.Zeng et al. proposes to carry out entity relation extraction using convolutional neural networks.They
Using the position vector of word and term vector as the input of convolutional neural networks, and in feature be added entity position vector and its
His relevant vocabulary, enables the entity information in sentence to be preferably applied in Relation extraction.It will but this above
The method that Entity recognition and Relation extraction are implemented separately ignores inner link between the two.Zheng et al. proposes one kind
Entity relationship combines abstracting method end to end, and joint is extracted and is converted into sequence labelling problem, passes through long memory network in short-term
Question sentence is coded and decoded, and adds biasing loss and finally obtains annotated sequence.This algorithm takes full advantage of context
Information, but apply on English data set, it is very different with Chinese corpus, and the model when list entries is very long
It is difficult to acquire reasonable vector expression, all contextual informations not distinguished, which limits the performances of model, lead to mould
The effect of type is poor.
Summary of the invention
The technical problem to be solved by the present invention is to propose that a kind of sequence labelling joint based on attention mechanism (ATT) is taken out
The method for taking entity relationship, the new dimension model proposed first according to Zheng et al., using natural language sentence as Seq2Seq
List entries, word, which is converted into vector, by embedding layers indicates, and uses two-way length memory network (LSTM) in short-term
It is encoded, the mark to relationship is added on the basis of former dimension model, is then equally using long short-term memory net
Network exports addition attention mechanism, the sequence marked finally by softmax layers when being decoded, and thus may be used
To obtain the annotated sequence to entire sentence, convenient for extracting entity relationship by the identification of sequence.
The present invention proposes a kind of method that joint extracts entity relationship towards magnanimity Chinese sentence corpus.First by magnanimity
Chinese sentence corpus carry out the pretreatment such as denoising, then segmented, to single word be converted into vector indicate, in this way may be used
Individual character is encoded using the input as two-way length memory network in short-term.Using two-way length, memory network can not only be learned in short-term
It practises for a long time with short-term Dependency Specification, can also simultaneously calculate the data of input layer by forwardly and rearwardly both direction, thus
Learn past contextual information and following contextual information, this is very useful to the sequence labelling of sentence.Then exist
Decoding layer introduces attention mechanism, so that decoding generates available front coding stage each character hidden layer when annotated sequence
Information vector, make full use of list entries carry information.The entity tagization for calculating each word finally by softmax is general
Rate can effectively obtain final annotated sequence and carry out the combination of entity and its corresponding relationship.
To achieve the goals above, the invention adopts the following technical scheme: extracting entity to combine by annotated sequence
Relationship realizes the conversion from statement sequence to annotated sequence using Seq2Seq model joint Attention mechanism.It first passes around
The embedding layers of high dimension vector by the expression word in higher dimensional space (dimension in space is usually the size of dictionary) is mapped to low
Vector in dimension (tens dimensions) continuous space, i.e., sentence is converted into vector indicates.Then using obtained term vector as two-way
The input of long memory network in short-term is encoded, including two parallel LSTM layers: positive LSTM layers and LSTM layers reversed, difference
Calculate forward and calculate backward.Finally by decoding layer length in short-term memory network by the vector generated before export at mark
Sequence is infused, and attention mechanism is added wherein, allows the mark of generation that can only instead of not pay close attention to global semantic coding vector,
It increases one " attention range ", indicates to pay close attention to which of list entries part when next output mark,
Then next output is generated according to the region of concern.Finally decoding obtained annotated sequence should include that this word is in entity
In position, corresponding relationship and the role in relationship combine to obtain entity relationship according to annotated sequence.This has just obtained one
It is a that abstracting method is combined based on the entity relationship of sequence labelling and LSTM attention mechanism.
The technical solution that this method uses extracts entity relationship for a kind of sequence labelling joint based on attention mechanism
Method, method includes the following steps:
Step 1, the entity relationship data set for obtaining Opening field are simultaneously pre-processed, and pretreated process is by the data
Collection is divided into training set and test set two parts, this two parts all includes sentence to be processed, is divided sentence wherein included
Word processing, so that sentence is converted into individual word.
The each word obtained in sentence after pretreatment is converted into vector expression by embeding layer, and is input to double by step 2
It is encoded into the long coding layer of memory network in short-term.
Memory network decodes in short-term by the length with attention mechanism for step 3, the output for obtaining coding layer, and wherein
Attention mechanism is added.
Step 4 exports entity tag probability, completion and composite entity based on mark predicted vector by softmax layers
And relationship, obtain triple.
Preferably, step 1 specifically includes the following steps:
Step 1.1, the entity relationship data set for obtaining Opening field, and all data concentrated to data carry out at denoising
Reason, including the useless blank character of removal, capitalization are converted into small letter etc.;
Data set is divided into training set and test set by step 1.2;
Step 1.3 establishes user's Custom Dictionaries, such as long word and proper noun, and using at Harbin Institute of Technology's natural language
Science and engineering tool LTP segments sentence;
Preferably, step 2 specifically includes the following steps:
Step 2.1, the corpus training Word2Vec term vector model using wikipedia, the dimension of term vector is 300;
Step 2.2, the term vector mapping matrix generated by Word2Vec, obtain the corresponding term vector of each word, whole
The term vector of a sentence is expressed as { w1,w2,......wn, wnIndicate that the term vector of n-th of word indicates;
The term vector that embeding layer is converted is input to two-way LSTM coding layer by step 2.3, it includes LSTM layers of forward direction, instead
To LSTM layers and articulamentum;
Step 2.4 encodes contextual information and semantic information by two-way LSTM coding layer, and positive LSTM is from w1
To wn, reversed LSTM is from wnTo w1, and the coding vector of entire sentence, calculation formula such as formula (1) are exported in the hidden layer of neuron
(2) shown in (3) (4) (5) (6).
it=δ (Wwiwt+Whiht-1+Wcict-1+bi) (1)
ft=δ (Wwfwt+Whfht-1+Wcfct-1+bf) (2)
zt=tanh (Wwcwt+Whcht-1+bc) (3)
ct=ftct-1+itzt (4)
ot=δ (Wwowt+Whoht-1+Wcoct+bo) (5)
ht=ot tanh(ct) (6)
I, f, z, o of formula (1) (2) (3) (5) are respectively input gate, forget door, update door, out gate, the c in formula (4)t
Indicate the cell state of t moment, the h in formula (6)tIndicate the output of t moment, W indicates relevant parameter, the W in formula (3)wcAnd Whc
The parameter of the parameter of the cell state of word and the cell state of output is respectively indicated, the W in (1) (2) (5)wx、WhxAnd WcxRespectively
Indicate the parameter of x words, the parameter of output and the parameter of cell state, wtIndicate that t-th of word, b indicate biasing loss, δ table
Show sigmoid activation primitive.
Preferably, step 3 specifically includes the following steps:
Output after two-way LSTM coding is input to LSTM decoding by step 3.1;
Attention mechanism is added in step 3.2, makes position of the model learning context in entity and in relationship
Role, the result y that final decoded result is predicted by last momentt-1, the input s at this momenttOn relevant with to this moment
Hereafter annotated sequence cseqtIt obtains, shown in calculation formula such as formula (7), (8).
p(yt|y1,y2,......,yt-1, cseq) and=g (yt-1,st,cseqt) (7)
L in formula (8)xIndicate the length of sentence, aseqtjIndicate the Automobile driving of j-th of word mark in read statement
Coefficient, hjIndicate the semantic coding of j-th of word, cseqtIndicate the relevant context annotated sequence of t moment.
Step 3.3, the decoded sequence of output.
Opposite with the prior art, the present invention has following clear superiority:
For the present invention when joint extracts entity relationship, will extract Task Switching is sequence labelling problem, using based on two-way
The Seq2Seq model of long memory network in short-term, and entity and relationship are marked word-based attention mechanism is wherein added
Note.Opposite other methods make to carry out joint in this way and extract to avoid first to extract entity and extract error caused by relationship again
Accumulation problem has deepened the inner link between entity and relationship, and the Feature Engineering for not needing complexity can learn to length
Short-term Dependency Specification.In addition, attention mechanism is introduced in decoding layer, can reduce the computation burden of processing higher-dimension list entries,
By the subset of the selection input of structuring, data dimension is reduced, while allowing task processing system to focus more on and finding input sequence
Significant useful information relevant to currently exporting in column, to improve the quality of output.In addition, sequence is added in attention mechanism
In decoding process in column mark task, the accuracy of sequence labelling can be effectively improved, improves whole efficiency.To sum up institute
It states, it is proposed in this paper to be had based on the entity relationship of sequence labelling and attention mechanism joint abstracting method towards magnanimity Chinese
Material, the advantage that entity relationship connection is close, accuracy rate is high.
Detailed description of the invention
Fig. 1 is the flow chart of method involved in the present invention.
Fig. 2 is the Seq2Seq model structure of sequence labelling task in the present invention.
Fig. 3 is the structure chart for the attention model that present invention decoding uses.
Fig. 4 is the present invention finally to the example of sentence annotated sequence.
Specific embodiment
Yi Xiajiehejutishishili,Bing Canzhaofutu,Dui Benfamingjinyibuxiangxishuoming.
Hardware device used in the present invention has PC machine 1;
As shown in Figure 1, the present invention provides a kind of side of sequence labelling joint extraction entity relationship based on attention mechanism
Method, specifically includes the following steps:
Step 1, the entity relationship data set for obtaining Opening field are simultaneously pre-processed, and pretreated process is by the data
Collection is divided into training set and test set two parts, this two parts all includes sentence to be processed, is divided sentence wherein included
Word processing, so that sentence is converted into individual word.
Step 1.1, the entity relationship data set of Opening field is obtained, and denoising is carried out to these data, including go
Except useless blank character, capitalization are converted into small letter etc.;
Step 1.2, data set is divided into training set and test set;
Step 1.3, user's Custom Dictionaries, such as long word and proper noun are established, and using at Harbin Institute of Technology's natural language
Science and engineering tool LTP segments sentence;
As shown in Fig. 2, obtaining word segmentation result after to Chinese sentence " founder of Apple Inc. is Qiao Busi " pretreatment
" founder of Apple Inc. is Qiao Busi ".
Step 2, each word in sentence after pretreatment is converted into vector by embeding layer indicates, and is input to two-way length
It is encoded in short-term memory network, the structure of whole Seq2Seq model is as shown in Figure 2.
Step 2.1, using the corpus training Word2Vec term vector model of wikipedia, the dimension of term vector is 300, will
The word segmentation result that step 1 obtains inputs in trained term vector model;
Step 2.2, the term vector mapping matrix generated by Word2Vec, obtains the corresponding term vector of each word, word
Vector can be expressed as { w1,w2,......,wn};
Step 2.3, the term vector that embeding layer is converted is input to two-way LSTM coding layer, it includes LSTM layers of forward direction, instead
To LSTM layers and articulamentum;
Step 2.4, contextual information and semantic information are encoded by two-way LSTM coding layer, positive LSTM is from w1
To wn, reversed LSTM is from wnTo w1, and the coding vector of entire sentence, calculation formula such as formula (1) are exported in the hidden layer of neuron
(2) shown in (3) (4) (5) (6).
it=δ (Wwiwt+Whiht-1+Wcict-1+bi) (1)
ft=δ (Wwfwt+Whfht-1+Wcfct-1+bf) (2)
zt=tanh (Wwcwt+Whcht-1+bc) (3)
ct=ftct-1+itzt (4)
ot=δ (Wwowt+Whoht-1+Wcoct+bo) (5)
ht=ot tanh(ct) (6)
I, f, z, o of formula (1) (2) (3) (5) are respectively input gate, forget door, update door, out gate, the c in (4)tTable
Show the cell state of t moment, the h in (6)tIndicate the output of t moment, W indicates relevant parameter, the W in (3)wcAnd WhcTable respectively
Show the parameter of the parameter of the cell state of word and the cell state of output, the W in (1) (2) (5)wx、WhxAnd WcxRespectively indicate x
Word parameter, the parameter of output and the parameter of cell state, wtIndicate that t-th of word, b indicate biasing loss, δ is indicated
Sigmoid activation primitive.
Step 3, by the length with attention mechanism, memory network decodes vector coding layer obtained in short-term, and wherein
Attention mechanism is added.
Step 3.1, the output after two-way LSTM coding is input to LSTM decoding;
As shown in figure 3, contextual information when by sequence labelling is imagined as being by a series of<Key, Value>word is to structure
At given some to be marked words Query is obtained by calculating the similitude or correlation of Query and each Key at this time
Each Key corresponds to the weight coefficient of Value, is then weighted summation to Value to get final Attention number has been arrived
Value.
Attention mechanism is added in step 3.2, makes position of the model learning context in entity and in relationship
Role, the result y that final decoded result is predicted by last momentt-1, the input s at this momenttOn relevant with to this moment
Hereafter annotated sequence cseqtIt obtains, shown in calculation formula such as formula (7), (8).
p(yt|y1,y2,......,yt-1, cseq) and=g (yt-1,st,cseqt) (7)
L in formula (8)xIndicate the length of sentence, aseqtjIndicate the Automobile driving of j-th of word mark in read statement
Coefficient, hjIndicate the semantic coding of j-th of word, cseqtIndicate the relevant context annotated sequence of t moment.
Step 3.3, decoded sequence is exported.
Step 4, entity tag probability is exported based on mark predicted vector by softmax layers.
As shown in figure 4, Seq2Seq model generates the annotated sequence of a sentence, and according to mark completion and composite entity
And relationship, finally obtain entity relationship triple.
Above embodiments are only exemplary embodiment of the present invention, are not used in the limitation present invention, protection scope of the present invention
It is defined by the claims.Those skilled in the art can within the spirit and scope of the present invention make respectively the present invention
Kind modification or equivalent replacement, this modification or equivalent replacement also should be regarded as being within the scope of the present invention.
Claims (4)
1. a kind of method that the sequence labelling joint based on attention mechanism extracts entity relationship, it is characterised in that:
Method includes the following steps:
Step 1, the entity relationship data set for obtaining Opening field are simultaneously pre-processed, and pretreated process is by the data set point
For training set and test set two parts, this two parts all includes sentence to be processed, is carried out at participle to sentence wherein included
Reason, so that sentence is converted into individual word;
The each word obtained in sentence after pretreatment is converted into vector expression by embeding layer, and is input to two-way length by step 2
It is encoded in the coding layer of short-term memory network;
Memory network decodes in short-term by the length with attention mechanism for step 3, the output for obtaining coding layer, and is added wherein
Attention mechanism;
Step 4 exports entity tag probability, completion and composite entity and pass based on mark predicted vector by softmax layers
System, obtains triple.
2. the method that a kind of sequence labelling joint based on attention mechanism according to claim 1 extracts entity relationship,
It is characterized by:
Step 1 specifically includes the following steps:
Step 1.1, the entity relationship data set for obtaining Opening field, and denoising is carried out to all data that data are concentrated,
Including removing useless blank character, capitalization is converted into small letter;
Data set is divided into training set and test set by step 1.2;
Step 1.3 establishes user's Custom Dictionaries, such as long word and proper noun, and utilizes Harbin Institute of Technology's natural language processing work
Tool LTP segments sentence.
3. the method that a kind of sequence labelling joint based on attention mechanism according to claim 1 extracts entity relationship,
It is characterized by: step 2 specifically includes the following steps:
Step 2.1, the corpus training Word2Vec term vector model using wikipedia, the dimension of term vector is 300;
Step 2.2, the term vector mapping matrix generated by Word2Vec, obtain the corresponding term vector of each word, entire language
The term vector of sentence is expressed as { w1,w2,......wn, wnIndicate that the term vector of n-th of word indicates;
The term vector that embeding layer is converted is input to two-way LSTM coding layer by step 2.3, it includes LSTM layers of forward direction, reversely
LSTM layers and articulamentum;
Step 2.4 encodes contextual information and semantic information by two-way LSTM coding layer, and positive LSTM is from w1To wn,
Reversed LSTM is from wnTo w1, and the coding vector of entire sentence, calculation formula such as formula (1) (2) are exported in the hidden layer of neuron
(3) shown in (4) (5) (6);
it=δ (Wwiwt+Whiht-1+Wcict-1+bi) (1)
ft=δ (Wwfwt+Whfht-1+Wcfct-1+bf) (2)
zt=tanh (Wwcwt+Whcht-1+bc) (3)
ct=ftct-1+itzt (4)
ot=δ (Wwowt+Whoht-1+Wcoct+bo) (5)
ht=ot tanh(ct) (6)
I, f, z, o of formula (1) (2) (3) (5) are respectively input gate, forget door, update door, out gate, the c in formula (4)tIndicate t
The cell state at moment, the h in (6)tIndicate the output of t moment, W indicates relevant parameter, the W in formula (3)wcAnd WhcIt respectively indicates
The parameter of the cell state of the parameter and output of the cell state of word, the W in formula (1) (2) (5)wx、WhxAnd WcxRespectively indicate x
Word parameter, the parameter of output and the parameter of cell state, wtIndicate that t-th of word, b indicate biasing loss, δ is indicated
Sigmoid activation primitive.
4. the method that a kind of sequence labelling joint based on attention mechanism according to claim 1 extracts entity relationship,
It is characterized by: step 3 specifically includes the following steps:
Output after two-way LSTM coding is input to LSTM decoding by step 3.1;
Attention mechanism is added in step 3.2, makes position of the model learning context in entity and the role in relationship,
The result y that final decoded result is predicted by last momentt-1, the input s at this momenttWith context relevant to this moment
Annotated sequence cseqtIt obtains, shown in calculation formula such as formula (7), (8);
p(yt|y1,y2,......,yt-1, cseq) and=g (yt-1,st,cseqt)(7)
L in formula (8)xIndicate the length of sentence, aseqtjIndicate the Automobile driving coefficient of j-th of word mark in read statement,
hjIndicate the semantic coding of j-th of word, cseqtIndicate the relevant context annotated sequence of t moment;
Step 3.3, the decoded sequence of output.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811157788.1A CN109408812A (en) | 2018-09-30 | 2018-09-30 | A method of the sequence labelling joint based on attention mechanism extracts entity relationship |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811157788.1A CN109408812A (en) | 2018-09-30 | 2018-09-30 | A method of the sequence labelling joint based on attention mechanism extracts entity relationship |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109408812A true CN109408812A (en) | 2019-03-01 |
Family
ID=65466650
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811157788.1A Pending CN109408812A (en) | 2018-09-30 | 2018-09-30 | A method of the sequence labelling joint based on attention mechanism extracts entity relationship |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109408812A (en) |
Cited By (64)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109871541A (en) * | 2019-03-06 | 2019-06-11 | 电子科技大学 | It is a kind of suitable for multilingual multi-field name entity recognition method |
CN109933806A (en) * | 2019-04-01 | 2019-06-25 | 长沙理工大学 | A kind of repetition generation method, system, equipment and computer readable storage medium |
CN109977861A (en) * | 2019-03-25 | 2019-07-05 | 中国科学技术大学 | Offline handwritten form method for identifying mathematical formula |
CN109992785A (en) * | 2019-04-09 | 2019-07-09 | 腾讯科技(深圳)有限公司 | Content calculation method, device and equipment based on machine learning |
CN110110061A (en) * | 2019-04-26 | 2019-08-09 | 同济大学 | Low-resource languages entity abstracting method based on bilingual term vector |
CN110196913A (en) * | 2019-05-23 | 2019-09-03 | 北京邮电大学 | Multiple entity relationship joint abstracting method and device based on text generation formula |
CN110209836A (en) * | 2019-05-17 | 2019-09-06 | 北京邮电大学 | Remote supervisory Relation extraction method and device |
CN110334219A (en) * | 2019-07-12 | 2019-10-15 | 电子科技大学 | The knowledge mapping for incorporating text semantic feature based on attention mechanism indicates learning method |
CN110334339A (en) * | 2019-04-30 | 2019-10-15 | 华中科技大学 | It is a kind of based on location aware from the sequence labelling model and mask method of attention mechanism |
CN110348021A (en) * | 2019-07-17 | 2019-10-18 | 湖北亿咖通科技有限公司 | Character string identification method, electronic equipment, storage medium based on name physical model |
CN110348019A (en) * | 2019-07-17 | 2019-10-18 | 南通大学 | A kind of medical bodies vector method for transformation based on attention mechanism |
CN110390021A (en) * | 2019-06-13 | 2019-10-29 | 平安科技(深圳)有限公司 | Drug knowledge mapping construction method, device, computer equipment and storage medium |
CN110399433A (en) * | 2019-07-23 | 2019-11-01 | 福建奇点时空数字科技有限公司 | A kind of data entity Relation extraction method based on deep learning |
CN110427605A (en) * | 2019-05-09 | 2019-11-08 | 苏州大学 | The Ellipsis recovering method understood towards short text |
CN110444259A (en) * | 2019-06-06 | 2019-11-12 | 昆明理工大学 | Traditional Chinese medical electronic case history entity relationship extracting method based on entity relationship mark strategy |
CN110502749A (en) * | 2019-08-02 | 2019-11-26 | 中国电子科技集团公司第二十八研究所 | A kind of text Relation extraction method based on the double-deck attention mechanism Yu two-way GRU |
CN110570920A (en) * | 2019-08-20 | 2019-12-13 | 华东理工大学 | Entity and relationship joint learning method based on attention focusing model |
CN110597948A (en) * | 2019-07-11 | 2019-12-20 | 东华大学 | Entity relation extraction method based on deep learning |
CN110598001A (en) * | 2019-08-05 | 2019-12-20 | 平安科技(深圳)有限公司 | Method, device and storage medium for extracting association entity relationship |
CN110705301A (en) * | 2019-09-30 | 2020-01-17 | 京东城市(北京)数字科技有限公司 | Entity relationship extraction method and device, storage medium and electronic equipment |
CN110765774A (en) * | 2019-10-08 | 2020-02-07 | 北京三快在线科技有限公司 | Training method and device of information extraction model and information extraction method and device |
CN110851620A (en) * | 2019-10-29 | 2020-02-28 | 天津大学 | Knowledge representation method based on combination of text embedding and structure embedding |
CN111008279A (en) * | 2019-11-27 | 2020-04-14 | 云知声智能科技股份有限公司 | Entity relationship extraction method and device |
CN111008283A (en) * | 2019-10-31 | 2020-04-14 | 中电药明数据科技(成都)有限公司 | Sequence labeling method and system based on composite boundary information |
CN111046668A (en) * | 2019-12-04 | 2020-04-21 | 北京信息科技大学 | Method and device for recognizing named entities of multi-modal cultural relic data |
CN111079433A (en) * | 2019-11-29 | 2020-04-28 | 北京奇艺世纪科技有限公司 | Event extraction method and device and electronic equipment |
CN111090724A (en) * | 2019-11-21 | 2020-05-01 | 新华智云科技有限公司 | Entity extraction method capable of judging relevance between text content and entity based on deep learning |
CN111125367A (en) * | 2019-12-26 | 2020-05-08 | 华南理工大学 | Multi-character relation extraction method based on multi-level attention mechanism |
CN111143691A (en) * | 2019-12-31 | 2020-05-12 | 四川长虹电器股份有限公司 | Joint information extraction method and device |
CN111241209A (en) * | 2020-01-03 | 2020-06-05 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
CN111275094A (en) * | 2020-01-17 | 2020-06-12 | 厦门快商通科技股份有限公司 | Data labeling method, device and equipment based on machine learning |
CN111274814A (en) * | 2019-12-26 | 2020-06-12 | 浙江大学 | Novel semi-supervised text entity information extraction method |
CN111274815A (en) * | 2020-01-15 | 2020-06-12 | 北京百度网讯科技有限公司 | Method and device for mining entity attention points in text |
CN111310472A (en) * | 2020-01-19 | 2020-06-19 | 合肥讯飞数码科技有限公司 | Alias generation method, device and equipment |
CN111368528A (en) * | 2020-03-09 | 2020-07-03 | 西南交通大学 | Entity relation joint extraction method for medical texts |
CN111460807A (en) * | 2020-03-13 | 2020-07-28 | 平安科技(深圳)有限公司 | Sequence labeling method and device, computer equipment and storage medium |
CN111488728A (en) * | 2020-03-12 | 2020-08-04 | 天闻数媒科技(北京)有限公司 | Labeling method, device and storage medium for unstructured test question data |
CN111640471A (en) * | 2020-05-27 | 2020-09-08 | 牛张明 | Method and system for predicting activity of drug micromolecules based on two-way long-short memory model |
CN111667238A (en) * | 2020-05-26 | 2020-09-15 | 南开大学 | Recruitment inspiring generation method based on skill perception multi-attention mechanism |
CN111767409A (en) * | 2020-06-14 | 2020-10-13 | 南开大学 | Entity relationship extraction method based on multi-head self-attention mechanism |
WO2020220636A1 (en) * | 2019-04-28 | 2020-11-05 | 平安科技(深圳)有限公司 | Text data enhancement method and apparatus, electronic device, and non-volatile computer-readable storage medium |
CN111914091A (en) * | 2019-05-07 | 2020-11-10 | 四川大学 | Entity and relation combined extraction method based on reinforcement learning |
CN111950278A (en) * | 2019-05-14 | 2020-11-17 | 株式会社理光 | Sequence labeling method and device and computer readable storage medium |
CN112015859A (en) * | 2019-05-31 | 2020-12-01 | 百度在线网络技术(北京)有限公司 | Text knowledge hierarchy extraction method and device, computer equipment and readable medium |
CN112163092A (en) * | 2020-10-10 | 2021-01-01 | 成都数之联科技有限公司 | Entity and relation extraction method, system, device and medium |
CN112329465A (en) * | 2019-07-18 | 2021-02-05 | 株式会社理光 | Named entity identification method and device and computer readable storage medium |
CN112597757A (en) * | 2020-12-04 | 2021-04-02 | 光大科技有限公司 | Word detection method and device, storage medium and electronic device |
CN112612871A (en) * | 2020-12-17 | 2021-04-06 | 浙江大学 | Multi-event detection method based on sequence generation model |
CN112667820A (en) * | 2020-12-08 | 2021-04-16 | 吉林省吉科软信息技术有限公司 | Deep learning construction method for full-process traceable ecological chain supervision knowledge map |
CN112765991A (en) * | 2021-01-14 | 2021-05-07 | 中山大学 | Deep dialogue semantic role labeling method and system based on knowledge enhancement |
CN112784597A (en) * | 2019-11-06 | 2021-05-11 | 阿里巴巴集团控股有限公司 | Method and device for evaluating quality of article |
CN112818124A (en) * | 2021-02-21 | 2021-05-18 | 昆明理工大学 | Entity relationship extraction method based on attention neural network |
CN112836482A (en) * | 2021-02-09 | 2021-05-25 | 浙江工商大学 | Method and device for generating problems by sequence generation model based on template |
CN113033189A (en) * | 2021-04-08 | 2021-06-25 | 北京理工大学 | Semantic coding method of long-short term memory network based on attention dispersion |
CN113191118A (en) * | 2021-05-08 | 2021-07-30 | 山东省计算中心(国家超级计算济南中心) | Text relation extraction method based on sequence labeling |
CN113553850A (en) * | 2021-03-30 | 2021-10-26 | 电子科技大学 | Entity relation extraction method based on ordered structure encoding pointer network decoding |
CN113642767A (en) * | 2021-07-09 | 2021-11-12 | 武汉科技大学 | Multi-dimensional feature combination prediction method based on MI-VMD-DA-EDLSTM-VEC |
CN113807079A (en) * | 2020-06-11 | 2021-12-17 | 四川大学 | End-to-end entity and relation combined extraction method based on sequence-to-sequence |
CN114580408A (en) * | 2022-03-10 | 2022-06-03 | 浙江理工大学 | Lower link generation method and device based on double-layer attention joint learning |
CN114625871A (en) * | 2020-12-14 | 2022-06-14 | 四川大学 | Triple classification method based on attention position joint coding |
CN115019893A (en) * | 2022-06-14 | 2022-09-06 | 邵阳学院 | Enhancer identification method based on bidirectional long-and-short-term memory and attention mechanism |
CN115510869A (en) * | 2022-05-30 | 2022-12-23 | 青海师范大学 | End-to-end Tibetan La lattice shallow semantic analysis method |
CN116227597A (en) * | 2023-05-05 | 2023-06-06 | 中国人民解放军国防科技大学 | Biomedical knowledge extraction method, device, computer equipment and storage medium |
CN117235286A (en) * | 2023-11-10 | 2023-12-15 | 昆明理工大学 | Attention-strengthening entity relation extraction model, construction method thereof and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018072563A1 (en) * | 2016-10-18 | 2018-04-26 | 中兴通讯股份有限公司 | Knowledge graph creation method, device, and system |
CN108280062A (en) * | 2018-01-19 | 2018-07-13 | 北京邮电大学 | Entity based on deep learning and entity-relationship recognition method and device |
CN108304911A (en) * | 2018-01-09 | 2018-07-20 | 中国科学院自动化研究所 | Knowledge Extraction Method and system based on Memory Neural Networks and equipment |
CN108460013A (en) * | 2018-01-30 | 2018-08-28 | 大连理工大学 | A kind of sequence labelling model based on fine granularity vocabulary representation model |
-
2018
- 2018-09-30 CN CN201811157788.1A patent/CN109408812A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018072563A1 (en) * | 2016-10-18 | 2018-04-26 | 中兴通讯股份有限公司 | Knowledge graph creation method, device, and system |
CN108304911A (en) * | 2018-01-09 | 2018-07-20 | 中国科学院自动化研究所 | Knowledge Extraction Method and system based on Memory Neural Networks and equipment |
CN108280062A (en) * | 2018-01-19 | 2018-07-13 | 北京邮电大学 | Entity based on deep learning and entity-relationship recognition method and device |
CN108460013A (en) * | 2018-01-30 | 2018-08-28 | 大连理工大学 | A kind of sequence labelling model based on fine granularity vocabulary representation model |
Non-Patent Citations (3)
Title |
---|
张晓斌等: "基于CNN和双向LSTM融合的实体关系抽取", 《网络与信息安全学报》 * |
李枫林等: "基于深度学习框架的实体关系抽取研究进展", 《情报科学》 * |
王红等: "基于注意力机制的LSTM 的语义关系抽取", 《计算机应用研究》 * |
Cited By (100)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109871541A (en) * | 2019-03-06 | 2019-06-11 | 电子科技大学 | It is a kind of suitable for multilingual multi-field name entity recognition method |
CN109977861A (en) * | 2019-03-25 | 2019-07-05 | 中国科学技术大学 | Offline handwritten form method for identifying mathematical formula |
CN109933806A (en) * | 2019-04-01 | 2019-06-25 | 长沙理工大学 | A kind of repetition generation method, system, equipment and computer readable storage medium |
CN109933806B (en) * | 2019-04-01 | 2024-01-30 | 长沙理工大学 | Method, system, equipment and computer readable storage medium for generating duplicate description |
CN109992785A (en) * | 2019-04-09 | 2019-07-09 | 腾讯科技(深圳)有限公司 | Content calculation method, device and equipment based on machine learning |
CN109992785B (en) * | 2019-04-09 | 2023-07-25 | 腾讯科技(深圳)有限公司 | Content calculation method, device and equipment based on machine learning |
CN110110061A (en) * | 2019-04-26 | 2019-08-09 | 同济大学 | Low-resource languages entity abstracting method based on bilingual term vector |
CN110110061B (en) * | 2019-04-26 | 2023-04-18 | 同济大学 | Low-resource language entity extraction method based on bilingual word vectors |
WO2020220636A1 (en) * | 2019-04-28 | 2020-11-05 | 平安科技(深圳)有限公司 | Text data enhancement method and apparatus, electronic device, and non-volatile computer-readable storage medium |
CN110334339A (en) * | 2019-04-30 | 2019-10-15 | 华中科技大学 | It is a kind of based on location aware from the sequence labelling model and mask method of attention mechanism |
CN111914091A (en) * | 2019-05-07 | 2020-11-10 | 四川大学 | Entity and relation combined extraction method based on reinforcement learning |
CN111914091B (en) * | 2019-05-07 | 2022-10-14 | 四川大学 | Entity and relation combined extraction method based on reinforcement learning |
CN110427605A (en) * | 2019-05-09 | 2019-11-08 | 苏州大学 | The Ellipsis recovering method understood towards short text |
CN111950278A (en) * | 2019-05-14 | 2020-11-17 | 株式会社理光 | Sequence labeling method and device and computer readable storage medium |
CN110209836B (en) * | 2019-05-17 | 2022-04-26 | 北京邮电大学 | Remote supervision relation extraction method and device |
CN110209836A (en) * | 2019-05-17 | 2019-09-06 | 北京邮电大学 | Remote supervisory Relation extraction method and device |
CN110196913A (en) * | 2019-05-23 | 2019-09-03 | 北京邮电大学 | Multiple entity relationship joint abstracting method and device based on text generation formula |
CN112015859B (en) * | 2019-05-31 | 2023-08-18 | 百度在线网络技术(北京)有限公司 | Knowledge hierarchy extraction method and device for text, computer equipment and readable medium |
CN112015859A (en) * | 2019-05-31 | 2020-12-01 | 百度在线网络技术(北京)有限公司 | Text knowledge hierarchy extraction method and device, computer equipment and readable medium |
CN110444259B (en) * | 2019-06-06 | 2022-09-23 | 昆明理工大学 | Entity relation extracting method of traditional Chinese medicine electronic medical record based on entity relation labeling strategy |
CN110444259A (en) * | 2019-06-06 | 2019-11-12 | 昆明理工大学 | Traditional Chinese medical electronic case history entity relationship extracting method based on entity relationship mark strategy |
CN110390021A (en) * | 2019-06-13 | 2019-10-29 | 平安科技(深圳)有限公司 | Drug knowledge mapping construction method, device, computer equipment and storage medium |
CN110597948A (en) * | 2019-07-11 | 2019-12-20 | 东华大学 | Entity relation extraction method based on deep learning |
CN110334219A (en) * | 2019-07-12 | 2019-10-15 | 电子科技大学 | The knowledge mapping for incorporating text semantic feature based on attention mechanism indicates learning method |
CN110348021A (en) * | 2019-07-17 | 2019-10-18 | 湖北亿咖通科技有限公司 | Character string identification method, electronic equipment, storage medium based on name physical model |
CN110348019A (en) * | 2019-07-17 | 2019-10-18 | 南通大学 | A kind of medical bodies vector method for transformation based on attention mechanism |
CN110348021B (en) * | 2019-07-17 | 2021-05-18 | 湖北亿咖通科技有限公司 | Character string recognition method based on named entity model, electronic device and storage medium |
CN112329465A (en) * | 2019-07-18 | 2021-02-05 | 株式会社理光 | Named entity identification method and device and computer readable storage medium |
CN110399433A (en) * | 2019-07-23 | 2019-11-01 | 福建奇点时空数字科技有限公司 | A kind of data entity Relation extraction method based on deep learning |
CN110502749A (en) * | 2019-08-02 | 2019-11-26 | 中国电子科技集团公司第二十八研究所 | A kind of text Relation extraction method based on the double-deck attention mechanism Yu two-way GRU |
CN110502749B (en) * | 2019-08-02 | 2023-10-03 | 中国电子科技集团公司第二十八研究所 | Text relation extraction method based on double-layer attention mechanism and bidirectional GRU |
CN110598001A (en) * | 2019-08-05 | 2019-12-20 | 平安科技(深圳)有限公司 | Method, device and storage medium for extracting association entity relationship |
CN110570920B (en) * | 2019-08-20 | 2023-07-14 | 华东理工大学 | Entity and relationship joint learning method based on concentration model |
CN110570920A (en) * | 2019-08-20 | 2019-12-13 | 华东理工大学 | Entity and relationship joint learning method based on attention focusing model |
CN110705301A (en) * | 2019-09-30 | 2020-01-17 | 京东城市(北京)数字科技有限公司 | Entity relationship extraction method and device, storage medium and electronic equipment |
CN110765774A (en) * | 2019-10-08 | 2020-02-07 | 北京三快在线科技有限公司 | Training method and device of information extraction model and information extraction method and device |
CN110765774B (en) * | 2019-10-08 | 2021-09-17 | 北京三快在线科技有限公司 | Training method and device of information extraction model and information extraction method and device |
CN110851620A (en) * | 2019-10-29 | 2020-02-28 | 天津大学 | Knowledge representation method based on combination of text embedding and structure embedding |
CN111008283B (en) * | 2019-10-31 | 2023-06-20 | 中电药明数据科技(成都)有限公司 | Sequence labeling method and system based on composite boundary information |
CN111008283A (en) * | 2019-10-31 | 2020-04-14 | 中电药明数据科技(成都)有限公司 | Sequence labeling method and system based on composite boundary information |
CN112784597A (en) * | 2019-11-06 | 2021-05-11 | 阿里巴巴集团控股有限公司 | Method and device for evaluating quality of article |
CN111090724A (en) * | 2019-11-21 | 2020-05-01 | 新华智云科技有限公司 | Entity extraction method capable of judging relevance between text content and entity based on deep learning |
CN111008279A (en) * | 2019-11-27 | 2020-04-14 | 云知声智能科技股份有限公司 | Entity relationship extraction method and device |
CN111008279B (en) * | 2019-11-27 | 2023-11-14 | 云知声智能科技股份有限公司 | Entity relation extraction method and device |
CN111079433A (en) * | 2019-11-29 | 2020-04-28 | 北京奇艺世纪科技有限公司 | Event extraction method and device and electronic equipment |
CN111079433B (en) * | 2019-11-29 | 2023-10-27 | 北京奇艺世纪科技有限公司 | Event extraction method and device and electronic equipment |
CN111046668A (en) * | 2019-12-04 | 2020-04-21 | 北京信息科技大学 | Method and device for recognizing named entities of multi-modal cultural relic data |
CN111046668B (en) * | 2019-12-04 | 2023-09-22 | 北京信息科技大学 | Named entity identification method and device for multi-mode cultural relic data |
CN111125367B (en) * | 2019-12-26 | 2023-05-23 | 华南理工大学 | Multi-character relation extraction method based on multi-level attention mechanism |
CN111274814A (en) * | 2019-12-26 | 2020-06-12 | 浙江大学 | Novel semi-supervised text entity information extraction method |
CN111274814B (en) * | 2019-12-26 | 2021-09-24 | 浙江大学 | Novel semi-supervised text entity information extraction method |
CN111125367A (en) * | 2019-12-26 | 2020-05-08 | 华南理工大学 | Multi-character relation extraction method based on multi-level attention mechanism |
CN111143691A (en) * | 2019-12-31 | 2020-05-12 | 四川长虹电器股份有限公司 | Joint information extraction method and device |
CN111241209A (en) * | 2020-01-03 | 2020-06-05 | 北京百度网讯科技有限公司 | Method and apparatus for generating information |
JP7112536B2 (en) | 2020-01-15 | 2022-08-03 | ベイジン バイドゥ ネットコム サイエンス テクノロジー カンパニー リミテッド | Method and apparatus for mining entity attention points in text, electronic device, computer-readable storage medium and computer program |
JP2021111413A (en) * | 2020-01-15 | 2021-08-02 | ベイジン バイドゥ ネットコム サイエンス アンド テクノロジー カンパニー リミテッド | Method and apparatus for mining entity focus in text, electronic device, computer-readable storage medium, and computer program |
US11775761B2 (en) | 2020-01-15 | 2023-10-03 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method and apparatus for mining entity focus in text |
CN111274815A (en) * | 2020-01-15 | 2020-06-12 | 北京百度网讯科技有限公司 | Method and device for mining entity attention points in text |
CN111274815B (en) * | 2020-01-15 | 2024-04-12 | 北京百度网讯科技有限公司 | Method and device for mining entity focus point in text |
CN111275094A (en) * | 2020-01-17 | 2020-06-12 | 厦门快商通科技股份有限公司 | Data labeling method, device and equipment based on machine learning |
CN111310472B (en) * | 2020-01-19 | 2024-02-09 | 合肥讯飞数码科技有限公司 | Alias generation method, device and equipment |
CN111310472A (en) * | 2020-01-19 | 2020-06-19 | 合肥讯飞数码科技有限公司 | Alias generation method, device and equipment |
CN111368528B (en) * | 2020-03-09 | 2022-07-08 | 西南交通大学 | Entity relation joint extraction method for medical texts |
CN111368528A (en) * | 2020-03-09 | 2020-07-03 | 西南交通大学 | Entity relation joint extraction method for medical texts |
CN111488728A (en) * | 2020-03-12 | 2020-08-04 | 天闻数媒科技(北京)有限公司 | Labeling method, device and storage medium for unstructured test question data |
CN111460807A (en) * | 2020-03-13 | 2020-07-28 | 平安科技(深圳)有限公司 | Sequence labeling method and device, computer equipment and storage medium |
CN111460807B (en) * | 2020-03-13 | 2024-03-12 | 平安科技(深圳)有限公司 | Sequence labeling method, device, computer equipment and storage medium |
CN111667238B (en) * | 2020-05-26 | 2022-07-29 | 南开大学 | Recruitment elicitation generation method based on skill perception multi-attention mechanism |
CN111667238A (en) * | 2020-05-26 | 2020-09-15 | 南开大学 | Recruitment inspiring generation method based on skill perception multi-attention mechanism |
CN111640471A (en) * | 2020-05-27 | 2020-09-08 | 牛张明 | Method and system for predicting activity of drug micromolecules based on two-way long-short memory model |
CN113807079A (en) * | 2020-06-11 | 2021-12-17 | 四川大学 | End-to-end entity and relation combined extraction method based on sequence-to-sequence |
CN113807079B (en) * | 2020-06-11 | 2023-06-23 | 四川大学 | Sequence-to-sequence-based end-to-end entity and relationship joint extraction method |
CN111767409A (en) * | 2020-06-14 | 2020-10-13 | 南开大学 | Entity relationship extraction method based on multi-head self-attention mechanism |
CN112163092A (en) * | 2020-10-10 | 2021-01-01 | 成都数之联科技有限公司 | Entity and relation extraction method, system, device and medium |
CN112163092B (en) * | 2020-10-10 | 2022-07-12 | 成都数之联科技股份有限公司 | Entity and relation extraction method, system, device and medium |
CN112597757A (en) * | 2020-12-04 | 2021-04-02 | 光大科技有限公司 | Word detection method and device, storage medium and electronic device |
CN112667820B (en) * | 2020-12-08 | 2023-04-18 | 吉林省吉科软信息技术有限公司 | Deep learning construction method for full-process traceable ecological chain supervision knowledge map |
CN112667820A (en) * | 2020-12-08 | 2021-04-16 | 吉林省吉科软信息技术有限公司 | Deep learning construction method for full-process traceable ecological chain supervision knowledge map |
CN114625871A (en) * | 2020-12-14 | 2022-06-14 | 四川大学 | Triple classification method based on attention position joint coding |
CN114625871B (en) * | 2020-12-14 | 2023-06-23 | 四川大学 | Ternary grouping method based on attention position joint coding |
CN112612871A (en) * | 2020-12-17 | 2021-04-06 | 浙江大学 | Multi-event detection method based on sequence generation model |
CN112612871B (en) * | 2020-12-17 | 2023-09-15 | 浙江大学 | Multi-event detection method based on sequence generation model |
CN112765991A (en) * | 2021-01-14 | 2021-05-07 | 中山大学 | Deep dialogue semantic role labeling method and system based on knowledge enhancement |
CN112765991B (en) * | 2021-01-14 | 2023-10-03 | 中山大学 | Knowledge enhancement-based deep dialogue semantic role labeling method and system |
CN112836482B (en) * | 2021-02-09 | 2024-02-23 | 浙江工商大学 | Method and device for generating problem by sequence generation model based on template |
CN112836482A (en) * | 2021-02-09 | 2021-05-25 | 浙江工商大学 | Method and device for generating problems by sequence generation model based on template |
CN112818124A (en) * | 2021-02-21 | 2021-05-18 | 昆明理工大学 | Entity relationship extraction method based on attention neural network |
CN113553850A (en) * | 2021-03-30 | 2021-10-26 | 电子科技大学 | Entity relation extraction method based on ordered structure encoding pointer network decoding |
CN113033189A (en) * | 2021-04-08 | 2021-06-25 | 北京理工大学 | Semantic coding method of long-short term memory network based on attention dispersion |
CN113191118B (en) * | 2021-05-08 | 2023-07-18 | 山东省计算中心(国家超级计算济南中心) | Text relation extraction method based on sequence annotation |
CN113191118A (en) * | 2021-05-08 | 2021-07-30 | 山东省计算中心(国家超级计算济南中心) | Text relation extraction method based on sequence labeling |
CN113642767A (en) * | 2021-07-09 | 2021-11-12 | 武汉科技大学 | Multi-dimensional feature combination prediction method based on MI-VMD-DA-EDLSTM-VEC |
CN114580408B (en) * | 2022-03-10 | 2024-05-07 | 浙江理工大学 | Method and device for generating downlink based on double-layer attention joint learning |
CN114580408A (en) * | 2022-03-10 | 2022-06-03 | 浙江理工大学 | Lower link generation method and device based on double-layer attention joint learning |
CN115510869A (en) * | 2022-05-30 | 2022-12-23 | 青海师范大学 | End-to-end Tibetan La lattice shallow semantic analysis method |
CN115510869B (en) * | 2022-05-30 | 2023-08-01 | 青海师范大学 | End-to-end Tibetan Lager shallow semantic analysis method |
CN115019893A (en) * | 2022-06-14 | 2022-09-06 | 邵阳学院 | Enhancer identification method based on bidirectional long-and-short-term memory and attention mechanism |
CN116227597A (en) * | 2023-05-05 | 2023-06-06 | 中国人民解放军国防科技大学 | Biomedical knowledge extraction method, device, computer equipment and storage medium |
CN117235286A (en) * | 2023-11-10 | 2023-12-15 | 昆明理工大学 | Attention-strengthening entity relation extraction model, construction method thereof and storage medium |
CN117235286B (en) * | 2023-11-10 | 2024-01-23 | 昆明理工大学 | Attention-strengthening entity relation extraction model, construction method thereof and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109408812A (en) | A method of the sequence labelling joint based on attention mechanism extracts entity relationship | |
CN108460013B (en) | Sequence labeling model and method based on fine-grained word representation model | |
US11194972B1 (en) | Semantic sentiment analysis method fusing in-depth features and time sequence models | |
CN109657239B (en) | Chinese named entity recognition method based on attention mechanism and language model learning | |
CN111241294B (en) | Relationship extraction method of graph convolution network based on dependency analysis and keywords | |
CN107526834B (en) | Word2vec improvement method for training correlation factors of united parts of speech and word order | |
CN109003601A (en) | A kind of across language end-to-end speech recognition methods for low-resource Tujia language | |
CN109800411A (en) | Clinical treatment entity and its attribute extraction method | |
CN111666758B (en) | Chinese word segmentation method, training device and computer readable storage medium | |
CN112784051A (en) | Patent term extraction method | |
CN110083710A (en) | It is a kind of that generation method is defined based on Recognition with Recurrent Neural Network and the word of latent variable structure | |
CN109086269B (en) | Semantic bilingual recognition method based on semantic resource word representation and collocation relationship | |
CN110263325A (en) | Chinese automatic word-cut | |
CN110688862A (en) | Mongolian-Chinese inter-translation method based on transfer learning | |
CN112765952A (en) | Conditional probability combined event extraction method under graph convolution attention mechanism | |
CN112966525B (en) | Law field event extraction method based on pre-training model and convolutional neural network algorithm | |
CN113255320A (en) | Entity relation extraction method and device based on syntax tree and graph attention machine mechanism | |
CN109766546A (en) | A kind of natural language inference method neural network based | |
CN113065349A (en) | Named entity recognition method based on conditional random field | |
CN111340006A (en) | Sign language identification method and system | |
CN113076718B (en) | Commodity attribute extraction method and system | |
Ma et al. | Joint pre-trained Chinese named entity recognition based on bi-directional language model | |
CN116186241A (en) | Event element extraction method and device based on semantic analysis and prompt learning, electronic equipment and storage medium | |
CN109960782A (en) | A kind of Tibetan language segmenting method and device based on deep neural network | |
CN115510230A (en) | Mongolian emotion analysis method based on multi-dimensional feature fusion and comparative reinforcement learning mechanism |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190301 |
|
RJ01 | Rejection of invention patent application after publication |