CN106934042A - A kind of knowledge mapping represents model and its method - Google Patents
A kind of knowledge mapping represents model and its method Download PDFInfo
- Publication number
- CN106934042A CN106934042A CN201710155940.1A CN201710155940A CN106934042A CN 106934042 A CN106934042 A CN 106934042A CN 201710155940 A CN201710155940 A CN 201710155940A CN 106934042 A CN106934042 A CN 106934042A
- Authority
- CN
- China
- Prior art keywords
- entity
- module
- model
- space
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Model and its method are represented the invention discloses a kind of knowledge mapping, it is related to knowledge mapping presentation technology field, the expression model includes entity space module, majorized function module and model training module;The entity space is used for the representation space of presentation-entity feature, and it includes intrinsic state space and mimicry space;The majorized function is used to represent different entities distance after translation that it to include that distance is calculated and weight vectors;The model training module is used for features training and exports training result, and the training result is used to carry out knowledge mapping prediction and classification.Translation model of the present invention based on dynamic representation space, one dynamic representation space is set to each relation, solve the problems, such as to represent that model cannot be distinguished by different relation spaces in the prior art, there is provided more fully reliable method for expressing, and the efficiency that the complexity of model improves algorithm is reduced, preferable effect is achieved in actual applications.
Description
Technical field
The present invention relates to knowledge mapping presentation technology field, and in particular to a kind of knowledge mapping represents model and its method.
Background technology
Current worldwide existing knowledge mapping method for expressing be concentrated mainly on using artificial constructed feature and
Feature based on RDF frame representations.There is inefficiency in these character representation methods, algorithm is complicated in terms of the representation of knowledge is carried out
The problems such as.In recent years, a series of knowledge representation method was proposed using the method for deep learning, but current training is known
How much higher in the presence of some model complexities know method for expressing, or the relatively low problem of training effectiveness.
Other knowledge mapping method for expressing both domestic and external represents sex work mainly includes TransE (based on the embedded of translation
Model) [1], TransH (the embedded model based on hyperplane) [2], TransR (the embedded moulds based on entity relationship space
Type) [3], CTransR (the embedded model based on cluster and entity relationship space) [3] and TransD (are based on dynamic mapping square
The embedded model of battle array) the method above methods such as [4] are collectively referred to as the Knowledge Representation Model based on translation.Mould based on translation
Type thinks, to each triple (h, r, t), relation r therein is a translation of from the beginning entity vector h to tail entity vector t
Operation, accordingly, Bordes et al. takes the lead in proposing TransE (the embedded model based on translation) knowledge representation method, TransE
(based on the embedded model of translation) weighs the semantic similarity between computational entity by the side-play amount in Euclidean distance, is
A kind of simple basic its optimization aims of knowledge representation method are to try to cause h+r=t, thus corresponding model study must
Function is divided to be fr(h, t)=| | h+r-t | |2, wherein | | h+r-t | |2It is the 2 rank norms of h+r-t, i.e. Euclidean distance.TransH
(the embedded model based on hyperplane) method establishes a hyperplane for facing relation, and it is by a normal vector nrAnd translation
Vectorial r represents that head entity vector h and tail entity vector t is projected to the hyperplane of relation first, obtains vectorial h⊥=h-
nr ThnrAnd t⊥=t-nr THnr. thus, TransH (based on hyperplane embedded model) optimization aim be changed into h⊥+ r=t⊥,
Corresponding its scoring function is revised as fr(h, t)=| | h⊥+r-t⊥||2.TransR (the embedded moulds based on entity relationship space
Type) and CTransR (based on cluster and entity relationship space embedded model) wish by setting up an image matrix MrWith
One vector r represents each relation r, and specifically, TransR (the embedded model based on entity relationship space) is by head reality
Body vector h and tail entity vector t are mapped on the level of relation vector r by matrix, obtain MrH+r=MrT, namely TransR
The optimization aim of (the embedded model based on entity relationship space), TransD (the embedded model based on dynamic mapping matrix)
The matrix that be instead of with vector operations in TransR (the embedded model based on entity relationship space) is operated with the multiplication of vector,
Improve efficiency of algorithm.
In actual applications, TransE (the embedded model based on translation) [1] achieves preferable prediction effect.
In TransE (the embedded model based on translation), for each triple (h, r, t), head entity vector h, tail entity vector t
N-dimensional vector h (t) and r are represented as with relation r.Embedded vector t is approximately equal to embedded h plus embedded r, i.e. h+r ≈ t,
TransE (the embedded model based on translation) can well process one-one relationship, but in treatment such as a pair of N, N is to one and N
There is an obvious shortcoming during to the complex relationship of N, specifically, during complex relationship is processed, different realities can be caused
Using identical vector, this does not meet actual conditions to body.TransH (the embedded model based on hyperplane) [2] is by inciting somebody to action
The hyperplane mapping ruler that head entity vector h and tail entity vector t is mapped to relation specificity hyperplane solves complex relationship
Problem.But entity and relation are two kinds of entirely different concepts, therefore it is incorrect to put them on same vector space
's.TransR (the embedded model based on entity relationship space)/CTransR (insertions based on cluster and entity relationship space
Formula model) [4] propose an entity and relation is placed on difference for [3] and TransD (the embedded model based on dynamic mapping matrix)
Two kinds of models of novelty of vector space, for example:Entity space and multirelation space, TransR (are based on entity relationship space
Embedded model) a mapping matrix Mr is set to each relation r, in then mapping entities to relation space with Mr.
In relation space, the entity vector sum relation vector r after being mapped with Mr can construct a gold triple, this triple
It is described as Mrh+r ≈ Mrt.As the extension to TransR (the embedded model based on entity relationship space), CTransR
(the embedded model based on cluster and entity relationship space) is using cluster algorithm to TransE (the embedded model based on translation)
Initial results split, be several subrelation rs by each relation r point.To a certain extent, r is replaced using rs to solve
The ambiguity problem of each relation.TransD (the embedded model based on dynamic mapping matrix) uses two vector ep and hp
It is each entity-relation to constructing dynamic mapping matrix.But TransR (the embedded model based on entity relationship space)/
The algorithm complex of CTransR (the embedded model based on cluster and entity relationship space) is higher, cannot apply in practice.
【1】Bordes A,Usunier N,Garcia-Duran A,et al.Translating embeddings for
modeling multi-relational data[C]//Proc of NIPS.Cambridge,MA:MIT Press,2013:
2787–2795
【2】Wang Zhen,Zhang Jianwen,Feng Jianlin,et al.Knowledge graph
embedding by translating on hyperplanes[C]//Proc of AAAI.Menlo Park,CA:AAAI,
2014:1112–1119
【3】Lin Yankai,Liu Zhiyuan,Sun Maosong,et al.Learning entity and
relation embeddings for knowledge graph completion[C]//Proc of AAAI.Menlo
Park,CA:AAAI,2015
【4】Ji Guoliang,He Shizhu,Xu Liheng,et al.Knowledge graph embedding
via dynamic mapping matrix[C]//Proc of ACL.Stroudsburg PA:ACL,2045:687–696
The content of the invention
A kind of knowledge mapping of the expression ability it is an object of the invention to propose that knowledge mapping can be improved represents model
And its method, its completion to knowledge mapping and verification also play the role of and its important.
In order to realize the purpose of the present invention, the first technical scheme of the invention is specific as follows:
A kind of knowledge spectrogram represents model, and the expression model includes entity space module, majorized function module, Yi Jimo
Type training module;
The entity space module is used for the representation space of presentation-entity feature, and it includes that intrinsic state space and mimicry are empty
Between;
The majorized function module is used to represent different entities distance after translation that it to include that distance is calculated and weight
Vector;
The model training module is used for features training and exports training result, and the training result is used to carry out knowledge graph
Spectrum prediction and classification.
Second technical scheme of the invention is specific as follows:
It is a kind of to build the method that the knowledge spectrogram represents model, the described method comprises the following steps:
1) entity space building process, it uses two vectors come presentation-entity and relation, and described two vectors include this
State vector sum mimicry vector is levied, for describing entity relationship eigenstate, the mimicry vector is for describing for the eigenstate vector
Entity relationship mimicry, the mimicry vector constitutes mimicry matrix, and mimicry vector sum eigenstate vector collectively forms entity space
Characteristic vector;
2) majorized function process, it includes calculating the range formula after head entity is translated with tail entity, using weight vectors
The weight of different dimensions is represented, to reach the purpose of optimization distance computing formula;
3) model training process, it includes the dynamic training of weight vectors and prevents the setting of over-fitting parameter.
3rd technical scheme of the invention is specific as follows:
Knowledge spectrogram represents the implementation of model as elucidated before, and it also includes data acquisition module, pretreatment mould
Block, feature extraction module, training module, knowledge mapping completion module and sort module, the implementation specifically include with
Lower step:
1) data in existing knowledge collection of illustrative plates are extracted using data acquisition module, using distributed reptile system to internet
Present in knowledge carry out distributed collection, and store it in distributed chart database;
2) structuring treatment carried out to the data for extracting using pretreatment module, the pretreatment module is to the number that collects
According to being filtered, be broadly divided into entity relationship duplicate removal, filter out do not meet Description standard entity relationship and filtering exist it is illegal
The part of entity relationship three of character;
3) data after being processed structuring using feature extraction module carry out feature extraction, are included in extraction knowledge mapping
Entity, relation, attribute, and it is described with the form of triple, and using the training module to the feature that extracts
It is trained;
4) by the knowledge mapping completion module and sort module to carry out knowledge mapping using the result for training pre-
Survey and classify, the knowledge mapping completion module and sort module are tested to verify model to the expression model for training
Validity, realize that the entity or relation that are lacked in knowledge mapping are recommended and carry out just existing triple
Whether true judgement.
Compared with prior art, the invention has the advantages that:
1st, translation model of the present invention based on dynamic representation space, a dynamic expression is set to each relation empty
Between, solve the problems, such as to represent that model cannot be distinguished by different relation spaces in the prior art, there is provided more fully reliable to represent
Method, and the efficiency that the complexity of model improves algorithm is reduced, preferable effect is achieved in actual applications.
2nd, the present invention proposes a TransDR model for novelty (the embedded model based on dynamic relationship space), its
For each relation constructs a dynamic relation space, for each relation vector space, it both provides a self adaptation and closes
It is weight;TransDR (the embedded model based on dynamic relationship space) can reduce the noise from other relations, improve not
With the discrimination between relation.
Brief description of the drawings
Fig. 1 is that the knowledge spectrogram in the present invention represents that the implementation of model constitutes structural representation.
Fig. 2 is the basic thought elaboration figure of TransDR models in the present invention.
Fig. 3 is that knowledge mapping represents that the entity space of model builds flow chart in the present invention.
Fig. 4 is that knowledge mapping represents that the majorized function of model builds flow chart in the present invention.
Fig. 5 is that knowledge mapping represents model training flow chart in the present invention.
Fig. 6 is that knowledge mapping represents model workflow diagram in the present invention.
Specific embodiment
The present invention is described in further details below in conjunction with the accompanying drawings.
This specific embodiment represents model and its method for a kind of knowledge mapping, as shown in figure 1, knowledge mapping of the present invention
Represent that model includes data acquisition module, pretreatment module, feature extraction module, knowledge mapping completion module and classification mould
Block.
The data acquisition module carries out distributed adopting using distributed reptile system to knowledge present in internet
Collection, and store it in distributed chart database;The pretreatment module is filtered to the data for collecting, main point
For entity relationship duplicate removal, filter out and do not meet Description standard entity relationship and filtering has the entity relationship three of forbidden character
Point;The feature extraction module is to represent that model is trained by knowledge mapping to the entity for extracting;The knowledge graph
Spectrum completion module and sort module are tested the expression model for training, and have verified that the validity of model.
The training module represents model for knowledge mapping, and it sets up the specific as follows of process:
It is triple (h, r, t) to define G, and wherein h represents an entity, and r represents relation, and t represents tail entity, boldface letter h, r,
T is the embedded expression of (h, r, t).
H=(h1, h2..., hi..., hn);
R=(r1, r2..., ri..., rn);
T=(t1, t2..., ti..., tn);
hi, ri, tiIt is respectively h, the ith feature of r, t, n is the length of feature, Δ represents correct triplet sets,
Δ ' represent incorrect triplet sets, therefore (h, r, t) ∈ Δs represent that triple is correct, (h', r', t') ∈ Δs ' table
Show that triple is incorrect;WrThe embedded expression of expression relation weight, Wr=(wr1, wr2..., wri..., wrn);
Knowledge mapping of the present invention represents that the structure of model specifically includes entity space structure, majorized function, model training.
Entity space builds
Due to potential feature being unevenly distributed in different dimensions of knowledge mapping, therefore it can not be each dimension
Equal weight is set.Then the present invention proposes a more flexible model, i.e. TransDR is (embedding based on dynamic relationship space
Enter formula model), that takes into account different types of relation space avoiding the uneven distribution of different dimensions.As shown in Fig. 2
TransDR (the embedded model based on dynamic relationship space) defines two vectors for each entity h (t), is that each relation r determines
Adopted three vectors, first the first of vector sum r vector representation physical feature of h (t), second vector sum r's of h (t)
Second vector representation is used to construct the surface of mapping matrix, and the relation in the 3rd vector representation each dimension of r is empty
Between weight.Entity is random unordered in entity space, but after being translated in relation space, entity has been reformed into order
's.Although
δi=hi⊥+r-ti⊥,Wherein δiRepresent between correct triple away from
From,Represent the distance between correct triple, but r1(r2) space can effectively recognize the tail entity that makes mistake and correct
Tail entity, specifically as shown in Figure 2.
The present invention comes presentation-entity and relation using two vectors, first eigenstate of vector description entity relationship, separately
One mimicry of vector description entity relationship, uses six vector he, hm, re, rm, te,tmA triple (h, r, t) is represented,
Wherein subscript e and m represent eigenstate and mimicry, and the true state of entity is collectively constituted by mimicry and eigenstate.
Wherein, he,hm,re,rm,te,tm∈Gm。
RmThe entity relationship space of m dimensions is represented, for each triple (h, r, t), the present invention defines two mimicry squares
Battle array Mh、MtHead entity and tail entity mimicry space are represented respectively, to Mh、MtIt is defined as follows:
Mh=rmhm T (1)
Mt=rmtm T (2)
Symmetrical mimicry matrix is constructed by the mimicry vector of entity and relation, therefore for each triple (h, r, t), h
All there is unique mimicry matrix with t.
In fact, the time of day of entity is collectively formed by mimicry and eigenstate, therefore present invention definition is real
Vector is as follows:
H=Mhhe+he (3)
T=Mtte+te (4)
Entity space model construction flow is specifically as shown in Figure 3.
Majorized function
For a gold triple, scoring function fr(h, t) should be relatively low, for an incorrect Huang
Should be able to be higher for golden triple, with L2As a example by norm, traditional scoring function is:
Traditional scoring function calculates the Euclidean distance (ED) between h+r and t, and Euclidean distance is considered often
Individual characteristic dimension is uniformly distributed, but it cannot be distinguished by the trend of different relation spaces, and in order to solve this problem, the present invention makes
ED is substituted with standardized Euclidean distance (SED).
Wherein, the function of SED is:
Characteristic vector before wherein X and X* represent normal state respectively and after normal state, μ and σ is respectively expectation and marks
Quasi- bias vector.SED can eliminate the uneven distribution of feature by normal stateization, therefore the distance between h+r and t are:
Wherein, σiIt is i-th dimensional characteristics of σ, takes Wr=1/ σ2, the present invention one new scoring function of proposition:
It is specifically as shown in Figure 4 that majorized function builds flow.
Model training
In order to obtain the differentiation between gold triple and incorrect triple, present invention definition is below based on limit
Ranking loss function:
Wherein,Δ and Δ ' be respectively positive triple and negative triple set, γ is positive and negative
The distance between sample.Because original triplet sets contains only positive triple, must be obtained by manually generated
Negative triple.
The present invention samples two kinds of strategies to build negative triplet sets using stochastical sampling and Bernoulli Jacob.
Constrained below considering when loss L is minimized:
| | h | |≤1, | | t | |≤1
||Wr| |=1 (7)
Wherein equation (7) ensure that relation weight vectors Wr is constant, is converted into by way of soft-constraint following
Without constraint loss function:
Wherein, λ and η are two hyper parameters to the weighting of soft-constraint importance.
Loss function is trained using improved stochastic gradient descent method (ADADELTA), constraint formula is lacked in formula (8)
(5) and formula (6), conversely, just directly meeting constraint formula (5) and formula (6) before each small lot is accessed;In order to accelerate to receive
Hold back, h is replaced using the result of Trans Ee, te, reRandom starting values, the flow of model training is specifically as shown in Figure 5.
Wherein, knowledge mapping of the present invention represents model workflow as shown in fig. 6, knowledge mapping of the present invention represents model
Implementation comprises the following steps:
1) data in existing knowledge collection of illustrative plates are extracted using data acquisition module, using distributed reptile system to internet
Present in knowledge carry out distributed collection, and store it in distributed chart database;
2) structuring treatment carried out to the data for extracting using pretreatment module, the pretreatment module is to the number that collects
According to being filtered, be broadly divided into entity relationship duplicate removal, filter out do not meet Description standard entity relationship and filtering exist it is illegal
The part of entity relationship three of character;
3) data after being processed structuring using feature extraction module carry out feature extraction, are included in extraction knowledge mapping
Entity, relation, attribute, and it is described with the form of triple, and represent model to taking out using the knowledge mapping
The feature for taking is trained;
4) by the knowledge mapping completion module and sort module to carry out knowledge mapping using the result for training pre-
Survey and classify, the knowledge mapping completion module and sort module are tested to verify model to the expression model for training
Validity, realize that the entity or relation that are lacked in knowledge mapping are recommended and carry out just existing triple
Whether true judgement.
Above example only plays a part of to explain technical solution of the present invention, protection domain of the presently claimed invention not office
It is limited to realize system and specific implementation step described in above-described embodiment.Therefore, only to specific formula in above-described embodiment and
Algorithm is simply replaced, but its substance technical scheme still consistent with the method for the invention, all should belong to this hair
Bright protection domain.
Claims (3)
1. a kind of knowledge spectrogram represents model, it is characterised in that the expression model includes entity space module, majorized function mould
Block and model training module;
The entity space module is used for the representation space of presentation-entity feature, and it includes intrinsic state space and mimicry space;
The majorized function module be used for represent different entities distance after translation, it include distance calculate and weight to
Amount;
The model training module is used for features training and exports training result, and the training result is pre- for carrying out knowledge mapping
Survey and classify.
2. it is a kind of to build the method that knowledge spectrogram as claimed in claim 1 represents model, it is characterised in that methods described bag
Include:
1) entity space building process, it uses two vectors come presentation-entity and relation, and described two vectors include eigenstate
Vector sum mimicry vector, for describing entity relationship eigenstate, the mimicry vector is for describing entity for the eigenstate vector
Relation mimicry, the mimicry vector constitutes mimicry matrix, and mimicry vector sum eigenstate vector collectively forms the feature of entity space
Vector;
2) majorized function process, it is included calculating the range formula after head entity is translated with tail entity, is assigned using weight vectors
The different dimensions of entity are with different weights end to end, to reach the purpose of optimization distance computing formula;
3) model training process, it includes the dynamic training of weight vectors and prevents the setting of over-fitting parameter.
3. a kind of knowledge spectrogram as claimed in claim 1 represents the implementation of model, it is characterised in that it also includes data
Acquisition module, pretreatment module, feature extraction module, knowledge mapping completion module and sort module, the implementation tool
Body is comprised the following steps:
1) data in existing knowledge collection of illustrative plates are extracted using data acquisition module, using distributed reptile system to being deposited in internet
Knowledge carry out distributed collection, and store it in distributed chart database;
2) structuring treatment is carried out to the data for extracting using pretreatment module, the pretreatment module is entered to the data for collecting
Row filtering, be broadly divided into entity relationship duplicate removal, filter out do not meet Description standard entity relationship and filtering there is forbidden character
The part of entity relationship three;
3) data after being processed structuring using feature extraction module carry out feature extraction, extract the reality included in knowledge mapping
Body, relation, attribute, and it is described with the form of triple, and represent model to extraction using the knowledge mapping
Feature is trained;
4) using the result for training by the knowledge mapping completion module and sort module carry out knowledge mapping predict and
Classification, the knowledge mapping completion module and sort module are tested to verify having for model to the expression model for training
Effect property, realize in knowledge mapping lack entity or relation recommended and existing triple is carried out correctly with
No judgement.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710155940.1A CN106934042B (en) | 2017-03-16 | 2017-03-16 | Knowledge graph representation system and implementation method thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710155940.1A CN106934042B (en) | 2017-03-16 | 2017-03-16 | Knowledge graph representation system and implementation method thereof |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106934042A true CN106934042A (en) | 2017-07-07 |
CN106934042B CN106934042B (en) | 2020-05-29 |
Family
ID=59432514
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710155940.1A Active CN106934042B (en) | 2017-03-16 | 2017-03-16 | Knowledge graph representation system and implementation method thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106934042B (en) |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107590237A (en) * | 2017-09-11 | 2018-01-16 | 桂林电子科技大学 | A kind of knowledge mapping based on dynamic translation principle represents learning method |
CN107943874A (en) * | 2017-11-13 | 2018-04-20 | 平安科技(深圳)有限公司 | Knowledge mapping processing method, device, computer equipment and storage medium |
CN108052683A (en) * | 2018-01-22 | 2018-05-18 | 桂林电子科技大学 | A kind of knowledge mapping based on cosine measurement rule represents learning method |
CN108376287A (en) * | 2018-03-02 | 2018-08-07 | 复旦大学 | Multi-valued attribute segmenting device based on CN-DBpedia and method |
CN109063021A (en) * | 2018-07-12 | 2018-12-21 | 浙江大学 | A kind of knowledge mapping distribution representation method for capableing of encoding relation semanteme Diversity structure |
CN109165337A (en) * | 2018-10-17 | 2019-01-08 | 珠海市智图数研信息技术有限公司 | A kind of method and system of knowledge based map construction bidding field association analysis |
CN109189937A (en) * | 2018-08-22 | 2019-01-11 | 阿里巴巴集团控股有限公司 | A kind of characteristic relation recommended method and device, a kind of calculating equipment and storage medium |
CN109213872A (en) * | 2018-09-11 | 2019-01-15 | 中国电子科技集团公司第二十八研究所 | Knowledge based indicates the entity relationship prediction technique and forecasting system of study |
CN109241290A (en) * | 2017-07-10 | 2019-01-18 | 华东师范大学 | A kind of knowledge mapping complementing method, device and storage medium |
CN109614501A (en) * | 2018-12-13 | 2019-04-12 | 浙江工商大学 | A kind of industrial hidden danger standardization report method and system of knowledge based map |
CN109753571A (en) * | 2019-01-11 | 2019-05-14 | 中山大学 | A kind of scene map lower dimensional space embedding grammar based on secondary theme space projection |
CN109922075A (en) * | 2019-03-22 | 2019-06-21 | 中国南方电网有限责任公司 | Network security knowledge map construction method and apparatus, computer equipment |
CN110019766A (en) * | 2017-09-25 | 2019-07-16 | 腾讯科技(深圳)有限公司 | Methods of exhibiting, device, mobile terminal and the readable storage medium storing program for executing of knowledge mapping |
CN110275966A (en) * | 2019-07-01 | 2019-09-24 | 科大讯飞(苏州)科技有限公司 | A kind of Knowledge Extraction Method and device |
CN111339321A (en) * | 2020-05-18 | 2020-06-26 | 中国人民解放军国防科技大学 | Triple authenticity detection method and device in knowledge graph |
CN111392538A (en) * | 2020-03-17 | 2020-07-10 | 浙江新再灵科技股份有限公司 | Elevator comprehensive fault early warning method based on multi-dimensional Internet of things atlas big data |
CN111680109A (en) * | 2020-04-22 | 2020-09-18 | 北京三快在线科技有限公司 | Knowledge graph representation learning model training method and device and electronic equipment |
CN112380298A (en) * | 2020-12-03 | 2021-02-19 | 成都航天科工大数据研究院有限公司 | Knowledge graph construction method and system for oil and gas pipeline defect repair |
CN112784059A (en) * | 2021-01-20 | 2021-05-11 | 和美(深圳)信息技术股份有限公司 | Knowledge graph representation learning method and device, electronic device and storage medium |
CN113033914A (en) * | 2021-04-16 | 2021-06-25 | 哈尔滨工业大学 | Entity and relation prediction method for machining process knowledge graph |
CN116705338A (en) * | 2023-08-08 | 2023-09-05 | 中国中医科学院中医药信息研究所 | Traditional Chinese medicine multi-mode knowledge graph reasoning method and device based on rules and paths |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7319998B2 (en) * | 2003-11-14 | 2008-01-15 | Universidade De Coimbra | Method and system for supporting symbolic serendipity |
CN103207380A (en) * | 2013-03-12 | 2013-07-17 | 西安电子科技大学 | Broadband target direction finding method based on two-dimensional frequency domain sparse constraint |
CN105630901A (en) * | 2015-12-21 | 2016-06-01 | 清华大学 | Knowledge graph representation learning method |
CN106156365A (en) * | 2016-08-03 | 2016-11-23 | 北京智能管家科技有限公司 | A kind of generation method and device of knowledge mapping |
-
2017
- 2017-03-16 CN CN201710155940.1A patent/CN106934042B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7319998B2 (en) * | 2003-11-14 | 2008-01-15 | Universidade De Coimbra | Method and system for supporting symbolic serendipity |
CN103207380A (en) * | 2013-03-12 | 2013-07-17 | 西安电子科技大学 | Broadband target direction finding method based on two-dimensional frequency domain sparse constraint |
CN105630901A (en) * | 2015-12-21 | 2016-06-01 | 清华大学 | Knowledge graph representation learning method |
CN106156365A (en) * | 2016-08-03 | 2016-11-23 | 北京智能管家科技有限公司 | A kind of generation method and device of knowledge mapping |
Non-Patent Citations (1)
Title |
---|
刘峤等: ""知识图谱构建技术综述"", 《计算机研究与发展》 * |
Cited By (32)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109241290A (en) * | 2017-07-10 | 2019-01-18 | 华东师范大学 | A kind of knowledge mapping complementing method, device and storage medium |
CN107590237A (en) * | 2017-09-11 | 2018-01-16 | 桂林电子科技大学 | A kind of knowledge mapping based on dynamic translation principle represents learning method |
CN107590237B (en) * | 2017-09-11 | 2021-04-30 | 桂林电子科技大学 | Knowledge graph representation learning method based on dynamic translation principle |
CN110019766B (en) * | 2017-09-25 | 2023-01-13 | 腾讯科技(深圳)有限公司 | Knowledge graph display method and device, mobile terminal and readable storage medium |
CN110019766A (en) * | 2017-09-25 | 2019-07-16 | 腾讯科技(深圳)有限公司 | Methods of exhibiting, device, mobile terminal and the readable storage medium storing program for executing of knowledge mapping |
CN107943874A (en) * | 2017-11-13 | 2018-04-20 | 平安科技(深圳)有限公司 | Knowledge mapping processing method, device, computer equipment and storage medium |
WO2019091019A1 (en) * | 2017-11-13 | 2019-05-16 | 平安科技(深圳)有限公司 | Knowledge graph processing method and device, computer device and computer storage medium |
CN108052683A (en) * | 2018-01-22 | 2018-05-18 | 桂林电子科技大学 | A kind of knowledge mapping based on cosine measurement rule represents learning method |
CN108052683B (en) * | 2018-01-22 | 2021-08-03 | 桂林电子科技大学 | Knowledge graph representation learning method based on cosine measurement rule |
CN108376287A (en) * | 2018-03-02 | 2018-08-07 | 复旦大学 | Multi-valued attribute segmenting device based on CN-DBpedia and method |
CN109063021A (en) * | 2018-07-12 | 2018-12-21 | 浙江大学 | A kind of knowledge mapping distribution representation method for capableing of encoding relation semanteme Diversity structure |
CN109063021B (en) * | 2018-07-12 | 2021-10-15 | 浙江大学 | Knowledge graph distributed expression method capable of coding relation semantic diversity structure |
CN109189937A (en) * | 2018-08-22 | 2019-01-11 | 阿里巴巴集团控股有限公司 | A kind of characteristic relation recommended method and device, a kind of calculating equipment and storage medium |
US11244232B2 (en) | 2018-08-22 | 2022-02-08 | Advanced New Technologies Co., Ltd. | Feature relationship recommendation method, apparatus, computing device, and storage medium |
CN109213872A (en) * | 2018-09-11 | 2019-01-15 | 中国电子科技集团公司第二十八研究所 | Knowledge based indicates the entity relationship prediction technique and forecasting system of study |
CN109165337B (en) * | 2018-10-17 | 2021-10-15 | 珠海市智图数研信息技术有限公司 | Method and system for establishing bid and ask field association analysis based on knowledge graph |
CN109165337A (en) * | 2018-10-17 | 2019-01-08 | 珠海市智图数研信息技术有限公司 | A kind of method and system of knowledge based map construction bidding field association analysis |
CN109614501A (en) * | 2018-12-13 | 2019-04-12 | 浙江工商大学 | A kind of industrial hidden danger standardization report method and system of knowledge based map |
CN109753571A (en) * | 2019-01-11 | 2019-05-14 | 中山大学 | A kind of scene map lower dimensional space embedding grammar based on secondary theme space projection |
CN109753571B (en) * | 2019-01-11 | 2022-04-19 | 中山大学 | Scene map low-dimensional space embedding method based on secondary theme space projection |
CN109922075A (en) * | 2019-03-22 | 2019-06-21 | 中国南方电网有限责任公司 | Network security knowledge map construction method and apparatus, computer equipment |
CN110275966A (en) * | 2019-07-01 | 2019-09-24 | 科大讯飞(苏州)科技有限公司 | A kind of Knowledge Extraction Method and device |
CN111392538A (en) * | 2020-03-17 | 2020-07-10 | 浙江新再灵科技股份有限公司 | Elevator comprehensive fault early warning method based on multi-dimensional Internet of things atlas big data |
CN111680109A (en) * | 2020-04-22 | 2020-09-18 | 北京三快在线科技有限公司 | Knowledge graph representation learning model training method and device and electronic equipment |
CN111680109B (en) * | 2020-04-22 | 2024-03-29 | 北京三快在线科技有限公司 | Knowledge graph representation learning model training method and device and electronic equipment |
CN111339321A (en) * | 2020-05-18 | 2020-06-26 | 中国人民解放军国防科技大学 | Triple authenticity detection method and device in knowledge graph |
CN111339321B (en) * | 2020-05-18 | 2020-08-21 | 中国人民解放军国防科技大学 | Triple authenticity detection method and device in knowledge graph |
CN112380298A (en) * | 2020-12-03 | 2021-02-19 | 成都航天科工大数据研究院有限公司 | Knowledge graph construction method and system for oil and gas pipeline defect repair |
CN112784059A (en) * | 2021-01-20 | 2021-05-11 | 和美(深圳)信息技术股份有限公司 | Knowledge graph representation learning method and device, electronic device and storage medium |
CN113033914A (en) * | 2021-04-16 | 2021-06-25 | 哈尔滨工业大学 | Entity and relation prediction method for machining process knowledge graph |
CN116705338A (en) * | 2023-08-08 | 2023-09-05 | 中国中医科学院中医药信息研究所 | Traditional Chinese medicine multi-mode knowledge graph reasoning method and device based on rules and paths |
CN116705338B (en) * | 2023-08-08 | 2023-12-08 | 中国中医科学院中医药信息研究所 | Traditional Chinese medicine multi-mode knowledge graph reasoning method and device based on rules and paths |
Also Published As
Publication number | Publication date |
---|---|
CN106934042B (en) | 2020-05-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106934042A (en) | A kind of knowledge mapping represents model and its method | |
CN106951499B (en) | A kind of knowledge mapping representation method based on translation model | |
CN107122375B (en) | Image subject identification method based on image features | |
CN109522956B (en) | Low-rank discriminant feature subspace learning method | |
CN103262118B (en) | Attribute value estimation device and property value method of estimation | |
CN108171318B (en) | Convolution neural network integration method based on simulated annealing-Gaussian function | |
CN107330355B (en) | Deep pedestrian re-identification method based on positive sample balance constraint | |
CN110826618A (en) | Personal credit risk assessment method based on random forest | |
CN111611877B (en) | Anti-age-interference face recognition method based on multi-time-space information fusion | |
CN112001422B (en) | Image mark estimation method based on deep Bayesian learning | |
CN114332568A (en) | Training method, system, equipment and storage medium of domain adaptive image classification network | |
CN110084149A (en) | A kind of face verification method based on difficult sample four-tuple dynamic boundary loss function | |
CN107133640A (en) | Image classification method based on topography's block description and Fei Sheer vectors | |
CN106951914A (en) | The Electronic Nose that a kind of Optimization of Fuzzy discriminant vectorses are extracted differentiates vinegar kind method | |
CN107220320B (en) | Emerging technology identification method based on patent citation | |
CN112084944A (en) | Method and system for identifying dynamically evolved expressions | |
CN111666984A (en) | Intelligent overvoltage identification method based on transfer learning | |
CN111428803A (en) | Wasserstein distance-based depth domain adaptive image classification method | |
CN103927540B (en) | A kind of invariant feature extraction method based on biological vision hierarchical mode | |
CN114881029B (en) | Chinese text readability evaluation method based on hybrid neural network | |
CN110502883A (en) | A kind of keystroke abnormal behavior detection method based on PCA | |
CN113505692B (en) | Face recognition method based on partial area optimization under working characteristic curve of subject | |
CN106650678A (en) | Face identification method through utilization of Gabor wavelet sub-band related structure | |
CN111950333B (en) | Electronic handwritten signature recognition method based on neural network | |
CN108629356A (en) | A kind of date storage method and device towards power load classification application |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |