CN110796254A - Knowledge graph reasoning method and device, computer equipment and storage medium - Google Patents
Knowledge graph reasoning method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110796254A CN110796254A CN201911044553.6A CN201911044553A CN110796254A CN 110796254 A CN110796254 A CN 110796254A CN 201911044553 A CN201911044553 A CN 201911044553A CN 110796254 A CN110796254 A CN 110796254A
- Authority
- CN
- China
- Prior art keywords
- vector
- knowledge
- knowledge graph
- entity
- graph
- 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
- 238000000034 method Methods 0.000 title claims abstract description 29
- 239000013598 vector Substances 0.000 claims abstract description 137
- 230000006870 function Effects 0.000 claims description 61
- 238000004590 computer program Methods 0.000 claims description 9
- 238000006467 substitution reaction Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 4
- 230000008520 organization Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000014616 translation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Databases & Information Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a knowledge graph reasoning method, a knowledge graph reasoning device, computer equipment and a storage medium. The method comprises the following steps: the knowledge inference platform infers a data set in the knowledge graph by using random gradient descent according to the loss function to obtain vector representations respectively corresponding to entities and relations in the data set; and the knowledge reasoning platform completes the knowledge graph by combining the scoring function according to the obtained entity vector and the relation vector. According to the knowledge graph reasoning method, the knowledge graph reasoning device, the computer equipment and the storage medium, the effect of the path with high support degree on the reasoning result is emphasized while considering the information of a plurality of paths between the entities, so that the reasoning result can be directly determined without considering other paths under the condition that the support degree of a certain path on the reasoning result is 100%, and the accuracy and the reasonability of the knowledge graph reasoning are improved.
Description
Technical Field
The invention relates to the field of natural language reasoning, in particular to a knowledge graph reasoning method, a knowledge graph reasoning device, computer equipment and a storage medium.
Background
With the development of the internet, the content of the network data presents an explosive growth situation. The characteristics of large scale, heterogeneous and multiple internet contents, loose organization structure and the like provide great challenges for people to effectively acquire information and knowledge. The Knowledge Graph (Knowledge Graph) lays a foundation for the intellectual organization and intelligent application of the internet era by virtue of the strong semantic reasoning ability and open organization ability of the Knowledge Graph. In a knowledge graph, these facts are typically represented as triples (head, relationship, tail). However, the coverage of these knowledge maps is far from sufficient compared to the objective world. Therefore, the completion of the knowledge-graph, i.e., the reasoning of knowledge, becomes an important part of the knowledge-graph research.
In recent years, methods of projecting entities and relationships of a knowledge graph to a multidimensional vector space to learn potential attributes of the entities and relationships have attracted extensive attention. Compared with the prior work, the knowledge embedding models have the advantages of low complexity, high reusability and the like. Typical of this type of method, transform-based Embedded, is a classical neural network-based model that represents entities and relationships as vectors by optimizing a boundary-based loss function, considers relationships as translations of head entity vectors to tail entity vectors, and uses a scoring function to evaluate the rationality of the existence of triples. TransH (transforming on Hyperplants) and TransR are representative variants of TransE that consider an entity from multiple aspects, with different relationships potentially emphasizing different aspects. However, with the improvement of TransE and the improvement of algorithm performance, the complexity of the algorithm is increased.
In recent research, a path inference algorithm PRA is introduced on the basis of an embedded model, and represents a multi-step relationship path between entities as a vector obtained through vector operation of the relationship contained in the path, so that the two methods are combined, the implicit information of the relationship path is considered on the basis of the embedded model, and the knowledge inference effect is greatly improved. PTransE (Path-based transform) and RPE (Path Ranking Algorithm) are typical of them, and the scoring function of these algorithms is divided into two parts, one part is the scoring function of the embedded model, i.e., the TransE or a variant of the TransE, and the other part is Path-based reasoning, and triple scoring is performed by comprehensively considering the relevant paths between the target entity pairs. However, it is not reasonable to say that these algorithms simply obtain final scores by weighted averaging in the path inference part, and neglect the situation that a single path plays a decisive role in the path inference.
Disclosure of Invention
In order to solve the problems, the invention provides a knowledge graph reasoning method which is characterized by comprising the following steps:
the knowledge inference platform infers a data set in the knowledge graph by using random gradient descent according to the loss function to obtain vector representations respectively corresponding to entities and relations in the data set, and the method specifically comprises the following steps:
embedding and converting the triple set in the knowledge graph into a low-dimensional space, namely generating a vector for each entity and relation in the data set, wherein the dimension of the vector is set by self;
extracting vector representations (h, r, t) of a specified number of triples from the dataset, and generating a corresponding vector representation (h ', r', t ') of a negative triplet for the vector representation (h, r, t) of each triplet by randomly replacing at least one of the entity vector and the relationship vector of the triplet, wherein h is a head entity vector, r is a relationship vector, t is a tail entity vector, h' is a head entity vector replacing the head entity vector h, r 'is a relationship vector replacing the relationship vector r, and t' is a tail entity vector replacing the tail entity vector t;
establishing a sum of loss functions according to the positive triples and the negative triples;
calculating the gradient of the sum of the loss functions, and updating the vectors corresponding to the entities and the relations according to the decrease of the random gradient;
repeating the steps until the sum of the loss functions is converged to obtain a final entity vector and a final relation vector;
the knowledge reasoning platform completes the knowledge graph by combining the scoring function according to the obtained entity vector and the relation vector, and the method specifically comprises the following steps:
randomly combining entity vectors and relationship vectors from a dataset of a knowledge graph to generate a vector representation of a new triplet;
establishing a scoring function, and calculating the score of the new triple;
judging the possibility of existence of the triples according to the scores, and adding the triples with high possibility of existence into the knowledge graph.
Further, the sum of the loss functions is specifically:
wherein,
wherein, L (h, r, t) represents a loss function of scores of the triples themselves, L (p, r) represents a loss function of scores of the paths, and L (p) represents a probability that the similarity of the paths p and the relation r is close to the relation deduced by the paths through constraint; γ is a boundary value that distinguishes the scores of a positive triplet from those of a negative triplet;
indicates the similarity of paths and relationships, [ x ]]+Max (0, x), P (r | P) represents the probability that a relationship is inferred by a path, obtained by traversing the knowledge-graph.
Further, the scoring function S (h, r, t) is specifically:
wherein (1- | II)p∈P(h,t)(1-sim (p, r)) is a path inference based score; sim (h + r, t) is a score based on the TransE (Embedded model) variants; mu is a weight parameter used for adjusting the proportion of the two scoring modes in the scoring function; p (h, t) represents the set of all paths P from the head entity vector h to the tail entity vector t, sim (P, r) represents the cosine similarity by path vector P and relationship vector r;
a knowledge-graph inference apparatus is also presented, the apparatus comprising:
the updating module is used for reasoning the data set in the knowledge graph by using random gradient descent according to the loss function, updating the vectors corresponding to the entities and the relations in the data set and obtaining the final entity vectors and the relation vectors;
the improvement module is used for calculating a scoring function and improving the knowledge graph through the scoring function;
further, the update module includes:
the first generation unit is used for generating corresponding entity vectors and relation vectors from the entities and the relations in the data sets of the knowledge graph; for obtaining a vector representation corresponding to a negative triplet corresponding to the vector representation of the extracted triplet;
the loss function establishing unit is used for establishing a loss function according to the vector representation of the current triple and the vector representation of the corresponding negative triple;
and the updating unit is used for calculating the gradient of the loss function and updating the entity and the vector corresponding to the relation according to the descending of the random gradient.
Further, the perfection module comprises:
the second generation unit randomly combines the entity vector and the relation vector from the data set of the knowledge graph to generate a new vector representation of the triple;
the scoring function establishing unit is used for calculating the scoring of the new triple;
and the perfecting unit is used for perfecting the knowledge graph according to the scores of the new triples.
A computer arrangement is also proposed, comprising a memory, an reasoner and a computer program stored on the memory and executable on the reasoner, the reasoner executing the computer program to implement the steps of the above-mentioned method of knowledge-graph inference.
A storage medium is also proposed, on which a computer program is stored which, when executed by an reasoner, implements the steps in the above-described knowledge-graph inference method.
Compared with the prior art, the invention has the beneficial effects that:
the invention emphasizes the function of the path with high support degree on the inference result while considering the information of a plurality of paths between the entities, so that the inference result can be directly determined without considering other paths under the condition that the support degree of a certain path on the inference result is 100 percent, and the accuracy and the rationality of the inference on the knowledge graph are improved.
Drawings
Fig. 1 is an application scenario diagram of the knowledge-graph inference method according to the first embodiment.
Fig. 2 is a flowchart of the knowledge-graph inference method of the second embodiment.
Fig. 3 is a schematic structural diagram of a knowledge-graph inference apparatus according to a third embodiment.
Detailed Description
The present invention is described in further detail below with reference to the attached drawing figures.
In this disclosure, aspects of the present invention are described with reference to the accompanying drawings, in which a number of illustrative embodiments are shown. Embodiments of the present disclosure are not necessarily intended to include all aspects of the invention. It should be appreciated that the various concepts and embodiments described above, as well as those described in greater detail below, may be implemented in any of numerous ways, as the disclosed concepts and embodiments are not limited to any one implementation. In addition, some aspects of the present disclosure may be used alone, or in any suitable combination with other aspects of the present disclosure.
Example one
FIG. 1 is a diagram of an application scenario of a knowledge-graph inference method, including a knowledge-graph inference platform and a knowledge-graph storage database, where an initial established knowledge-graph is stored in the knowledge-graph storage database, the knowledge-graph inference platform can obtain the initial established knowledge-graph from the knowledge-graph storage database, and the knowledge-graph inference platform uses random gradient descent to infer a data set in the knowledge-graph according to a loss function to obtain vector representations of entities and relationships in the data set; and the knowledge graph reasoning platform completes the knowledge graph by combining the scoring function according to the obtained entity vector and the relation vector. The improved knowledge graph can be stored in a knowledge graph storage database, a user can send a knowledge graph query request to the knowledge graph storage database through a knowledge graph reasoning platform, and then the knowledge graph can be checked through a display interface of the knowledge graph platform to systematically express the relationship between data.
Example two
Fig. 2 is a flowchart of a knowledge-graph inference method, which is applied to the knowledge-graph inference platform in fig. 1, and a knowledge-graph inference program is run on the platform, and the knowledge-graph inference program implements the knowledge-graph inference. The method comprises the following steps:
s1: the knowledge inference platform infers a data set in the knowledge graph by using random gradient descent according to the loss function to obtain vector representation of entities and relations in the data set, and the method specifically comprises the following steps:
s11: embedding and converting the triple set in the knowledge graph into a low-dimensional space, namely generating a vector for each entity and relation in the data set, wherein the dimension of the vector is set by self;
s12: extracting vector representations (h, r, t) of a specified number of triples from the data set, generating a corresponding vector representation (h ', r', t ') of a negative triplet for each triplet by randomly replacing at least one of the entity vector and the relationship vector of the triplet, wherein h is a head entity vector, r is a relationship vector, t is a tail entity vector, h' is a head entity vector after immediate replacement, r 'is a relationship vector after immediate replacement, t'
The tail entity vector after the random substitution is obtained;
s13: calculate the sum of the loss functions:
wherein,
l (h, r, t) represents a loss function of scores for the triples themselves, L (p, r) represents a loss function of scores for the paths, and L (p) represents a probability that the similarity of the paths and the relationships is close to the relationship inferred by the paths by the constraints. γ is a boundary value that distinguishes the scores of a positive triplet from those of a negative triplet. sim (h + r, t) is a score based on the TransE (Embedded model) variants; sim (h ' + r ', t ') is the TransE (Embedded model) variant-based score after three random substitutions of h, r, t;
representing the similarity of the path vector p and the relationship vector r,
representing the similarity of the path vector p 'and the relation vector r' after random replacement; [ x ] of]+Max (0, x), P (r | P) represents the probability of reasoning out a relationship from a path, obtained by traversing the knowledge-graph; s is the set of knowledge-graphs in the correct triples, S-The value { (h ', r, t) ∪ (h, r ', t) ∪ (h, r, t ') } is obtained by randomly replacing at least one of h, r, t of a triplet to obtain a set of error triplets, and P (h, t) represents all paths from h to tAnd (4) collecting.
S14: calculating the gradient of the loss function, and updating the vector corresponding to the entity and the relation according to the decrease of the random gradient;
s15: repeating S12 to S14 until the sum of the loss functions is converged to obtain a final entity vector and a final relation vector;
s2: the knowledge reasoning platform completes the knowledge graph by combining the scoring function according to the obtained entity vector and the relation vector, and the method specifically comprises the following steps:
s21: randomly combining entity vectors and relationship vectors from a dataset of a knowledge graph to generate a vector representation of a new triplet;
s22: calculating the score of the new triple, wherein the score is obtained by a score function S (h, r, t), and the score is specifically as follows:
wherein (1- | II)p∈P(h,t)(1-sim (p, r)) is a path inference based score; sim (h + r, t) is a score based on the TransE (Embedded model) variants; mu is a weight parameter used for adjusting the proportion of the two scoring modes in the scoring function; p (h, t) represents the set of all paths P from the head entity vector h to the tail entity vector t, sim (P, r) represents the cosine similarity by the path vector P and the relation vector r.
S23: judging the possibility of existence of the triples according to the scores, and adding the triples with high possibility of existence into the knowledge graph.
EXAMPLE III
Fig. 3 is a schematic diagram of a knowledge-graph inference apparatus, which includes:
the updating module is configured to perform inference on a data set in the knowledge graph by using random gradient descent according to the loss function, update vectors corresponding to the entities and the relationships in the data set, and obtain a final entity vector and a final relationship vector, and specifically includes:
the first generation unit is used for generating corresponding entity vectors and relation vectors from the entities and the relations in the data sets of the knowledge graph; for obtaining a vector representation corresponding to a negative triplet corresponding to the vector representation of the extracted triplet;
the loss function establishing unit is used for establishing a loss function according to the vector representation of the current triple and the vector representation of the corresponding negative triple;
the updating unit is used for calculating the gradient of the loss function and updating the entity and the vector corresponding to the relation according to the decrease of the random gradient;
the perfecting module is used for calculating a scoring function and perfecting the knowledge graph through the scoring function, and specifically comprises the following steps:
the second generation unit randomly combines the entity vector and the relation vector from the data set of the knowledge graph to generate a new vector representation of the triple;
the scoring function establishing unit is used for establishing a scoring function and calculating the score of the new triple;
and the perfecting unit is used for perfecting the knowledge graph according to the scores of the new triples.
The above specific limitations on the knowledge graph inference apparatus can be referred to the limitations on the knowledge graph inference method, and are not described herein again. The modules in the knowledge-graph inference apparatus can be wholly or partially implemented by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a reasoner in computer equipment, and can also be stored in a memory in the computer equipment in a software form, so that the reasoner can call and execute operations corresponding to the modules. The reasoner can be a central inference unit (CPU), a micro-reasoner, a single chip microcomputer and the like. The above-described knowledge-graph inference apparatus may be implemented in the form of a computer program that is executable on an event knowledge-graph inference platform as shown in fig. 1.
Example four
A computer device for performing the knowledge-graph building, which may be a conventional terminal or any other suitable computer device, including a memory, an inference engine, an operating system, a database, and a knowledge-graph inference program stored on the memory and operable on the inference engine, wherein the memory may include an internal memory providing a cached operating environment for the operating system, the database, and the computer-executable program in a non-volatile storage medium, and the inference engine implements the steps of embodiment two when executing the knowledge-graph building program.
EXAMPLE five
A storage medium having stored thereon a computer program which, when executed by an reasoner, performs the steps of embodiment two.
The present invention has application to, but is not limited to, search engines, expert systems, web searches, and knowledge question answering.
The invention emphasizes the function of the path with high support degree on the inference result while considering the information of a plurality of paths between the entities, so that the inference result can be directly determined without considering other paths under the condition that the support degree of a certain path on the inference result is 100 percent, and the accuracy and the rationality of the inference on the knowledge graph are improved.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (8)
1. A knowledge graph reasoning method is characterized by comprising the following steps:
the knowledge inference platform infers a data set in the knowledge graph by using random gradient descent according to the loss function to obtain vector representations respectively corresponding to entities and relations in the data set, and the method specifically comprises the following steps:
embedding and converting the triple set in the knowledge graph into a low-dimensional space, namely generating a vector for each entity and relation in the data set, wherein the dimension of the vector is set by self;
extracting vector representations (h, r, t) of a specified number of triples from the data set, and generating a corresponding vector representation (h ', r', t ') of a negative triplet for the vector representation (h, r, t) of each triplet by randomly replacing at least one of the entity vectors and the relationship vectors of the triples, wherein h is a head entity vector, r is a relationship vector, t is a tail entity vector, h' is a head entity vector after immediate replacement, r 'is a relationship vector after immediate replacement, and t' is a tail entity vector after immediate replacement;
establishing a sum of loss functions according to the positive triples and the negative triples;
calculating the gradient of the sum of the loss functions, and updating the vectors corresponding to the entities and the relations according to the decrease of the random gradient;
repeating the steps until the sum of the loss functions is converged to obtain a final entity vector and a final relation vector;
the knowledge reasoning platform completes the knowledge graph by combining the scoring function according to the obtained entity vector and the relation vector, and the method specifically comprises the following steps:
randomly combining entity vectors and relationship vectors from a dataset of a knowledge graph to generate a vector representation of a new triplet;
establishing a scoring function, and calculating the score of the new triple;
judging the possibility of existence of the triples according to the scores, and adding the triples with high possibility of existence into the knowledge graph.
2. The method of knowledge-graph inference according to claim 1, characterized in that said sum of loss functions is specifically:
wherein,
wherein the ratio of L (h, r,t) represents the loss function of the score of the triplet itself, L (p, r) represents the loss function of the score of the path, L (p) represents the probability of approximating the similarity of the path p and the relation r to the relation deduced by the path through constraint; γ is a boundary value that distinguishes the scores of a positive triplet from those of a negative triplet; sim (h + r, t) is a score based on the TransE (Embedded model) variants; sim (h ' + r ', t ') is the TransE (Embedded model) variant-based score after three random substitutions of h, r, t;
representing the similarity of the path vector p and the relationship vector r,
representing the similarity of the path vector p 'and the relation vector r' after random replacement; [ x ] of]+Max (0, x), P (r | P) represents the probability of reasoning out a relationship from a path, obtained by traversing the knowledge-graph; s is the set of knowledge-graphs in the correct triples, S-The value { (h ', r, t) ∪ (h, r ', t) ∪ (h, r, t ') } is the set of error triples obtained by randomly replacing at least one of the triples h, r, t, and P (h, t) represents the set of all paths from h to t.
3. The knowledgegraph inference method according to claim 1, characterised in that the scoring function S (h, r, t) is specifically:
wherein (1-pi)p∈P(h，t)(1-sim (p, r)) is a path inference based score; mu is a weight parameter used for adjusting the proportion of the two scoring modes in the scoring function; sim (p, r) represents the similarity between the path vector p and the relationship vector r.
4. A knowledge-graph reasoning apparatus, the apparatus comprising:
the updating module is used for reasoning the data set in the knowledge graph by using random gradient descent according to the loss function, updating the vectors corresponding to the entities and the relations in the data set and obtaining the final entity vectors and the relation vectors;
and the perfecting module is used for calculating a scoring function and perfecting the knowledge graph through the scoring function.
5. The knowledgegraph inference device according to claim 4, wherein said update module comprises:
the first generation unit is used for generating corresponding entity vectors and relation vectors from the entities and the relations in the data sets of the knowledge graph; for obtaining a vector representation corresponding to a negative triplet corresponding to the vector representation of the extracted triplet;
the loss function establishing unit is used for establishing a loss function according to the vector representation of the current triple and the vector representation of the corresponding negative triple;
and the updating unit is used for calculating the gradient of the loss function and updating the entity and the vector corresponding to the relation according to the descending of the random gradient.
6. The knowledgegraph inference device according to claim 5, wherein said perfecting module comprises:
the second generation unit randomly combines the entity vector and the relation vector from the data set of the knowledge graph to generate a new vector representation of the triple;
the scoring function establishing unit is used for calculating the scoring of the new triple;
and the perfecting unit is used for perfecting the knowledge graph according to the scores of the new triples.
7. A computer device comprising a memory, a reasoner, and a computer program stored on the memory and executable on the reasoner, the reasoner executing the computer program to perform the steps of the method of any one of claims 1-3.
8. A storage medium having stored thereon a computer program, characterized in that the computer program, when executed by an reasoner, carries out the steps of the method according to any one of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911044553.6A CN110796254B (en) | 2019-10-30 | 2019-10-30 | Knowledge graph reasoning method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911044553.6A CN110796254B (en) | 2019-10-30 | 2019-10-30 | Knowledge graph reasoning method and device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110796254A true CN110796254A (en) | 2020-02-14 |
| CN110796254B CN110796254B (en) | 2024-02-27 |
Family
ID=69442081
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911044553.6A Active CN110796254B (en) | 2019-10-30 | 2019-10-30 | Knowledge graph reasoning method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110796254B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111581343A (en) * | 2020-04-24 | 2020-08-25 | 北京航空航天大学 | Reinforced learning knowledge graph reasoning method and device based on graph convolution neural network |
| CN112069249A (en) * | 2020-10-14 | 2020-12-11 | 中国平安人寿保险股份有限公司 | Knowledge graph relation mining method and device, computer equipment and storage medium |
| CN112596031A (en) * | 2020-12-22 | 2021-04-02 | 电子科技大学 | Target radar threat degree assessment method based on knowledge graph |
| CN112818137A (en) * | 2021-04-19 | 2021-05-18 | 中国科学院自动化研究所 | Entity alignment-based multi-source heterogeneous knowledge graph collaborative reasoning method and device |
| CN112819164A (en) * | 2021-02-02 | 2021-05-18 | 京东数科海益信息科技有限公司 | Inference method and device of affair map and computer equipment |
| CN113128689A (en) * | 2021-04-27 | 2021-07-16 | 中国电力科学研究院有限公司 | Entity relationship path reasoning method and system for regulating knowledge graph |
| CN113220833A (en) * | 2021-05-07 | 2021-08-06 | 支付宝(杭州)信息技术有限公司 | Entity association degree identification method and device |
| CN113672741A (en) * | 2021-08-19 | 2021-11-19 | 支付宝(杭州)信息技术有限公司 | Information processing method, device and equipment |
| CN113688249A (en) * | 2020-05-18 | 2021-11-23 | 第四范式（北京）技术有限公司 | Knowledge graph embedding method and system based on relation cognition |
| CN114219089A (en) * | 2021-11-11 | 2022-03-22 | 山东人才发展集团信息技术有限公司 | Construction method and equipment of new-generation information technology industry knowledge graph |
| CN114745256A (en) * | 2022-02-28 | 2022-07-12 | 国家电网有限公司信息通信分公司 | Server alarm tracing method and device and storage medium |
| CN117634599A (en) * | 2023-10-17 | 2024-03-01 | 中国电子信息产业集团有限公司第六研究所 | Path reasoning method and device based on knowledge graph, electronic equipment and medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106909622A (en) * | 2017-01-20 | 2017-06-30 | 中国科学院计算技术研究所 | Knowledge mapping vector representation method, knowledge mapping relation inference method and system |
| CN108446769A (en) * | 2018-01-23 | 2018-08-24 | 深圳市阿西莫夫科技有限公司 | Knowledge mapping relation inference method, apparatus, computer equipment and storage medium |
| CN108664652A (en) * | 2018-01-24 | 2018-10-16 | 北京理工大学 | Expression learning method based on Path selection under a kind of complex network |
| CN109325129A (en) * | 2018-06-08 | 2019-02-12 | 浙江捷尚人工智能研究发展有限公司 | A kind of knowledge mapping inference method, electronic equipment, storage medium and system |
| CN109543043A (en) * | 2018-05-10 | 2019-03-29 | 国网江西省电力有限公司信息通信分公司 | A kind of automatic monitoring method of the power communication big data mistake of knowledge based map reasoning |
-
2019
- 2019-10-30 CN CN201911044553.6A patent/CN110796254B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106909622A (en) * | 2017-01-20 | 2017-06-30 | 中国科学院计算技术研究所 | Knowledge mapping vector representation method, knowledge mapping relation inference method and system |
| CN108446769A (en) * | 2018-01-23 | 2018-08-24 | 深圳市阿西莫夫科技有限公司 | Knowledge mapping relation inference method, apparatus, computer equipment and storage medium |
| CN108664652A (en) * | 2018-01-24 | 2018-10-16 | 北京理工大学 | Expression learning method based on Path selection under a kind of complex network |
| CN109543043A (en) * | 2018-05-10 | 2019-03-29 | 国网江西省电力有限公司信息通信分公司 | A kind of automatic monitoring method of the power communication big data mistake of knowledge based map reasoning |
| CN109325129A (en) * | 2018-06-08 | 2019-02-12 | 浙江捷尚人工智能研究发展有限公司 | A kind of knowledge mapping inference method, electronic equipment, storage medium and system |
Non-Patent Citations (2)
| Title |
|---|
| ZIKANG WANG，ETC: "Attention-based Multi-hop Reasoning for Knowledge Graph", 《2018 IEEE INTERNATIONAL CONFERENCE ON INTELLIGENCE AND SECURITY INFORMATICS (ISI)》 * |
| 官赛萍等: "面向知识图谱的知识推理研究进展", 《软件学报》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111581343A (en) * | 2020-04-24 | 2020-08-25 | 北京航空航天大学 | Reinforced learning knowledge graph reasoning method and device based on graph convolution neural network |
| CN113688249A (en) * | 2020-05-18 | 2021-11-23 | 第四范式（北京）技术有限公司 | Knowledge graph embedding method and system based on relation cognition |
| CN112069249A (en) * | 2020-10-14 | 2020-12-11 | 中国平安人寿保险股份有限公司 | Knowledge graph relation mining method and device, computer equipment and storage medium |
| CN112596031A (en) * | 2020-12-22 | 2021-04-02 | 电子科技大学 | Target radar threat degree assessment method based on knowledge graph |
| CN112819164A (en) * | 2021-02-02 | 2021-05-18 | 京东数科海益信息科技有限公司 | Inference method and device of affair map and computer equipment |
| CN112819164B (en) * | 2021-02-02 | 2024-05-24 | 京东科技信息技术有限公司 | Logic map reasoning method, logic map reasoning device and computer equipment |
| CN112818137B (en) * | 2021-04-19 | 2022-04-08 | 中国科学院自动化研究所 | Entity alignment-based multi-source heterogeneous knowledge graph collaborative reasoning method and device |
| CN112818137A (en) * | 2021-04-19 | 2021-05-18 | 中国科学院自动化研究所 | Entity alignment-based multi-source heterogeneous knowledge graph collaborative reasoning method and device |
| CN113128689A (en) * | 2021-04-27 | 2021-07-16 | 中国电力科学研究院有限公司 | Entity relationship path reasoning method and system for regulating knowledge graph |
| CN113220833A (en) * | 2021-05-07 | 2021-08-06 | 支付宝(杭州)信息技术有限公司 | Entity association degree identification method and device |
| CN113672741A (en) * | 2021-08-19 | 2021-11-19 | 支付宝(杭州)信息技术有限公司 | Information processing method, device and equipment |
| CN114219089A (en) * | 2021-11-11 | 2022-03-22 | 山东人才发展集团信息技术有限公司 | Construction method and equipment of new-generation information technology industry knowledge graph |
| CN114745256A (en) * | 2022-02-28 | 2022-07-12 | 国家电网有限公司信息通信分公司 | Server alarm tracing method and device and storage medium |
| CN114745256B (en) * | 2022-02-28 | 2024-05-24 | 国家电网有限公司信息通信分公司 | Server alarm tracing method, device and storage medium |
| CN117634599A (en) * | 2023-10-17 | 2024-03-01 | 中国电子信息产业集团有限公司第六研究所 | Path reasoning method and device based on knowledge graph, electronic equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110796254B (en) | 2024-02-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110796254A (en) | Knowledge graph reasoning method and device, computer equipment and storage medium | |
| CN112434169B (en) | Knowledge graph construction method and system and computer equipment thereof | |
| CN110866190B (en) | Method and device for training neural network model for representing knowledge graph | |
| JP5147162B2 (en) | Method and system for determining object similarity based on heterogeneous relationships | |
| CN111709476A (en) | Knowledge distillation-based small classification model training method and device | |
| CN108986872B (en) | Multi-granularity attribute weight Spark method for big data electronic medical record reduction | |
| Gasse et al. | An experimental comparison of hybrid algorithms for Bayesian network structure learning | |
| US11797507B2 (en) | Relation-enhancement knowledge graph embedding method and system | |
| CN110119399B (en) | Business process optimization method based on machine learning | |
| CN115080761B (en) | Low-resource knowledge graph entity alignment method based on semantic perception | |
| CN112115971B (en) | Method and system for carrying out student portrait based on heterogeneous academic network | |
| Uzhinskiy et al. | Multifunctional platform and mobile application for plant disease detection | |
| EP3847583A1 (en) | Determining control policies by minimizing the impact of delusion | |
| CN105933172A (en) | Cloud computing based disease self-diagnosis service construction system | |
| CN114417986A (en) | Artificial intelligence-based medicine characteristic information determination method and device | |
| CN114780879A (en) | Interpretable link prediction method for knowledge hypergraph | |
| CN117852673A (en) | Federal learning method, system, equipment and storage medium | |
| CN110119268B (en) | Workflow optimization method based on artificial intelligence | |
| CN115529350B (en) | Parameter optimization method, device, electronic equipment and readable storage medium | |
| US20240005170A1 (en) | Recommendation method, apparatus, electronic device, and storage medium | |
| CN111291196B (en) | Knowledge graph perfecting method and device, and data processing method and device | |
| CN110135747B (en) | Flow customization method based on neural network | |
| CN110765345A (en) | Searching method, device and equipment | |
| CN117786416B (en) | Model training method, device, equipment, storage medium and product | |
| US20220245469A1 (en) | Decision Making Using Integrated Machine Learning Models and Knowledge Graphs |
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 |