CN113722501B - Knowledge graph construction method, device and storage medium based on deep learning - Google Patents
Knowledge graph construction method, device and storage medium based on deep learning Download PDFInfo
- Publication number
- CN113722501B CN113722501B CN202110902458.6A CN202110902458A CN113722501B CN 113722501 B CN113722501 B CN 113722501B CN 202110902458 A CN202110902458 A CN 202110902458A CN 113722501 B CN113722501 B CN 113722501B
- Authority
- CN
- China
- Prior art keywords
- entity information
- knowledge graph
- deep learning
- sentences
- preset
- 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.)
- Active
Links
- 238000013135 deep learning Methods 0.000 title claims abstract description 40
- 238000010276 construction Methods 0.000 title claims abstract description 37
- 238000013136 deep learning model Methods 0.000 claims abstract description 36
- 238000012549 training Methods 0.000 claims description 11
- 238000000034 method Methods 0.000 claims description 7
- 241001122767 Theaceae Species 0.000 description 20
- 239000008267 milk Substances 0.000 description 15
- 210000004080 milk Anatomy 0.000 description 15
- 235000013336 milk Nutrition 0.000 description 15
- 230000035622 drinking Effects 0.000 description 11
- 238000012545 processing Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 239000007858 starting material Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000007723 transport mechanism Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- 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
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The application discloses a knowledge graph construction method, equipment and a storage medium based on deep learning, wherein the knowledge graph construction method based on deep learning comprises the following steps: acquiring a constructed knowledge graph, and extracting first entity information in the constructed knowledge graph; collecting information related to the first entity information in the constructed knowledge graph to obtain a data set; acquiring a plurality of natural segment sentences related to the first entity information in a data set; classifying the plurality of natural section sentences according to a preset non-supervision deep learning model to obtain a plurality of classification results; and obtaining natural section sentences of which the classification results meet preset conditions in the plurality of classification results to obtain correlation sentences, and adding second entity information and relation information corresponding to the first entity information in the constructed knowledge graph according to the correlation sentences. The application modifies the constructed knowledge graph according to the newly obtained second entity information and relation information, thereby obtaining a more accurate and complete knowledge graph.
Description
Technical Field
The application relates to the technical field of deep learning, in particular to a knowledge graph construction method, equipment and storage medium based on deep learning.
Background
Knowledge Graph (knowledgegraph) generally refers to a semantic network capable of revealing relationships between entities, and based on means such as data mining, information processing, graph drawing and the like, a visual Graph is utilized to vividly display a complex Knowledge field, so that a development rule of the Knowledge field can be embodied to a certain extent.
Along with the development of the big data age, the requirements of people on the knowledge graph are not simple relation chains any more, and high requirements on the completeness and accuracy of the knowledge graph are also provided, but the knowledge graph in the related technology has high construction cost and difficult guarantee of the completeness, so that the accuracy of knowledge reasoning calculation is low.
Disclosure of Invention
The present application aims to solve at least one of the technical problems existing in the prior art. Therefore, the application provides a knowledge graph construction method based on deep learning, which can improve the integrity and accuracy of the knowledge graph.
The application also provides electronic control equipment.
The application also proposes a computer readable storage medium.
In a first aspect, an embodiment of the present application provides a knowledge graph construction method based on deep learning, including:
acquiring a constructed knowledge graph, and extracting first entity information in the constructed knowledge graph;
collecting information related to the first entity information in the constructed knowledge graph to obtain a data set;
acquiring a plurality of natural segment sentences related to the first entity information in the data set;
classifying the natural section sentences according to a preset non-supervision deep learning model to obtain a plurality of classification results;
and acquiring the natural section sentences of which the classification results meet preset conditions in the classification results to obtain correlation sentences, and adding second entity information and relation information corresponding to the first entity information in the constructed knowledge graph according to the correlation sentences.
The knowledge graph construction method based on deep learning provided by the embodiment of the application has at least the following beneficial effects: and modifying the relation structure of the first entity information in the constructed knowledge graph according to the newly obtained second entity information and relation information so as to continuously perfect the constructed knowledge graph, thereby obtaining a more accurate and complete knowledge graph.
According to some other embodiments of the application, the method for constructing a deep learning-based knowledge graph classifies the natural section sentences according to a preset unsupervised deep learning model to obtain a plurality of classification results, including:
the preset non-supervision deep learning model splits a plurality of natural section sentences into a preset mode structure to obtain sentence mode structures, wherein the preset mode structure is first entity information and/or relationship information and/or second entity information;
calculating the credibility of the sentence pattern structure and the first entity information according to the preset unsupervised deep learning model;
and determining the classification result according to the sentence pattern structure of the plurality of natural-segment sentences and the corresponding credibility.
According to other embodiments of the present application, the knowledge graph construction method based on deep learning, the preset conditions include: reference mode structure, preset credibility threshold.
According to other embodiments of the present application, the knowledge graph construction method based on deep learning further includes:
and if the sentence pattern structure of the natural segment sentence does not accord with the reference pattern structure and/or the credibility is lower than the preset credibility threshold value, rejecting the corresponding natural segment sentence.
According to further embodiments of the present application, the method for constructing a deep learning-based knowledge graph, which obtains the natural-segment sentences of the classification results satisfying a preset condition in the classification results to obtain relevance sentences, and adds second entity information and relationship information corresponding to the first entity information in the constructed knowledge graph according to the relevance sentences, includes:
acquiring the natural section sentence with the sentence pattern structure conforming to the reference pattern structure and the credibility being greater than the preset credibility threshold value to obtain the correlation sentence;
and adding second entity information and relation information corresponding to the first entity information in the constructed knowledge graph according to the correlation statement.
According to some other embodiments of the present application, the method for constructing a deep learning-based knowledge graph, adding second entity information and relationship information corresponding to the first entity information in the constructed knowledge graph according to the relevance statement, includes:
extracting the second entity information which is different from the first entity information in the correlation statement;
extracting the relation information associated with the first entity information and the second entity information in the correlation statement;
adding the extracted second entity information and the relation information into a graph structure of the first entity information in the constructed knowledge graph.
According to other embodiments of the present application, the knowledge graph construction method based on deep learning further includes:
collecting sentences conforming to the reference pattern structure to obtain a corpus training set;
substituting the corpus training set into the preset non-supervised deep learning model to adjust parameters of the preset non-supervised deep learning model to obtain the optimized preset non-supervised deep learning model.
According to the knowledge graph construction method based on deep learning of other embodiments of the present application, the preset unsupervised deep learning model is a mask language model.
In a second aspect, one embodiment of the present application provides an electronic control apparatus including:
at least one processor, and,
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the deep learning based knowledge graph construction method of the first aspect.
In a third aspect, an embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions for causing a computer to perform the deep learning-based knowledge-graph construction method according to the first aspect.
Additional features and advantages of the application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application. The objectives and other advantages of the application may be realized and attained by the structure particularly pointed out in the written description and drawings.
Drawings
FIG. 1 is a schematic flow chart of a knowledge graph construction method based on deep learning according to an embodiment of the application;
FIG. 2 is a schematic flow chart of another embodiment of a knowledge graph construction method based on deep learning in an embodiment of the application;
FIG. 3 is a flowchart of another embodiment of a knowledge graph construction method based on deep learning according to an embodiment of the present application;
FIG. 4 is a flowchart of another embodiment of a knowledge graph construction method based on deep learning according to an embodiment of the application;
FIG. 5 is a flowchart of another embodiment of a knowledge graph construction method based on deep learning according to an embodiment of the application;
FIG. 6 is a flowchart of another embodiment of a knowledge graph construction method based on deep learning according to an embodiment of the application;
fig. 7 is a block diagram of an embodiment of an electronic control device according to an embodiment of the present application.
Detailed Description
The conception and the technical effects produced by the present application will be clearly and completely described in conjunction with the embodiments below to fully understand the objects, features and effects of the present application. It is apparent that the described embodiments are only some embodiments of the present application, but not all embodiments, and that other embodiments obtained by those skilled in the art without inventive effort are within the scope of the present application based on the embodiments of the present application.
In the description of the present application, if an orientation description such as "upper", "lower", "front", "rear", "left", "right", etc. is referred to, it is merely for convenience of description and simplification of the description, and does not indicate or imply that the apparatus or element referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the application. If a feature is referred to as being "disposed," "secured," "connected," or "mounted" on another feature, it can be directly disposed, secured, or connected to the other feature or be indirectly disposed, secured, connected, or mounted on the other feature.
In the description of the embodiments of the present application, if "several" is referred to, it means more than one, if "multiple" is referred to, it is understood that the number is not included if "greater than", "less than", "exceeding", and it is understood that the number is included if "above", "below", "within" is referred to. If reference is made to "first", "second" it is to be understood as being used for distinguishing technical features and not as indicating or implying relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.
Along with the development of internet technology, the knowledge graph is also converted from manual production into a mode of automatically generating a computer program, but a traditional computer program constructs an information database according to automatically acquired information, then carries out data analysis according to the information in the information database to determine entity information and relation information in sentences, and then constructs the knowledge graph according to the entity information and the relation information corresponding to the entity information. However, the integrity and accuracy of the knowledge graph completed according to one-time construction are low, so that the knowledge graph needs to be continuously enhanced to construct a more perfect and accurate knowledge graph.
Based on the method, the device and the storage medium for constructing the knowledge graph based on the deep learning, the constructed knowledge graph can be reinforced according to a trained unsupervised deep learning model, so that a more perfect and accurate knowledge graph can be obtained.
In a first aspect, referring to fig. 1, an embodiment of the present application discloses a knowledge graph construction method based on deep learning, including:
s100, acquiring a constructed knowledge graph, and extracting first entity information in the constructed knowledge graph;
s200, acquiring information related to the first entity information in the constructed knowledge graph to obtain a data set;
s300, acquiring a plurality of natural segment sentences related to first entity information in a data set;
s400, classifying the plurality of natural section sentences according to a preset non-supervision deep learning model to obtain a plurality of classification results;
s500, obtaining natural section sentences of which the classification results meet preset conditions in the plurality of classification results to obtain correlation sentences, and adding second entity information and relation information corresponding to the first entity information in the constructed knowledge graph according to the correlation sentences.
Extracting first entity information from the constructed knowledge graph, acquiring natural section sentences related to the first entity information from a data set, classifying the natural section sentences according to a preset unsupervised deep learning model to obtain a classification result, judging whether the classification result accords with preset conditions, acquiring natural section sentences of which the classification result accords with the preset conditions to obtain correlation sentences, determining second entity information and relation information corresponding to the first entity information according to the correlation sentences, and modifying the relation structure of the first entity information in the constructed knowledge graph according to the newly acquired second entity information and relation information so as to continuously perfect the constructed knowledge graph, thereby obtaining a more accurate and complete knowledge graph.
The constructed knowledge graph is updated according to a preset time interval period, so that the integrity and accuracy of the knowledge graph are continuously improved.
Referring to fig. 2, in some embodiments, step S400 includes:
s410, a preset non-supervision deep learning model splits a plurality of natural section sentences into a preset mode structure to obtain sentence mode structures, wherein the preset mode structure is first entity information and/or relationship information and/or second entity information;
s420, calculating the credibility of the sentence pattern structure and the first entity information according to a preset non-supervision deep learning model;
s430, determining a classification result according to the sentence pattern structure of the plurality of natural-segment sentences and the corresponding credibility.
Since the data set is obtained by collecting information related to the first entity information in the constructed knowledge-graph, the information related to the first entity information may be related information related to the first entity information or different second entity information in the constructed knowledge-graph. Therefore, according to the preset non-supervision deep learning model, splitting the plurality of natural section sentences into a preset mode structure to obtain a sentence mode structure, namely splitting the natural section sentences into first entity information and/or relationship information and/or second entity information to obtain the sentence mode structure of the natural section sentences, wherein the sentence mode structure can be the first entity information and the relationship information and/or the second entity information. And then calculating the credibility of the sentence pattern structure of each natural section sentence and the first entity information by adopting a preset non-supervision deep learning model, namely splitting the natural section sentence into relation information and/or second entity information, then calculating the credibility between the relation information and/or the second entity information and the first entity information, then determining a classification result according to the sentence pattern structures of a plurality of natural section sentences and the credibility corresponding to the natural section sentences, and calculating to obtain the classification result of the natural section sentences so as to judge whether the classification result meets preset conditions or not, and further judging that the plurality of natural section sentences can be used as natural section sentences for perfecting the constructed knowledge graph, thereby improving the integrity and accuracy of the constructed knowledge graph.
For example: searching a natural section sentence matched with first entity information in a data set, searching a natural section sentence matched with Zhou Jielun in the data set if the first entity information is Zhou Jielun, searching three natural section sentences which are related to Zhou Jielun and respectively are Zhou Jielun daily drinking milk tea, zhou Jielun legend Youmei milk tea and Zhou Jielun favorite starter, and splitting the three natural section sentences into preset model structures to obtain sentence pattern structures of the three natural section sentences, wherein the sentence pattern structures are Zhou Jielun + drinking +milk tea, zhou Jielun + introduction +milk tea and Zhou Jielun + favorite starter. And then respectively calculating the credibility of the three sentence pattern structures and Zhou Jielun to be 0.8, 0.6 and 0.7 through a preset non-supervision deep learning model, and then determining a classification result according to the sentence pattern structures of the three natural section sentences and the corresponding credibility, so that which natural section sentence can be determined according to the classification result of the three natural section sentences to perfect the constructed knowledge graph, thereby improving the accuracy and the integrity of the constructed knowledge graph.
In some embodiments, the preset conditions include: reference mode structure, preset credibility threshold. Because the classification result is the sentence pattern structure of the natural section sentence and the corresponding credibility, whether the classification result meets the preset condition or not is judged, namely whether the sentence pattern structure is the reference pattern structure or not is judged, and the credibility is compared with the preset credibility threshold value.
Referring to fig. 3, in some embodiments, the deep learning-based knowledge graph construction method further includes:
s600, if the sentence pattern structure of the natural section sentence does not accord with the reference pattern structure and/or the credibility is lower than a preset credibility threshold, eliminating the corresponding natural section sentence.
The natural section sentences matched with the first entity information are acquired through the data set and stored in a plurality of ways, but not all natural section sentences can be used for modifying the constructed knowledge graph in order to improve the integrity and the accuracy of the constructed knowledge graph, so that only the natural section sentences which are most satisfactory in the plurality of natural section sentences can be selected. If the sentence pattern structure of the natural section sentence does not accord with the reference pattern, that is, the relation structure related to the first entity information in the constructed knowledge graph cannot be perfected according to the natural section sentence, the natural section sentence needs to be removed. Or the credibility of the natural section sentences is lower than a preset credibility threshold, namely the correlation between the natural section sentences and the first entity information is not high, so that the accuracy and the integrity of the constructed knowledge graph are not affected by eliminating the natural section sentences, the number of the natural section sentences participating in subsequent screening can be saved, and the screening efficiency of the natural section sentences is improved.
Referring to fig. 4, step S500 includes:
s510, acquiring a natural section sentence with a sentence pattern structure conforming to a reference pattern structure and reliability greater than a preset reliability threshold value to obtain a correlation sentence;
s520, adding second entity information and relation information corresponding to the first entity information in the constructed knowledge graph according to the correlation statement.
And rejecting the natural section sentences with sentence pattern structures which do not accord with the reference pattern structure and/or with credibility lower than a preset credibility threshold value in the plurality of natural section sentences, wherein the sentence pattern structures of the rest natural section sentences accord with the reference pattern structure and the credibility is higher than the preset credibility threshold value, defining the rest natural section sentences as related sentences, and then perfecting the sentences of the constructed knowledge graph according to the related sentences, thereby improving the integrity and the accuracy of the constructed knowledge graph.
For example, if the preset credibility threshold is 0.7, the sentence pattern structures of the natural-segment sentences are "Zhou Jielun +drinking+milky tea", "Zhou Jielun +drinking+milky tea", "Zhou Jielun +like+composing", and the credibility of the three sentence pattern structures and "Zhou Jielun" are 0.8, 0.6 and 0.7, respectively, the reference pattern structure is the first entity information+the relation information+the second entity information, so that the pattern structures of the three natural-segment sentences all conform to the reference pattern structure, and the natural-segment sentences with the credibility greater than the preset credibility threshold are only "Zhou Jielun drinking milky tea", so that the "Zhou Jielun drinking milky tea" is defined as the correlation sentence, and then the second entity information and the relation information corresponding to the first entity information in the constructed knowledge graph are increased according to the "Zhou Jielun drinking milky tea", thereby improving the accuracy and the integrity of the constructed knowledge graph and obtaining a more complete and accurate knowledge graph.
Referring to fig. 5, in some embodiments, step S530 includes:
s531, extracting second entity information which is different from the first entity information in the correlation statement;
s532, extracting relation information associated with the first entity information and the second entity information in the correlation statement;
s533, adding the extracted second entity information and relation information into a map structure of the first entity information in the constructed knowledge map.
And adding the determined relevance statement into a relation structure related to the first entity information in the constructed knowledge graph, and extracting the second entity information and the relation information in the relevance statement because the statement pattern structure of the obtained relevance statement accords with the reference pattern structure. And adding the extracted relation information and the second entity information into a graph structure related to the first entity information in the constructed knowledge graph to obtain a more complete and accurate knowledge graph.
For example, if the relevance sentence is "Zhou Jielun drink milk tea", the second entity information and the relationship information are respectively "milk tea" and "drink" from the relevance sentence, and then the information with the sentence pattern structure of "Zhou Jielun" + "drink" + "milk tea" is added to the constructed knowledge graph, so that the accuracy and the integrity of the knowledge graph are improved.
Referring to fig. 6, in some embodiments, the deep learning-based knowledge graph construction method further includes:
s700, collecting sentences conforming to a reference mode structure to obtain a corpus training set;
s800, substituting the corpus training set into a preset non-supervision deep learning model to adjust parameters of the preset non-supervision deep learning model so as to obtain the optimized preset non-supervision deep learning model.
In order to improve accuracy of the preset non-supervised deep learning model, a corpus training set needs to be updated regularly, sentences stored in the corpus training set are sentences which are collected by a third-party platform regularly and meet a standard mode structure, so that the preset non-supervised deep learning model is trained according to the corpus training set through regularly perfecting the corpus training set, and accuracy of calculation reliability of the preset non-supervised deep learning model is improved. The corpus training set stores the matching information of sentences and credibility, then the sentences are brought into a preset unsupervised deep learning model to obtain credibility containing an unknown function, then the specific value of the function is determined according to the credibility of sentence matching and the credibility containing the unknown function, and then parameters of the preset unsupervised deep learning model are determined according to the specific value of the function, so that accuracy of calculating credibility of the unsupervised deep learning model is improved.
The preset non-supervision deep learning model is a shielding language model.
The pre-set unsupervised deep learning model is a masked language model in which we typically mask words of a given confidence in a given sentence, the model expects to predict these masked words based on other words in the sentence. Therefore, by adjusting parameters of the shielding language model to obtain an optimized shielding language model through statement information with given credibility, accuracy of the shielding language model in calculating credibility of the natural-segment statements and the first entity information can be improved.
A knowledge-graph construction method based on deep learning according to an embodiment of the present application will be described in detail with reference to fig. 1 to 6. It is to be understood that the following description is exemplary only and is not intended to limit the application in any way.
Extracting first entity information from the constructed knowledge graph, if the first entity information is Zhou Jielun, searching natural section sentences matched with Zhou Jielun in a data set, if three natural section sentences related to Zhou Jielun are found, and are respectively 'Zhou Jielun drink milk tea every day', 'Zhou Jielun say Youle-Mei milk tea', 'Zhou Jielun like-composer', and then respectively splitting the three natural section sentences into preset model structures to obtain sentence pattern structures of the three natural section sentences, namely 'Zhou Jielun + drink +milk tea', 'Zhou Jielun + say +milk tea', 'Zhou Jielun + like-composer'. And then respectively calculating the credibility of the three sentence pattern structures and 'Zhou Jielun' to be 0.8, 0.6 and 0.7 through a preset non-supervision deep learning model. If the preset credibility threshold value is 0.7, the natural sentence with credibility larger than the preset credibility threshold value is only "Zhou Jielun drinking milk tea", so "Zhou Jielun drinking milk tea" is defined as a correlation sentence. And extracting second entity information and relation information from the correlation sentences, namely 'milk tea' and 'drinking' respectively, and adding the information with sentence pattern structures of 'Zhou Jielun', 'drinking', 'milk tea' into the constructed knowledge graph, thereby improving the accuracy and the integrity of the knowledge graph.
In a second aspect, referring to fig. 7, an embodiment of the present application further discloses an electronic control device, including: at least one processor 100, and a memory 200 communicatively coupled to the at least one processor 100; wherein the memory 200 stores instructions executable by the at least one processor 100 to enable the at least one processor 100 to perform the deep learning based knowledge graph construction method as described in the first aspect.
The electronic control device may be a mobile terminal device or a non-mobile terminal device. The mobile terminal device may be a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle-mounted terminal device, a wearable device, an ultra mobile personal computer, a netbook, a personal digital assistant, a CPE, a UFI (wireless hotspot device), etc.; the non-mobile terminal equipment can be a personal computer, a television, a teller machine, a self-service machine or the like; the embodiment of the present application is not particularly limited.
Processor 100 may include one or more processing units, such as: processor 100 may include an application processor (application processor, AP), a modem processor, a graphics processor (graphics processing unit, GPU), an image signal processor (image signal processor, ISP), a controller, a memory, a video codec, a digital signal processor (digital signal processor, DSP), a baseband processor, and/or a neural network processor (neural-network processing unit, NPU), etc. Wherein the different processing units may be separate devices or may be integrated in one or more processors.
Memory 200 is a cache memory. The memory 200 may hold instructions or data that the processor has just used or recycled. If the processor 100 needs to reuse the instruction or data, it may be called directly from the memory 200. Repeated accesses are avoided and the latency of the processor 100 is reduced, thereby improving the efficiency of the system.
In a third aspect, an embodiment of the present application further discloses a computer-readable storage medium, where computer-executable instructions are stored, where the computer-executable instructions are configured to cause a computer to perform the knowledge graph construction method based on deep learning according to the first aspect.
The above described apparatus embodiments are merely illustrative, wherein the units illustrated as separate components may or may not be physically separate, i.e. may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
Those of ordinary skill in the art will appreciate that all or some of the steps, systems, and methods disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as known to those skilled in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by a computer. Furthermore, as is well known to those of ordinary skill in the art, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media.
The embodiments of the present application have been described in detail with reference to the accompanying drawings, but the present application is not limited to the above embodiments, and various changes can be made within the knowledge of one of ordinary skill in the art without departing from the spirit of the present application. Furthermore, embodiments of the application and features of the embodiments may be combined with each other without conflict.
Claims (9)
1. The knowledge graph construction method based on deep learning is characterized by comprising the following steps of:
acquiring a constructed knowledge graph, and extracting first entity information in the constructed knowledge graph;
collecting information related to the first entity information in the constructed knowledge graph to obtain a data set;
acquiring a plurality of natural segment sentences related to the first entity information in the data set;
classifying the plurality of natural segment sentences according to a preset non-supervision deep learning model to obtain a plurality of classification results, wherein the method specifically comprises the following steps of:
the preset non-supervision deep learning model splits a plurality of natural section sentences into a preset mode structure to obtain sentence mode structures, wherein the preset mode structure is first entity information and/or relationship information and/or second entity information;
calculating the credibility of the sentence pattern structure and the first entity information according to the preset unsupervised deep learning model;
determining the classification result according to the sentence pattern structures of the plurality of natural-segment sentences and the corresponding credibility;
and acquiring the natural section sentences of which the classification results meet preset conditions in the classification results to obtain correlation sentences, and adding second entity information and relation information corresponding to the first entity information in the constructed knowledge graph according to the correlation sentences.
2. The knowledge graph construction method based on deep learning according to claim 1, wherein the preset conditions include: reference mode structure, preset credibility threshold.
3. The knowledge-graph construction method based on deep learning according to claim 2, further comprising:
and if the sentence pattern structure of the natural segment sentence does not accord with the reference pattern structure and/or the credibility is lower than the preset credibility threshold value, rejecting the corresponding natural segment sentence.
4. The deep learning-based knowledge graph construction method according to claim 2 or 3, wherein the obtaining the natural segment sentence of the classification result satisfying a preset condition in the plurality of classification results to obtain a correlation sentence, and adding second entity information and relationship information corresponding to the first entity information in the constructed knowledge graph according to the correlation sentence, includes:
acquiring the natural section sentence with the sentence pattern structure conforming to the reference pattern structure and the credibility being greater than the preset credibility threshold value to obtain the correlation sentence;
and adding second entity information and relation information corresponding to the first entity information in the constructed knowledge graph according to the correlation statement.
5. The deep learning-based knowledge graph construction method according to claim 4, wherein the adding second entity information and relationship information corresponding to the first entity information in the constructed knowledge graph according to the relevance statement includes:
extracting the second entity information which is different from the first entity information in the correlation statement;
extracting the relation information associated with the first entity information and the second entity information in the correlation statement;
adding the extracted second entity information and the relation information into a graph structure of the first entity information in the constructed knowledge graph.
6. The knowledge-graph construction method based on deep learning according to claim 4, further comprising:
collecting sentences conforming to the reference pattern structure to obtain a corpus training set;
substituting the corpus training set into the preset non-supervised deep learning model to adjust parameters of the preset non-supervised deep learning model to obtain the optimized preset non-supervised deep learning model.
7. The knowledge graph construction method based on deep learning according to claim 1, wherein the preset unsupervised deep learning model is a mask language model.
8. An electronic control apparatus, characterized by comprising:
at least one processor, and,
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the deep learning based knowledge-graph construction method of any one of claims 1 to 7.
9. A computer-readable storage medium storing computer-executable instructions for causing a computer to perform the deep learning-based knowledge-graph construction method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110902458.6A CN113722501B (en) | 2021-08-06 | 2021-08-06 | Knowledge graph construction method, device and storage medium based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110902458.6A CN113722501B (en) | 2021-08-06 | 2021-08-06 | Knowledge graph construction method, device and storage medium based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113722501A CN113722501A (en) | 2021-11-30 |
| CN113722501B true CN113722501B (en) | 2023-09-22 |
Family
ID=78675087
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110902458.6A Active CN113722501B (en) | 2021-08-06 | 2021-08-06 | Knowledge graph construction method, device and storage medium based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113722501B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114357192B (en) * | 2021-12-31 | 2024-04-19 | 海南大学 | DIKW-based content integrity modeling and judging method |
| CN114595686B (en) * | 2022-03-11 | 2023-02-03 | 北京百度网讯科技有限公司 | Knowledge extraction method, and training method and device of knowledge extraction model |
| CN115098629B (en) * | 2022-06-22 | 2024-09-17 | 马上消费金融股份有限公司 | File processing method, device, server and readable storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107589826A (en) * | 2016-07-07 | 2018-01-16 | 深圳狗尾草智能科技有限公司 | The man-machine interaction method and system of knowledge based collection of illustrative plates |
| CN109213854A (en) * | 2018-09-05 | 2019-01-15 | 平安科技（深圳）有限公司 | Knowledge mapping approaches to IM, device, computer equipment and storage medium |
| CN110110092A (en) * | 2018-09-30 | 2019-08-09 | 北京国双科技有限公司 | A kind of knowledge mapping construction method and relevant device |
| CN111914550A (en) * | 2020-07-16 | 2020-11-10 | 华中师范大学 | Knowledge graph updating method and system for limited field |
| CN112100343A (en) * | 2020-08-17 | 2020-12-18 | 深圳数联天下智能科技有限公司 | Method for expanding knowledge graph, electronic equipment and storage medium |
| CN112989002A (en) * | 2021-03-31 | 2021-06-18 | 中国工商银行股份有限公司 | Question-answer processing method, device and equipment based on knowledge graph |
| CN113128233A (en) * | 2021-05-11 | 2021-07-16 | 济南大学 | Construction method and system of mental disease knowledge map |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9037529B2 (en) * | 2011-06-15 | 2015-05-19 | Ceresis, Llc | Method for generating visual mapping of knowledge information from parsing of text inputs for subjects and predicates |
-
2021
- 2021-08-06 CN CN202110902458.6A patent/CN113722501B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107589826A (en) * | 2016-07-07 | 2018-01-16 | 深圳狗尾草智能科技有限公司 | The man-machine interaction method and system of knowledge based collection of illustrative plates |
| CN109213854A (en) * | 2018-09-05 | 2019-01-15 | 平安科技（深圳）有限公司 | Knowledge mapping approaches to IM, device, computer equipment and storage medium |
| CN110110092A (en) * | 2018-09-30 | 2019-08-09 | 北京国双科技有限公司 | A kind of knowledge mapping construction method and relevant device |
| CN111914550A (en) * | 2020-07-16 | 2020-11-10 | 华中师范大学 | Knowledge graph updating method and system for limited field |
| CN112100343A (en) * | 2020-08-17 | 2020-12-18 | 深圳数联天下智能科技有限公司 | Method for expanding knowledge graph, electronic equipment and storage medium |
| CN112989002A (en) * | 2021-03-31 | 2021-06-18 | 中国工商银行股份有限公司 | Question-answer processing method, device and equipment based on knowledge graph |
| CN113128233A (en) * | 2021-05-11 | 2021-07-16 | 济南大学 | Construction method and system of mental disease knowledge map |
Non-Patent Citations (1)
| Title |
|---|
| 面向深度学习的动态知识图谱建构模型及评测;姜强;药文静;赵蔚;李松;;《电化教育研究》(第3期);87-94 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113722501A (en) | 2021-11-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113722501B (en) | Knowledge graph construction method, device and storage medium based on deep learning | |
| EP3872652B1 (en) | Method and apparatus for processing video, electronic device, medium and product | |
| CN112613293B (en) | Digest generation method, digest generation device, electronic equipment and storage medium | |
| CN110362814A (en) | A kind of name entity recognition method and device based on improvement loss function | |
| CN111653274B (en) | Wake-up word recognition method, device and storage medium | |
| US20220351634A1 (en) | Question answering systems | |
| CN111710337A (en) | Voice data processing method and device, computer readable medium and electronic equipment | |
| EP3832485A1 (en) | Question answering systems | |
| CN111444906B (en) | Image recognition method and related device based on artificial intelligence | |
| CN116958512A (en) | Target detection method, target detection device, computer readable medium and electronic equipment | |
| CN113850383A (en) | Text matching model training method and device, electronic equipment and storage medium | |
| CN109657710B (en) | Data screening method and device, server and storage medium | |
| CN116628507A (en) | Data processing method, device, equipment and readable storage medium | |
| CN116912478A (en) | Object detection model construction, image classification method and electronic equipment | |
| CN113722500B (en) | Knowledge graph construction method, system, equipment and storage medium | |
| CN114996466A (en) | Method and system for establishing medical standard mapping model and using method | |
| CN111382322B (en) | Method and device for determining similarity of character strings | |
| CN113722502B (en) | Knowledge graph construction method, system and storage medium based on deep learning | |
| CN115098680B (en) | Data processing method, device, electronic equipment, medium and program product | |
| CN113449728B (en) | Character recognition method and related equipment thereof | |
| CN117649933B (en) | Online consultation assistance method and device, electronic equipment and storage medium | |
| CN112926309B (en) | Safety information distinguishing method and device and electronic equipment | |
| CN114595331A (en) | Intention identification method and device, electronic equipment and storage medium | |
| CN116152558A (en) | Model training and object detection method and device, equipment and storage medium | |
| CN117746440A (en) | Sensitive data identification method and electronic equipment |
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 |