Atlas expansion method and device, electronic device, computer readable storage medium

Abstract

Embodiments of the present disclosure provide a graph expansion method and device, electronic equipment and computer readable storage medium, and the specific implementation scheme is: obtaining a knowledge graph comprising a first entity, the first entity being obtained through metadata of unstructured data; in response to receiving an entity expansion request for the knowledge graph, obtaining a data subject of the unstructured data; based on the data subject and a visual operation interface, obtaining an expansion entity related to the first entity and an entity relationship related to the expansion entity; and based on the expansion entity, the entity relationship and the knowledge graph, obtaining an expansion graph. Through the present embodiment, the fusion capability of multi-source heterogeneous data in the knowledge graph is improved.

Description

Technical Field

[0001] The embodiments of this disclosure relate to the field of computer technology, and more specifically, to a map expansion method and apparatus, electronic device, and computer-readable storage medium. Background Technology

[0002] When performing data fusion analysis on data from multiple industries, it is often necessary to integrate data sources with multiple structures and analyze their inherent relationships.

[0003] Traditional data fusion analysis primarily involves the joint operations of structured data tables. Analyzing correlations requires separate knowledge modeling of the structured data before importing it into a graph database. For business data analysts, the technical barrier to entry is high when it comes to analyzing unstructured data, making effective data fusion analysis difficult. Summary of the Invention

[0004] The embodiments described herein provide a map expansion method and apparatus, an electronic device, and a computer-readable storage medium storing a computer program.

[0005] According to a first aspect of this disclosure, a knowledge graph extension method is provided. In this method, a knowledge graph including a first entity is obtained, the first entity being obtained through metadata of unstructured data; in response to receiving a request for entity extension of the knowledge graph, a data body of the unstructured data is obtained; based on the data body and a visual interface, extended entities related to the first entity and entity relationships related to the extended entities are obtained; and based on the extended entities, entity relationships, and the knowledge graph, an extended knowledge graph is obtained.

[0006] In some embodiments of this disclosure, obtaining the extended entities related to the first entity and the entity relationships related to the extended entities based on the data subject and the visual operation interface includes: determining a preset artificial intelligence algorithm corresponding to the type of the data subject, and displaying the name of the artificial intelligence algorithm on the visual operation interface; determining the selected artificial intelligence algorithm based on the operation information obtained from the visual operation interface; processing the data subject using the selected artificial intelligence algorithm to obtain the extended entities related to the first entity and the attribute values ​​of the extended entities; and determining the entity relationships related to the extended entities based on the extended entities, the attribute values ​​of the extended entities, and the knowledge graph.

[0007] In some embodiments of this disclosure, the data subject is of type text. The data subject is processed using a selected artificial intelligence algorithm to obtain extended entities and attribute values ​​of the extended entities related to the first entity, including: extracting extended entities and attribute values ​​of the extended entities from the data subject using a material recognition algorithm or an entity extraction algorithm.

[0008] In some embodiments of this disclosure, the data subject is of the type of video. The selected artificial intelligence algorithm is used to process the data subject to obtain the extended entity and the attribute value of the extended entity related to the first entity. This includes: using a machine vision algorithm to extract the video entity and the attribute value of the video entity from the data subject, taking the video entity as the extended entity related to the first entity, and taking the attribute value of the video entity as the attribute value of the extended entity.

[0009] In some embodiments of this disclosure, the machine vision algorithm includes: a human pose prediction algorithm and a face recognition algorithm; the extraction of video entities and attributes of video entities from the data subject using the machine vision algorithm includes: extracting at least one event entity and attribute values ​​of each event entity from the data subject using the human pose prediction algorithm; and extracting at least one person entity and attribute values ​​of each person entity from the data subject using the face recognition algorithm.

[0010] In some embodiments of this disclosure, the above-mentioned determination of entity relationships related to an extended entity based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph includes: in response to the extended entity only including event entities, determining a first entity relationship between the extended entities based on the inherent connections between the event entities; in response to the event entity's attribute values ​​being the same as or similar to the attribute values ​​of entities in the knowledge graph, determining a second entity relationship between the entity in the knowledge graph and the event entity; and using the first entity relationship and the second entity relationship together as the entity relationship related to the extended entity.

[0011] In some embodiments of this disclosure, the above-mentioned determination of entity relationships related to an extended entity based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph includes: in response to the extended entity's attribute values ​​being the same or similar to the attribute values ​​of an entity in the knowledge graph, determining that the entity in the knowledge graph and the extended entity have an entity relationship.

[0012] According to a second aspect of this disclosure, a knowledge graph expansion apparatus is provided. The apparatus includes: a knowledge graph acquisition unit configured to acquire a knowledge graph including a first entity, the first entity being obtained through metadata of unstructured data; a subject acquisition unit configured to acquire a data subject of unstructured data in response to receiving a request for entity expansion of the knowledge graph; an acquisition unit configured to acquire extended entities related to the first entity and entity relationships related to the extended entities based on the data subject and a visual operation interface; and an expansion unit configured to acquire an extended knowledge graph based on the extended entities, entity relationships, and the knowledge graph.

[0013] In some embodiments of this disclosure, the aforementioned obtaining unit is further configured to: determine a preset artificial intelligence algorithm corresponding to the type of the data subject, and display the name of the artificial intelligence algorithm on a visual operation interface; determine the selected artificial intelligence algorithm based on the operation information obtained from the visual operation interface; process the data subject using the selected artificial intelligence algorithm to obtain an extended entity and attribute values ​​of the extended entity related to the first entity; and determine entity relationships related to the extended entity based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph.

[0014] In some embodiments of this disclosure, the data subject is of type text, and the obtaining unit is further configured to: extract extended entities and attribute values ​​of extended entities from the data subject using a material recognition algorithm or an entity extraction algorithm.

[0015] In some embodiments of this disclosure, the data subject is of the type of video, and the obtaining unit is further configured to: extract video entities and attribute values ​​of video entities from the data subject using a machine vision algorithm, treat the video entities as extended entities related to the first entity, and treat the attribute values ​​of the video entities as attribute values ​​of the extended entities.

[0016] In some embodiments of this disclosure, the machine vision algorithm includes: a human pose prediction algorithm and a face recognition algorithm; the obtaining unit is further configured to: use the human pose prediction algorithm to extract at least one event entity and the attribute values ​​of each event entity from the data subject; and use the face recognition algorithm to extract at least one person entity and the attribute values ​​of each person entity from the data subject.

[0017] In some embodiments of this disclosure, the above-mentioned obtaining unit is further configured to: in response to the extended entity including only event entities, determine a first entity relationship between the extended entities based on the inherent connections between the event entities; in response to the attribute value of the event entity being the same as or similar to the attribute value of the entity in the knowledge graph, determine a second entity relationship between the entity in the knowledge graph and the event entity; and use the first entity relationship and the second entity relationship together as the entity relationship related to the extended entity.

[0018] In some embodiments of this disclosure, the above-described obtaining unit is further configured to: determine that the entity in the knowledge graph and the extended entity have an entity relationship in response to the attribute value of the extended entity being the same as or similar to the attribute value of the entity in the knowledge graph.

[0019] According to a third aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and at least one memory storing a computer program; wherein, when the computer program is executed by the at least one processor, the device performs the steps of the method according to a first aspect of this disclosure.

[0020] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the method according to a first aspect of this disclosure.

[0021] The knowledge graph expansion method and apparatus disclosed herein first acquire a knowledge graph including a first entity, which is obtained through metadata of unstructured data; second, in response to receiving an entity expansion request for the knowledge graph, acquire the data body of the unstructured data; third, based on the data body and a visual operation interface, obtain the expanded entities related to the first entity and the entity relationships related to the expanded entities; finally, based on the expanded entities, entity relationships, and the knowledge graph, obtain the expanded knowledge graph. Thus, by processing the data body using a visual operation interface to obtain the expanded entities and their entity relationships, efficient and low-tech-threshold knowledge graph data processing of multi-source heterogeneous data is achieved, forming diverse knowledge graph entities and relationships, and enhancing the data fusion and analysis capabilities of the knowledge graph. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. It should be understood that the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure, wherein:

[0023] Figure 1 This is a flowchart of an embodiment of the atlas expansion method according to the present disclosure;

[0024] Figure 2 This is a flowchart of an embodiment of the method for extending entities and relationships with entities according to this disclosure;

[0025] Figure 3 This is a relationship diagram of various extended entities in the visual operation interface of this embodiment;

[0026] Figure 4 This is a schematic diagram of a structure according to an embodiment of the atlas extension device of this disclosure; and

[0027] Figure 5 This is a block diagram of an electronic device used to implement the map expansion method of the embodiments of this disclosure. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are also within the scope of protection of this disclosure.

[0029] See Figure 1 The diagram illustrates a flow 100 of an embodiment of a map expansion method according to the present disclosure, which includes the following steps:

[0030] Step 101: Obtain the knowledge graph including the first entity.

[0031] In this embodiment, the first entity is an entity in the knowledge graph, and it is obtained through the metadata of unstructured data. Since unstructured data is data without a fixed pattern, it includes metadata and a data body. Metadata describes the unstructured data, while the data body is the actual content of the unstructured data. Metadata can be obtained when constructing a knowledge graph, and the first entity in the knowledge graph is generated based on this metadata. The first entity includes an entity name and entity attribute values. Each attribute value can be the value of at least one attribute. For example, if the entity name is "student," the attributes of the first entity include age, and the attribute value is a specific age value, such as "5 years old." The first entity is a concrete example of metadata. For instance, if the metadata describes the author of a document, the first entity could be "Zhang San" or "Li Si," where both are the author's names. The attributes of the first entity can be age and gender. The age attribute value is the age value (e.g., 1-120 years old), and the gender attribute value is "male" or "female."

[0032] In this embodiment, unstructured data can be stored in a database, and unstructured data can be associated with a knowledge graph. This association can be represented by a metadata and unstructured data correspondence table. The correspondence table can represent the correspondence between metadata, the first entity, the name of the unstructured data, and the name of the database where the unstructured data is located. When it is necessary to obtain the unstructured data of the first entity, the unstructured data name corresponding to the first entity and the name of the database where the unstructured data is located can be directly queried from the correspondence table, and the unstructured data corresponding to the first entity can be obtained directly.

[0033] Step 102: In response to receiving a request to expand the entity of the knowledge graph, obtain the data body of the unstructured data.

[0034] In this embodiment, the knowledge graph is a graph with fixed entities and entity relationships. When it is necessary to expand the graph (e.g., add entities and entity relationships to the knowledge graph), the entities and entity relationships can be expanded based on unstructured data corresponding to the metadata. The knowledge graph includes a first entity obtained through unstructured data, and the knowledge graph may also include entities obtained through structured data.

[0035] In this embodiment, the entity expansion request for the knowledge graph is an example of expanding the knowledge graph. The specific requests for expanding the knowledge graph include: entity expansion requests and relation expansion requests. The entity expansion request is a request to expand the entities in the knowledge graph, and the relation expansion request is a request to expand the entity relations.

[0036] In this embodiment, the entity expansion request may include: a request to expand the first entity and a request to expand entities in the knowledge graph other than the first entity. After receiving the entity expansion request, the data body of the unstructured data associated with the first entity is obtained. The data body can be represented by various data types, such as text documents, video clips, and audio clips.

[0037] Step 103: Based on the data subject and the visual operation interface, obtain the extended entities related to the first entity and the entity relationships related to the extended entities.

[0038] In this embodiment, the visual operation interface displays the data subject and allows operations on the corresponding data (e.g., segmenting the data subject to obtain extended entities; performing semantic analysis on the data subject to determine the attributes and attribute values ​​of the extended entities; and analyzing the attribute values ​​of different extended entities to determine the entity relationships between them). Users can convert the data subject into a data format suitable for the knowledge graph and transform it into multiple extended entities by operating different controls displayed on the visual operation interface. Furthermore, the user can use the visual operation interface to find entity relationships between extended entities and between extended entities and entities in the knowledge graph. Specifically, the controls on the visual operation interface may include: a loader for loading the data subject onto the visual interface; for example, if the data subject is a Word document, the user can load the data subject and then display it in Word format on the visual operation interface. The controls on the visual operation interface may also include: a selector and a drag-and-drop tool; the selector is used to select text in the data subject that corresponds to an extended entity, and after text processing, the extended entity is obtained; the drag-and-drop tool is used to drag multiple extended entities displayed on the visual operation interface and entities in the knowledge graph together to establish entity relationships.

[0039] Optionally, through a visual user interface, users can select different entity processing algorithms to process the data subject and obtain extended entities.

[0040] Step 104: Based on the extended entities, entity relationships, and knowledge graph, the extended graph is obtained.

[0041] In this embodiment, entity relationships describe the relationships between extended entities and between extended entities and various entities in the knowledge graph. After obtaining the extended entities and entity relationships, the execution entity running on the graph extension method converts the obtained extended entities and entity relationships into triples based on the inconsistency in the description of the data subject between the knowledge graph and the database containing unstructured data. These triples are then fused with the entities in the knowledge graph, mainly for entity alignment and integration with the data model. After fusion, a standard data representation is formed, ultimately resulting in fused extended entities and fused entity relationships. The fused extended entities are added after the first entity in the knowledge graph, and the relationships between the fused extended entities and the entities in the knowledge graph are established through the processed entity relationships, thus obtaining the extended graph.

[0042] The aforementioned entity alignment falls under the category of knowledge fusion, which involves merging descriptions of entities from different knowledge bases to obtain a complete description of the entity. For example, different knowledge bases may have different descriptions of a historical figure XX. The knowledge from different knowledge bases can be complemented and merged to form a comprehensive, accurate, and complete extended entity.

[0043] The knowledge graph expansion method provided in this embodiment first obtains a knowledge graph including a first entity, which is obtained through metadata of unstructured data. Second, in response to receiving a request to expand the knowledge graph, it obtains the data body of the unstructured data. Third, based on the data body and a visual interface, it obtains expanded entities related to the first entity and entity relationships related to the expanded entities. Finally, based on the expanded entities, entity relationships, and the knowledge graph, it obtains the expanded knowledge graph. Thus, by processing the data body using a visual interface to obtain expanded entities and their entity relationships, it achieves efficient and low-tech-threshold knowledge graph data processing for multi-source heterogeneous data, forming diverse knowledge graph entities and relationships, and enhancing the data fusion and analysis capabilities of the knowledge graph.

[0044] The above embodiments disclose the process of expanding entities. When expanding a knowledge graph, the entities and their attributes can be extended without changing them, only the relationships between entities can be expanded to obtain an expanded knowledge graph. In another embodiment of this disclosure, after obtaining the knowledge graph, the above graph expansion method may further include: in response to not receiving an entity expansion request for the knowledge graph, determining whether a relationship expansion request for the knowledge graph has been received; if a relationship expansion request for the knowledge graph has been received, determining at least two entities in the knowledge graph that need relationship expansion based on the relationship expansion request, displaying the at least two entities on a visual operation interface; editing the original entity relationship of the at least two entities based on the operation information of the visual operation interface to obtain the target entity relationship corresponding to the relationship expansion request, adding the target entity relationship to the knowledge graph to obtain the expanded graph.

[0045] In this embodiment, the relationship expansion request includes at least two entities whose relationship needs to be added, modified, or deleted, and the target entity relationship to be added, modified, or deleted for the at least two entities (the relationship that the relationship expansion request needs to modify). The relationship expansion request may also include: the original entity relationship of the at least two entities. By adding, modifying, or deleting the original entity relationship, the target entity relationship can be obtained. The target entity relationship is added to the knowledge graph to obtain the expanded graph. Compared to the knowledge graph, the expanded graph obtained in this embodiment does not change the entities or entity attribute values; only the entity relationships between the entities are changed.

[0046] In this embodiment, the operation information of the visual operation interface includes: the content of the knowledge graph displayed on the visual operation interface; and the user's operations such as moving or connecting the content of the knowledge graph to add, modify, and delete entities and entity relationships. In this embodiment, the entity that needs to expand its relationships can be any entity in the knowledge graph. The content of the knowledge graph can be all the entities displayed in the knowledge graph and the attribute values ​​of the entities. The user drags and drops the frame corresponding to the entity and connects the attribute values ​​displayed in the frame together.

[0047] To achieve unified control over multi-source heterogeneous data, in another embodiment of this disclosure, the above-mentioned graph expansion method may further include: normalizing the data entering the knowledge graph to obtain data that meets the normalization requirements; and normalizing the extended entities and entity relationships in response to the extended entities and entity relationships not meeting the normalization requirements.

[0048] In this embodiment, the data entering the knowledge graph is multi-source heterogeneous data of entities and entity relationships obtained after processing. Multi-source heterogeneous data may include unstructured data and database tables, etc. The multi-source heterogeneous data is mapped to a normalized data model for normalization processing to obtain data that meets the normalization requirements output by the normalized data model.

[0049] Similarly, when extended entities and entity relationships do not meet the normalization requirements, the extended entities and entity relationships are input into the normalized data model and normalized to obtain normalized extended entities and entity relationships.

[0050] To better obtain extended entities and entity relationships, this disclosure provides a method for obtaining extended entities related to a first entity and entity relationships related to the extended entities. See [link to relevant documentation]. Figure 2 The document illustrates a flow 200 of an embodiment of a method for obtaining extended entities and entity relationships according to this disclosure. The method for obtaining extended entities associated with a first entity and entity relationships associated with the extended entities includes the following steps:

[0051] Step 201: Based on the type of the data subject, determine the preset artificial intelligence algorithm corresponding to the type, and display the name of the artificial intelligence algorithm on the visual operation interface.

[0052] In this embodiment, for various types of unstructured data, there are corresponding artificial intelligence algorithms. These artificial intelligence algorithms can process unstructured data to obtain structured data of entities and entity attribute values ​​used to realize knowledge graphs.

[0053] In this embodiment, the data subject can be text, video, or voice, etc. For different types of data subjects, the graph expansion method and its execution entity have corresponding testing tools for different types of data subjects. For example, for the text type, a natural language model is used to identify whether the data subject can output text results.

[0054] Step 202: Based on the operation information obtained from the visual operation interface, determine the selected artificial intelligence algorithm.

[0055] In this embodiment, the visual operation interface displays all artificial intelligence algorithms corresponding to the current type of data subject. When the user has different entity processing needs, they can select the name of the artificial intelligence algorithm by operating the controls on the visual operation interface (such as the selection box in front of the name of different artificial intelligence algorithms) and obtain the operation information of the selected artificial intelligence algorithm. The execution subject running on it in the graph expansion method determines the selected artificial intelligence algorithm based on the operation information.

[0056] Optionally, when multiple artificial intelligence algorithms are selected, the order of the selected artificial intelligence algorithms can be determined by the user's operation sequence on the visual operation interface. When the selected artificial intelligence algorithm is used to process the data subject, the data subject is processed in sequence according to the artificial intelligence algorithm corresponding to that order to obtain the extended entity and the attribute value of the extended entity related to the first entity.

[0057] Step 203: The selected artificial intelligence algorithm is used to process the data subject to obtain the extended entity and the attribute values ​​of the extended entity related to the first entity.

[0058] In this embodiment, the extended entity can be a physical entity such as a person, equipment, or vehicle, or an abstract entity with elements such as time, place, and people, such as a traffic accident or company establishment; the data subject is the specific content of unstructured data files such as text documents, audio / video files, and image files. For example... Figure 3 As shown, the data body includes: video body S and text body W. The ship entity T in video body S is extracted using an artificial intelligence algorithm, and the vehicle entity Y and its attribute values, as well as the person entity and its attribute values, are extracted from text body W using the same algorithm. Figure 3 In the process, based on the attribute values ​​of character entities R1, R2, and R3, it can be concluded that all three entities have an entity relationship with vehicle entity Y.

[0059] In this embodiment, the artificial intelligence algorithm is an algorithm for extracting entities and entity attribute values ​​from a data subject. For example, for a data subject that is text-type, the artificial intelligence algorithm can be a keyword extraction algorithm and an entity recognition algorithm. The keyword extraction algorithm can extract keywords from the sentence, and the entity recognition algorithm can identify the entities in the sentence by recognizing the keywords.

[0060] Step 204: Based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph, determine the entity relationships related to the extended entity.

[0061] In this embodiment, after obtaining the attribute values ​​of the extended entity, the attribute values ​​of the extended entity are matched with the attribute values ​​of entities in the knowledge graph. If the attribute values ​​of the extended entity and the attribute values ​​of entities in the knowledge graph have matching fields, it is determined that the extended entity and the entity in the knowledge graph have an entity relationship where the content of a certain field appears. For example, if the extended entity is a director, the attribute value of the director is "directing movie Y"; if the entity in the knowledge graph is an actor, the attribute value of the actor is "participating in movie Y".

[0062] The method for obtaining extended entities and entity relationships provided in this embodiment displays the name of the artificial intelligence algorithm corresponding to the type of data subject on a visual operation interface for users to operate, thereby obtaining the selected artificial intelligence algorithm. This allows users to decide on the processing of the data subject according to their actual needs, satisfying user requirements and improving user experience.

[0063] In some optional implementations of this embodiment, the data subject is of type text, and the selected artificial intelligence algorithm is used to process the data subject to obtain the extended entity and the attribute value of the extended entity related to the first entity, including: using a material recognition algorithm or an entity extraction algorithm to extract the extended entity and the attribute value of the extended entity in the data subject.

[0064] In this embodiment, each entity in the knowledge graph has at least one attribute, and each attribute has a corresponding attribute value. For example, the entity is: student, and the attribute of student includes gender, and the attribute value of gender includes male and female.

[0065] In this embodiment, the material recognition algorithm is used to identify materials in the text data, such as plastic, steel, and wood. The entity extraction algorithm is used to extract entities from the text data, such as people, ships, vehicles, and events.

[0066] The method for processing data subjects provided in this embodiment uses material recognition algorithms or entity extraction algorithms to extract extended entities and attribute values ​​of extended entities for text-type data subjects, providing an optional implementation method for processing text-type data subjects and improving the reliability of data subject information extraction.

[0067] Optionally, in order to process the text-type data subject, before using material recognition algorithms or entity extraction algorithms to extract extended entities and attribute values ​​of extended entities from the data subject, the method further includes: preprocessing the data subject, wherein the preprocessing includes: removing punctuation marks, removing numbers, and segmenting sentences, etc. By preprocessing the data subject, the content of the data subject can be effectively obtained, and the accuracy of entity acquisition can be improved.

[0068] In one embodiment of this disclosure, the data subject is of type video. The selected artificial intelligence algorithm is used to process the data subject to obtain extended entities and attribute values ​​of the extended entities related to the first entity. This includes: using a machine vision algorithm to extract video entities and attribute values ​​of the video entities from the data subject, using the video entities as extended entities related to the first entity, and using the attribute values ​​of the video entities as attribute values ​​of the extended entities.

[0069] In this embodiment, the machine vision algorithm is an algorithm for extracting entities from video data. Through the machine vision algorithm, entities such as people, objects, and animals appearing in video data can be automatically extracted.

[0070] The method for obtaining extended entities and their attribute values ​​provided in this embodiment uses machine vision algorithms to extract video entities and their attribute values ​​from the data subject, providing an optional implementation method for processing video-type data subjects and improving the reliability of data subject information extraction.

[0071] Optionally, the data subject can also be speech, and there are corresponding artificial intelligence algorithms for speech processing, such as speech recognition algorithms and speech synthesis algorithms. The above-mentioned use of selected artificial intelligence algorithms to process the data subject and obtain extended entities and attribute values ​​of extended entities related to the first entity includes: using a speech recognition algorithm to extract speech entities and attribute values ​​of speech entities from the data subject, using the speech entities as extended entities related to the first entity, and using the attribute values ​​of the speech entities as attribute values ​​of the extended entities.

[0072] In some optional implementations of this embodiment, the machine vision algorithm includes: a human pose prediction algorithm and a face recognition algorithm; the extraction of video entities and attributes of video entities from the data subject using the machine vision algorithm includes: extracting at least one event entity and the attribute values ​​of each event entity from the data subject using the human pose prediction algorithm; and extracting at least one person entity and the attribute values ​​of each person entity from the data subject using the face recognition algorithm.

[0073] In this embodiment, an event entity is an abstract entity that includes elements such as time, location, and people, encompassing events that are currently occurring, have occurred historically, or are predicted to occur. The human pose prediction algorithm is used to predict the poses of various individuals in time-frequency data, determining their activities based on their poses. An event entity can also be identified as an event where a person is falling, as determined by the human pose prediction algorithm. The face recognition algorithm is used to identify people in video data and annotate them as entity objects. It should be noted that both the human pose prediction algorithm and the face recognition algorithm are mature technologies, and their specific implementation details will not be elaborated upon here.

[0074] The method for extracting video entities and their attributes provided in this embodiment employs a human pose prediction algorithm to extract at least one event entity and the attribute values ​​of each event entity from the data subject; and employs a face recognition algorithm to extract at least one person entity and the attribute values ​​of each person entity from the data subject. By extracting event entities and people from video-type data subjects respectively, the reliability of video entity extraction is improved.

[0075] In some optional implementations of this embodiment, determining the entity relationships related to the extended entity based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph includes:

[0076] In response to the fact that the extended entities only include event entities, a first entity relationship between the extended entities is determined based on the inherent connections between the event entities; in response to the fact that the attribute values ​​of the event entities are the same as or similar to the attribute values ​​of the entities in the knowledge graph, a second entity relationship between the entities in the knowledge graph and the event entities is determined; the first entity relationship and the second entity relationship are used together as the entity relationship related to the extended entities.

[0077] In this embodiment, after multiple event entities are determined, each event entity has a fixed relationship with the others. For example, the event entity "harvesting rice" has an inherent relationship with "harvesting wheat", as both are the harvesting of crops.

[0078] In this embodiment, when the attribute values ​​of an event entity and an entity in the knowledge graph are the same or similar, it is determined that the event entity and the entity in the knowledge graph have the same or similar attributes. Attributes are used to characterize the specific features of an entity, and specific features can characterize the inherent qualities of things. Therefore, event entities with the same features and entities in the knowledge graph must have an entity relationship.

[0079] The method for determining entity relationships related to extended entities provided by this optional implementation determines the first entity relationship between extended entities based on the inherent connections between event entities when the extended entities only include event entities. It also determines the second entity relationship between the event entities and the event entities based on the fact that the attribute values ​​of the event entities are the same or similar to the attribute values ​​of the entities in the knowledge graph. The first and second entity relationships are used together as the entity relationships related to the extended entities. Thus, the entity relationships of event entities are analyzed in a comprehensive manner, improving the reliability of the determination of entity relationships of event entities.

[0080] In some optional implementations of this disclosure, the determination of entity relationships related to the extended entity based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph includes:

[0081] In response to the fact that the attribute values ​​of the extended entity are the same as or similar to the attribute values ​​of the entity in the knowledge graph, it is determined that the entity in the knowledge graph and the extended entity have an entity relationship.

[0082] Optionally, determining entity relationships related to an extended entity based on the extended entity, its attribute values, and the knowledge graph may further include: determining that different extended entities have entity relationships in response to the same or similar attribute values.

[0083] It should be noted that entity relationships are not limited to equality or similarity based on attribute values. Entity relationships can also include contractual relationships, hierarchical relationships, etc. For example, the hierarchical relationship between companies.

[0084] The method for determining entity relationships provided by this optional implementation improves the reliability of determining entity relationships by searching the relationship between the attribute values ​​of the extended entity and the entities in the knowledge graph.

[0085] In this embodiment, the visual operation interface supports modification of entity names, entity attribute values, entity relationships, relationship names, and other content. Specifically, the visual operation interface is equipped with modification controls. After the user clicks the modification controls, the content on the extended graph is displayed on the visual operation interface. Furthermore, on the visual operation interface, the entity names, entity attribute values, entity relationships, relationship names, and other information in the extended graph can be deleted, modified, added, or queried.

[0086] In some embodiments of this disclosure, in response to the extended entity including the event entity, rich attributes can be added to the event entity based on the geographic location information and time information in the metadata corresponding to the first entity, such as the event entity's time and the event entity's location; the attribute values ​​after adding attributes to the event entity are determined by artificial intelligence algorithms.

[0087] In some embodiments of this disclosure, the above-mentioned graph expansion method further includes: using a graph execution algorithm to process the expanded graph to obtain a processed expanded graph. The graph execution algorithm can greatly expand the connectivity of each entity in the graph. Specifically, the graph execution algorithm may include: finding the shortest path between different entities, an entity centrality algorithm in the graph, a community detection algorithm, etc.

[0088] See also Figure 4 As a response to the above Figure 1 To implement the method shown, this application provides a map expansion device, which is related to... Figure 1 Corresponding to the method embodiments shown, this device can be applied to various electronic devices.

[0089] like Figure 4As shown, the knowledge graph expansion device 400 of this embodiment may include: a knowledge graph acquisition unit 401, a subject acquisition unit 402, a obtaining unit 403, and an expansion unit 404. The knowledge graph acquisition unit 401 can be configured to acquire a knowledge graph including a first entity, the first entity being obtained through metadata of unstructured data. The subject acquisition unit 402 can be configured to acquire the data subject of unstructured data in response to receiving an entity expansion request for the knowledge graph. The obtaining unit 403 can be configured to obtain extended entities related to the first entity and entity relationships related to the extended entities based on the data subject and a visual operation interface. The expansion unit 404 can be configured to obtain an extended knowledge graph based on the extended entities, entity relationships, and the knowledge graph.

[0090] In some embodiments of this disclosure, the aforementioned obtaining unit 403 is further configured to: determine a preset artificial intelligence algorithm corresponding to the type of the data subject, and display the name of the artificial intelligence algorithm on a visual operation interface; determine the selected artificial intelligence algorithm based on the operation information obtained from the visual operation interface; process the data subject using the selected artificial intelligence algorithm to obtain an extended entity and attribute values ​​of the extended entity related to the first entity; and determine the entity relationship related to the extended entity based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph.

[0091] In some embodiments of this disclosure, the data subject is of type text, and the obtaining unit 403 is further configured to: extract extended entities and attribute values ​​of extended entities from the data subject using a material recognition algorithm or an entity extraction algorithm.

[0092] In some embodiments of this disclosure, the data subject is of the type of video, and the obtaining unit 403 is further configured to: extract video entities and attribute values ​​of video entities from the data subject using a machine vision algorithm, use video entities as extended entities related to the first entity, and use the attribute values ​​of video entities as attribute values ​​of the extended entities.

[0093] In some embodiments of this disclosure, the machine vision algorithm includes: a human pose prediction algorithm and a face recognition algorithm; the obtaining unit 403 is further configured to: use the human pose prediction algorithm to extract at least one event entity and the attribute values ​​of each event entity from the data subject; and use the face recognition algorithm to extract at least one person entity and the attribute values ​​of each person entity from the data subject.

[0094] In some embodiments of this disclosure, the above-mentioned obtaining unit 403 is further configured to: in response to the extended entity only including event entities, determine a first entity relationship between the extended entities based on the inherent connection between the event entities; in response to the attribute value of the event entity being the same as or similar to the attribute value of the entity in the knowledge graph, determine a second entity relationship between the entity in the knowledge graph and the event entity; and use the first entity relationship and the second entity relationship together as the entity relationship related to the extended entity.

[0095] In some embodiments of this disclosure, the above-mentioned obtaining unit 403 is further configured to: determine that the entity in the knowledge graph and the extended entity have an entity relationship in response to the attribute value of the extended entity being the same as or similar to the attribute value of the entity in the knowledge graph.

[0096] The knowledge graph expansion device provided in this embodiment firstly, the knowledge graph acquisition unit 401 acquires a knowledge graph including a first entity, which is obtained through metadata of unstructured data; secondly, the subject acquisition unit 402, in response to receiving a request for entity expansion of the knowledge graph, acquires the data subject of the unstructured data; thirdly, the acquisition unit 403, based on the data subject and a visual operation interface, obtains the expanded entities related to the first entity and the entity relationships related to the expanded entities; finally, the expansion unit 404, based on the expanded entities, entity relationships, and the knowledge graph, obtains the expanded graph. Thus, by processing the data subject using a visual operation interface to obtain the expanded entities and their entity relationships, efficient and low-tech knowledge graph data processing is achieved for multi-source heterogeneous data, forming diverse knowledge graph entities and relationships, and enhancing the data fusion and analysis capabilities of the knowledge graph.

[0097] Figure 5 A schematic block diagram of an electronic device 500 illustrating a map expansion method according to an embodiment of the present disclosure is shown. Figure 5 As shown, the device 500 may include a processor 501 and a memory 502 storing a computer program. When the computer program is executed by the processor 501, the device 500 is made capable of performing actions such as... Figure 1 or Figure 2 The steps of the method are shown. In one example, device 500 may be a computer device or a cloud computing node.

[0098] In embodiments of this disclosure, processor 501 may be, for example, a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a processor based on a multi-core processor architecture, etc. Memory 502 may be any type of memory implemented using data storage technologies, including but not limited to random access memory, read-only memory, semiconductor-based memory, flash memory, disk storage, etc.

[0099] Furthermore, in embodiments of this disclosure, the device 500 may also include an input device 503, such as a microphone, keyboard, mouse, etc., for inputting multiple multimedia files to be mixed. Additionally, the device 500 may also include an output device 504, such as a loudspeaker, display, etc., for outputting the mixed multimedia files.

[0100] The electronic device using the map expansion method provided in this disclosure can be applied to any product with display functionality, such as electronic paper, mobile phones, tablet computers, televisions, laptops, digital photo frames, wearable devices, or navigators.

[0101] In other embodiments of this disclosure, a computer-readable storage medium storing a computer program is also provided, wherein the computer program, when executed by a processor, is capable of performing the following functions: Figures 1 to 2 The steps of the method shown.

[0102] The knowledge graph extension method disclosed herein first obtains a knowledge graph including a first entity, which is obtained through metadata of unstructured data. Second, in response to receiving a request for entity extension of the knowledge graph, it obtains the data body of the unstructured data. Third, based on the data body and a visual interface, it obtains extended entities related to the first entity and entity relationships related to the extended entities. Finally, based on the extended entities, entity relationships, and the knowledge graph, it obtains the extended knowledge graph. Thus, by processing the data body using a visual interface to obtain extended entities and their entity relationships, it achieves efficient and low-tech-threshold knowledge graph data processing for multi-source heterogeneous data, forming diverse knowledge graph entities and relationships, and enhancing the data fusion and analysis capabilities of the knowledge graph.

[0103] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatuses and methods according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0104] Unless otherwise expressly indicated by the context, the singular form of words used herein and in the appended claims includes the plural form, and vice versa. Thus, when referring to the singular, the plural form of the corresponding term is generally included. Similarly, the terms “comprising” and “including” shall be interpreted as including rather than exclusively. Likewise, the terms “including” and “or” shall be interpreted as including unless such interpretation is expressly prohibited herein. Where the term “example” is used herein, particularly when it follows a set of terms, “example” is merely exemplary and illustrative and should not be considered exclusive or extensive.

[0105] Further aspects and scope of adaptation become apparent from the description provided herein. It should be understood that various aspects of this disclosure may be implemented individually or in combination with one or more other aspects. It should also be understood that the descriptions and specific embodiments herein are for illustrative purposes only and are not intended to limit the scope of this disclosure.

[0106] Several embodiments of this disclosure have been described in detail above. However, it is obvious that those skilled in the art can make various modifications and variations to the embodiments of this disclosure without departing from the spirit and scope of this disclosure. The scope of protection of this disclosure is defined by the appended claims.

Claims

1. A method for map expansion, the method comprising: Obtain a knowledge graph including a first entity, which is obtained through metadata of unstructured data; In response to receiving a request to expand the entity of the knowledge graph, the data body of the unstructured data is obtained; Based on the type of the data subject, a preset artificial intelligence algorithm corresponding to the type is determined, and the name of the artificial intelligence algorithm is displayed on the visual operation interface; Based on the operation information obtained from the visual operation interface, the selected artificial intelligence algorithm is determined; The selected artificial intelligence algorithm is used to process the data subject to obtain extended entities related to the first entity and the attribute values ​​of the extended entities; Based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph, determine the entity relationships related to the extended entity; The step of determining the entity relationships related to the extended entity based on the extended entity, the attribute values ​​of the extended entity, and the knowledge graph includes: In response to the fact that the extended entity only includes event entities, a first entity relationship between the extended entities is determined based on the inherent relationships between the event entities; In response to the fact that the attribute value of the event entity is the same as or similar to the attribute value of the entity in the knowledge graph, a second entity relationship between the entity in the knowledge graph and the event entity is determined; The first entity relationship and the second entity relationship are collectively considered as the entity relationship related to the extended entity. Based on the extended entity, the entity relationship, and the knowledge graph, an extended graph is obtained; The data subject is of type text. The selected artificial intelligence algorithm is used to process the data subject to obtain extended entities related to the first entity and the attribute values ​​of the extended entities, including: The extended entities and attribute values ​​of the extended entities in the data body are extracted using material identification algorithms or entity extraction algorithms; Machine vision algorithms include: human pose prediction algorithms and face recognition algorithms; the extraction of video entities and their attributes from the data subject using these machine vision algorithms includes: The human pose prediction algorithm is used to extract at least one event entity and the attribute values ​​of each event entity from the data subject. The face recognition algorithm is used to extract at least one human entity and the attribute values ​​of each human entity from the data subject.

2. The method according to claim 1, wherein, The data subject is a video, and the selected artificial intelligence algorithm is used to process the data subject to obtain extended entities related to the first entity and the attribute values ​​of the extended entities, including: The machine vision algorithm is used to extract video entities and attribute values ​​from the data body. The video entities are used as extended entities related to the first entity, and the attribute values ​​of the video entities are used as attribute values ​​of the extended entities.

3. The method according to claim 2, wherein determining the entity relationship related to the extended entity based on the extended entity, the attribute value of the extended entity, and the knowledge graph includes: In response to the fact that the attribute value of the extended entity is the same as or similar to the attribute value of the entity in the knowledge graph, it is determined that the entity in the knowledge graph and the extended entity have an entity relationship.

4. A map expansion apparatus, the apparatus being used to perform a map expansion method as described in any one of claims 1-3, comprising: The graph acquisition unit is configured to acquire a knowledge graph including a first entity, which is obtained through metadata of unstructured data; The subject acquisition unit is configured to acquire the data subject of the unstructured data in response to receiving an entity expansion request for the knowledge graph; The unit is configured to obtain extended entities related to the first entity and entity relationships related to the extended entities based on the data subject and the visual operation interface; The extension unit is configured to obtain an extended graph based on the extended entity, the entity relationship, and the knowledge graph.

5. An electronic device, comprising: At least one processor; And at least one memory storing a computer program; When the computer program is executed by the at least one processor, the apparatus performs the steps of the method according to any one of claims 1 to 3.

6. A computer-readable storage medium storing a computer program, wherein, The computer program, when executed by a processor, implements the steps of the method according to any one of claims 1 to 3.

Images