A spatiotemporal embedding knowledge graph construction method and device, and a computing device

Abstract

The application provides a spatiotemporal embedded knowledge graph construction method and device and computing equipment, the method comprising: obtaining situation trajectory data and spatiotemporal intelligence corpus of a target; performing knowledge extraction and entity recognition on the situation trajectory data and spatiotemporal intelligence corpus to obtain knowledge layer information of the target; wherein the knowledge layer information comprises geographic space layer information, spatial trajectory layer information and spatial event layer information, and the knowledge layer information is represented by entities, relationships and attributes corresponding to the target; determining the event intent of the target according to the geographic space layer information and the spatial trajectory layer information; fusing and reasoning the event intent and the spatial event layer information to obtain the implied relationship between entities; and constructing a knowledge graph according to the knowledge layer information and the implied relationship. The spatiotemporal embedded knowledge graph construction method provided by the scheme realizes complex reasoning based on spatial relationships, and improves the logical reasoning capability and potential intelligence acquisition capability of the knowledge graph.

Description

Technical Field

[0001] This invention relates to the field of knowledge graph technology, and in particular to a spatiotemporal embedded knowledge graph construction method, apparatus, and computing device. Background Technology

[0002] Traditional situational trajectory data is primarily stored in structured databases and triplet formats, with retrieval as the main processing method. It often relies on rapid data processing and integration in the human mind to form a situational description based on spatial, temporal, and causal relationships. However, relying on manual processing for large amounts of target situational data and unstructured text corpora makes it difficult to guarantee accuracy. Furthermore, due to the diverse sources of the data, it is impossible to fuse them or perform complex logical reasoning based on spatial relationships. Consequently, it is difficult to effectively support comprehensive intelligence analysis such as target threat, situational awareness, and intent analysis, and cannot promptly assist humans in making rapid responses.

[0003] Because the target's intent is largely hidden in the spatial patterns and event sequences of situational data, and traditional knowledge graphs are mainly based on retrieval, they lack the ability to express spatial relationships such as distance, orientation, and topology during the spatiotemporal data modeling process. This makes it impossible to perform complex reasoning based on spatial relationships, resulting in poor logical reasoning ability and weak potential intelligence acquisition capability. Summary of the Invention

[0004] This invention provides a spatiotemporal embedded knowledge graph construction method, apparatus, and computing device, which realizes complex reasoning based on spatial relationships and improves the logical reasoning ability and potential intelligence acquisition ability of the knowledge graph.

[0005] In a first aspect, embodiments of the present invention provide a method for constructing a spatiotemporal embedded knowledge graph, including:

[0006] Acquire target situational trajectory data and spatiotemporal intelligence corpus;

[0007] Knowledge extraction and entity recognition are performed on the situation trajectory data and the spatiotemporal intelligence corpus to obtain the knowledge layer information of the target; wherein, the knowledge layer information includes geospatial layer information, spatial trajectory layer information and spatial event layer information, and the knowledge layer information is represented by entities, relationships and attributes corresponding to the target;

[0008] Based on the geospatial layer information and the spatial trajectory layer information, the event intent of the target is determined;

[0009] By fusing and reasoning the event intent with the spatial event layer information, the implicit relationships between entities can be obtained;

[0010] A knowledge graph is constructed based on the knowledge layer information and the implicit relationships.

[0011] Optionally, the step of extracting knowledge and identifying entities from the situation trajectory data and the spatiotemporal intelligence corpus to obtain the knowledge layer information of the target includes:

[0012] Knowledge extraction is performed on the aforementioned situational trajectory data to obtain geospatial information and spatial trajectory information;

[0013] The spatial region where the target is located is determined based on the geospatial information;

[0014] Discretize the spatial region into a grid to obtain a spatial entity;

[0015] Based on the spatial trajectory information of the target, the transfer relationships between the spatial entities, the association relationships between the target entity and the spatial entities, and the attributes of the target are determined; wherein, the attributes include motion sequence information, motion time, velocity, acceleration, direction, starting point, and ending point;

[0016] Based on the spatial entity, the target entity, the transfer relationship, the association relationship, and the attributes of the target, determine the geospatial layer information and the spatial trajectory layer information;

[0017] Knowledge extraction and entity recognition are performed on the spatiotemporal intelligence corpus to obtain the spatial event layer information; wherein, the entities in the spatial event information include spatial entities, target entities, and event entities, the relationships include coreference relationships, temporal relationships, causal relationships, and sub-event relationships between entities, and the attributes include time, location, and event type; the event entities include event trigger words, event subjects, and event objects.

[0018] Optionally, after acquiring the target's situational trajectory data and spatiotemporal intelligence corpus, and before determining the target's event intent, the method further includes:

[0019] The situational trajectory data of the target acquired by at least two detection devices are completed to obtain at least two completed trajectory information;

[0020] The complete trajectory information of the target is obtained by fusing at least two of the completed trajectory information.

[0021] Optionally, determining the event intent of the target includes:

[0022] The knowledge extraction and entity recognition are performed on the full trajectory information to obtain the geospatial layer information and spatial trajectory layer information of the corresponding spatial entities;

[0023] Each spatial entity containing the geospatial layer information and the spatial trajectory layer information is represented by a vector to obtain a semantic vector;

[0024] The semantic vector of each spatial entity is input into a pre-trained trajectory sequence model in chronological order according to the full trajectory information, and the event intent of the target is output.

[0025] Optionally, the trajectory sequence model is trained using at least two sets of samples, each set of samples including a sequence of semantic vectors of the target as input within a first historical time period and the event intent of the target as output; wherein the sequence of semantic vectors is a sequence of semantic vectors arranged chronologically according to the full trajectory information of the target.

[0026] Optionally, obtaining the semantic vector includes:

[0027] For each of the aforementioned spatial entities, the following is performed:

[0028] Identify the adjacent spatial entities that have a spatial adjacency relationship with the spatial entity, and aggregate the spatial entity with the adjacent spatial entities to obtain the semantic vector corresponding to the spatial entity;

[0029] The spatial entity is aggregated with the adjacent spatial entities to obtain an aggregated vector;

[0030] Obtain the attribute values ​​of the spatial entity, and encode the attribute values ​​to obtain an entity attribute encoding vector; wherein, the attribute values ​​include the time, velocity, acceleration, direction, starting point and ending point of the target corresponding to the spatial entity;

[0031] The spatial location of the spatial entity is obtained, and the spatial location is vectorized using a preset granularity to obtain a spatial feature encoding vector; wherein the preset granularity is different from the granularity of the spatial entity.

[0032] The semantic vector is obtained by concatenating the aggregation vector, the entity attribute encoding vector, and the spatial feature encoding vector.

[0033] Optionally, the step of fusing and reasoning the event intent with the spatial event layer information to obtain the implicit relationships between entities includes:

[0034] Entity recognition is performed on the event intent to obtain the first event entity;

[0035] Based on the geospatial layer information and the spatial trajectory layer information, a second entity that is related to the first event entity is determined;

[0036] Determine whether the first event entity exists in the spatial event layer information;

[0037] If it exists, then obtain the third entity in the spatial event layer information that is related to the first event entity, and determine that there is an implicit relationship between the second entity and the third entity.

[0038] Secondly, embodiments of the present invention also provide a spatiotemporal embedded knowledge graph construction apparatus, comprising:

[0039] The acquisition module is used to acquire the target's situational trajectory data and spatiotemporal intelligence corpus;

[0040] The extraction and recognition module is used to extract knowledge and recognize entities from the situation trajectory data and the spatiotemporal intelligence corpus to obtain the knowledge layer information of the target; wherein, the knowledge layer information includes geospatial layer information, spatial trajectory layer information and spatial event layer information, and the knowledge layer information is represented by entities, relationships and attributes corresponding to the target;

[0041] The determination module is used to determine the event intent of the target based on the geospatial layer information and the spatial trajectory layer information;

[0042] The fusion reasoning module is used to perform fusion reasoning on the event intent and the spatial event layer information to obtain the implicit relationships between entities;

[0043] A construction module is used to construct a knowledge graph based on the knowledge layer information and the implicit relationships.

[0044] Thirdly, embodiments of the present invention also provide a computing device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the spatiotemporal embedded knowledge graph construction method described in any of the above claims.

[0045] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program thereon, which, when executed in a computer, causes the computer to execute the spatiotemporal embedded knowledge graph construction method described in any of the preceding claims.

[0046] This invention provides a method, apparatus, and computing device for constructing a spatiotemporally embedded knowledge graph. The method extracts knowledge and identifies entities from situational trajectory data and spatiotemporal intelligence corpora, expressing the two types of target data using entities, relationships, and attributes. This embeds spatiotemporal relationships into the knowledge graph, effectively organizing massive amounts of data. Simultaneously, based on geospatial and spatial trajectory layer information, the method determines the target's event intent. Through reasoning and fusion of event intent and spatial event layer information, it further infers the implicit relationships between the target and related events, ultimately constructing a knowledge graph containing knowledge layer information and implicit relationships. Thus, the knowledge graph constructed by this invention effectively organizes massive amounts of data, not only realizing the spatiotemporal association between targets and events but also enabling complex reasoning based on spatial relationships, thereby improving the logical reasoning ability and potential intelligence acquisition capability of the knowledge graph. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart of a spatiotemporal embedded knowledge graph construction method provided in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of mapping the GIS trajectory of a target ship to a spatial entity according to an embodiment of the present invention;

[0050] Figure 3 This is a schematic diagram of a fusion reasoning process provided in an embodiment of the present invention;

[0051] Figure 4 This is a hardware architecture diagram of a computing device provided in an embodiment of the present invention;

[0052] Figure 5 This is a structural diagram of a spatiotemporal embedded knowledge graph construction device provided in an embodiment of the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0054] Although command and decision-makers have accumulated a wealth of expert knowledge, they still struggle to make effective judgments and decisions in real time when faced with massive amounts of situational information. Therefore, there is an urgent need to explore more effective knowledge storage and representation methods to connect the "data-knowledge-application" link and improve the level of target situational awareness. Simultaneously, since target intentions are largely hidden in the spatial patterns and time series of situational data, and the space itself, in addition to latitude and longitude coordinates, also possesses environmental attribute information and is correlated through the movement relationships of targets, there is an urgent need to explore ontology modeling methods based on spatial entities to enhance the semantic expression capabilities of spatial entities. Furthermore, based on multi-modal integrated graph logical reasoning, the ability to acquire potential intelligence and logical links can be improved.

[0055] However, existing traditional intelligence knowledge graphs primarily rely on retrieval, lacking the ability to express spatial relationships such as distance, orientation, and topology in their spatiotemporal data modeling process, thus hindering complex reasoning based on spatial relationships. Therefore, this application integrates knowledge graphs with spatiotemporal associations, combining large-scale graph structure analysis, geospatial analysis, and event sequence analysis techniques to analyze locations, targets, and events directly related to space. This allows for the extraction of high-value information to grasp the target's activity trajectory, analyze spatial trends of events, control related objects and events within a region, and achieve logical spatiotemporal connections between the target itself and target events.

[0056] Please refer to Figure 1 This invention provides a method for constructing a spatiotemporal embedded knowledge graph, the method comprising:

[0057] Step 100: Obtain the target's situational trajectory data and spatiotemporal intelligence corpus;

[0058] Step 102: Extract knowledge and identify entities from the situation trajectory data and spatiotemporal intelligence corpus to obtain the target's knowledge layer information; wherein, the knowledge layer information includes geospatial layer information, spatial trajectory layer information and spatial event layer information, and the knowledge layer information is represented by entities, relationships and attributes corresponding to the target;

[0059] Step 104: Determine the target's event intent based on geospatial layer information and spatial trajectory layer information;

[0060] Step 106: Integrate the event intent with the spatial event layer information to obtain the implicit relationships between entities;

[0061] Step 108: Construct a knowledge graph based on knowledge layer information and implicit relationships.

[0062] In this embodiment of the invention, the method extracts knowledge and identifies entities from situational trajectory data and spatiotemporal intelligence corpora, expressing the two types of target data using entities, relationships, and attributes. This embeds spatiotemporal relationships into a knowledge graph, effectively organizing massive amounts of data. Simultaneously, based on geospatial and spatial trajectory information, the method determines the target's event intent. Through reasoning and fusion of event intent and spatial event information, it further infers the implicit relationships between the target and related events, ultimately constructing a knowledge graph containing knowledge layer information and implicit relationships. Thus, the knowledge graph constructed by this invention effectively organizes massive amounts of data, not only realizing the spatiotemporal association between targets and events but also enabling complex reasoning based on spatial relationships, thereby improving the logical reasoning ability and potential intelligence acquisition capability of the knowledge graph.

[0063] The following description Figure 1 The execution method for each step is shown.

[0064] First, for step 100, acquire the target's situational trajectory data and spatiotemporal intelligence corpus. Specifically, the situational trajectory data includes real-time motion trajectory data of different targets and real-time motion trajectory data of the same target monitored by different sensors or detection devices; the spatiotemporal intelligence corpus includes the occurrence of different types of events and the logical relationships between events. Acquisition methods include, but are not limited to, crawling relevant data using web crawling technology.

[0065] After step 100, the method further includes: completing the target situation trajectory data obtained by at least two detection devices to obtain at least two completed trajectory information;

[0066] At least two complete trajectory information are fused to obtain the target's complete trajectory information.

[0067] In this invention, by performing trajectory completion and multi-source trajectory fusion on the trajectory data of the target, support can be provided for subsequent analysis using the trajectory sequence model, thereby improving the accuracy of event intent classification and trajectory prediction for future moments in the trajectory sequence model.

[0068] Then, for step 102, knowledge extraction and entity recognition are performed on the situational trajectory data and spatiotemporal intelligence corpus to obtain the target's knowledge layer information; wherein, the knowledge layer information includes geospatial layer information, spatial trajectory layer information, and spatial event layer information, and the knowledge layer information is represented by entities, relationships, and attributes corresponding to the target, including:

[0069] Knowledge extraction is performed on the situation trajectory data to obtain geospatial information and spatial trajectory information;

[0070] Determine the spatial region where the target is located based on geospatial information;

[0071] Discretize the spatial region into a grid to obtain a spatial entity;

[0072] Based on the target's spatial trajectory information, determine the transfer relationships between spatial entities, the association relationships between the target entity and spatial entities, and the target's attributes; among which, the attributes include motion sequence information, motion time, velocity, acceleration, direction, starting point, and ending point;

[0073] Based on the attributes of spatial entities, target entities, transfer relationships, association relationships, and targets, determine the geospatial layer information and spatial trajectory layer information;

[0074] Knowledge extraction and entity recognition are performed on the spatiotemporal intelligence corpus to obtain spatial event layer information. Among them, the entities in the spatial event information include spatial entities, target entities, and event entities. The relationships include coreference, temporal, causal, and sub-event relationships between entities. The attributes include time, location, and event type. Event entities include event trigger words, event subjects, and event objects.

[0075] It should be noted that the Earth's surface region is discretized into a grid, and the grid region is then abstracted into entities to obtain spatial entities. Spatial entities are connected by bidirectional directed edges based on physical adjacency relationships, representing the reachability relationships between spatial entity regions.

[0076] Specifically, the steps for extracting knowledge and events from intelligence corpora include event entity extraction, event trigger word extraction, element role extraction, event extraction, and event logical relationship extraction, followed by merging and disambiguation of similar events. Element roles refer to the relationship between event trigger words and event entities, extracting meta-events. In event extraction, an event is composed of an event trigger word as its core, along with one or more event elements. The event trigger word is the word that best represents the occurrence of the event, and the event elements are the participants in the event; the role played by the event elements in the entire event is the corresponding element role. The main function of event relationship extraction is to extract the relationships between events within a sentence, on a sentence-by-sentence basis. The most important difference between event relationships and entity relationships is that entity relationships are specific relationship types between specific entity categories, while event relationships are a finite number of event relationship types between any two types of events, including but not limited to logical relationships such as causal relationships and sequential relationships.

[0077] In this invention, a knowledge graph framework for target situation trajectory data is constructed based on spatial entities, target entities, transfer relationships, association relationships, and target attributes, divided into spatial and trajectory layers. This framework comprehensively considers temporal and spatial characteristics, enhancing the ability to express spatial relationships such as distance, orientation, and topology. Simultaneously, a spatiotemporal intelligence corpus is constructed based on a knowledge graph framework for spatial entities, target entities, event entities, coreference relationships between entities, temporal relationships, causal relationships, and sub-event relationships, as well as attributes such as time, location, and event type. This ensures that the knowledge graph simultaneously includes situation trajectory data and spatiotemporal intelligence corpus, enriching its data information and effectively supporting comprehensive intelligence analysis.

[0078] In some preferred embodiments, the spatial region is discretized into a grid to obtain a spatial entity, including:

[0079] Determine the data accuracy of the situation trajectory data;

[0080] The target granularity for dividing the spatial region is determined based on the size of the spatial region and the data accuracy.

[0081] The spatial region is discretized and meshed according to the target granularity to obtain the spatial entity.

[0082] In this invention, the target granularity for dividing the spatial region is determined based on the size of the spatial region and the data precision. The spatial region is then discretized and gridded according to the target granularity to obtain spatial entities. This allows for the reasonable allocation and determination of the data corresponding to each spatial entity, ensuring that the amount of data information within each spatial entity is neither redundant nor excessively lacking, which is more conducive to the subsequent construction of knowledge graphs and trajectory sequence models.

[0083] Specifically, such as Figure 2 The t1 to t2 shown n A schematic diagram showing the GIS trajectory of the target ship near the Taiwan Strait and its mapping to spatial entities within a given time period, where t1=T1, t n =T n The data precision of the GIS trajectory and the size of the Taiwan Strait region are determined, the target granularity for dividing the spatial region is determined, and the Taiwan Strait is discretized into a grid to obtain n spatial entities. Here, S1 is the spatial entity corresponding to time T1, and S... n For T n The spatial entity corresponding to a given moment.

[0084] In step 104, the event intent of the target is determined, including:

[0085] Knowledge extraction and entity recognition are performed on the full trajectory information to obtain the geospatial layer information and spatial trajectory layer information of the corresponding spatial entities;

[0086] Each spatial entity containing geospatial layer information and spatial trajectory layer information is represented by a vector to obtain a semantic vector;

[0087] The semantic vector of each spatial entity is input into a pre-trained trajectory sequence model in chronological order according to the full trajectory information, and the event intent of the target is output.

[0088] In a preferred embodiment, the trajectory sequence model is trained using at least two sample sets. Each sample set includes a sequence of semantic vectors of the target as input within a first historical time period and the target's event intent as output. The semantic vector sequence is a sequence of semantic vectors arranged chronologically according to the target's full trajectory information. Specifically, after determining a target motion sequence, the target's full trajectory information is extracted into a target meta-event. The meta-event includes necessary event elements such as the event subject, event time, and event location. The target meta-event is matched against expert knowledge to determine if it is an intent event, and then labeled and used to train the trajectory sequence model.

[0089] In this invention, the trajectory time of each spatial entity on the spatial trajectory is determined according to the motion time of the target's spatial trajectory. All semantic vectors corresponding to the target are arranged in the time order of the motion trajectory, and the trajectory of the target on the timeline is retained to obtain the temporal trajectory of the semantic vectors. This is beneficial for accurately capturing the target's intent using a pre-trained trajectory sequence model, and can even predict its dynamic changes at the next moment.

[0090] In a preferred embodiment, the semantic vector is obtained, including:

[0091] For each spatial entity, execute:

[0092] Identify the adjacent spatial entities that have a spatial adjacency relationship with the spatial entity, and aggregate the spatial entity and the adjacent spatial entities to obtain the semantic vector corresponding to the spatial entity;

[0093] The spatial entity is aggregated with adjacent spatial entities to obtain an aggregated vector;

[0094] Obtain the attribute values ​​of the spatial entity and encode the attribute values ​​to obtain the entity attribute encoding vector; wherein, the attribute values ​​include the time, velocity, acceleration, direction, starting point and ending point of the target when it corresponds to the spatial entity;

[0095] The spatial location of the spatial entity is obtained, and the spatial location is vectorized using a preset granularity to obtain a spatial feature encoding vector; wherein, the preset granularity is different from the granularity of the spatial entity.

[0096] The semantic vector is obtained by concatenating the aggregation vector, entity attribute encoding vector, and spatial feature encoding vector.

[0097] In this invention, the semantic vector contains the attribute values ​​of spatial entities and combines the representations of spatial entities at different granularities with the representations of spatial neighbor nodes, thus enriching the feature representation of spatial entities. Specifically, for each spatial entity, its neighboring spatial entities are first aggregated using graph aggregation methods, including average aggregation, max pooling aggregation, average pooling aggregation, GCN aggregation, LSTM aggregation, etc., to obtain an aggregated vector. Then, the attribute values ​​of the spatial entities are grouped, and the features within each group are embedded and then average pooled. The vector representations of each group are combined and encoded into a fixed dimension by an encoder as the node feature output, thus obtaining the entity attribute encoding vector. Next, the spatial entity vector representations at different granularities at the same spatial location are calculated to obtain the spatial feature encoding vector. Finally, the above three types of vectors are concatenated and a feature cross is performed to output a fixed-dimensional vector representation as the representation method of the spatial entity, thus obtaining the semantic vector.

[0098] In some preferred embodiments, the BERT4Rec model is used, which is a user behavior sequence modeling method based on bidirectional self-attention and Cloze task. Deep sequence model and Cloze task are introduced in the pre-training process. U={u1,u2,...,u|U|} is defined as the target set, V={v1,v2,...,v|V|} is the spatial entity set, and Su=[v1(u),...,vt(u),...,vnu(u)] is the semantic vector sequence within the historical time period.

[0099] Next, for step 106, the event intent and spatial event layer information are fused and reasoned to obtain the implicit relationships between entities, including:

[0100] Entity identification is performed on the event intent to obtain the first event entity;

[0101] Based on geospatial layer information and spatial trajectory layer information, identify the second entity that is related to the first event entity;

[0102] Determine whether a first event entity exists in the spatial event layer information;

[0103] If it exists, then obtain the third entity in the spatial event layer information that is related to the first event entity, and determine that there is an implicit relationship between the second entity and the third entity.

[0104] For example, such as Figure 3The diagram illustrating the fusion reasoning process shows that by analyzing the situation trajectory data of a target, its event intent is determined to be event B, and the event with a causal relationship to event B in the situation trajectory data is event A. In the spatial event layer information obtained from the spatiotemporal intelligence corpus, event B has a causal relationship with intelligence C. Therefore, it can be determined that there is an implicit causal relationship between event A and intelligence C.

[0105] In this invention, the event intent obtained through reasoning and the spatial event layer information are fused together to determine the implicit relationship between the first event entity contained in the event intent and the third entity in the spatial event layer information that has no direct relationship. The second entity and the third entity and their implicit relationship are assigned to the top-level event layer to construct a knowledge graph that includes the top-level event layer and knowledge layer information, thereby improving the logical reasoning ability and potential intelligence acquisition ability of the knowledge graph.

[0106] Finally, step 108 constructs a knowledge graph based on the knowledge layer information and implicit relationships.

[0107] like Figure 4 , Figure 5 As shown, this invention provides a spatiotemporal embedded knowledge graph construction device. The device can be implemented in software, hardware, or a combination of both. From a hardware perspective, as... Figure 4 The diagram shown is a hardware architecture diagram of a computing device housing a spatiotemporal embedded knowledge graph construction device provided in an embodiment of the present invention. (Except for...) Figure 4 In addition to the processor, memory, network interface, and non-volatile memory shown, the computing device in the embodiment may also include other hardware, such as a forwarding chip responsible for processing packets. Taking software implementation as an example, such as... Figure 5 As shown, as a logical device, it is formed by the CPU of its computing device reading the corresponding computer program from the non-volatile memory into memory and running it. This embodiment provides a spatiotemporal embedded knowledge graph construction device, including: an acquisition module 500, an extraction and identification module 502, a determination module 504, a fusion reasoning module 506, and a construction module 508;

[0108] The acquisition module 500 is used to acquire the target's situational trajectory data and spatiotemporal intelligence corpus;

[0109] The extraction and recognition module 502 is used to extract knowledge and recognize entities from situational trajectory data and spatiotemporal intelligence corpus to obtain the knowledge layer information of the target; wherein, the knowledge layer information includes geospatial layer information, spatial trajectory layer information and spatial event layer information, and the knowledge layer information is represented by entities, relationships and attributes corresponding to the target;

[0110] The determination module 504 is used to determine the event intent of the target based on geospatial layer information and spatial trajectory layer information;

[0111] The fusion reasoning module 506 is used to perform fusion reasoning on event intent and spatial event layer information to obtain implicit relationships between entities;

[0112] Module 508 is used to construct a knowledge graph based on knowledge layer information and implicit relationships.

[0113] In some specific implementations, the acquisition module 500 can be used to perform the above step 100, the extraction and identification module 502 can be used to perform the above step 102, the determination module 504 can be used to perform the above step 104, the fusion reasoning module 506 can be used to perform the above step 106, and the construction module 508 can be used to perform the above step 108.

[0114] In some specific implementations, the extraction and identification module 502 is also used to perform the following operations:

[0115] Knowledge extraction is performed on the situation trajectory data to obtain geospatial information and spatial trajectory information;

[0116] Determine the spatial region where the target is located based on geospatial information;

[0117] Discretize the spatial region into a grid to obtain a spatial entity;

[0118] Based on the target's spatial trajectory information, determine the transfer relationships between spatial entities, the association relationships between the target entity and spatial entities, and the target's attributes; among which, the attributes include motion sequence information, motion time, velocity, acceleration, direction, starting point, and ending point;

[0119] Based on the attributes of spatial entities, target entities, transfer relationships, association relationships, and targets, determine the geospatial layer information and spatial trajectory layer information;

[0120] Knowledge extraction and entity recognition are performed on the spatiotemporal intelligence corpus to obtain spatial event layer information. Among them, the entities in the spatial event information include spatial entities, target entities, and event entities. The relationships include coreference, temporal, causal, and sub-event relationships between entities. The attributes include time, location, and event type. Event entities include event trigger words, event subjects, and event objects.

[0121] In some specific embodiments, the device further includes a completion and fusion module, which performs the following operations:

[0122] The situational trajectory data of the target acquired by at least two detection devices are supplemented to obtain at least two supplementary trajectory information;

[0123] At least two complete trajectory information are fused to obtain the target's complete trajectory information.

[0124] In some specific implementations, the determining module 504 is also used to perform the following operations:

[0125] Knowledge extraction and entity recognition are performed on the full trajectory information to obtain the geospatial layer information and spatial trajectory layer information of the corresponding spatial entities;

[0126] Each spatial entity containing geospatial layer information and spatial trajectory layer information is represented by a vector to obtain a semantic vector;

[0127] The semantic vector of each spatial entity is input into a pre-trained trajectory sequence model in chronological order according to the full trajectory information, and the event intent of the target is output.

[0128] In some specific implementations, the trajectory sequence model is trained using at least two sets of samples, each set of samples including a sequence of semantic vectors of the target as input within a first historical time period and the event intent of the target as output; wherein, the sequence of semantic vectors is a sequence of semantic vectors arranged in chronological order according to the full trajectory information of the target.

[0129] In some specific implementations, the determining module 504 is also used to perform the following operations:

[0130] For each spatial entity, execute:

[0131] Identify the adjacent spatial entities that have a spatial adjacency relationship with the spatial entity, and aggregate the spatial entity and the adjacent spatial entities to obtain the semantic vector corresponding to the spatial entity;

[0132] The spatial entity is aggregated with adjacent spatial entities to obtain an aggregated vector;

[0133] Obtain the attribute values ​​of the spatial entity and encode the attribute values ​​to obtain the entity attribute encoding vector; wherein, the attribute values ​​include the time, velocity, acceleration, direction, starting point and ending point of the target when it corresponds to the spatial entity;

[0134] The spatial location of the spatial entity is obtained, and the spatial location is vectorized using a preset granularity to obtain a spatial feature encoding vector; wherein, the preset granularity is different from the granularity of the spatial entity.

[0135] The semantic vector is obtained by concatenating the aggregation vector, entity attribute encoding vector, and spatial feature encoding vector.

[0136] In some specific implementations, the fusion inference module 506 is also used to perform the following operations:

[0137] Entity identification is performed on the event intent to obtain the first event entity;

[0138] Based on geospatial layer information and spatial trajectory layer information, identify the second entity that is related to the first event entity;

[0139] Determine whether a first event entity exists in the spatial event layer information;

[0140] If it exists, then obtain the third entity in the spatial event layer information that is related to the first event entity, and determine that there is an implicit relationship between the second entity and the third entity.

[0141] It is understood that the structures illustrated in the embodiments of the present invention do not constitute a specific limitation on a spatiotemporal embedded knowledge graph construction device. In other embodiments of the present invention, a spatiotemporal embedded knowledge graph construction device may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0142] The information interaction and execution process between the modules in the above-mentioned device are based on the same concept as the method embodiment of the present invention, and the specific details can be found in the description of the method embodiment of the present invention, and will not be repeated here.

[0143] This invention also provides a computing device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a spatiotemporal embedded knowledge graph construction method according to any embodiment of this invention.

[0144] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform a spatiotemporal embedded knowledge graph construction method according to any embodiment of this invention.

[0145] Specifically, a system or apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer (or CPU or MPU) of the system or apparatus may read and execute the program code stored in the storage medium.

[0146] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.

[0147] Storage media embodiments for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0148] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0149] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion module connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion module execute some and all of the actual operations, thereby realizing the function of any of the above embodiments.

[0150] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0151] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as ROM, RAM, magnetic disk, or optical disk.

[0152] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a spatiotemporal embedded knowledge graph, characterized in that, include: Acquire target situational trajectory data and spatiotemporal intelligence corpus; Knowledge extraction and entity recognition are performed on the situation trajectory data and the spatiotemporal intelligence corpus to obtain the knowledge layer information of the target; wherein, the knowledge layer information includes geospatial layer information, spatial trajectory layer information and spatial event layer information, and the knowledge layer information is represented by entities, relationships and attributes corresponding to the target; The situational trajectory data of the target acquired by at least two detection devices is supplemented to obtain at least two supplemented trajectory information; the at least two supplemented trajectory information are fused to obtain the full trajectory information of the target; Based on the geospatial layer information and the spatial trajectory layer information, the event intent of the target is determined; By fusing and reasoning the event intent with the spatial event layer information, the implicit relationships between entities can be obtained; Construct a knowledge graph based on the knowledge layer information and the implicit relationships; The event intent for determining the target includes: The knowledge extraction and entity recognition are performed on the full trajectory information to obtain the geospatial layer information and spatial trajectory layer information of the corresponding spatial entities; Each spatial entity containing the geospatial layer information and the spatial trajectory layer information is represented by a vector to obtain a semantic vector; The semantic vector of each spatial entity is input into a pre-trained trajectory sequence model in chronological order according to the full trajectory information, and the event intent of the target is output. The obtained semantic vector includes: For each of the aforementioned spatial entities, the following is performed: Identify the adjacent spatial entities that have a spatial adjacency relationship with the spatial entity, and aggregate the spatial entity with the adjacent spatial entities to obtain the semantic vector corresponding to the spatial entity; The spatial entity is aggregated with the adjacent spatial entities to obtain an aggregated vector; Obtain the attribute values ​​of the spatial entity, and encode the attribute values ​​to obtain an entity attribute encoding vector; wherein, the attribute values ​​include the time, velocity, acceleration, direction, starting point and ending point of the target corresponding to the spatial entity; The spatial location of the spatial entity is obtained, and the spatial location is vectorized using a preset granularity to obtain a spatial feature encoding vector; wherein the preset granularity is different from the granularity of the spatial entity. The semantic vector is obtained by concatenating the aggregation vector, the entity attribute encoding vector, and the spatial feature encoding vector.

2. The method according to claim 1, characterized in that, The process of extracting knowledge and identifying entities from the situational trajectory data and the spatiotemporal intelligence corpus to obtain the knowledge layer information of the target includes: Knowledge extraction is performed on the aforementioned situational trajectory data to obtain geospatial information and spatial trajectory information; The spatial region where the target is located is determined based on the geospatial information; Discretize the spatial region into a grid to obtain a spatial entity; Based on the spatial trajectory information of the target, the transfer relationships between the spatial entities, the association relationships between the target entity and the spatial entities, and the attributes of the target are determined; wherein, the attributes include motion sequence information, motion time, velocity, acceleration, direction, starting point, and ending point; Based on the spatial entity, the target entity, the transfer relationship, the association relationship, and the attributes of the target, determine the geospatial layer information and the spatial trajectory layer information; Knowledge extraction and entity recognition are performed on the spatiotemporal intelligence corpus to obtain the spatial event layer information; wherein, the entities in the spatial event layer information include spatial entities, target entities, and event entities, the relationships include coreference relationships, temporal relationships, causal relationships, and sub-event relationships between entities, and the attributes include time, location, and event type; the event entities include event trigger words, event subjects, and event objects.

3. The method according to claim 1, characterized in that, The trajectory sequence model is trained using at least two sets of samples. Each set of samples includes a sequence of semantic vectors of the target as input within a first historical time period and the event intent of the target as output. The sequence of semantic vectors is a sequence of semantic vectors arranged chronologically according to the full trajectory information of the target.

4. The method according to any one of claims 1 to 3, characterized in that, The step of fusing and reasoning the event intent with the spatial event layer information to obtain the implicit relationships between entities includes: Entity recognition is performed on the event intent to obtain the first event entity; Based on the geospatial layer information and the spatial trajectory layer information, a second entity that is related to the first event entity is determined; Determine whether the first event entity exists in the spatial event layer information; If it exists, then obtain the third entity in the spatial event layer information that is related to the first event entity, and determine that there is an implicit relationship between the second entity and the third entity.

5. A spatiotemporal embedded knowledge graph construction device, characterized in that, For implementing the method as described in any one of claims 1 to 4, comprising: The acquisition module is used to acquire the target's situational trajectory data and spatiotemporal intelligence corpus; The extraction and recognition module is used to extract knowledge and recognize entities from the situation trajectory data and the spatiotemporal intelligence corpus to obtain the knowledge layer information of the target; wherein, the knowledge layer information includes geospatial layer information, spatial trajectory layer information and spatial event layer information, and the knowledge layer information is represented by entities, relationships and attributes corresponding to the target; The determination module is used to determine the event intent of the target based on the geospatial layer information and the spatial trajectory layer information; The fusion reasoning module is used to perform fusion reasoning on the event intent and the spatial event layer information to obtain the implicit relationships between entities; A construction module is used to construct a knowledge graph based on the knowledge layer information and the implicit relationships.

6. A computing device comprising a memory and a processor, wherein the memory stores a computer program, and the processor, when executing the computer program, implements the method as described in any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method of any one of claims 1-4.

Images