Travel graph construction method and device, computer equipment, readable storage medium and program product
By obtaining and processing vehicle operating status data for traffic emergencies, building a rational map, the safety issues of vehicle driver decisions in emergencies are solved, and driving safety and data processing efficiency are improved.
Patent Information
- Application Number
- CN202411068891.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-06
- Publication Date
- 2024-11-08
AI Technical Summary
The lack of a vehicle driver decision corpus and a rational map for emergencies in the prior art, making it difficult to ensure the safety of vehicle driving.
By obtaining the vehicle operation status data of traffic emergencies, extracting the characteristics of the event data, forming event groups and fusion characteristics, building a rational map, and ensuring the safety of vehicle driving under emergencies.
It enriches the driver's decision corpus in emergencies, improves vehicle driving safety, reduces the repetition of data processing, and improves the speed of construction of the logic map.
Smart Images

Figure CN118917400A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of vehicle technology, and in particular to a method, device, computer equipment, computer-readable storage medium and computer program product for constructing a causal graph. Background Art
[0002] In today's society, the field of automotive intelligence is developing rapidly, and L2-level driver assistance systems are becoming more and more popular. Industry-related forces are competing fiercely in the fields of sensors, algorithms, product forms, etc. However, faced with increasingly complex driver assistance systems and complex road conditions, it is unknown whether drivers can make correct decisions when emergencies occur, and what impacts will be caused if wrong decisions are made.
[0003] The event graph is a logical knowledge base that describes the evolutionary laws and patterns between events. At present, the event graph is mainly used in various fields such as power grid and transportation. However, the corpus used for driver decision-making in emergencies is very scarce, and there is no corresponding event graph, which makes it impossible to ensure vehicle driving safety in emergencies. Summary of the invention
[0004] Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product for constructing a causal graph that can ensure vehicle driving safety in emergencies in response to the above-mentioned technical problems.
[0005] In a first aspect, the present application provides a method for constructing a truth graph, comprising:
[0006] Acquire a plurality of event data, wherein the event data is control data of vehicle operation status for traffic emergencies;
[0007] Extract features from each event data respectively;
[0008] Based on the extracted features of each event data, determining matching event data from the multiple event data, and taking the matching event data as an event group;
[0009] Fusing features corresponding to the event data in the event group to obtain fused features of the event group;
[0010] Based on the fusion features of the event group and the features of the first event data among the multiple event data, a matter graph is constructed; the first event data is the event data among the multiple event data except the event data in the event group.
[0011] In one embodiment, extracting features from each event data separately includes:
[0012] For each event data, based on a pre-built logical keyword library, a target keyword in the current event data that matches a logical keyword in the logical keyword library is determined, event data containing the target keyword is extracted from the current event data, and explicit features of the current event data are obtained based on the event data containing the target keyword;
[0013] The latent features of the current event data are extracted through a long short-term memory network model.
[0014] In one embodiment, the determining matching event data from the plurality of event data based on the extracted features of each event data includes:
[0015] For each event data, the explicit features and implicit features of the current event data are fused to obtain fused features of the current event data;
[0016] Based on the fusion features of each event data, matching event data are determined from the multiple event data.
[0017] In one embodiment, the determining matching event data from the plurality of event data based on the fusion features of each event data includes:
[0018] For each event data, calculating the feature similarity between the fused features of the current event data and the fused features of each second event data in the multiple event data; the second event data is the event data in the multiple event data except the current event data;
[0019] The second event data whose corresponding feature similarity is greater than a preset feature similarity threshold and the current event data are used as the matched event data.
[0020] In one embodiment, after constructing the event graph based on the fusion features of the event group and the features of the first event data among the multiple event data, the method further includes:
[0021] Based on a preset graphic database, the event graph is visualized.
[0022] In one embodiment, extracting features from each event data separately includes:
[0023] Convert each event data into vector data;
[0024] For the vector data of each event data, calculate the vector similarity between the current vector data and each other vector data in the plurality of vector data except the current vector data; determine the target vector data in the other vector data whose corresponding vector similarity is greater than a preset vector similarity threshold; determine the repeated event data based on the event data of the current vector data and the event data of the target vector data;
[0025] Eliminate duplicate event data corresponding to each event data from the plurality of event data to obtain processed event data;
[0026] Perform feature extraction on the processed event data.
[0027] In a second aspect, the present application also provides a device for constructing a theory graph, the device comprising:
[0028] An acquisition module, used for acquiring a plurality of event data, wherein the event data is control data of a vehicle operation state for a traffic emergency;
[0029] An extraction module, used for extracting features from each event data;
[0030] A determination module, configured to determine matching event data from the plurality of event data based on the features of each event data extracted, and use the matching event data as an event group;
[0031] A fusion module, used for fusing features corresponding to the event data in the event group to obtain fusion features of the event group;
[0032] A construction module is used to construct an event graph based on the fusion features of the event group and the features of the first event data among the multiple event data; the first event data is the event data among the multiple event data except the event data in the event group.
[0033] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method of each embodiment when executing the computer program.
[0034] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method of each embodiment when executed by a processor.
[0035] In a fifth aspect, the present application also provides a computer program product, including a computer program, which implements the steps of the methods of each embodiment when executed by a processor.
[0036] The above-mentioned method, device, computer equipment, computer-readable storage medium and computer program product for constructing a cause-and-effect graph can collect control data on the vehicle operating status of traffic emergencies, enrich the corpus of driver decision-making under emergencies, and then extract features from each event data respectively. Based on the extracted features of each event data, matching event data are determined from multiple event data, and the matching event data are used as an event group. The features corresponding to the event data in the event group are fused to obtain the fusion features of the event group. Based on the fusion features of the event group and the features of the first event data in multiple event data, a cause-and-effect graph is constructed. In this way, a cause-and-effect graph for traffic emergencies can be constructed to ensure vehicle driving safety under emergencies. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0038] Figure 1 A schematic diagram of a process of constructing a truth graph in one embodiment;
[0039] Figure 2 A schematic diagram of a process of constructing a truth graph in one embodiment;
[0040] Figure 3 A schematic diagram of a process flow of visualizing a matter graph in an embodiment;
[0041] Figure 4 A structural block diagram of a device for constructing a truth graph in an embodiment;
[0042] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0043] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0044] In one embodiment, Figure 1 As shown, a method for constructing a causal graph is provided. This embodiment takes the method applied to a vehicle as an example for explanation. In this embodiment, the method includes the following steps:
[0045] Step 102: Acquire multiple event data, where the event data is control data of vehicle operation status for traffic emergencies;
[0046] The scope of event data collection is determined to be social networking platforms, news media, and academic journals.
[0047] The driver controls the vehicle when facing a traffic emergency, so the vehicle generates control data of the vehicle's operating status based on the driver's control. The control data can be crawled through the crawler framework based on the Python language. If it is an unmanned vehicle, the vehicle automatically controls the vehicle to generate control data of the vehicle's operating status when a traffic emergency occurs.
[0048] Traffic emergencies include incidents caused by other traffic participants violating traffic rules, incidents caused by obstacles not marked in high-precision maps, or other unconventional traffic incidents.
[0049] The control data of the vehicle operation state includes the operation parameter data of the vehicle, which may be the vehicle speed, position, acceleration, wheel direction angle, etc.;
[0050] Step 104: extract features from each event data respectively;
[0051] Feature extraction is performed on each event data to obtain explicit features and implicit features of each event data, and the explicit features and implicit features are used as the extracted features of each event data.
[0052] Step 106: Based on the extracted features of each event data, matching event data are determined from the multiple event data, and the matching event data are used as an event group;
[0053] Based on the features of each event data in the multiple event data, the similarity between two event data is calculated, two event data with similarity higher than the corresponding threshold are regarded as matched event data, and an event group is constructed based on the matched event data;
[0054] It should be noted that if a certain event data matches one or more other event data, then an event group is formed based on the certain event data and the other one or more event data.
[0055] Step 108: Fusing the features corresponding to the event data in the event group to obtain the fused features of the event group;
[0056] For two or more event data in an event group, explicit features and implicit features corresponding to the two or more event data are fused to obtain a fused feature of the event group.
[0057] For example, the event group includes event data 1 and event data 2. The explicit features and implicit features corresponding to event data 1 and the explicit features and implicit features corresponding to event data 2 are fused to obtain the fused features of the event group.
[0058] Step 110: construct an event graph based on the fusion features of the event group and the features of the first event data among the multiple event data; the first event data is the event data among the multiple event data except the event data in the event group.
[0059] The event data other than the event data in the event group among the multiple event data are referred to as first event data. The characteristics of the first event data include explicit characteristics and implicit characteristics of the first event data.
[0060] In this embodiment, control data on the vehicle operating status of traffic emergencies can be collected to enrich the corpus of driver decision-making under emergencies, and then feature extraction is performed on each event data respectively. Based on the extracted features of each event data, matching event data are determined from multiple event data, and the matching event data are used as an event group. The features corresponding to the event data in the event group are fused to obtain the fusion features of the event group. Based on the fusion features of the event group and the features of the first event data in multiple event data, a causal graph is constructed. In this way, a causal graph for traffic emergencies can be constructed to ensure vehicle driving safety under emergencies.
[0061] At the same time, by fusing the features of the matching event data to obtain the fused features of the event group, a causal graph can be constructed directly based on the fused features, thereby avoiding separate processing of the features of the matching event data to construct a causal graph, avoiding the duplication of data processing, and speeding up the construction of the causal graph.
[0062] Event feature extraction is a key link in natural language processing. The quality of event feature extraction directly affects the quality of the event graph to a certain extent. Event feature extraction is mainly performed on unstructured text. Event relationships are divided into explicit relationships that contain event relationship prompt words and implicit relationships that do not contain relationship prompt words. When establishing an event graph, these two types of relationships need to be extracted separately. Explicit feature extraction is obtained based on a logical keyword library, while implicit feature extraction relies on a neural network model to identify events.
[0063] In one of the embodiments, the feature extraction for each event data respectively includes: for each event data, based on a pre-constructed logical keyword library, determining a target keyword in the current event data that matches the logical keyword in the logical keyword library, extracting event data containing the target keyword from the current event data, and obtaining the explicit feature of the current event data based on the event data containing the target keyword; and extracting the implicit feature of the current event data through a long short-term memory network model.
[0064] The text with explicit relations contains logical keywords used to represent relational sentences. The extraction of explicit relations can be directly matched based on the logical keyword library. Specifically: traverse each event data to determine whether there is a target keyword matching the logical keyword in the pre-built logical keyword library in the event data. The target keyword matching the logical keyword can be the same target keyword as the logical keyword. The event data can be event text. If it exists, extract the part of the event text containing the target keyword from the event text as the extracted event data containing the target keyword.
[0065] In this embodiment, the logical keyword library includes causal logic keywords and sequential logic keywords. The causal logic keyword can be "because" and the sequential logic keyword can be "lead to". The determined logical keyword library is input into the unstructured text (i.e., event data) for matching. If the match is successful, the causal event sentence and the sequential event sentence in the unstructured text are returned.
[0066] Obtaining explicit features of current event data based on event data containing target keywords includes: performing trigger word and event argument processing on the event data containing the target keywords to obtain explicit features of the current event data.
[0067] However, sentences with explicit relationships account for a small proportion of the text, so it is necessary to extract implicit relationships. For implicit relationship extraction, we consider using bidirectional encoder representations from transformers (BERT), long short-term memory networks (LSTM), attention mechanisms, and normalized exponential function models to extract events from unstructured text.
[0068] In this implementation, possible logical keywords in the driver's behavioral decision-making in the face of emergencies are considered to improve the efficiency of extracting explicit features of event data.
[0069] The long short-term memory network model is used to extract latent features, which overcomes the gradient vanishing problem caused by the long sequence of the traditional recurrent neural network (RNN) model and has a good effect on long text processing.
[0070] In this embodiment, for each event data, the explicit features and implicit features of the current event data can also be fused to obtain the fused features of the current event data. The above-mentioned fusion features based on the event group and the features of the first event data among the multiple event data construct a cause-and-effect graph, including: constructing a cause-and-effect graph based on the fusion features of the event group and the fusion features of the first event data among the multiple event data.
[0071] This application can construct a high-quality reasoning graph through explicit feature extraction and implicit feature extraction, providing strong support for the subsequent construction of a question-answering system.
[0072] When performing unstructured extraction on the same or different texts, you may encounter situations where the same meaning is used but different expressions are used. If the deep learning model is not applicable, the events can be standardized and the fused results can be obtained by ignoring, resolving or avoiding conflicts. The standardization process includes segmenting the event description, removing stop words and function word components to obtain a relatively dry result. Then, event coreference resolution is performed. Event coreference resolution is a type of technology that fuses and updates events with a unified direction.
[0073] In one embodiment, matching event data is determined from the multiple event data, including: for each event data, fusing the explicit features and implicit features of the current event data to obtain a fused feature of the current event data; and determining matching event data from the multiple event data based on the fused features of each event data.
[0074] In one of the embodiments, the determining of matching event data from the multiple event data based on the fusion features of each event data includes: for each event data, calculating the feature similarity between the fusion features of the current event data and the fusion features of each second event data in the multiple event data; the second event data is the event data in the multiple event data except the current event data; and taking the second event data whose corresponding feature similarity is greater than a preset feature similarity threshold and the current event data as the matching event data.
[0075] In this embodiment, after obtaining the fusion features of the current event data, the method further includes: converting the fusion features of each event data into a feature vector;
[0076] The method of calculating the feature similarity between the fused features of the current event data and the fused features of each second event data is as follows: calculating the similarity between the feature vector corresponding to the current event data and the feature vector corresponding to each second event data to obtain the feature similarity;
[0077] In this embodiment, the method for calculating the similarity may adopt a cosine similarity algorithm.
[0078] The number of second event data whose corresponding feature similarity is greater than a preset feature similarity threshold can be one or more. For example, if the second event data 1, the second event data 2 and the second event data 3 are all similar to the current event data, then the second event data 1, the second event data 2, the second event data 3 and the current event data are regarded as matching event data.
[0079] One scenario is:
[0080] (1) Put the event data after feature extraction into the event data list S = {E1, E2,..., En}, where n is the total number of extracted events;
[0081] (2) By traversing the entire event data list S, the similarity between the fusion features of the event data is calculated to determine whether the event data are of the same type. Event data of the same type are grouped into event groups and labeled until all events are traversed;
[0082] (3) If the event data are of the same type, event fusion is required and the label is -1; if there is no relationship between the event data, fusion is not required and the label is 0.
[0083] The present application may regard event data with the same meaning but different expressions as matching event data, and then fuse the matching event data to obtain the fusion features of the event group. A cause-and-effect graph may be constructed directly based on the fusion features, thereby avoiding processing the features of the matching event data separately to construct a cause-and-effect graph, avoiding the duplication of data processing, and speeding up the construction of the cause-and-effect graph.
[0084] Data from news, social media platforms, etc. often contain content that is irrelevant to sudden events, or bugs may occur in the crawler during data capture due to programming problems, resulting in data crawling errors or duplicate capture. Therefore, before extracting features from each event data, it is necessary to preprocess the event data, clean the captured event data, and remove noise data in the corpus (such as invalid data such as duplicates). Since the captured data is unstructured text data, in order to delete duplicate crawled data, the captured data records are merged and sorted, the text is converted through word vectors, and based on this, the text is tested for text similarity using standardized Euclidean distance to filter out duplicates. Specifically:
[0085] In one of the embodiments, the feature extraction of each event data respectively includes: converting each event data into vector data; for the vector data of each event data, calculating the vector similarity between the current vector data and each other vector data in a plurality of vector data except the current vector data; determining the target vector data in the other vector data whose corresponding vector similarity is greater than a preset vector similarity threshold; determining repeated event data based on the event data of the current vector data and the event data of the target vector data; eliminating the repeated event data corresponding to each event data respectively from the plurality of event data to obtain processed event data; and extracting features from the processed event data.
[0086] In this embodiment, the method of calculating the vector similarity between the current vector data and each other vector data may be: calculating the Euclidean distance between the current vector data and each other vector data, and obtaining the vector similarity based on the Euclidean distance.
[0087] If the vector similarity between the target vector data and the current vector data is greater than a preset vector similarity threshold, the event data of the target vector data and the event data of the current vector data are repeated. The repeated event data is used as the repeated event data of the current event data, and the repeated event data corresponding to each event data is removed from the multiple event data to obtain multiple event data after processing.
[0088] Perform data cleaning on the processed multiple event data, and perform natural language processing on the cleaned unstructured text data, including part-of-speech tagging, Chinese word segmentation, dependency syntax analysis, etc. Perform feature extraction on the cleaned event data.
[0089] It can be seen that the present application can clear duplicate event data in multiple event data, thereby reducing the amount of data to be processed and speeding up the data processing speed.
[0090] In one of the embodiments, after constructing the event graph based on the fusion features of the event group and the features of the first event data among the multiple event data, the method further includes: visualizing the event graph based on a preset graphic database.
[0091] The present application can utilize a graph database to realize visualization of a matter graph, and the graph database may be Neo4j. By inputting the matter graph data into a preset graph database, visualization of the matter graph can be realized.
[0092] In summary, refer to Figure 2 The overall plan of this application is as follows:
[0093] Step 1: Establish an event dataset based on driver decisions that is authentic and timely.
[0094] Obtain unstructured data (i.e., event data) from the corresponding graph and construct a dataset.
[0095] Step 2: Build a common logical keyword library and templates related to drivers' behavioral decisions in emergency situations.
[0096] In the corresponding graph, explicit relation extraction is performed on the event data. Explicit relation extraction identifies the target keywords in the event data through the logical keyword library, extracts the event data containing the target keywords from the event data, and extracts event arguments from the event data containing the target keywords to obtain explicit features.
[0097] Step S3, determining the implicit relationship extraction model.
[0098] Perform hidden relationship extraction on event data, and the LATM model is used for hidden relationship extraction;
[0099] Step S4: perform data fusion and conflict resolution.
[0100] The explicit features extracted from the explicit relationship and the implicit features extracted from the implicit relationship are fused to obtain fused features, and event coreference resolution is performed. The specific implementation of event coreference resolution is referred to the above embodiment, which will not be described in detail here.
[0101] Step S4, building a fact graph and related question-answering system based on Neo4j.
[0102] Construct a reasoning graph. Visualize the reasoning graph and apply it to the vehicle question-answering system.
[0103] Figure 3To create a specific flow chart for visualizing the event graph, first create a project database and start the project, import data (the imported data is the event graph data), install py2neo through the operation component, connect to python and neo4j, and import the neo graph database.
[0104] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0105] Based on the same inventive concept, the embodiment of the present application also provides a device for constructing a theory map for implementing the method for constructing a theory map involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the theory map construction device provided below can refer to the limitations of the theory map construction method above, and will not be repeated here.
[0106] In an exemplary embodiment, Figure 4 As shown, a device 400 for constructing a matter graph is provided, comprising:
[0107] An acquisition module 401 is used to acquire a plurality of event data, wherein the event data is control data of a vehicle operation state for a traffic emergency;
[0108] Extraction module 402, used to extract features from each event data;
[0109] A determination module 403, configured to determine matching event data from the plurality of event data based on the features of each event data extracted, and use the matching event data as an event group;
[0110] A fusion module 404 is used to fuse the features corresponding to the event data in the event group to obtain a fusion feature of the event group;
[0111] The construction module 405 is used to construct an event graph based on the fusion features of the event group and the features of the first event data among the multiple event data; the first event data is the event data among the multiple event data except the event data in the event group.
[0112] In one embodiment, the extraction module 402 specifically includes:
[0113] For each event data, based on a pre-built logical keyword library, a target keyword in the current event data that matches a logical keyword in the logical keyword library is determined, event data containing the target keyword is extracted from the current event data, and explicit features of the current event data are obtained based on the event data containing the target keyword;
[0114] The latent features of the current event data are extracted through a long short-term memory network model.
[0115] In one embodiment, when determining the matching event data from the plurality of event data based on the features of each event data extracted, the determination module 403 is specifically configured to:
[0116] For each event data, the explicit features and implicit features of the current event data are fused to obtain fused features of the current event data;
[0117] Based on the fusion features of each event data, matching event data are determined from the multiple event data.
[0118] In one embodiment, when determining the matching event data from the plurality of event data based on the fusion features of each event data, the determination module 403 is specifically configured to:
[0119] For each event data, calculating the feature similarity between the fused features of the current event data and the fused features of each second event data in the multiple event data; the second event data is the event data in the multiple event data except the current event data;
[0120] The second event data whose corresponding feature similarity is greater than a preset feature similarity threshold and the current event data are used as the matched event data.
[0121] In one of the embodiments, the device further comprises a visualization module;
[0122] After the construction module 405 constructs the event graph based on the fusion features of the event group and the features of the first event data among the multiple event data; a visualization module is used to visualize the event graph based on a preset graphic database.
[0123] In one embodiment, the extraction module 402 is specifically used to: convert each event data into vector data; for the vector data of each event data, calculate the vector similarity between the current vector data and each other vector data in multiple vector data except the current vector data; determine the target vector data in the other vector data whose corresponding vector similarity is greater than a preset vector similarity threshold; determine repeated event data based on the event data of the current vector data and the event data of the target vector data; eliminate the repeated event data corresponding to each event data from the multiple event data to obtain processed event data; and perform feature extraction on the processed event data.
[0124] Each module in the above-mentioned event graph construction device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0125] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, a mobile cellular network, near field communication (Near Field Communication, NFC) or other technologies. When the computer program is executed by the processor, a method for constructing a causal map is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.
[0126] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0127] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps of the method of each embodiment are implemented.
[0128] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps of the method of each embodiment are implemented.
[0129] In one embodiment, a computer program product is provided, including a computer program, which implements the following steps of the method of each embodiment when executed by a processor.
[0130] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0131] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.
[0132] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.
[0133] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A method for constructing a matter graph, characterized in that: The method comprises: Acquire a plurality of event data, wherein the event data is control data of vehicle operation status for traffic emergencies; Extract features from each event data respectively; Based on the extracted features of each event data, determining matching event data from the multiple event data, and taking the matching event data as an event group; Fusing features corresponding to the event data in the event group to obtain fused features of the event group; Based on the fusion features of the event group and the features of the first event data among the multiple event data, a matter graph is constructed; the first event data is the event data among the multiple event data except the event data in the event group.
2. The method according to claim 1, characterized in that The feature extraction of each event data respectively includes: For each event data, based on a pre-built logical keyword library, a target keyword in the current event data that matches a logical keyword in the logical keyword library is determined, event data containing the target keyword is extracted from the current event data, and explicit features of the current event data are obtained based on the event data containing the target keyword; The latent features of the current event data are extracted through a long short-term memory network model.
3. The method according to claim 2, characterized in that The step of determining matching event data from the plurality of event data based on the extracted features of each event data includes: For each event data, the explicit features and implicit features of the current event data are fused to obtain fused features of the current event data; Based on the fusion features of each event data, matching event data are determined from the multiple event data.
4. The method according to claim 3, characterized in that: The determining matching event data from the plurality of event data based on the fusion features of each event data includes: For each event data, calculating the feature similarity between the fused features of the current event data and the fused features of each second event data in the multiple event data; the second event data is the event data in the multiple event data except the current event data; The second event data whose corresponding feature similarity is greater than a preset feature similarity threshold and the current event data are used as the matched event data.
5. The method according to any one of claims 1 to 4, characterized in that After constructing the event graph based on the fusion feature of the event group and the feature of the first event data among the multiple event data, the method further includes: Based on a preset graphic database, the event graph is visualized.
6. The method according to any one of claims 1 to 4, characterized in that The feature extraction of each event data respectively includes: Convert each event data into vector data; For the vector data of each event data, calculate the vector similarity between the current vector data and each other vector data in the plurality of vector data except the current vector data; determine the target vector data in the other vector data whose corresponding vector similarity is greater than a preset vector similarity threshold; determine the repeated event data based on the event data of the current vector data and the event data of the target vector data; Eliminate duplicate event data corresponding to each event data from the plurality of event data to obtain processed event data; Perform feature extraction on the processed event data.
7. A device for constructing a matter-of-fact graph, characterized in that: The device comprises: An acquisition module, used for acquiring a plurality of event data, wherein the event data is control data of a vehicle operation state for a traffic emergency; An extraction module, used for extracting features from each event data; A determination module, configured to determine matching event data from the plurality of event data based on the features of each event data extracted, and use the matching event data as an event group; A fusion module, used for fusing features corresponding to the event data in the event group to obtain fusion features of the event group; A construction module is used to construct an event graph based on the fusion features of the event group and the features of the first event data among the multiple event data; the first event data is the event data among the multiple event data except the event data in the event group.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
