Technical requirements generation method and equipment based on perception task scenario

Abstract

The present invention proposes a method and device for generating technical requirements based on perceiving task scenarios. The method for generating technical requirements includes: obtaining task scenarios; extracting scene feature information according to the task scenarios; and analyzing the scene feature information to generate technical requirements. The present invention is conducive to improving the efficiency of obtaining technical requirements.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method and device for generating technical requirements based on perception task scenarios. Background Art

[0002] The product design process usually involves stages such as demand analysis, conceptual design, technical design, and detailed design. In product design, R&D personnel determine technical requirements based on the product's application scenarios, then generate product functions based on technical requirements, and finally design product implementation plans based on product functions.

[0003] Usually, R&D personnel conduct research on application scenarios and complete demand analysis based on their personal experience. This method of demand analysis will consume a lot of time of R&D personnel, reduce R&D efficiency, increase labor costs, and have limitations due to insufficient personal knowledge. Therefore, how to quickly obtain technical requirements is particularly important for improving product development efficiency. Summary of the invention

[0004] The present invention proposes a method and device for generating technical requirements based on perception task scenarios, which is beneficial to reducing the time for obtaining technical requirements and improving product development efficiency.

[0005] In a first aspect, a method for generating technical requirements is provided, which includes:

[0006] Get the task scenario;

[0007] Extracting scene feature information according to the task scene;

[0008] The scene feature information is analyzed to generate technical requirements.

[0009] In one embodiment, in the method for generating technical requirements, the task scenario includes at least one of a task scenario described in text, a task scenario described in image, and a task scenario described in video.

[0010] In one embodiment, the method for generating technical requirements, wherein the step of extracting scene feature information according to the task scenario includes:

[0011] When the task scenario is a task scenario described in text, knowledge entities and relationships between the knowledge entities are extracted from the task scenario described in text to obtain scenario feature information;

[0012] When the task scene is described by an image or a video, video sample information or image sample information matching the task scene is obtained, and scene feature information is obtained according to the video sample information or image sample information.

[0013] In one embodiment, the method for generating technical requirements, wherein the analyzing the scene feature information to generate technical requirements, comprises:

[0014] Building a scenario database, wherein the scenario database includes a scenario knowledge graph;

[0015] Acquire graph information of a preset scene from the scene knowledge graph to obtain graph information of the preset scene;

[0016] Obtaining matching scene information according to the similarity between the scene feature information and the preset scene graph information;

[0017] Acquire preset requirement information corresponding to the matching scene information from the scene knowledge graph;

[0018] Comparing the scene feature information with the matching scene information to determine difference information between the scene feature information and the matching scene information;

[0019] The preset requirement information is modified according to the difference information to generate technical requirements.

[0020] In one embodiment, the method for generating technical requirements, wherein obtaining matching scene information according to the similarity between the scene feature information and the preset scene graph information, comprises:

[0021] Constructing scene feature graph information according to the knowledge entities in the scene feature information and the relationships between the knowledge entities;

[0022] Respectively calculating the similarities between the scene feature graph information and a plurality of the preset scene graph information to obtain a plurality of similarity values;

[0023] When the maximum similarity value among the multiple similarity values ​​is greater than or equal to the preset threshold, the preset scene graph information corresponding to the maximum similarity value is the most matching scene information;

[0024] When the maximum similarity value among the multiple similarity values ​​is less than a preset threshold, multiple preset scene graph information are selected according to the similarity values ​​for fusion to obtain matching scene information.

[0025] In one embodiment, the method for generating technical requirements, wherein the similarities between the scene feature graph information and the plurality of preset scene graph information are respectively calculated to obtain a plurality of similarity values, comprises:

[0026] According to the entity node in the scene feature graph information, query the nodes matching the entity node in the preset scene graph information to obtain the common nodes;

[0027] According to the connection relationship of the entity nodes in the scene feature graph information, searching the preset scene graph information for a connection relationship that matches the connection relationship of the entity nodes to obtain a common connection relationship;

[0028] Constructing a common subgraph between the scene feature graph information and the preset scene graph information according to the common nodes and the common connection relationship;

[0029] Calculating a similarity value based on the common subgraph;

[0030] The calculation formula of the similarity value is: S=w1×NSimavg+w2×SSim;

[0031] Among them, S is the similarity value, w1 and w2 are the weights of entity node similarity and common subgraph similarity respectively, NSimavg is the entity node similarity, SSim represents the common subgraph similarity, the entity node similarity is the average value of the name similarity of all common nodes, SSim=NComMax / Min(NNumT,NNumG), NComMax is the number of entity nodes of the common subgraph, NNumT is the number of entity nodes of the scene feature graph information, NNumG is the number of entity nodes of the preset scene graph information, Min(NNumT,NNumG) is the minimum value between the number of entity nodes of the scene feature graph information and the number of entity nodes of the preset scene graph information.

[0032] In one embodiment, the method for generating technical requirements, wherein the step of modifying the preset requirement information according to the difference information to generate the technical requirements, includes:

[0033] Determine the modification method corresponding to the difference information through the preset scene knowledge graph information;

[0034] Modify the preset requirement information according to the modification method to obtain multiple candidate technical requirements;

[0035] Obtaining a criticality score for each of the candidate technical requirements;

[0036] The candidate technical requirements are screened according to the criticality scores of the candidate technical requirements to obtain technical requirements.

[0037] In one embodiment, the method for generating technical requirements, wherein obtaining the criticality score of each candidate technical requirement includes:

[0038] Determining current technologies corresponding to the candidate technology requirements;

[0039] Determining a maturity score based on the maturity of the current technology;

[0040] Determine a technology route score according to the evolution route of the current technology;

[0041] Determine a gap score according to the technology gap between the candidate technology requirement and the current technology;

[0042] The criticality score of each of the candidate technical requirements is determined based at least on the maturity score, the technical route score, and the gap score.

[0043] Second, a method for generating technical requirements for a perception task is provided, including:

[0044] Acquire the perception task scenario;

[0045] The perception task scenario is processed using the method for generating technical requirements as described above to obtain perception technical requirements.

[0046] In a third aspect, a system for generating technical requirements is provided, including:

[0047] Acquisition module, used to acquire task scenarios;

[0048] An extraction module, used to extract scene feature information according to the task scene;

[0049] A generation module is used to analyze the scene feature information and generate technical requirements.

[0050] In a fourth aspect, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, wherein when the processor executes the program, the steps of the method for generating technical requirements as described above are implemented.

[0051] In a fifth aspect, a storage medium is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the method for generating technical requirements as described above are implemented.

[0052] The present invention can extract scene feature information according to the task scene, and then generate corresponding technical requirements according to the scene feature information, thereby reducing the time-consuming and inaccurate situations existing in manual demand analysis and improving the efficiency of research and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] By reading the detailed description of the preferred specific embodiments below, various other advantages and benefits of the present invention will become clear to those of ordinary skill in the art. The drawings in the specification are only used to illustrate the preferred embodiments and are not considered to be limitations of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative work.

[0054] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0055] Figure 1 It is a flowchart of a method for generating technical requirements of an embodiment of the present invention;

[0056] Figure 2 is a use case diagram for demand generation based on task scenarios according to an embodiment of the present invention;

[0057] Figure 3 is a schematic diagram of preset scene spectrogram information according to an embodiment of the present invention;

[0058] Figure 4 It is a schematic diagram of the structure of a system for generating technical requirements of an embodiment of the present invention;

[0059] Figure 5 It is a schematic diagram of the structure of an analysis system for sensing technical requirements according to an embodiment of the present invention;

[0060] Figure 6 The figure is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0061] It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present invention is further described in detail below in conjunction with the drawings and specific embodiments.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0063] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0064] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0065] This embodiment takes the technical requirements analysis of sensing devices as an example to illustrate the technical solution of the present invention. Sensing devices play a vital role in many fields such as autonomous driving, robots, the Internet of Things, smart homes, and security monitoring. Sensing devices are responsible for collecting data about the surrounding environment and providing a basis for decision-making for the system. In the field of autonomous driving, sensing devices generally include cameras, lidars, millimeter-wave radars, ultrasonic radars, etc., and different sensing devices have different ranging accuracy, detection distances, and environmental adaptability.

[0066] Figure 1 This is a flowchart of the method for generating technical requirements in this implementation. Figure 1 The method for generating technical requirements includes: step 10, step 20 and step 30.

[0067] Step 10: Get the task scenario.

[0068] The mission scenario is the scenario information faced when executing the mission. It is usually the usage scenario of the target device or device. For example, for autonomous driving, the mission scenario is generally the complex road scene that the car may face when driving.

[0069] Step 20: Extract scene feature information based on the task scenario.

[0070] Scene feature information is the feature of the task scene, that is, the task scene is described in the form of feature information. For example, for autonomous driving, feature extraction is performed on the autonomous driving road scene to obtain features that can describe the corresponding features of the autonomous driving road scene: straight roads, curves, uphills, downhills, tunnels, lane merging and other scene features. In addition, the features of the autonomous driving task scene also include: driving time, driving speed, driving direction and other task features.

[0071] Step 30: Analyze the scene feature information and generate technical requirements.

[0072] Technical requirements are requirements for devices or equipment in terms of technology. Devices or equipment that meet technical requirements should be devices or equipment that have the ability to perform corresponding tasks in corresponding task scenarios.

[0073] Analyze the scene feature information to determine the correspondence between the scene feature information and the technical requirements; then generate the technical requirements based on the correspondence between the scene feature information and the technical requirements. For example, for the autonomous driving scene, the corresponding relevant perception capability requirements are obtained based on the scene feature information. For example, based on the unmanned driving scene feature information, the lane perception capability requirements, obstacle perception requirements, and vehicle speed perception capability requirements are obtained.

[0074] This embodiment can extract scene feature information according to the task scenario, and then generate corresponding technical requirements according to the scene feature information, thereby reducing the time-consuming and inaccurate situation of manual demand analysis and improving the efficiency of research and development.

[0075] In one implementation manner, a method for generating technical requirements, wherein obtaining a task scenario includes: adding a task scenario, deleting a task scenario, modifying a task scenario, and querying a task scenario.

[0076] like Figure 2 As shown in the figure, a use case diagram is generated based on the needs of the task scenario. Users can add, delete, modify, and query the perceived task scenarios in the system. To add a task scenario, you need to set the corresponding task scenario name, task scenario content, and the capabilities required to complete the task, and then you can generate a task scenario; to delete a task, you need to select the task scenario you want to delete, and the system will delete the selected task scenario after confirming the deletion information; to modify a task scenario, you need to select the corresponding task scenario, and you can modify the corresponding task scenario name, task scenario content, and other information in the visual interface. After confirmation, the system will update the task scenario information; to query the task scenario information, you need to enter the task scenario name, and click the search button to find the specified task scenario information.

[0077] In one embodiment, a method for generating technical requirements, wherein the task scenario includes at least one of a task scenario described in text, a task scenario described in image, and a task scenario described in video.

[0078] There are various forms of task scenarios, mainly including: task scenarios described in text, task scenarios described in images, and task scenarios described in videos. Among them, the task scenario described in text is a paragraph of text that describes the task scenario. The task scenario described in images is a task scenario image, that is, the task scenario is presented through an image. The task scenario described in videos is a task scenario video, that is, the task scenario is presented through a video, such as a demonstration or simulation video of the task scenario.

[0079] In one implementation manner, a method for generating technical requirements, wherein scene feature information is extracted based on a task scenario, includes: step 210 and step 220.

[0080] Step 210: When the task scenario is a task scenario described in text, knowledge entities and relationships between knowledge entities are extracted from the task scenario described in text to obtain scenario feature information.

[0081] The task scenario described in text is referred to as scenario description text.

[0082] Scene feature information is information that can reflect the overall meaning of the task scene. In this embodiment, knowledge entities and the relationships between knowledge entities are extracted from the task scene described in text, so that information on scene features can be reflected through knowledge entities and the relationships between knowledge entities.

[0083] Optionally, a knowledge extraction method is used to extract knowledge entities and the relationships between knowledge entities. Knowledge entities are identified based on the scene description text to extract knowledge entities. Knowledge entity identification is implemented using rule-based and statistical machine learning methods. Knowledge entity relationship extraction methods include dictionary-driven, pipeline model-based, and other methods. First, the bidirectional long short-term memory conditional random field method (BiLSTM-CRF) is used to extract knowledge entities; then, through the analysis of text syntax, for a sentence containing entity attribute relationships, the relationship between entities is extracted based on this.

[0084] For example, for the sentence "the speed limit is 50km / h", the knowledge entities extracted are: "speed limit", "50km / h", and the knowledge entity relationship extracted is: "is".

[0085] Step 220: When a task scene is described by an image or a task scene is described by a video, video sample information or image sample information matching the task scene is obtained, and scene feature information is obtained based on the video sample information or the image sample information.

[0086] Task scenes described by images or by videos need to be converted into scene feature information. Pre-prepared video sample information or image sample information can be stored in a database. The image sample information contains the correspondence between the image and the image description text. The image sample most similar to the task scene is determined from the database through an image similarity algorithm. According to the correspondence between the image sample and the image description text, the image description text corresponding to the most similar image sample is obtained, and the scene feature information is obtained. The method of extracting scene feature information from a video is similar to the method of extracting scene feature information from a picture, and will not be repeated here.

[0087] In one embodiment, a method for generating technical requirements, wherein scene feature information is analyzed to generate technical requirements, includes: step 310, step 320, step 330, step 340, step 350, and step 360.

[0088] Step 310: Build a scenario database, which includes a scenario knowledge graph.

[0089] The scenario database is a collection of information about the task scenario. The scenario database can be pre-built or built in real time. The scenario knowledge graph is one of the most important data in the scenario database. The scenario knowledge graph is a collection of task scenario knowledge entities and the relationships between knowledge entities. The scenario knowledge graph has a graph structure that can intuitively display the relationship between knowledge entities, which is conducive to demand analysis.

[0090] Step 320: Obtain graph information of a preset scene from the scene knowledge graph to obtain graph information of the preset scene.

[0091] The preset scene graph information is the graph information corresponding to the pre-stored task scene. The preset scene is generally a representative task scene. The preset scene can be determined manually or according to a rule (the task scene with the most occurrences in a scene category is used as the preset scene).

[0092] The scene database has multiple preset scenes for selection. Specifically, the preset scene graph information is stored in the scene knowledge graph in the form of a knowledge graph. For example, see Figure 3In the scene knowledge graph, there is a first knowledge entity 401 of a preset scene (such as an autonomous driving scene). The first knowledge entity 401 is connected to multiple subdivided second knowledge entities 402 (such as driving in snow, driving in rainy days, driving in foggy days, etc.), and the second knowledge entity 402 is further connected to related third knowledge entities 403 (such as the speed limit and perceived distance corresponding to the scene), thus forming a tree-structured scene knowledge graph.

[0093] Step 330: Obtain matching scene information based on the similarity between the scene feature information and the preset scene graph information.

[0094] The similarity between the scene feature information and each preset scene graph information is calculated using the information similarity algorithm. According to the similarity between the scene feature information and each preset scene graph information, the preset scenes are sorted in descending order of similarity, and the corresponding graph information of the first preset scene is the matching scene information.

[0095] Step 340: Obtain preset requirement information corresponding to the matching scene information from the scene knowledge graph.

[0096] The preset demand information is pre-stored demand information. The preset demand information is set by the user based on experience or automatically generated according to certain rules by using knowledge extraction based on previous demand solutions.

[0097] In the scenario database, the preset scenario information, preset demand information and their corresponding relationships are represented by the scenario knowledge graph, that is, Figure 3 As shown, each preset scenario knowledge entity 401 is connected to one or more corresponding requirement knowledge entities 404 .

[0098] In the scene knowledge graph, the preset scenes are connected with the corresponding preset demand information. After the matching scene information is obtained, the corresponding preset demand information is obtained according to the corresponding relationship between the preset scene and the preset demand information.

[0099] Step 350: Compare the scene feature information with the matching scene information to determine difference information between the scene feature information and the matching scene information.

[0100] In rare cases, the scene feature information is exactly the same as the matching scene information, and the preset requirement information is the generated technical requirement. Usually, there are differences between the scene feature information and the matching scene information, and the preset requirement information needs to be modified.

[0101] Compare the knowledge entities in the scene feature information with the knowledge entities of the matching scene information. Divide the knowledge entities into multiple levels according to the comparison results to obtain multiple levels of knowledge entities, which are the difference information. The multiple levels of knowledge entities include: first-level knowledge entities, second-level knowledge entities, and third-level knowledge entities. The first-level knowledge entity is a knowledge entity whose similarity between the knowledge entity in the scene feature information and the knowledge entity of the matching scene information is greater than a first preset similarity value. The second-level knowledge entity is a knowledge entity whose similarity between the knowledge entity in the scene feature information and the knowledge entity of the matching scene information is greater than a second preset similarity value. The first preset similarity value is greater than the second preset similarity value. Specifically, the first-level knowledge entity is a knowledge entity whose scene feature information is exactly the same as the matching scene information, and the second-level knowledge entity is a knowledge entity whose scene feature information is relatively high in similarity with the matching scene information.

[0102] Step 360: Modify the preset requirement information according to the difference information to generate technical requirements.

[0103] The preset requirements provided by the preset scenario map may not be the requirements required by the task scenario, and the preset requirement information needs to be modified. During the modification process, the preset requirements that are different from the task scenario are mainly modified, while the requirements that are not different do not need to be modified, thereby generating new requirements. The modification methods include adding new requirements to the preset requirements, deleting redundant requirements from the preset requirements, and making other adjustments to the preset requirements.

[0104] In one embodiment, a method for generating technical requirements, wherein matching scene information is obtained based on the similarity between scene feature information and preset scene graph information, includes: step 331, step 332, step 333, and step 334.

[0105] Step 331: Construct scene feature graph information based on the knowledge entities in the scene feature information and the relationships between the knowledge entities.

[0106] The scene feature graph information is a tree-like entity relationship graph constructed based on the knowledge entities extracted from the scene feature information, which is equivalent to a small knowledge graph. For example, the knowledge entities extracted from the scene feature information are used as nodes of the scene feature graph, and the relationships between knowledge entities are used as edges to form a tree-like entity relationship graph.

[0107] Step 332: Calculate the similarity between the scene feature graph information and multiple preset scene graph information respectively to obtain multiple similarity values.

[0108] The preset scene graph information may be a partial graph with preset scene information intercepted from the scene knowledge graph. Therefore, the knowledge entities and knowledge entity relationships in the preset scene graph also have a tree graph structure.

[0109] Based on the similarity of the tree graph, the scene feature information is compared with the preset scene information to determine the similarity between the two. In a simpler way, the similarity can be determined by the number of identical entity nodes in the scene feature graph information and the preset scene graph information. In a more complex way, the similarity can be determined by the identical entity nodes in the scene feature graph information and the preset scene graph information and the corresponding connection relationship.

[0110] Step 333: When the maximum similarity value among the multiple similarity values ​​is greater than or equal to the preset threshold, the preset scene graph information corresponding to the maximum similarity value is the matching scene information.

[0111] The maximum similarity value is obtained from multiple similarity values. When the maximum similarity value satisfies or is greater than or equal to a preset threshold, it means that the preset scene corresponding to the maximum similarity value has a high similarity with the current task scene. Then the preset scene graph information corresponding to the maximum similarity value can be the matching scene information.

[0112] Step 334: When the maximum similarity value among multiple similarity values ​​is less than a preset threshold, multiple preset scene graph information are selected according to the similarity values ​​for fusion to obtain matching scene information.

[0113] When all similarity values ​​are less than the preset threshold, several preset scene graphs with higher similarity values ​​are fused to obtain matching scene information with higher similarity. For example, 3 to 5 preset scene graphs with the highest similarity values ​​are obtained as the preset scene graphs to be fused. During the fusion process, the preset scene graph to be fused corresponding to the maximum similarity value is used as the basic graph; the entities in the basic graph are modified and adjusted according to the remaining preset scene graphs to be fused until the similarity between the fused preset scene graph and the scene feature graph information is greater than the preset threshold or other termination conditions are met. Specifically, the knowledge entities in the basic graph whose similarity with the scene feature graph is lower than the first threshold are deleted; the knowledge entities in the remaining preset scene graphs to be fused whose similarity with the scene feature graph is higher than the second threshold and the demand entities connected to the knowledge entities are obtained to obtain the entity tree to be added; the entity tree to be added is added to the basic graph to complete the fusion and obtain the matching scene information.

[0114] In one embodiment, a method for generating technical requirements, wherein the similarities between scene feature graph information and multiple preset scene graph information are calculated respectively to obtain multiple similarity values, includes: step 3321, step 3322, step 3323 and step 3324.

[0115] Step 3321: According to the entity nodes in the scene feature graph information, search for nodes matching the entity nodes in the preset scene graph information to obtain the common nodes.

[0116] According to the name string similarity of the entity node (such as Jaccard coefficient), predefined entity type, or embedded vector (such as Word2Vec, TransE, etc.), the entity node features are determined for each entity node of the graph. According to the similarity between the entity node features, the common nodes of the two spectra (scene feature graph information and preset scene graph information) are determined. Among them, the common nodes refer to two entity nodes whose similarity is greater than the third preset threshold, that is, the common nodes are two identical entity nodes, respectively in the scene feature graph information and the preset scene graph information.

[0117] Step 3322: According to the entity node connection relationship in the scene feature graph information, query the connection relationship that matches the entity node connection relationship in the preset scene graph information to obtain the common connection relationship.

[0118] The connection relationship between entity nodes is represented as an edge in the graph. The edge features are defined according to the relationship type and direction of the edge in the graph. According to the edge features, the matching edges are queried in the preset scene graph information and the graph information to obtain the common connection relationship. Generally speaking, two edges with the same entity nodes at both ends are matched edges, that is, a common connection relationship in the two graphs.

[0119] Step 3323: Construct a common subgraph between the scene feature graph information and the preset scene graph information based on the common nodes and common connection relationships.

[0120] The common subgraph is a graph composed of all common nodes and common connection relationships between the scene feature graph information and the preset scene graph information. The common subgraph is used as a basic indicator to measure the degree of matching between two graphs.

[0121] Step 3324: Calculate the similarity value. Specifically, combine the features of the entity nodes and edges, as well as the common subgraphs, to calculate the overall similarity.

[0122] For example, the calculation formula of the similarity value is: S=w1×NSimavg+w2×SSim;

[0123] Where S is the similarity value, w1 and w2 are the weights of entity node similarity and common subgraph similarity respectively, NSimavg is the entity node similarity, SSim represents the common subgraph similarity, entity node similarity is the average value of the name similarity of all common nodes, SSim=NComMax / Min(NNumT,NNumG), NComMax is the number of common subgraph nodes, NNumT is the number of entity nodes of scene feature graph information, NNumG is the number of entity nodes of preset scene graph information, Min(NNumT,NNumG) is the minimum value between the number of entity nodes of scene feature graph information and the number of nodes of preset scene information.

[0124] Specifically, the node similarity is measured by the Jaccard coefficient to measure the name similarity. The subgraph similarity can be determined by the ratio of the number of entity nodes in the common subgraph to the number of entity nodes in the smaller graph.

[0125] In some embodiments, the method for generating technical requirements further includes calculating overall similarity based on edge similarity. The edge similarity is determined based on the consistency of the relationship type, and the types of each edge are compared in turn to see if they are the same. If the edge types are the same, the value is 1, otherwise it is 0, and finally the average value of all matching edges is taken comprehensively.

[0126] The calculation formula of the similarity value is: S=w1×NSimavg+w2×SSim+w3×ESim; among them, ESim is the edge similarity and w3 is the weight of the edge similarity.

[0127] It should be noted that the above formula is a feasible calculation method that comprehensively considers the similarity of nodes, edges and graphs. In the application process, more complex models and algorithms can be used to deal with fuzzy matching, uncertainty and complex structures, and the choice of similarity calculation method should be flexibly adjusted according to specific task requirements.

[0128] In one implementation manner, a method for generating technical requirements, wherein preset requirement information is modified according to difference information to generate technical requirements, includes: step 361, step 362, step 363, and step 364.

[0129] Step 361: Determine the modification method corresponding to the difference information through the preset scene knowledge graph information.

[0130] Determine the corresponding preset modification method based on the difference information. According to the size of the difference, the difference information is divided into multiple levels of entities. Among them, since the difference between the scene feature information and the first level knowledge entity corresponding to the matching scene information is not large, the preset modification method is to keep the original preset requirement unchanged; since there are certain differences between the scene feature information and the second level knowledge entity corresponding to the matching scene information, it is necessary to determine the modification method based on the difference attributes; since there are large differences between the scene feature information and the third level knowledge entity corresponding to the matching scene information, it is determined that the preset requirement cannot be modified. Among them, the difference attributes are directly obtained from the attributes of the corresponding second level knowledge entity (such as Figure 3 As shown, the requirement knowledge entity 404 is connected to the attribute knowledge entity 405 , and different difference attributes have different preset modification methods.

[0131] For example, the scene feature information contains the knowledge entity of "Speed ​​Limit on Roads in City A". In the preset scene information, it is found that "Speed ​​Limit on Roads in City B" is similar to "Speed ​​Limit on Roads in City A", but they are not the same, and there is difference information. At the same time, the preset requirement "100km / h" corresponding to "Road in City B" is obtained in the preset scene information. Then, the corresponding difference attribute (equivalent to the connection relationship) can be obtained as the speed limit according to the attribute of "Speed ​​Limit on Roads in City A", and a feasible speed limit modification method (such as search and replace) can be determined according to the speed limit attribute.

[0132] Step 362: Modify the preset requirement information according to the modification method to obtain multiple candidate technical requirements.

[0133] There are many ways to modify, the common modification methods include adding new technical requirements, deleting redundant technical requirements, and replacing requirements. Adding technical requirements means adding requirements that meet the requirements to the preset requirement information. Deleting redundant technical requirements means first comparing and identifying redundant technical requirements, and then deleting them.

[0134] For example, after determining the modification method, the preset demand information is modified according to the modification method to generate technical requirements. For example, in the search and replace method, the speed limit value corresponding to the speed limit of the road in city A is found to be "90km / h", and the original preset requirement "100km / h" is modified to "90km / h" as the corresponding requirement of the road in city A.

[0135] In addition, demand mining will be carried out in combination with maturity prediction knowledge graphs, evolutionary path prediction knowledge graphs, etc., and the preset demand information will be modified according to the mined needs to predict future needs and mine hidden needs.

[0136] Step 363: Obtain the criticality score of each candidate technical requirement.

[0137] There are many candidate technical requirements, which are redundant. We need to select the candidate technical requirements and identify the key technical requirements. Therefore, we calculate the criticality score of each candidate technical requirement and use the criticality score as the screening criterion.

[0138] Step 364: Screen the candidate technical requirements according to the criticality score of each candidate technical requirement to obtain the technical requirements.

[0139] The higher the criticality score, the more critical the technical requirement. Therefore, according to the rule that the higher the criticality score, the easier it is to be selected as the final technical requirement, the requirement is selected from multiple candidate technical requirements as the final technical requirement.

[0140] The remaining non-critical requirements in the candidate technologies will serve as reference requirements and can be displayed to users for reference and selection when necessary.

[0141] In one implementation manner, a method for generating technical requirements, wherein a criticality score of each candidate technical requirement is obtained, includes: step 3631 , step 3632 , step 3633 , step 3634 , and step 3635 .

[0142] Step 3631: Determine the current technology corresponding to the candidate technology requirements.

[0143] Current technology can be understood as existing technology that is used to meet the requirements of candidate technologies. For example, to meet the technical requirements of obstacle recognition, the corresponding existing technology includes an obstacle perception algorithm based on a neural network model.

[0144] Step 3632: Determine a maturity score based on the maturity of the current technology.

[0145] The maturity score is mainly used to score the maturity of the current technology. The development of technology usually includes infancy, growth, maturity and decline. The maturity score is determined according to the development stage of the current technology. The maturity score is stored in the current technology knowledge graph. The maturity score is generally related to the time when the current technology appeared, and can also be set by technical experts in related fields.

[0146] Step 3633: Determine the technology route score based on the evolutionary route of the current technology.

[0147] The technical route score is mainly used to score the technical route of the current technology. There may be multiple technical routes for the same demand, and the development trends of each technical route are different. The technical route score is used to evaluate the prospects of the current technology. The technical route score is stored in the current technical knowledge graph. The technical route score is generally related to the research popularity. It can be scored by the number of related papers published, or it can be set by technical experts in related fields.

[0148] Step 3634: Determine a gap score based on the technology gap between the candidate technology requirements and the current technology.

[0149] In general, the current technology may not be able to meet the candidate technology requirements, and research and development on the basis of the current technology is required to meet the candidate technology requirements. If the gap between the candidate technology requirements and the current technology is large, then developing new technologies on the basis of current technology will take a lot of time and may not be practical. For this reason, it is necessary to use the gap score to evaluate the gap between the candidate technology requirements and the current technology, estimate the difficulty and time of research and development, and help rationalize the requirements.

[0150] The gap score needs to be determined based on the current development law of technology. For example, the gap score can be estimated based on the estimated development time. The longer the estimated development time, the greater the gap score. The gap score is generally a score of the difficulty of the technology gap based on the current technology development trend. It can also be set by technical experts in related fields.

[0151] Optionally, taking the perception technology requirements as an example, the capability gap is calculated based on the current status of the perception capability (i.e., current technology) and the task capability requirements (i.e., technology requirements), and the importance of the capability requirement gap is determined. By comparing the difference between the current status of the perception capability and the task capability requirements, the capability gap is quantitatively analyzed. Specifically, according to the current status of the perception capability and the task capability requirements analysis capability, the capability gap is accurately calculated, and a judgment on the importance of the capability requirement gap can be provided. Through an efficient comparison between the current status of the perception capability and the task capability requirements, a comprehensive quantitative analysis of the capability gap is achieved. In this process, the importance of different capability requirement gaps can be set according to the gap between the current level of the perception capability and the level of the perception capability requirements. This setting can determine which capability gaps are the most important and should be solved first. At the same time, this setting of the importance of the capability requirement gap also makes the entire decision-making process more user-friendly.

[0152] Step 3635: Determine the criticality score of each candidate technical requirement based on at least the maturity score, the technical path score, and the gap score.

[0153] Optionally, the criticality score is positively correlated with the maturity score, the criticality score is positively correlated with the technical route score, and the criticality score is negatively correlated with the gap score.

[0154] For example, S=λ1S1+λ2S2-λ3S3; S is the criticality score, S1 is the maturity score, S1 is the technical route score, S1 is the gap score, and λ1, λ2, and λ3 are the weight coefficients of the maturity score, technical route score, and gap score, respectively.

[0155] This embodiment also provides a method for generating technical requirements of a perception task, including: step 40 and step 50.

[0156] Step 40: Obtain the perception task scenario.

[0157] The perception task scenario is the scenario corresponding to the execution of the perception task. For example, when the perception task is the task of a car sensing surrounding obstacles while driving, the perception task scenario is various possible driving scenarios of various cars.

[0158] Step 50: Use the above method for generating technical requirements to process the perception task scenario and obtain the perception technical requirements.

[0159] The technical requirement generation system of this embodiment can extract scene feature information based on the perception task scenario, and then generate corresponding perception task technical requirements based on the scene feature information, thereby reducing the time-consuming and inaccurate situation of manual demand analysis and improving the efficiency of research and development.

[0160] See also Figure 4 This embodiment also provides a system 60 for generating technical requirements, including: an acquisition module 610, an extraction module 620 and a generation module 630.

[0161] The acquisition module 610 is used to acquire the task scenario.

[0162] The extraction module 620 is used to extract scene feature information according to the task scene.

[0163] The generation module 630 is used to analyze the scene feature information and generate technical requirements.

[0164] The technical requirement generation system 60 of this embodiment can extract scene feature information based on the task scenario, and then generate corresponding technical requirements based on the scene feature information, thereby reducing the time-consuming and inaccurate situation of manual requirement analysis and improving the efficiency of research and development.

[0165] In one embodiment, the extraction module 620 is also used to extract knowledge entities and the relationships between knowledge entities from the task scene described in text to obtain scene feature information when the task scene is a task scene described in text; when the task scene is a task scene described in image or a task scene described in video, video sample information or image sample information matching the task scene is obtained, and scene feature information is obtained based on the video sample information or image sample information.

[0166] In one embodiment, the generation module 630 is also used to construct a scene database, which includes a scene knowledge graph; obtain graph information of a preset scene from the scene knowledge graph to obtain preset scene graph information; obtain matching scene information based on the similarity between the scene feature information and the preset scene graph information; obtain preset requirement information corresponding to the matching scene information from the scene knowledge graph; compare the scene feature information with the matching scene information to determine the difference information between the scene feature information and the matching scene information; modify the preset requirement information based on the difference information to generate technical requirements.

[0167] In one embodiment, the generation module 630 is also used to construct scene feature graph information based on knowledge entities in the scene feature information and the relationship between knowledge entities; calculate the similarity between the scene feature graph information and multiple preset scene graph information respectively to obtain multiple similarity values; when the maximum similarity value among the multiple similarity values ​​is greater than or equal to the preset threshold, the preset scene graph information corresponding to the maximum similarity value is the most matching scene information; when the maximum similarity value among the multiple similarity values ​​is less than the preset threshold, multiple preset scene graph information are selected according to the similarity value for fusion to obtain matching scene information.

[0168] In one embodiment, the generation module 630 is also used to query the preset scene graph information for nodes that match the entity nodes in the scene feature graph information to obtain common nodes; query the preset scene graph information for connection relationships that match the connection relationships of the entity nodes in the scene feature graph information to obtain common connection relationships; and construct a common subgraph between the scene feature graph information and the preset scene graph information based on the common nodes and the common connection relationships.

[0169] In one embodiment, the generation module 630 is also used to determine the modification method corresponding to the difference information through the preset scenario knowledge graph information; modify the preset requirement information according to the modification method to obtain multiple candidate technical requirements; obtain the criticality score of each candidate technical requirement; and screen the candidate technical requirements according to the criticality score of each candidate technical requirement to obtain the technical requirements.

[0170] In one embodiment, the generation module 630 is also used to determine the current technology corresponding to the candidate technology requirement; determine the maturity score based on the maturity of the current technology; determine the technology route score based on the evolutionary path of the current technology; determine the gap score based on the technology gap between the candidate technology requirement and the current technology; determine the criticality score of each candidate technology requirement based on at least the maturity score, the technology route score, and the gap score.

[0171] It should be noted that the technical requirement generation system 60 of this embodiment is a system corresponding to the above-mentioned technical requirement generation method. The operating principle of the technical requirement generation system 60 can refer to the above-mentioned technical requirement generation method, so it will not be repeated here.

[0172] Specifically, Figure 5 As shown, the generation system of perception technology requirements consists of data storage layer, data access layer, service layer and user interaction layer.

[0173] Data storage layer: This layer is mainly responsible for storing various capability data, image data, technical data, etc. These data can come from multiple different sources and are usually organized and stored in the form of knowledge graphs.

[0174] Data access layer: This layer is mainly responsible for interacting with the data storage layer and providing operations such as data reading, querying, and updating. In addition, the data access layer also needs to be able to process various data formats and convert them into a format that the system can recognize.

[0175] Service layer: This layer is mainly responsible for processing business logic, including demand analysis and demand generation. The service layer needs to call the interface provided by the data access layer to obtain the required data and process the generated results.

[0176] User interaction layer: This layer is mainly responsible for interacting with users, including displaying the user interface, receiving user input, and displaying processing results. The user interaction layer needs to provide a friendly user interface so that users can operate conveniently. In addition, the user interaction layer also needs to be able to handle user error input and provide clear error prompts.

[0177] The interaction relationship between the above four layers is as follows: the user operates in the user interaction layer, the user interaction layer passes the user's request to the service layer, the service layer calls the interface provided by the data access layer, obtains the required data, and then processes it. The processing results are displayed to the user through the user interaction layer.

[0178] See also Figure 6 , Figure 6 7 is a schematic diagram of the structure of an electronic device of this embodiment. The electronic device includes a memory 702 and a processor 703. The memory 702 stores a computer program that can be run on the processor. When the processor 703 executes the program, the steps of the method for generating the above technical requirements are implemented.

[0179] The electronic device includes a memory 702 and a processor 703 which are connected to each other through a system bus 701 for communication. It should be noted that the figure only shows an electronic device having a system bus 701, a memory 702 and a processor 703, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead. Among them, those skilled in the art can understand that the electronic device here is an electronic device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASIC), programmable gate arrays (FPGA), digital processors (DSP), embedded devices, etc.

[0180] Electronic devices can be computing devices such as desktop computers, notebooks, PDAs, and cloud servers. The devices can interact with users through keyboards, mice, remote controls, touchpads, or voice-controlled devices.

[0181] The memory 702 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 702 can be an internal memory 702 of the device, such as a hard disk or memory of the device. In other embodiments, the memory 702 can also be an external storage device of the device, such as a plug-in hard disk equipped on the device, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 702 can also include both the internal storage unit of the device and its external storage device. In this embodiment, the memory 702 is generally used to store the operating system and various application software installed on the device, such as computer-readable instructions of the target detection method, etc. In addition, the memory 702 can also be used to temporarily store various types of data that have been output or are to be output.

[0182] The processor 703 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 703 is generally used to control the overall operation of the device. In this embodiment, the processor 703 is used to run computer-readable instructions stored in the memory 400 or process data, such as computer-readable instructions for running a target detection method.

[0183] This embodiment provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for generating the above technical requirements are implemented.

[0184] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods of each embodiment of the present application.

[0185] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for generating technical requirements, characterized in that: include: Get the task scenario; The task scenario is the scenario information faced in executing the task, the task is a perception task, and obtaining the task scenario includes: adding a task scenario, deleting a task scenario, modifying a task scenario, and querying a task scenario; Extracting scene feature information according to the task scene; Analyze the scene feature information to generate technical requirements in the product design process; The analyzing the scene feature information to generate technical requirements includes: Building a scenario database, wherein the scenario database includes a scenario knowledge graph; Acquire graph information of a preset scene from the scene knowledge graph to obtain graph information of the preset scene; Obtaining matching scene information according to the similarity between the scene feature information and the preset scene graph information; Acquire preset requirement information corresponding to the matching scene information from the scene knowledge graph; Comparing the scene feature information with the matching scene information to determine difference information between the scene feature information and the matching scene information; Modify the preset requirement information according to the difference information to generate technical requirements; The obtaining matching scene information according to the similarity between the scene feature information and the preset scene graph information includes: Constructing scene feature graph information according to the knowledge entities in the scene feature information and the relationships between the knowledge entities; Respectively calculating the similarities between the scene feature graph information and a plurality of the preset scene graph information to obtain a plurality of similarity values; When the maximum similarity value among the multiple similarity values ​​is greater than or equal to the preset threshold, the preset scene graph information corresponding to the maximum similarity value is the matching scene information; When the maximum similarity value among the multiple similarity values ​​is less than a preset threshold, multiple preset scene graph information are selected according to the multiple similarity values ​​for fusion to obtain matching scene information; In the fusion process, the preset scene graph to be fused corresponding to the maximum similarity value is used as the basic graph; the knowledge entities in the basic graph whose similarity with the scene feature graph is lower than the first threshold are deleted; the remaining knowledge entities in the preset scene graph to be fused whose similarity with the scene feature graph is higher than the second threshold and the demand entities connected to the knowledge entities are obtained to obtain the entity tree to be added; the entity tree to be added is added to the basic graph to complete the fusion and obtain the matching scene information; the entities in the basic graph are modified and adjusted according to the remaining preset scene graphs to be fused until the similarity between the fused preset scene graph and the scene feature graph information is greater than the preset threshold or other termination conditions are met; The respectively calculating the similarities between the scene feature graph information and the plurality of preset scene graph information to obtain a plurality of similarity values ​​includes: According to the entity node in the scene feature graph information, query the nodes matching the entity node in the preset scene graph information to obtain the common nodes; According to the entity node connection relationship in the scene feature graph information, searching the preset scene graph information for a connection relationship that matches the entity node connection relationship to obtain a common connection relationship; Constructing a common subgraph between the scene feature graph information and the preset scene graph information according to the common nodes and the common connection relationship; Calculating a similarity value based on the common subgraph; The calculation formula of the similarity value is: S=w1×NSimavg+w2×SSim; Wherein, S is the similarity value, w1 and w2 are the weights of entity node similarity and common subgraph similarity respectively, NSimavg is the entity node similarity, SSim represents the common subgraph similarity, the entity node similarity is the average value of the name similarity of all common nodes, SSim=NComMax / Min(NNumT,NNumG), NComMax is the number of entity nodes of the common subgraph, NNumT is the number of entity nodes of the scene feature graph information, NNumG is the number of entity nodes of the preset scene graph information, Min(NNumT,NNumG) is the minimum value between the number of entity nodes of the scene feature graph information and the number of entity nodes of the preset scene graph information; The step of modifying the preset requirement information according to the difference information to generate technical requirements includes: Determine the modification method corresponding to the difference information through the preset scene knowledge graph information; Modify the preset requirement information according to the modification method to obtain multiple candidate technical requirements; Obtaining a criticality score for each of the candidate technical requirements; Screening the candidate technical requirements according to the criticality scores of the candidate technical requirements to obtain technical requirements; The obtaining of the criticality score of each candidate technical requirement includes: Determining current technologies corresponding to the candidate technology requirements; Determining a maturity score based on the maturity of the current technology; Determine a technology route score according to the evolution route of the current technology; Determine a gap score according to the technology gap between the candidate technology requirement and the current technology; Determine the criticality score of each of the candidate technical requirements based at least on the maturity score, the technical route score, and the gap score; The criticality score is positively correlated with the maturity score, the criticality score is positively correlated with the technical route score, and the criticality score is negatively correlated with the gap score.

2. The method for generating technical requirements according to claim 1, characterized in that: The task scenario includes at least one of a task scenario described in text, a task scenario described in image, and a task scenario described in video.

3. The method for generating technical requirements according to claim 2, characterized in that: The extracting scene feature information according to the task scene includes: When the task scenario is a task scenario described in text, knowledge entities and relationships between the knowledge entities are extracted from the task scenario described in text to obtain scenario feature information; When the task scene is described by an image or a video, video sample information or image sample information matching the task scene is obtained, and scene feature information is obtained according to the video sample information or image sample information.

4. A method for generating technical requirements for a perception task, characterized in that: include: Acquire the perception task scenario; The perception task scenario is processed using the method for generating technical requirements as described in any one of claims 1 to 3 to obtain perception technical requirements.

5. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, characterized in that: When the processor executes the program, the steps of the method for generating technical requirements according to any one of claims 1 to 3 are implemented.

Images