Vehicle-mounted chip type selection method, device and equipment and storage medium

Abstract

A vehicle-mounted chip type selection method, apparatus and device, and a storage medium, the method comprising: continuously acquiring related information of a vehicle-mounted chip according to a network crawling tool and an internal data source to update a local multi-dimensional feature library for constructing the vehicle-mounted chip, and constructing a local vector multivariate database; receiving a natural language question input by a user for vehicle-mounted chip type selection, performing natural language processing and feature extraction on the natural language question, and generating a feature vector of the natural language question; performing similarity retrieval on the feature vectors of the natural language question and vectors in the local vector multivariate database, and sorting candidate chips according to feature weights; and inputting the sorted candidate chips and the matching features thereof into a preset generation model, and generating a vehicle-mounted chip type selection suggestion, thereby solving the technical problem that a traditional vehicle-mounted chip type selection method mainly depends on artificial experience and a simple retrieval mechanism and is difficult to meet efficient and accurate decision-making requirements of modern industry.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a method, apparatus, device, and computer-readable storage medium for selecting automotive chips. Background Technology

[0002] The automotive industry faces the need for transformation towards new energy and connectivity, requiring significant adjustments to vehicle controllers and infotainment systems. Compared to traditional gasoline vehicles, new energy vehicles require numerous vehicle situational awareness controllers and energy controllers; the performance and stability of these controllers determine the vehicle's safety and reliability. The core component of high-performance controllers is a high-performance automotive-grade chip.

[0003] New energy vehicle chips are characterized by numerous parameters, high value, and rapid changes. Traditional methods for selecting automotive chips mainly rely on manual experience and simple search mechanisms, which are insufficient to meet the modern industrial demand for efficient and accurate decision-making. With the surge in data volume and increasing professional requirements, traditional knowledge base retrieval methods struggle to handle complex queries and ambiguous statements, failing to meet the needs of accurate and rapid question-and-answer systems.

[0004] Meanwhile, due to the significant impact of specialized fields on overall vehicle products, the relevant knowledge bases of each vehicle manufacturer are relatively closed, making it difficult to keep up with technological updates in real time. Furthermore, because technical personnel in these specialized fields are relatively isolated across companies, they cannot update their knowledge or access the latest product information. This results in a narrow and closed product selection pool, slow knowledge updates, low selection efficiency, high costs, and a high error rate. Summary of the Invention

[0005] This application provides a method, apparatus, device, and computer-readable storage medium for selecting automotive chips, which can solve the technical problem that traditional automotive chip selection methods in the prior art mainly rely on manual experience and simple search mechanisms, making it difficult to meet the modern industrial demand for efficient and accurate decision-making.

[0006] In a first aspect, embodiments of this application provide a method for selecting automotive chips, characterized in that it includes: We continuously acquire relevant information about automotive chips using web crawling tools and internal data sources to update the local multi-dimensional feature library for automotive chips and build a local vector multivariate database. The system receives a natural language question from a user regarding the selection of an onboard chip, performs natural language processing and feature extraction on the natural language question, and generates a feature vector for the natural language question. The feature vectors of the natural language problem are compared with the vectors in the local vector multivariate database for similarity retrieval. The candidate chips are then sorted according to feature weights by combining the KNN algorithm and semantic similarity calculation. The sorted candidate chips and their matching features are input into a pre-set generation model. The pre-set generation model integrates multi-dimensional matching information to generate vehicle chip selection recommendations.

[0007] In conjunction with the first aspect, in one implementation, the step of continuously acquiring relevant information about the vehicle-mounted chip based on web crawling tools and internal data sources to update the local multi-dimensional feature library for the vehicle-mounted chip and construct a local vector multivariate database includes: By acquiring the basic parameters and technical characteristics data of automotive chips, organizing chip features according to preset dimensions, and assigning weights to each dimension and feature item, a local multi-dimensional feature library for automotive chips is constructed. The preset dimensions include a first layer dimension and a second layer dimension. The first layer dimension includes business dimension, technical dimension, and policy dimension; the second layer dimension includes manufacturer, model, key parameters, historical usage of the unit, and industry usage. The system continuously acquires relevant data from automotive chips through web crawling tools and internal data sources. It then preprocesses, segments, and vectorizes the acquired data to update the local multi-dimensional feature library and construct a local vector multivariate database.

[0008] In conjunction with the first aspect, in one implementation, the step of continuously acquiring relevant data from the vehicle-mounted chip through web crawling tools and internal data sources, preprocessing, segmenting, and vectorizing the acquired relevant data, updating the local multi-dimensional feature library, and constructing a local vector multivariate database includes: Data was obtained from internal company historical data, industry standard documents, publicly available parameters from chip manufacturers, and publicly available statistical reports. We periodically access chip manufacturers' official websites, automotive industry vertical websites, and industry media reports through web scraping tools to obtain the latest chip parameter changes and design import information. By using natural language segmentation technology to remove stop words and punctuation, and simplifying long texts with an attention mechanism, standardized text is generated. The standardized text is segmented into blocks according to a preset granularity; The sliced ​​text is vectorized using an embedding model to generate a vector representation, which, along with the original text, structured feature values, source identifier, acquisition time, dimension label, and current weight value, is stored in the local multidimensional feature library to update the local vector multidimensional database, thus constructing the local vector multidimensional database.

[0009] In conjunction with the first aspect, in one implementation, receiving a user input of a natural language question about selecting an in-vehicle chip, performing natural language processing and feature extraction on the natural language question to generate a feature vector for the natural language question, includes: The natural language processing is performed on the user input question about selecting vehicle chips; Extract business, technical, or policy features from the natural language processing problem that are relevant to the selection of the vehicle-mounted chip; Construct semantic representations based on the extracted business, technology, or policy features and their interrelationships; The semantic representation is used to generate the feature vector of the natural language problem.

[0010] In conjunction with the first aspect, in one implementation, the step of performing similarity retrieval between the feature vector of the natural language problem and the vectors in the local vector multivariate database, and combining the KNN algorithm and semantic similarity calculation to rank the candidate chips according to feature weights includes: Calculate the similarity between the feature vector of the natural language problem and each vector in the local vector multivariate database; The KNN algorithm is used to determine the number of records most similar to the natural language problem. The similarity is weighted according to the current weight of each dimension of features; Candidate chips are sorted in descending order based on weighted overall similarity.

[0011] In conjunction with the first aspect, in one implementation, the step of inputting the sorted candidate chips and their matching features into a preset generation model, and integrating multi-dimensional matching information through the preset generation model to generate vehicle chip selection recommendations, includes: The top-ranked candidate chips and their matching feature lists in each dimension are used as context inputs to pre-generate the model. The multi-source matching features are weighted and fused using the attention mechanism of the pre-built generation model; Generate natural language selection suggestions that include recommended chip priorities, key matching features for each chip, and prompts for unmet requirements.

[0012] In conjunction with the first aspect, in one implementation, after inputting the sorted candidate chips and their matching features into a preset generation model, and integrating multi-dimensional matching information through the preset generation model to generate vehicle chip selection recommendations, the method further includes: The matching results of recommended chips are displayed according to business, technology, and policy dimensions. A ranking chart showing the contribution weights and influence of each dimension of features on the final ranking; Provides an interface for users to input their acceptance of the recommendation results, their approval of the matching degree of each feature, and their opinions on feature weight adjustments; Receive and record user-submitted feedback data; Analyze user feedback regarding weight adjustments for specific features; Modify the weight coefficients of the corresponding features in the local multi-dimensional feature library; The adjusted weight values ​​are synchronously written into the weight metadata field of the relevant records in the local vector multivariate database; Based on the adjustment directions that appeared most frequently in the feedback, the weight templates of similar features were migrated and adjusted in batches.

[0013] Secondly, embodiments of this application provide an in-vehicle chip selection device, the in-vehicle chip selection device comprising: The module is used to continuously obtain relevant information about the vehicle chip based on web crawling tools and internal data sources, so as to update the local multi-dimensional feature library for the vehicle chip and build a local vector multivariate database. The first generation module is used to receive a natural language question input by the user regarding the selection of an onboard chip, perform natural language processing and feature extraction on the natural language question, and generate a feature vector of the natural language question. The sorting module is used to perform similarity retrieval between the feature vector of the natural language problem and the vector in the local vector multivariate database, and sort the candidate chips according to the feature weights by combining the KNN algorithm and semantic similarity calculation. The second generation module is used to input the sorted candidate chips and their matching features into a preset generation model, and to integrate multi-dimensional matching information through the preset generation model to generate vehicle chip selection suggestions.

[0014] Thirdly, this application provides an in-vehicle chip selection device, which includes a processor, a memory, and an in-vehicle chip selection program stored in the memory and executable by the processor. When the in-vehicle chip selection program is executed by the processor, it implements the steps of the in-vehicle chip selection method described above.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium storing an in-vehicle chip selection program, wherein when the in-vehicle chip selection program is executed by a processor, it implements the steps of the in-vehicle chip selection method described above.

[0016] The beneficial effects of the technical solutions provided in this application include: By continuously acquiring relevant information about automotive chips using web crawling tools and internal data sources, the local multi-dimensional feature library for automotive chips is updated, and a local vector multivariate database is constructed. The system receives natural language questions from users regarding automotive chip selection, performs natural language processing and feature extraction on these questions, and generates feature vectors for each question. The feature vectors of the natural language questions are then compared with vectors in the local vector multivariate database for similarity retrieval. Combining the KNN algorithm and semantic similarity calculation, candidate chips are ranked according to feature weights. The ranked candidate chips and their matching features are input into a pre-set generation model. This model integrates multi-dimensional matching information to generate automotive chip selection recommendations. This approach solves the technical problem that traditional automotive chip selection methods rely heavily on human experience and simple retrieval mechanisms, making it difficult to meet the modern industrial demand for efficient and accurate decision-making. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the first embodiment of the vehicle-mounted chip selection method of this application; Figure 2 This is a functional module diagram of an embodiment of the vehicle chip selection device of this application; Figure 3 This is a schematic diagram of the hardware structure of the vehicle chip selection device involved in the embodiments of this application. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0019] First, some of the technical terms used in this application will be explained to help those skilled in the art understand this application.

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0021] Firstly, embodiments of this application provide a method for selecting automotive chips.

[0022] In one embodiment, reference is made to Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the vehicle-mounted chip selection method of this application. Figure 1 As shown, the selection method for automotive chips includes: Step S10: Continuously obtain relevant information about the vehicle chip based on web crawling tools and internal data sources to update the local multi-dimensional feature library for the vehicle chip and build a local vector multivariate database. As an example, a web crawling tool is launched, set to crawl once per hour, automatically accessing data sources such as the official websites of mainstream chip manufacturers, technical forums, and industry databases to obtain the latest technical specifications, performance parameters, and application scenarios of automotive chips. Simultaneously, the system connects to internal data sources, including historical project databases, test report databases, and supplier information databases, to continuously collect automotive chip-related information. The acquired information is categorized and organized according to multiple dimensions such as chip type, performance indicators, power consumption characteristics, temperature range, packaging form, and price information to construct a local multi-dimensional feature library. The feature library adopts a hierarchical storage structure, storing the chip's basic attributes, performance parameters, and application features in different data layers for easy subsequent rapid retrieval and matching. Based on the feature library, word embedding technology is used to convert the multi-dimensional features of each chip into high-dimensional vector representations, with the vector dimension set to 512 dimensions, covering all key feature information of the chip. These feature vectors are stored in a local vector multivariate database, which adopts a distributed index structure to support efficient similarity calculation and retrieval operations.

[0023] Specifically, by acquiring basic parameters and technical characteristic data of automotive chips, chip features are organized according to preset dimensions, and weights are assigned to each dimension and feature item to construct a local multi-dimensional feature library for automotive chips. The preset dimensions include a first-layer dimension and a second-layer dimension. The first-layer dimension includes business, technical, and policy dimensions; the second-layer dimension includes manufacturer, model, key parameters, historical usage by the organization, and industry usage. Relevant data on automotive chips is continuously acquired through web crawling tools and internal data sources. This acquired data is preprocessed, segmented, and vectorized to update the local multi-dimensional feature library and construct a local vector multivariate database. For example, basic parameters and technical characteristic data of automotive chips are acquired from industry associations, standards bodies, and internal company documents. A multi-dimensional basic feature library is established according to preset dimensions. NLP-word segmentation technology is used to preprocess and clean the continuously acquired external data, removing stop words, punctuation, and numbers. Text slicing is performed based on algorithms, and new features are extracted through comparison. Based on dimension matching and feature similarity judgment, the feature library values ​​are updated.

[0024] Data is obtained from internal historical data, industry standard documents, publicly available parameters from chip manufacturers, and publicly available statistical reports. Web scraping tools are used to periodically access chip manufacturer websites, automotive industry vertical websites, and industry media reports to obtain the latest chip parameter changes and design import information. Natural language processing (NLP) word segmentation technology is used to remove stop words and punctuation, and long texts are simplified using an attention mechanism to generate standardized text. The standardized text is then segmented according to a preset granularity. An embedding model is used to vectorize the segmented text, generating vector representations, which, along with the original text, structured feature values, source identifiers, collection time, dimension labels, and current weight values, are stored in the local multi-dimensional feature library to update the local vector multivariate database, thus constructing the local vector multivariate database. For example, the latest online information is crawled from manufacturer website reports, enterprise information databases (such as Qichacha), and online searches. NLP technology is used to simplify long texts, removing repetitive or meaningless words to obtain standardized text and label relevant features. External industry standard data (total chip sales, latest industry design parameters, chip type information) is accessed through specific ports. A local professional knowledge base is imported to build the foundation for the multi-source database. Data is segmented according to preset token values.

[0025] By combining the EMBEDDING model to vectorize information data, a vector database is completed, providing a standardized retrieval interface.

[0026] Step S20: Receive a natural language question from the user regarding the selection of an onboard chip, perform natural language processing and feature extraction on the natural language question, and generate a feature vector for the natural language question; As an example, when a user inputs a natural language question about selecting an in-vehicle chip, the system first preprocesses the input text, including word segmentation, part-of-speech tagging, and named entity recognition. For instance, if a user inputs "I need a low-power processor chip suitable for autonomous driving systems, with an operating temperature range of -40°C to 85°C," the system identifies key information: the application scenario is autonomous driving, the power requirement is low power consumption, the chip type is a processor, and the temperature range is -40°C to 85°C. A BERT pre-trained language model is used to perform deep semantic understanding of the natural language question, extracting explicit and implicit features from the user's needs. Explicit features include technical parameters explicitly mentioned by the user, while implicit features are obtained through semantic reasoning; for example, "autonomous driving system" implies high requirements for real-time performance and reliability. Through the feature extraction module, the system converts the user's question into a feature vector of the same dimension as the chip feature library. During feature extraction, the system assigns different weights to different types of needs: performance requirements have a weight of 0.4, application scenarios have a weight of 0.3, cost requirements have a weight of 0.2, and other factors have a weight of 0.1, generating a 512-dimensional feature vector for the user's question.

[0027] Specifically, the natural language question of selecting an in-vehicle chip, input by the user, is processed through word segmentation; business, technical, or policy features related to the selection of the in-vehicle chip are extracted from the natural language question; a semantic representation is constructed based on the extracted business, technical, or policy features and their interrelationships; and an embedding model is used to generate a feature vector for the natural language question from the semantic representation. For example, Step S30: Perform similarity retrieval between the feature vector of the natural language problem and the vector in the local vector multivariate database, and sort the candidate chips according to the feature weights by combining the KNN algorithm and semantic similarity calculation. As an example, the system calculates the similarity between the feature vector of the user's question and all chip vectors in the local multivariate vector database, using cosine similarity as the basic metric. To improve retrieval efficiency, the system first uses an approximate nearest neighbor search algorithm to quickly filter candidate chips with a similarity greater than 0.6, typically yielding 200-500 preliminary candidates. Next, the system combines the KNN algorithm for precise matching, setting K to 50, and selecting the 50 candidate chips with the highest similarity for detailed analysis. In the KNN calculation process, the system considers not only the Euclidean distance between vectors but also introduces semantic similarity calculation, using a word vector model to calculate the semantic matching degree between the user's requirement description and the chip application description. The system comprehensively scores and ranks the candidate chips according to preset feature weights. Performance matching accounts for 40% of the weight, application scenario matching accounts for 30%, cost-effectiveness accounts for 20%, and supply stability accounts for 10%. The final score of each candidate chip is calculated by weighted average, and the system ranks the candidate chips from highest to lowest score, selecting the top 10 as the final recommended candidates.

[0028] Specifically, the similarity between the feature vector of the natural language problem and each vector in the local vector multivariate database is calculated; the KNN algorithm is used to determine the number of records most similar to the natural language problem; the similarity is weighted according to the current weight of each dimension feature; and the candidate chips are sorted in descending order according to the weighted comprehensive similarity.

[0029] Step S40: Input the sorted candidate chips and their matching features into the preset generation model, and integrate multi-dimensional matching information through the preset generation model to generate vehicle chip selection suggestions.

[0030] As an example, the ranked candidate chips and their matching feature information are input into a pre-built generative model. This generative model is built on the Transformer architecture and trained on a large number of automotive chip selection cases, possessing powerful multi-dimensional information integration capabilities. After receiving the input, the generative model first performs a comprehensive analysis of the technical specifications, performance advantages, and application adaptability of each candidate chip. The model compares key parameters such as chip processing power, power consumption level, temperature adaptability, and packaging compatibility with user requirements item by item, identifying the best matching point and potential risk points.

[0031] Based on the integration of multi-dimensional matching information, the system generates structured automotive chip selection recommendations. These recommendations include a ranking of recommended chips, a detailed comparison of the technical parameters of each chip, a matching degree analysis report, and optimization suggestions tailored to the user's specific application scenario. The system also provides alternative solutions and risk warnings to ensure users make the most suitable selection decision. The generated selection recommendations are presented in a visually appealing format, including chip performance comparison charts, application architecture diagrams, cost-benefit analyses, and other visualizations to help users intuitively understand the advantages and disadvantages of each candidate chip, ultimately leading to accurate automotive chip selection.

[0032] Specifically, the top-ranked candidate chips and their matching feature lists in various dimensions are used as context input to a pre-built generation model; the multi-source matching features are weighted and fused through the attention mechanism of the pre-built generation model; and natural language selection suggestions are generated, including the priority of recommended chips, key matching features of each chip, and prompts for unmet requirements.

[0033] Specifically, after inputting the sorted candidate chips and their matching features into a pre-set generation model, and integrating multi-dimensional matching information through the pre-set generation model to generate vehicle chip selection suggestions, the process further includes: The system displays the matching status of recommended chips according to business, technology, and policy dimensions; it shows the contribution weight of each dimension's features to the final ranking and the ranking chart; it provides an interface for users to input their acceptance of the recommendation results, the degree of recognition of each feature's matching degree, and suggestions for feature weight adjustments; it receives and records user-submitted feedback data; it parses user feedback regarding weight adjustments for specific features; it modifies the weight coefficients of corresponding features in the local multi-dimensional feature library; it synchronously writes the adjusted weight values ​​into the weight metadata fields of relevant records in the local vector multivariate database; and it performs batch migration and adjustment of weight templates for similar features based on the adjustment directions that appear most frequently in multiple feedback responses.

[0034] For automotive chip selection scenarios, a local multi-dimensional chip feature library is built and continuously maintained. This library, combined with a local vector multivariate database and RAG mechanism, achieves accurate semantic mapping from user natural language selection intent to multi-dimensional chip features. Through dynamic feature weight allocation and user feedback loop, the local feature library and matching model are continuously optimized. The entire process maintains data locality and does not rely on real-time inference from large cloud models. Highly specialized, fragmented, and rapidly updated multi-source features in the automotive chip field are unified into a local multi-layered weighted feature system. Through structured feature extraction based on a local vector database and user selection intent, and feedback-driven adaptive weight adjustment, a closed-loop, self-evolving micro-domain chip selection knowledge system is formed.

[0035] In this embodiment, relevant information about automotive chips is continuously acquired using web crawling tools and internal data sources to update the local multi-dimensional feature library for automotive chips and construct a local vector multivariate database. The system receives a natural language question from a user regarding automotive chip selection, performs natural language processing and feature extraction on the question, and generates a feature vector for the question. The feature vector of the question is then compared with the vectors in the local vector multivariate database for similarity retrieval. Combined with the KNN algorithm and semantic similarity calculation, candidate chips are ranked according to feature weights. The ranked candidate chips and their matching features are input into a pre-set generation model. This model integrates multi-dimensional matching information to generate automotive chip selection recommendations. This approach solves the technical problem that traditional automotive chip selection methods rely heavily on human experience and simple retrieval mechanisms, making it difficult to meet the modern industrial demand for efficient and accurate decision-making.

[0036] Secondly, embodiments of this application also provide an in-vehicle chip selection device.

[0037] In one embodiment, reference is made to Figure 2 , Figure 2 This is a functional module diagram of an embodiment of the vehicle-mounted chip selection device of this application. Figure 2 As shown, the vehicle-mounted chip selection device includes: Module 10 is used to continuously obtain relevant information about the vehicle chip based on web crawling tools and internal data sources, so as to update the local multi-dimensional feature library for the vehicle chip and build a local vector multivariate database. The first generation module 20 is used to receive a natural language question input by the user regarding the selection of an on-board chip, perform natural language processing and feature extraction on the natural language question, and generate a feature vector of the natural language question. The sorting module 30 is used to perform similarity retrieval between the feature vector of the natural language problem and the vector in the local vector multivariate database, and sort the candidate chips according to the feature weights by combining the KNN algorithm and semantic similarity calculation. The second generation module 40 is used to input the sorted candidate chips and their matching features into a preset generation model, and to integrate multi-dimensional matching information through the preset generation model to generate vehicle chip selection suggestions.

[0038] Furthermore, in one embodiment, the construction module 10 is used for: By acquiring the basic parameters and technical characteristics data of automotive chips, organizing chip features according to preset dimensions, and assigning weights to each dimension and feature item, a local multi-dimensional feature library for automotive chips is constructed. The preset dimensions include a first layer dimension and a second layer dimension. The first layer dimension includes business dimension, technical dimension, and policy dimension; the second layer dimension includes manufacturer, model, key parameters, historical usage of the unit, and industry usage. The system continuously acquires relevant data from automotive chips through web crawling tools and internal data sources. It then preprocesses, segments, and vectorizes the acquired data to update the local multi-dimensional feature library and construct a local vector multivariate database.

[0039] Furthermore, in one embodiment, the vehicle-mounted chip selection device further includes a new module for: Data was obtained from internal company historical data, industry standard documents, publicly available parameters from chip manufacturers, and publicly available statistical reports. We periodically access chip manufacturers' official websites, automotive industry vertical websites, and industry media reports through web scraping tools to obtain the latest chip parameter changes and design import information. By using natural language segmentation technology to remove stop words and punctuation, and simplifying long texts with an attention mechanism, standardized text is generated. The standardized text is segmented into blocks according to a preset granularity; The sliced ​​text is vectorized using an embedding model to generate a vector representation, which, along with the original text, structured feature values, source identifier, acquisition time, dimension label, and current weight value, is stored in the local multidimensional feature library to update the local vector multidimensional database, thus constructing the local vector multidimensional database.

[0040] Furthermore, in one embodiment, the first generation module 20 is used to: The natural language processing is performed on the user input question about selecting vehicle chips; Extract business, technical, or policy features from the natural language processing problem that are relevant to the selection of the vehicle-mounted chip; Construct semantic representations based on the extracted business, technology, or policy features and their interrelationships; The semantic representation is used to generate the feature vector of the natural language problem.

[0041] Furthermore, in one embodiment, the sorting module 30 is used for: Calculate the similarity between the feature vector of the natural language problem and each vector in the local vector multivariate database; The KNN algorithm is used to determine the number of records most similar to the natural language problem. The similarity is weighted according to the current weight of each dimension of features; Candidate chips are sorted in descending order based on weighted overall similarity.

[0042] Furthermore, in one embodiment, the second generation module 40 is used for: The top-ranked candidate chips and their matching feature lists in each dimension are used as context inputs to pre-generate the model. The multi-source matching features are weighted and fused using the attention mechanism of the pre-built generation model; Generate natural language selection suggestions that include recommended chip priorities, key matching features for each chip, and prompts for unmet requirements.

[0043] Furthermore, in one embodiment, the vehicle-mounted chip selection device further includes a new module for: The matching results of recommended chips are displayed according to business, technology, and policy dimensions. A ranking chart showing the contribution weights and influence of each dimension of features on the final ranking; Provides an interface for users to input their acceptance of the recommendation results, their approval of the matching degree of each feature, and their opinions on feature weight adjustments; Receive and record user-submitted feedback data; Analyze user feedback regarding weight adjustments for specific features; Modify the weight coefficients of the corresponding features in the local multi-dimensional feature library; The adjusted weight values ​​are synchronously written into the weight metadata field of the relevant records in the local vector multivariate database; Based on the adjustment directions that appeared most frequently in the feedback, the weight templates of similar features were migrated and adjusted in batches.

[0044] The functions of each module in the above-mentioned vehicle chip selection device correspond to the steps in the above-mentioned vehicle chip selection method embodiment, and their functions and implementation processes will not be described in detail here.

[0045] Thirdly, embodiments of this application provide an in-vehicle chip selection device, which can be a personal computer (PC), laptop computer, server, or other device with data processing capabilities.

[0046] Reference Figure 3 , Figure 3 This is a schematic diagram of the hardware structure of the vehicle-mounted chip selection device involved in the embodiments of this application. In the embodiments of this application, the vehicle-mounted chip selection device may include a processor, a memory, a communication interface, and a communication bus.

[0047] The communication bus can be of any type and is used to interconnect the processor, memory, and communication interface.

[0048] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces used for interconnecting devices within the vehicle-mounted chip selection device, as well as interfaces used for interconnecting the vehicle-mounted chip selection device with other devices (such as other computing devices or user equipment). Physical interfaces can be Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user equipment can be displays, keyboards, etc.

[0049] Memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0050] The processor can be a general-purpose processor, which can call the vehicle chip selection program stored in the memory and execute the vehicle chip selection method provided in the embodiments of this application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the vehicle chip selection program is called can be referred to in the various embodiments of the vehicle chip selection method of this application, and will not be repeated here.

[0051] Those skilled in the art will understand that Figure 3 The hardware structure shown does not constitute a limitation of this application and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0052] Fourthly, embodiments of this application also provide a computer-readable storage medium.

[0053] The present application provides a computer-readable storage medium storing an in-vehicle chip selection program, wherein when the in-vehicle chip selection program is executed by a processor, it implements the steps of the in-vehicle chip selection method described above.

[0054] The method implemented when the vehicle chip selection procedure is executed can be referred to in various embodiments of the vehicle chip selection method of this application, and will not be repeated here.

[0055] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0056] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0057] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0058] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0059] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0060] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0061] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for selecting automotive-grade chips, characterized in that, include: We continuously acquire relevant information about automotive chips using web crawling tools and internal data sources to update the local multi-dimensional feature library for automotive chips and build a local vector multivariate database. The system receives a natural language question from a user regarding the selection of an onboard chip, performs natural language processing and feature extraction on the natural language question, and generates a feature vector for the natural language question. The feature vectors of the natural language problem are compared with the vectors in the local vector multivariate database for similarity retrieval. The candidate chips are then sorted according to feature weights by combining the KNN algorithm and semantic similarity calculation. The sorted candidate chips and their matching features are input into a pre-set generation model. The pre-set generation model integrates multi-dimensional matching information to generate vehicle chip selection recommendations.

2. The vehicle-mounted chip selection method as described in claim 1, characterized in that, The process of continuously acquiring relevant information about the vehicle-mounted chip using web crawling tools and internal data sources to update the local multi-dimensional feature library for the vehicle-mounted chip and construct a local vector multivariate database includes: By acquiring the basic parameters and technical characteristics data of automotive chips, organizing chip features according to preset dimensions, and assigning weights to each dimension and feature item, a local multi-dimensional feature library for automotive chips is constructed. The preset dimensions include a first layer dimension and a second layer dimension. The first layer dimension includes business dimension, technical dimension, and policy dimension; the second layer dimension includes manufacturer, model, key parameters, historical usage of the unit, and industry usage. The system continuously acquires relevant data from automotive chips through web crawling tools and internal data sources. It then preprocesses, segments, and vectorizes the acquired data to update the local multi-dimensional feature library and construct a local vector multivariate database.

3. The vehicle-mounted chip selection method as described in claim 2, characterized in that, The process involves continuously acquiring relevant data from the vehicle-mounted chip using web crawling tools and internal data sources, preprocessing, segmenting, and vectorizing the acquired data, updating the local multi-dimensional feature library, and constructing a local vector multivariate database, including: Data was obtained from internal company historical data, industry standard documents, publicly available parameters from chip manufacturers, and publicly available statistical reports. We periodically access chip manufacturers' official websites, automotive industry vertical websites, and industry media reports through web scraping tools to obtain the latest chip parameter changes and design import information. By using natural language segmentation technology to remove stop words and punctuation, and simplifying long texts with an attention mechanism, standardized text is generated. The standardized text is segmented into blocks according to a preset granularity; The sliced ​​text is vectorized using an embedding model to generate a vector representation, which, along with the original text, structured feature values, source identifier, acquisition time, dimension label, and current weight value, is stored in the local multidimensional feature library to update the local vector multidimensional database, thus constructing the local vector multidimensional database.

4. The vehicle-mounted chip selection method as described in claim 1, characterized in that, The process of receiving a user-input natural language question about selecting an onboard chip involves natural language processing and feature extraction to generate a feature vector for the natural language question, including: The natural language processing is performed on the user input question about selecting vehicle chips; Extract business, technical, or policy features from the natural language processing problem that are relevant to the selection of the vehicle-mounted chip; Construct semantic representations based on the extracted business, technology, or policy features and their interrelationships; The semantic representation is used to generate the feature vector of the natural language problem.

5. The vehicle-mounted chip selection method as described in claim 1, characterized in that, The step of performing similarity retrieval between the feature vector of the natural language problem and the vectors in the local vector multivariate database, combined with the KNN algorithm and semantic similarity calculation, and ranking the candidate chips according to feature weights includes: Calculate the similarity between the feature vector of the natural language problem and each vector in the local vector multivariate database; The KNN algorithm is used to determine the number of records most similar to the natural language problem. The similarity is weighted according to the current weight of each dimension of features; Candidate chips are sorted in descending order based on weighted overall similarity.

6. The method for selecting automotive chips as described in claim 1, characterized in that, The process involves inputting the sorted candidate chips and their matching features into a pre-set generation model, and then integrating multi-dimensional matching information through the pre-set generation model to generate vehicle chip selection recommendations, including: The top-ranked candidate chips and their matching feature lists in each dimension are used as context inputs to pre-generate the model. The multi-source matching features are weighted and fused using the attention mechanism of the pre-built generation model; Generate natural language selection suggestions that include recommended chip priorities, key matching features for each chip, and prompts for unmet requirements.

7. The method for selecting automotive chips as described in claim 1, characterized in that, After inputting the sorted candidate chips and their matching features into a pre-set generation model, and integrating multi-dimensional matching information through the pre-set generation model to generate vehicle chip selection suggestions, the process further includes: The matching results of recommended chips are displayed according to business, technology, and policy dimensions. A ranking chart showing the contribution weights and influence of each dimension of features on the final ranking; Provides an interface for users to input their acceptance of the recommendation results, their approval of the matching degree of each feature, and their opinions on feature weight adjustments; Receive and record user-submitted feedback data; Analyze user feedback regarding weight adjustments for specific features; Modify the weight coefficients of the corresponding features in the local multi-dimensional feature library; The adjusted weight values ​​are synchronously written into the weight metadata field of the relevant records in the local vector multivariate database; Based on the adjustment directions that appeared most frequently in the feedback, the weight templates of similar features were migrated and adjusted in batches.

8. A vehicle-mounted chip selection device, characterized in that, The vehicle-mounted chip selection device includes: The module is used to continuously obtain relevant information about the vehicle chip based on web crawling tools and internal data sources, so as to update the local multi-dimensional feature library for the vehicle chip and build a local vector multivariate database. The first generation module is used to receive a natural language question input by the user regarding the selection of an onboard chip, perform natural language processing and feature extraction on the natural language question, and generate a feature vector of the natural language question. The sorting module is used to perform similarity retrieval between the feature vector of the natural language problem and the vector in the local vector multivariate database, and sort the candidate chips according to the feature weights by combining the KNN algorithm and semantic similarity calculation. The second generation module is used to input the sorted candidate chips and their matching features into a preset generation model, and to integrate multi-dimensional matching information through the preset generation model to generate vehicle chip selection suggestions.

9. A vehicle-mounted chip selection device, characterized in that, The vehicle chip selection device includes a processor, a memory, and a vehicle chip selection program stored in the memory and executable by the processor, wherein when the vehicle chip selection program is executed by the processor, it implements the steps of the vehicle chip selection method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an in-vehicle chip selection program, wherein when the in-vehicle chip selection program is executed by a processor, it implements the steps of the in-vehicle chip selection method as described in any one of claims 1 to 7.

Images